2013-10-30 Vladimir Makarov <vmakarov@redhat.com>
[official-gcc.git] / gcc / ChangeLog
blob966b4a5ffe08602564c0bc98c5bd6ea3b86a5ad0
1 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
3         PR bootstrap/58933
4         * ira-color.c (update_costs_from_copies): Add new parameter.  Use
5         it for calling update_costs_from_allocno.
6         (assign_hard_reg): Call restore_costs_from_copies only for
7         !retry_p.  Pass new argument to update_costs_from_copies.
8         (color_pass): Pass new argument to update_costs_from_copies.
9         (ira_mark_allocation_change): Ditto.
11 2013-10-30  Sharad Singhai  <singhai@google.com>
13         PR middle-end/58134
14         * opts.c (common_handle_option): Remove deprecated option
15         -ftree-vectorizer-verbose.
16         * doc/invoke.texi (Debugging Options): Ditto.
17         * opts-global.c (handle_common_deferred_options): Ditto.
18         (dump_remap_tree_vectorizer_verbose): Delete.
19         * common.opt: Set -ftree-vectorizer-verbose as an ignored option.
21 2013-10-30  DJ Delorie  <dj@redhat.com>
23         * config/rx/rx.c (ADD_RX_BUILTIN0): New macro, used for builtins
24         that take no arguments.
26 2013-10-30  Joern Rennecke  <joern.rennecke@embecosm.com>
28         PR other/58545
29         * reload1.c (update_eliminables_and_spill): New function, broken
30         out of reload.
31         (reload): Use it.  Check for frame size change after frame
32         size alignment, and call update_eliminables_and_spill first
33         if continue-ing.
35 2013-10-30  Cong Hou  <congh@google.com>
37         PR target/58762
38         * config/i386/i386-protos.h (ix86_expand_sse2_abs): New function.
39         * config/i386/i386.c (ix86_expand_sse2_abs): New function.
40         * config/i386/sse.md: Add SSE2 support to abs (8/16/32-bit-int).
42 2013-10-18  Mikael Pettersson  <mikpelinux@gmail.com>
44         PR rtl-optimization/58369
45         * reload1.c (compute_reload_subreg_offset): New function.
46         (choose_reload_regs): Use it to pass endian-correct
47         offset to subreg_regno_offset.
49 2013-10-30  Tobias Burnus  <burnus@net-b.de>
51         PR other/33426
52         * gcc/tree-cfg.c (replace_loop_annotate): Replace warning by
53         warning_at.
55 2013-10-30  Jason Merrill  <jason@redhat.com>
57         * configure.ac (loose_warn): Add -Wno-format if
58         --disable-build-format-warnings.
60 2013-10-30  David Malcolm  <dmalcolm@redhat.com>
62         * cgraphunit.c (analyze_functions): Split symtab_node
63         declarations onto multiple lines to make things easier
64         for rename_symtab.py.
66         * symtab.c (symtab_dissolve_same_comdat_group_list): Likewise.
67         (symtab_semantically_equivalent_p): Likewise.
69 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
71         PR target/58784
72         * lra.c (check_rtl): Remove address check before LRA work.
74 2013-10-30  Marc Glisse  <marc.glisse@inria.fr>
76         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for a
77         POINTER_PLUS_EXPR in the defining statement.
79 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
81         * regmove.c: Remove.
82         * tree-pass.h (make_pass_regmove): Remove.
83         * timevar.def (TV_REGMOVE): Remove.
84         * passes.def (pass_regmove): Remove.
85         * opts.c (default_options_table): Remove entry for regmove.
86         * doc/passes.texi: Remove regmove pass description.
87         * doc/invoke.texi (-foptimize-register-move, -fregmove): Remove
88         options.
89         (-fdump-rtl-regmove): Ditto.
90         * common.opt (foptimize-register-move, fregmove): Ignore.
91         * Makefile.in (OBJS): Remove regmove.o.
92         * regmove.c: Remove.
93         * ira-int.h (struct ira_allocno_pref, ira_pref_t): New structure
94         and type.
95         (struct ira_allocno) New member allocno_prefs.
96         (ALLOCNO_PREFS): New macro.
97         (ira_prefs, ira_prefs_num): New external vars.
98         (ira_setup_alts, ira_get_dup_out_num, ira_debug_pref): New
99         prototypes.
100         (ira_debug_prefs, ira_debug_allocno_prefs, ira_create_pref):
101         Ditto.
102         (ira_add_allocno_pref, ira_remove_pref, ira_remove_allocno_prefs):
103         Ditto.
104         (ira_add_allocno_copy_to_list): Remove prototype.
105         (ira_swap_allocno_copy_ends_if_necessary): Ditto.
106         (ira_pref_iterator): New type.
107         (ira_pref_iter_init, ira_pref_iter_cond): New functions.
108         (FOR_EACH_PREF): New macro.
109         * ira.c (commutative_constraint_p): Move from ira-conflicts.c.
110         (ira_get_dup_out_num): Ditto. Rename from get_dup_num.  Modify the
111         code.
112         (ira_setup_alts): New function.
113         (decrease_live_ranges_number): New function.
114         (ira): Call the above function.
115         * ira-build.c (ira_prefs, ira_prefs_num): New global vars.
116         (ira_create_allocno): Initialize allocno prefs.
117         (pref_pool, pref_vec): New static vars.
118         (initiate_prefs, find_allocno_pref, ira_create_pref): New
119         functions.
120         (add_allocno_pref_to_list, ira_add_allocno_pref, print_pref): Ditto.
121         (ira_debug_pref, print_prefs, ira_debug_prefs): Ditto.
122         (print_allocno_prefs, ira_debug_allocno_prefs, finish_pref): Ditto.
123         (ira_remove_pref, ira_remove_allocno_prefs, finish_prefs): Ditto.
124         (ira_add_allocno_copy_to_list): Make static.  Rename to
125         add_allocno_copy_to_list.
126         (ira_swap_allocno_copy_ends_if_necessary): Make static.  Rename to
127         swap_allocno_copy_ends_if_necessary.
128         (remove_unnecessary_allocnos, remove_low_level_allocnos): Call
129         ira_remove_allocno_prefs.
130         (ira_flattening): Ditto.
131         (ira_build): Call initiate_prefs, print_prefs.
132         (ira_destroy): Call finish_prefs.
133         * ira-color.c (struct update_cost_record): New.
134         (struct allocno_color_data): Add new member update_cost_records.
135         (update_cost_record_pool): New static var.
136         (init_update_cost_records, get_update_cost_record): New functions.
137         (free_update_cost_record_list, finish_update_cost_records): Ditto.
138         (struct update_cost_queue_elem): Add member from.
139         (initiate_cost_update): Call init_update_cost_records.
140         (finish_cost_update): Call finish_update_cost_records.
141         (queue_update_cost, get_next_update_cost): Add new param from.
142         (Update_allocno_cost, update_costs_from_allocno): New functions.
143         (update_costs_from_prefs): Ditto.
144         (update_copy_costs): Rename to update_costs_from_copies.
145         (restore_costs_from_copies): New function.
146         (update_conflict_hard_regno_costs): Don't go back.
147         (assign_hard_reg): Call restore_costs_from_copies.  Add printing
148         more debug info.
149         (pop_allocnos): Add priniting more debug info.
150         (color_allocnos): Remove prefs for conflicting hard regs.
151         Call update_costs_from_prefs.
152         * ira-conflicts.c (commutative_constraint_p): Move to ira.c
153         (get_dup_num): Rename, modify, and move to ira.c
154         (process_regs_for_copy): Add prefs.
155         (add_insn_allocno_copies): Put src as first arg of
156         process_regs_for_copy.  Remove dead code.  Call ira_setup_alts.
157         * ira-costs.c (record_reg_classes): Modify and move code into
158         record_operands_costs.
159         (find_costs_and_classes): Create prefs for the hard reg of small
160         reg class.
161         (process_bb_node_for_hard_reg_moves): Add prefs.
163 2013-10-30  Richard Biener  <rguenther@suse.de>
165         PR middle-end/57100
166         * basic-block.h (pre_and_rev_post_order_compute_fn): New function.
167         * cfganal.c (pre_and_rev_post_order_compute_fn): New function
168         as worker for ...
169         (pre_and_rev_post_order_compute): ... which now wraps it.
170         * graph.c (draw_cfg_nodes_no_loops): Use
171         pre_and_rev_post_order_compute_fn to avoid ICEing and dependence
172         on cfun.
174 2013-10-30  Christian Bruel  <christian.bruel@st.com>
176         * gcc/config/sh/sh-mem.cc (sh_expand_cmpnstr): New function.
177         (sh_expand_cmpstr): Handle known align and schedule improvements.
178         * gcc/config/sh/sh-protos.h (sh_expand_cmpstrn): Declare.
179         * gcc/config/sh/sh.md (cmpstrnsi): New pattern.
181 2013-10-30  Martin Jambor  <mjambor@suse.cz>
183         PR rtl-optimization/10474
184         * ira.c (find_moveable_pseudos): Do not calculate dominance info
185         nor df analysis.
186         (interesting_dest_for_shprep): New function.
187         (split_live_ranges_for_shrink_wrap): Likewise.
188         (ira): Calculate dominance info and df analysis. Call
189         split_live_ranges_for_shrink_wrap.
191 2013-10-30  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
193         PR target/58854
194         * config/arm/arm.c (arm_expand_epilogue_apcs_frame): Emit blockage.
196 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
198         * tree-core.h (tree_index): Add TI_POINTER_BOUNDS_TYPE.
199         * tree.h (POINTER_BOUNDS_P): New.
200         (BOUNDED_TYPE_P): New.
201         (BOUNDED_P): New.
202         (pointer_bounds_type_node): New.
203         * tree.c (build_common_tree_nodes): Initialize
204         pointer_bounds_type_node.
205         * gimple.h (gimple_call_get_nobnd_arg_index): New.
206         (gimple_call_num_nobnd_args): New.
207         (gimple_call_nobnd_arg): New.
208         (gimple_return_retbnd): New.
209         (gimple_return_set_retbnd): New
210         * gimple.c (gimple_build_return): Increase number of ops
211         for return statement.
212         (gimple_call_get_nobnd_arg_index): New.
213         * gimple-pretty-print.c (dump_gimple_return): Print second op.
215 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
217         * ipa.c (cgraph_build_static_cdtor_1): Support contructors
218         with "chkp ctor" and "bnd_legacy" attributes.
219         * gimplify.c (gimplify_init_constructor): Avoid infinite
220         loop during gimplification of bounds initializer.
222 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
224         * c-family/c-common.c (handle_bnd_variable_size_attribute): New.
225         (handle_bnd_legacy): New.
226         (c_common_attribute_table): Add bnd_variable_size and bnd_legacy.
227         * doc/extend.texi: Document bnd_variable_size and bnd_legacy
228         attributes.
230 2013-10-29  Ilya Enkovich  <ilya.enkovich@intel.com>
232         * builtin-types.def (BT_FN_VOID_CONST_PTR): New.
233         (BT_FN_PTR_CONST_PTR): New.
234         (BT_FN_CONST_PTR_CONST_PTR): New.
235         (BT_FN_PTR_CONST_PTR_SIZE): New.
236         (BT_FN_PTR_CONST_PTR_CONST_PTR): New.
237         (BT_FN_VOID_PTRPTR_CONST_PTR): New.
238         (BT_FN_VOID_CONST_PTR_SIZE): New.
239         (BT_FN_PTR_CONST_PTR_CONST_PTR_SIZE): New.
240         * chkp-builtins.def: New.
241         * builtins.def: include chkp-builtins.def.
242         (DEF_CHKP_BUILTIN): New.
243         * builtins.c (expand_builtin): Support BUILT_IN_CHKP_INIT_PTR_BOUNDS,
244         BUILT_IN_CHKP_NULL_PTR_BOUNDS, BUILT_IN_CHKP_COPY_PTR_BOUNDS,
245         BUILT_IN_CHKP_CHECK_PTR_LBOUNDS, BUILT_IN_CHKP_CHECK_PTR_UBOUNDS,
246         BUILT_IN_CHKP_CHECK_PTR_BOUNDS, BUILT_IN_CHKP_SET_PTR_BOUNDS,
247         BUILT_IN_CHKP_NARROW_PTR_BOUNDS, BUILT_IN_CHKP_STORE_PTR_BOUNDS,
248         BUILT_IN_CHKP_GET_PTR_LBOUND, BUILT_IN_CHKP_GET_PTR_UBOUND,
249         BUILT_IN_CHKP_BNDMK, BUILT_IN_CHKP_BNDSTX, BUILT_IN_CHKP_BNDCL,
250         BUILT_IN_CHKP_BNDCU, BUILT_IN_CHKP_BNDLDX, BUILT_IN_CHKP_BNDRET,
251         BUILT_IN_CHKP_INTERSECT, BUILT_IN_CHKP_ARG_BND, BUILT_IN_CHKP_NARROW,
252         BUILT_IN_CHKP_EXTRACT_LOWER, BUILT_IN_CHKP_EXTRACT_UPPER.
253         * common.opt (fcheck-pointer-bounds): New.
254         * toplev.c (process_options): Check Pointer Bounds Checker is supported.
255         * doc/extend.texi: Document Pointer Bounds Checker built-in functions.
257 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
259         * target.def (builtin_chkp_function): New.
260         (chkp_bound_type): New.
261         (chkp_bound_mode): New.
262         (fn_abi_va_list_bounds_size): New.
263         (load_bounds_for_arg): New.
264         (store_bounds_for_arg): New.
265         * targhooks.h (default_load_bounds_for_arg): New.
266         (default_store_bounds_for_arg): New.
267         (default_fn_abi_va_list_bounds_size): New.
268         (default_chkp_bound_type): New.
269         (default_chkp_bound_mode): New.
270         (default_builtin_chkp_function): New.
271         * targhooks.c (default_load_bounds_for_arg): New.
272         (default_store_bounds_for_arg): New.
273         (default_fn_abi_va_list_bounds_size): New.
274         (default_chkp_bound_type): New.
275         (default_chkp_bound_mode); New.
276         (default_builtin_chkp_function): New.
277         * doc/tm.texi.in (TARGET_FN_ABI_VA_LIST_BOUNDS_SIZE): New.
278         (TARGET_LOAD_BOUNDS_FOR_ARG): New.
279         (TARGET_STORE_BOUNDS_FOR_ARG): New.
280         (TARGET_BUILTIN_CHKP_FUNCTION): New.
281         (TARGET_CHKP_BOUND_TYPE): New.
282         (TARGET_CHKP_BOUND_MODE): New.
283         * doc/tm.texi: Regenerated.
284         * langhooks.h (lang_hooks): Add chkp_supported field.
285         * langhooks-def.h (LANG_HOOKS_CHKP_SUPPORTED): New.
286         (LANG_HOOKS_INITIALIZER); Add LANG_HOOKS_CHKP_SUPPORTED.
288 2013-10-29  Andrew Pinski <apinski@cavium.com>
290         * tree-ssa-ifcombine.c: Include rtl.h and tm_p.h.
291         (ifcombine_ifandif): Handle cases where
292         maybe_fold_and_comparisons fails, combining the branches
293         anyways.
294         (tree_ssa_ifcombine): Inverse the order of
295         the basic block walk, increases the number of combinings.
296         * gimple.h (gsi_start_nondebug_after_labels_bb): New function.
298 2013-10-29  Mike Stump  <mikestump@comcast.net>
300         * machmode.def (PARTIAL_INT_MODE): Add precision and name.
301         * genmodes.c (PARTIAL_INT_MODE): Add precision and name.
302         (make_vector_mode): Increase namebuf to 16.
303         (emit_insn_modes_h): When processing BImode, don't
304         also match partial int modes.
305         (emit_class_narrowest_mode): Likewise.
307         * config/bfin/bfin-modes.def: Add precision to PDI.
308         * config/m32c/m32c-modes.def: Add precision to PSI.
309         * config/msp430/msp430-modes.def: Add precision to PSI.
310         * config/rs6000/rs6000-modes.def: Add precision to PTI.
311         * config/sh/sh-modes.def: Add precision to PSI and PDI.
313 2013-10-29  Oleg Endo  <olegendo@gcc.gnu.org>
315         PR target/54236
316         * config/sh/sh.md (*addc): Rename existing variations to ...
317         (*addc_r_r_1, *addc_2r_1, *addc_r_1): ... these.
318         (*addc_r_lsb, *addc_r_r_lsb, *addc_r_lsb_r, *addc_2r_lsb, *addc_r_msb,
319         *addc_r_r_msb, *addc_2r_msb): New insn_and_split patterns.
320         * config/sh/sh.c (addsubcosts): Handle some addc special cases.
322 2013-10-29  Teresa Johnson  <tejohnson@google.com>
324         PR ipa/58862
325         * tree-ssa-tail-merge.c (replace_block_by): Tolerate profile
326         insanities when updating probabilities.
328 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
330         * gdbhooks.py (CGraphNodePrinter.to_string): Update gdb
331         prettyprinter for cgraph_node to reflect the conversion of the
332         symtable types to a C++ class hierarchy: it now *is* a
333         symtab_node_base, rather than having one (named "symbol").
335 2013-10-29  Balaji V. Iyer  <balaji.v.iyer@intel.com>
337         * builtins.c (is_builtin_name): Added a check for __cilkrts_detach and
338         __cilkrts_pop_frame.  If matched, then return true for built-in
339         function name.
340         (expand_builtin): Added BUILT_IN_CILK_DETACH and
341         BUILT_IN_CILK_POP_FRAME case.
342         * langhooks-def.h (lhd_install_body_with_frame_cleanup): New prototype.
343         (lhs_cilk_detect_spawn): Likewise.
344         (LANG_HOOKS_DECLS): Added LANG_HOOKS_CILKPLUS.
345         (LANG_HOOKS_CILKPLUS_DETECT_SPAWN_AND_UNWRAP): New #define.
346         (LANG_HOOKS_CILKPLUS_FRAME_CLEANUP): Likewise.
347         (LANG_HOOKS_CILKPLUS_GIMPLIFY_SPAWN): Likewise.
348         (LANG_HOOKS_CILKPLUS): Likewise.
349         * tree.h (CILK_SPAWN_FN): Likewise.
350         * builtin.def (DEF_CILK_BUILTIN_STUB): Likewise.
351         * Makefile.in (C_COMMON_OBJS): Added c-family/cilk.o.
352         (OBJS): Added cilk-common.o.
353         (BUILTINS_DEF): Added cilk-builtins.def.
354         * langhooks.c (lhd_install_body_with_frame_cleanup): New function.
355         (lhd_cilk_detect_spawn): Likewise.
356         * langhooks.h (lang_hooks_for_cilkplus): New struct.
357         (struct lang_hooks): Added new field called "cilkplus."
358         * cilk-common.c: New file.
359         * cilk.h: Likewise.
360         * cilk-builtins.def: Likewise.
361         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Added
362         "__cilk" macro and set it to 200.
363         * function.h (struct function::cilk_frame_decl): New field.
364         (struct function::is_cilk_function): Likewise.
365         (struct function::calls_cilk_spawn): Likewise.
366         * gimplify.c (gimplify_call_expr): Added a check if the function call
367         being gimplified is a spawn detach point.  If so, then add pop_frame
368         and detach function calls.
369         (gimplify_expr): Added a CILK_SPAWN_STMT and CILK_SYNC_STMT case
370         for gimplifying _Cilk_spawn and _Cilk_sync statements.
371         (gimplify_return_expr): Added a check for _Cilk_spawn usage in
372         function.  If so, added a _Cilk_sync and gimplified it.
373         (gimplify_modify_expr): Added a check for _Cilk_spawn in MODIFY and
374         INIT_EXPRs.  If so, then call gimplify_cilk_spawn.
375         * ipa-inline-analysis (initialize_inline_failed): Prevent inlining of
376         spawner function.
377         (can_inline_edge_p): Prevent inling of spawnee function.
378         * ira.c (ira_setup_eliminable_regset): Force usage of frame pointer
379         for functions that use Cilk keywords.
380         * tree-inline.h (struct copy_body_data::remap_var_for_cilk): New field.
381         * tree-pretty-print.c (dump_generic_node): Added CILK_SPAWN_STMT and
382         CILK_SYNC_STMT cases.
383         * tree.def (DEFTREECODE): Added CILK_SPAWN_STMT and CILK_SYNC_STMT
384         trees.
385         * generic.texi (CILK_SPAWN_STMT): Added documentation for _Cilk_spawn.
386         (CILK_SYNC_STMT): Added documentation for _Cilk_sync.
387         * passes.texi (Cilk Keywords): New section that describes the compiler
388         code changes for handling Cilk Keywords.
389         
390 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
392         Patch autogenerated by refactor_symtab.py from
393         https://github.com/davidmalcolm/gcc-refactoring-scripts
394         revision 58bb219cc090b2f4516a9297d868c245495ee622
396         * asan.c (asan_finish_file): Update for conversion of symtab types to
397         a true class hierarchy.
398         * cfgexpand.c (estimated_stack_frame_size): Likewise.
399         * cgraph.c (cgraph_get_body): Likewise.
400         (cgraph_get_create_real_symbol_node): Likewise.
401         (verify_cgraph_node): Likewise.
402         (verify_edge_corresponds_to_fndecl): Likewise.
403         (verify_edge_count_and_frequency): Likewise.
404         (cgraph_will_be_removed_from_program_if_no_direct_calls): Likewise.
405         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
406         (cgraph_can_remove_if_no_direct_calls_and_refs_p): Likewise.
407         (cgraph_node_cannot_return): Likewise.
408         (cgraph_set_pure_flag_1): Likewise.
409         (cgraph_set_const_flag_1): Likewise.
410         (cgraph_set_nothrow_flag_1): Likewise.
411         (cgraph_make_node_local_1): Likewise.
412         (cgraph_for_node_and_aliases): Likewise.
413         (cgraph_for_node_thunks_and_aliases): Likewise.
414         (cgraph_node_can_be_local_p): Likewise.
415         (cgraph_node_cannot_be_local_p_1): Likewise.
416         (cgraph_function_body_availability): Likewise.
417         (dump_cgraph_node): Likewise.
418         (cgraph_rtl_info): Likewise.
419         (cgraph_mark_address_taken_node): Likewise.
420         (cgraph_remove_node): Likewise.
421         (cgraph_release_function_body): Likewise.
422         (cgraph_update_edges_for_call_stmt_node): Likewise.
423         (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
424         (cgraph_make_edge_direct): Likewise.
425         (cgraph_resolve_speculation): Likewise.
426         (cgraph_speculative_call_info): Likewise.
427         (cgraph_turn_edge_to_speculative): Likewise.
428         (cgraph_create_edge_1): Likewise.
429         (cgraph_set_call_stmt): Likewise.
430         (cgraph_node_for_asm): Likewise.
431         (cgraph_add_thunk): Likewise.
432         (cgraph_same_body_alias): Likewise.
433         (cgraph_create_function_alias): Likewise.
434         (cgraph_create_node): Likewise.
435         (cgraph_create_empty_node): Likewise.
436         (record_function_versions): Likewise.
437         (used_from_object_file_p): Likewise.
438         * cgraph.h (symtab_can_be_discarded): Likewise.
439         (symtab_real_symbol_p): Likewise.
440         (cgraph_mark_force_output_node): Likewise.
441         (cgraph_edge_recursive_p): Likewise.
442         (symtab_alias_target): Likewise.
443         (varpool_all_refs_explicit_p): Likewise.
444         (varpool_can_remove_if_no_refs): Likewise.
445         (cgraph_only_called_directly_or_aliased_p): Likewise.
446         (cgraph_next_function_with_gimple_body): Likewise.
447         (cgraph_first_function_with_gimple_body): Likewise.
448         (cgraph_function_with_gimple_body_p): Likewise.
449         (cgraph_next_function): Likewise.
450         (cgraph_first_function): Likewise.
451         (cgraph_next_defined_function): Likewise.
452         (cgraph_first_defined_function): Likewise.
453         (varpool_next_defined_variable): Likewise.
454         (varpool_first_defined_variable): Likewise.
455         (varpool_next_static_initializer): Likewise.
456         (varpool_first_static_initializer): Likewise.
457         (varpool_next_variable): Likewise.
458         (varpool_first_variable): Likewise.
459         (varpool_node_name): Likewise.
460         (varpool): Likewise.
461         (cgraph): Likewise.
462         (is_a_helper <varpool_node>::test): Likewise.
463         (is_a_helper <cgraph_node>::test): Likewise.
464         (varpool_variable_node): Likewise.
465         (cgraph_function_or_thunk_node): Likewise.
466         (varpool_alias_target): Likewise.
467         (cgraph_alias_target): Likewise.
468         (cgraph_node_name): Likewise.
469         (varpool_node_asm_name): Likewise.
470         (cgraph_node_asm_name): Likewise.
471         * cgraphbuild.c (remove_cgraph_callee_edges): Likewise.
472         (cgraph_rebuild_references): Likewise.
473         (rebuild_cgraph_edges): Likewise.
474         (record_eh_tables): Likewise.
475         (build_cgraph_edges): Likewise.
476         (mark_store): Likewise.
477         (mark_load): Likewise.
478         (mark_address): Likewise.
479         (record_type_list): Likewise.
480         (record_reference): Likewise.
481         * cgraphclones.c (cgraph_materialize_all_clones): Likewise.
482         (cgraph_materialize_clone): Likewise.
483         (cgraph_function_versioning): Likewise.
484         (cgraph_copy_node_for_versioning): Likewise.
485         (update_call_expr): Likewise.
486         (cgraph_find_replacement_node): Likewise.
487         (cgraph_create_virtual_clone): Likewise.
488         (cgraph_clone_node): Likewise.
489         * cgraphunit.c (compile): Likewise.
490         (output_weakrefs): Likewise.
491         (output_in_order): Likewise.
492         (expand_function): Likewise.
493         (assemble_thunks_and_aliases): Likewise.
494         (expand_thunk): Likewise.
495         (mark_functions_to_output): Likewise.
496         (handle_alias_pairs): Likewise.
497         (analyze_functions): Likewise.
498         (walk_polymorphic_call_targets): Likewise.
499         (varpool_finalize_decl): Likewise.
500         (process_function_and_variable_attributes): Likewise.
501         (cgraph_process_same_body_aliases): Likewise.
502         (analyze_function): Likewise.
503         (cgraph_add_new_function): Likewise.
504         (cgraph_finalize_function): Likewise.
505         (referred_to_p): Likewise.
506         (cgraph_reset_node): Likewise.
507         (cgraph_process_new_functions): Likewise.
508         (enqueue_node): Likewise.
509         (decide_is_symbol_needed): Likewise.
510         * coverage.c (coverage_compute_profile_id): Likewise.
511         * dbxout.c (dbxout_expand_expr): Likewise.
512         * dwarf2out.c (premark_types_used_by_global_vars_helper): Likewise.
513         (reference_to_unused): Likewise.
514         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
515         * gimplify.c (unvisit_body): Likewise.
516         (unshare_body): Likewise.
517         * ipa-cp.c (ipcp_generate_summary): Likewise.
518         (ipcp_decision_stage): Likewise.
519         (identify_dead_nodes): Likewise.
520         (decide_whether_version_node): Likewise.
521         (decide_about_value): Likewise.
522         (perhaps_add_new_callers): Likewise.
523         (create_specialized_node): Likewise.
524         (update_profiling_info): Likewise.
525         (ipcp_propagate_stage): Likewise.
526         (estimate_local_effects): Likewise.
527         (good_cloning_opportunity_p): Likewise.
528         (devirtualization_time_bonus): Likewise.
529         (propagate_constants_accross_call): Likewise.
530         (initialize_node_lattices): Likewise.
531         (ipcp_cloning_candidate_p): Likewise.
532         (determine_versionability): Likewise.
533         (print_all_lattices): Likewise.
534         (print_lattice): Likewise.
535         (ipcp_discover_new_direct_edges): Likewise.
536         * ipa-devirt.c (ipa_devirt): Likewise.
537         (likely_target_p): Likewise.
538         (update_type_inheritance_graph): Likewise.
539         (possible_polymorphic_call_target_p): Likewise.
540         (dump_possible_polymorphic_call_targets): Likewise.
541         (devirt_variable_node_removal_hook): Likewise.
542         (record_binfo): Likewise.
543         (maybe_record_node): Likewise.
544         (build_type_inheritance_graph): Likewise.
545         * ipa-inline-analysis.c (inline_write_summary): Likewise.
546         (inline_generate_summary): Likewise.
547         (inline_analyze_function): Likewise.
548         (do_estimate_growth): Likewise.
549         (simple_edge_hints): Likewise.
550         (estimate_node_size_and_time): Likewise.
551         (estimate_edge_devirt_benefit): Likewise.
552         (compute_inline_parameters): Likewise.
553         (estimate_function_body_sizes): Likewise.
554         (compute_bb_predicates): Likewise.
555         (initialize_inline_failed): Likewise.
556         (dump_inline_summary): Likewise.
557         (dump_inline_edge_summary): Likewise.
558         * ipa-inline-transform.c (inline_transform): Likewise.
559         (preserve_function_body_p): Likewise.
560         (save_inline_function_body): Likewise.
561         (inline_call): Likewise.
562         (clone_inlined_nodes): Likewise.
563         (can_remove_node_now_p): Likewise.
564         (can_remove_node_now_p_1): Likewise.
565         * ipa-inline.c (early_inliner): Likewise.
566         (early_inline_small_functions): Likewise.
567         (inline_always_inline_functions): Likewise.
568         (ipa_inline): Likewise.
569         (flatten_function): Likewise.
570         (inline_small_functions): Likewise.
571         (speculation_useful_p): Likewise.
572         (recursive_inlining): Likewise.
573         (update_caller_keys): Likewise.
574         (reset_edge_caches): Likewise.
575         (update_edge_key): Likewise.
576         (edge_badness): Likewise.
577         (relative_time_benefit): Likewise.
578         (want_inline_self_recursive_call_p): Likewise.
579         (want_inline_small_function_p): Likewise.
580         (want_early_inline_function_p): Likewise.
581         (num_calls): Likewise.
582         (can_early_inline_edge_p): Likewise.
583         (can_inline_edge_p): Likewise.
584         (report_inline_failed_reason): Likewise.
585         * ipa-profile.c (ipa_profile): Likewise.
586         (ipa_propagate_frequency): Likewise.
587         (ipa_propagate_frequency_1): Likewise.
588         (ipa_profile_generate_summary): Likewise.
589         * ipa-prop.c (ipcp_transform_function): Likewise.
590         (read_replacements_section): Likewise.
591         (ipa_prop_read_section): Likewise.
592         (ipa_modify_call_arguments): Likewise.
593         (ipa_print_node_params): Likewise.
594         (propagate_controlled_uses): Likewise.
595         (update_indirect_edges_after_inlining): Likewise.
596         (remove_described_reference): Likewise.
597         (ipa_make_edge_direct_to_target): Likewise.
598         (ipa_analyze_node): Likewise.
599         (ipa_analyze_params_uses): Likewise.
600         (ipa_compute_jump_functions): Likewise.
601         (ipa_get_callee_param_type): Likewise.
602         (ipa_print_node_jump_functions): Likewise.
603         (ipa_initialize_node_params): Likewise.
604         (ipa_populate_param_decls): Likewise.
605         (ipa_func_spec_opts_forbid_analysis_p): Likewise.
606         (write_agg_replacement_chain): Likewise.
607         (ipa_write_node_info): Likewise.
608         (ipa_edge_duplication_hook): Likewise.
609         (try_decrement_rdesc_refcount): Likewise.
610         * ipa-pure-const.c (propagate_nothrow): Likewise.
611         (propagate_pure_const): Likewise.
612         (pure_const_read_summary): Likewise.
613         (pure_const_write_summary): Likewise.
614         (analyze_function): Likewise.
615         * ipa-ref-inline.h (ipa_ref_referred_ref_list): Likewise.
616         (ipa_ref_referring_ref_list): Likewise.
617         * ipa-ref.c (ipa_clear_stmts_in_references): Likewise.
618         (ipa_remove_stmt_references): Likewise.
619         (ipa_find_reference): Likewise.
620         (ipa_dump_referring): Likewise.
621         (ipa_dump_references): Likewise.
622         (ipa_record_reference): Likewise.
623         * ipa-reference.c (ipa_reference_read_optimization_summary): Likewise.
624         (ipa_reference_write_optimization_summary): Likewise.
625         (write_node_summary_p): Likewise.
626         (propagate): Likewise.
627         (read_write_all_from_decl): Likewise.
628         (generate_summary): Likewise.
629         (analyze_function): Likewise.
630         (propagate_bits): Likewise.
631         (ipa_reference_get_not_written_global): Likewise.
632         (ipa_reference_get_not_read_global): Likewise.
633         * ipa-split.c (execute_split_functions): Likewise.
634         (split_function): Likewise.
635         * ipa-utils.c (ipa_merge_profiles): Likewise.
636         (dump_cgraph_node_set): Likewise.
637         (ipa_reverse_postorder): Likewise.
638         (ipa_edge_within_scc): Likewise.
639         (ipa_get_nodes_in_cycle): Likewise.
640         (ipa_free_postorder_info): Likewise.
641         (ipa_reduced_postorder): Likewise.
642         (searchc): Likewise.
643         (recursive_call_p): Likewise.
644         * ipa.c (ipa_cdtor_merge): Likewise.
645         (record_cdtor_fn): Likewise.
646         (function_and_variable_visibility): Likewise.
647         (varpool_externally_visible_p): Likewise.
648         (cgraph_externally_visible_p): Likewise.
649         (comdat_can_be_unshared_p): Likewise.
650         (comdat_can_be_unshared_p_1): Likewise.
651         (address_taken_from_non_vtable_p): Likewise.
652         (ipa_discover_readonly_nonaddressable_vars): Likewise.
653         (symtab_remove_unreachable_nodes): Likewise.
654         (walk_polymorphic_call_targets): Likewise.
655         (process_references): Likewise.
656         (enqueue_node): Likewise.
657         (has_addr_references_p): Likewise.
658         (cgraph_non_local_node_p_1): Likewise.
659         * is-a.h (varpool_analyze_node): Likewise.
660         * lto-cgraph.c (input_symtab): Likewise.
661         (merge_profile_summaries): Likewise.
662         (input_cgraph_1): Likewise.
663         (input_edge): Likewise.
664         (input_varpool_node): Likewise.
665         (input_node): Likewise.
666         (input_overwrite_node): Likewise.
667         (compute_ltrans_boundary): Likewise.
668         (output_refs): Likewise.
669         (lto_output_varpool_node): Likewise.
670         (lto_output_node): Likewise.
671         (reachable_from_other_partition_p): Likewise.
672         (referenced_from_other_partition_p): Likewise.
673         (lto_output_edge): Likewise.
674         (output_node_opt_summary): Likewise.
675         (add_node_to): Likewise.
676         (reachable_from_this_partition_p): Likewise.
677         (lto_set_symtab_encoder_in_partition): Likewise.
678         (lto_symtab_encoder_in_partition_p): Likewise.
679         (lto_set_symtab_encoder_encode_initializer): Likewise.
680         (lto_symtab_encoder_encode_initializer_p): Likewise.
681         (lto_set_symtab_encoder_encode_body): Likewise.
682         (lto_symtab_encoder_encode_body_p): Likewise.
683         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
684         Likewise.
685         * lto-streamer-in.c (lto_read_body): Likewise.
686         (fixup_call_stmt_edges): Likewise.
687         (fixup_call_stmt_edges_1): Likewise.
688         * lto-streamer-out.c (produce_symtab): Likewise.
689         (output_symbol_p): Likewise.
690         (write_symbol): Likewise.
691         (lto_output): Likewise.
692         (copy_function): Likewise.
693         (output_function): Likewise.
694         * passes.c (function_called_by_processed_nodes_p): Likewise.
695         (ipa_write_optimization_summaries): Likewise.
696         (ipa_write_summaries): Likewise.
697         (do_per_function_toporder): Likewise.
698         (do_per_function): Likewise.
699         (dump_passes): Likewise.
700         * symtab.c (symtab_semantically_equivalent_p): Likewise.
701         (symtab_nonoverwritable_alias): Likewise.
702         (symtab_nonoverwritable_alias_1): Likewise.
703         (symtab_for_node_and_aliases): Likewise.
704         (symtab_resolve_alias): Likewise.
705         (fixup_same_cpp_alias_visibility): Likewise.
706         (symtab_alias_ultimate_target): Likewise.
707         (symtab_used_from_object_file_p): Likewise.
708         (verify_symtab_base): Likewise.
709         (dump_symtab_base): Likewise.
710         (symtab_node_name): Likewise.
711         (symtab_node_asm_name): Likewise.
712         (symtab_dissolve_same_comdat_group_list): Likewise.
713         (symtab_add_to_same_comdat_group): Likewise.
714         (symtab_unregister_node): Likewise.
715         (symtab_insert_node_to_hashtable): Likewise.
716         (symtab_register_node): Likewise.
717         (unlink_from_assembler_name_hash): Likewise.
718         (insert_to_assembler_name_hash): Likewise.
719         (eq_assembler_name): Likewise.
720         (hash_node_by_assembler_name): Likewise.
721         (eq_node): Likewise.
722         (hash_node): Likewise.
723         * toplev.c (wrapup_global_declaration_2): Likewise.
724         * trans-mem.c (ipa_tm_execute): Likewise.
725         (ipa_tm_transform_clone): Likewise.
726         (ipa_tm_transform_transaction): Likewise.
727         (ipa_tm_transform_calls_redirect): Likewise.
728         (ipa_tm_insert_gettmclone_call): Likewise.
729         (ipa_tm_insert_irr_call): Likewise.
730         (ipa_tm_create_version): Likewise.
731         (ipa_tm_create_version_alias): Likewise.
732         (ipa_tm_mark_forced_by_abi_node): Likewise.
733         (ipa_tm_mark_force_output_node): Likewise.
734         (ipa_tm_diagnose_tm_safe): Likewise.
735         (ipa_tm_mayenterirr_function): Likewise.
736         (ipa_tm_scan_irr_function): Likewise.
737         (ipa_tm_note_irrevocable): Likewise.
738         (ipa_tm_scan_calls_clone): Likewise.
739         (get_cg_data): Likewise.
740         * tree-eh.c (tree_could_trap_p): Likewise.
741         * tree-emutls.c (ipa_lower_emutls): Likewise.
742         (create_emultls_var): Likewise.
743         (lower_emutls_function_body): Likewise.
744         (gen_emutls_addr): Likewise.
745         (emutls_decl): Likewise.
746         (new_emutls_decl): Likewise.
747         * tree-inline.c (tree_function_versioning): Likewise.
748         (optimize_inline_calls): Likewise.
749         (expand_call_inline): Likewise.
750         (estimate_num_insns): Likewise.
751         (copy_bb): Likewise.
752         (delete_unreachable_blocks_update_callgraph): Likewise.
753         * tree-nested.c (gimplify_all_functions): Likewise.
754         (create_nesting_tree): Likewise.
755         (check_for_nested_with_variably_modified): Likewise.
756         * tree-pretty-print.c (dump_function_header): Likewise.
757         * tree-profile.c (tree_profiling): Likewise.
758         * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
759         (modify_function): Likewise.
760         (convert_callers): Likewise.
761         (convert_callers_for_node): Likewise.
762         * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
763         (associate_varinfo_to_alias): Likewise.
764         (create_variable_info_for): Likewise.
765         (get_constraint_for_ssa_var): Likewise.
766         * tree-vectorizer.c (increase_alignment): Likewise.
767         * tree.c (find_decls_types_in_var): Likewise.
768         (find_decls_types_in_node): Likewise.
769         (free_lang_data_in_decl): Likewise.
770         * value-prof.c (gimple_ic_transform): Likewise.
771         (gimple_ic): Likewise.
772         (check_ic_target): Likewise.
773         (init_node_map): Likewise.
774         * varasm.c (decl_binds_to_current_def_p): Likewise.
775         (default_binds_local_p_1): Likewise.
776         (dump_tm_clone_pairs): Likewise.
777         (assemble_alias): Likewise.
778         (find_decl): Likewise.
779         (mark_decl_referenced): Likewise.
780         * varpool.c (varpool_for_node_and_aliases): Likewise.
781         (varpool_extra_name_alias): Likewise.
782         (varpool_create_variable_alias): Likewise.
783         (add_new_static_var): Likewise.
784         (varpool_finalize_named_section_flags): Likewise.
785         (varpool_remove_unreferenced_decls): Likewise.
786         (enqueue_node): Likewise.
787         (varpool_assemble_decl): Likewise.
788         (assemble_aliases): Likewise.
789         (varpool_analyze_node): Likewise.
790         (cgraph_variable_initializer_availability): Likewise.
791         (varpool_add_new_variable): Likewise.
792         (ctor_for_folding): Likewise.
793         (dump_varpool_node): Likewise.
794         (varpool_remove_initializer): Likewise.
795         (varpool_remove_node): Likewise.
796         (varpool_node_for_decl): Likewise.
797         (varpool_create_empty_node): Likewise.
798         * config/i386/i386.c (ix86_generate_version_dispatcher_body):
799         Likewise.
800         (ix86_get_function_versions_dispatcher): Likewise.
802 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
804         * cgraph.h (symtab_node_base): Convert to a class;
805         add GTY((desc ("%h.type"), tag ("SYMTAB_SYMBOL"))), and take
806         chain_next/prev from symtab_node_def.
807         (cgraph_node): Inherit from symtab_node; add GTY option
808         tag ("SYMTAB_FUNCTION").
809         (varpool_node): Inherit from symtab_node; add GTY option
810         tag ("SYMTAB_VARIABLE").
811         (symtab_node_def): Remove.
812         (is_a_helper <cgraph_node>::test (symtab_node_def *)): Convert to...
813         (is_a_helper <cgraph_node>::test (symtab_node_base *)): ...this.
814         (is_a_helper <varpool_node>::test (symtab_node_def *)): Convert to...
815         (is_a_helper <varpool_node>::test (symtab_node_base *)): ...this.
817         * ipa-ref.h (symtab_node_def): Drop.
818         (symtab_node): Change underlying type from symtab_node_def to
819         symtab_node_base.
820         (const_symtab_node): Likwise.
822         * is-a.h: Update examples in comment.
824         * symtab.c (symtab_hash): Change symtab_node_def to symtab_node_base.
825         (assembler_name_hash): Likewise.
827 2013-10-29  Martin Liska <marxin.liska@gmail.com>
829        * doc/tree-ssa.texi (gimple_phi_result): Document.
830        (gimple_phi_num_args, gimple_phi_arg): Likewise.
831        (gimple_phi_arg_edge, gimple_phi_arg_def): Likewise.
832        (PHI_RESULT, PHI_NUM_ARGS): Remove.
833        (PHI_ARG_ELT, PHI_ARG_EDGE, PHI_ARG_DEF): Likewise.
835 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
837         * expr.h: Revert change and include tree-core.h.
838         * rtl.h: Revert change and don't include tree-core.h.
840 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
842         * config/darwin.c: Include gimple.h.
843         * config/i386/winnt.c: Likewise.
845 2013-10-29  Marc Glisse  <marc.glisse@inria.fr>
847         PR tree-optimization/19831
848         * tree-ssa-alias.c (stmt_kills_ref_p_1): Handle BUILT_IN_FREE.
850 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
852         * tree-outof-ssa.h: Remove include files.
853         * tree-outof-ssa.c: Add required include files from tree-outof-ssa.h.
854         * expr.c: Likewise.
855         * tree-ssa-coalesce.c: Likewise.
856         * cfgexpand.c: Likewise.
857         * tree-ssa-ter.c: Likewise.
858         * ipa-prop.h: Remove gimple.h and tree-core.h from include list.
859         * lto-streamer.h: Likewise.
860         * cgraphbuild.c: Add gimple.h to include list.
861         * data-streamer-in.c: Likewise.
862         * ipa-cp.c: Likewise.
863         * tree-streamer.c: Likewise.
864         * lto-compress.c: Likewise.
865         * ipa-reference.c: Likewise.
866         * data-streamer-out.c: Likewise.
867         * lto-cgraph.c: Likewise.
868         * cgraphclones.c: Likewise.
869         * ipa-utils.c: Likewise.
870         * data-streamer.c: Likewise.
871         * ipa-split.c: Likewise.
872         * lto-section-in.c: Likewise.
873         * tree-streamer-out.c: Likewise.
874         * ipa-prop.c: Likewise.
875         * tree-streamer-in.c: Likewise.
876         * symtab.c: Likewise.
877         * opts-global.c: Likewise.
878         * lto-opts.c: Likewise.
879         * lto-section-out.c: Likewise.
880         * lto-streamer.c: Likewise.
881         * rtl.h: Add tree-core.h to include list.
882         * expr.h: Remove tree-core.h from include list.
883         * gimple.h: Likewise.
884         * ipa-utils.h: Likewise.
885         * streamer-hooks.h: Likewise.
886         * streamer-hooks.c: Include input.h.
888 2013-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
890         * config/arm/arm.c (cortexa7_extra_costs): New table.
891         (arm_cortex_a7_tune): New.
892         * config/arm/arm-cores.def: Use cortex_a7 tuning for cortex-a7.
894 2013-10-29  Eric Botcazou  <ebotcazou@adacore.com>
896         * expr.c (expand_expr_real_1) <MEM_EXPR>: Eliminate small redundancy.
898 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
900         * doc/gty.texi ("Inheritance and GTY"): Make it clear that
901         to use autogenerated markers for a class-hierarchy, every class
902         must have a GTY marker.
903         * gengtype.h (struct type): Add linked list of subclasses to
904         the "s" member of the union.
905         (add_subclass): New decl.
906         * gengtype-state.c (read_state_struct_type): Set up subclass
907         linked list.
908         * gengtype.c (get_ultimate_base_class): New.
909         (add_subclass): New.
910         (new_structure): Set up subclass linked list.
911         (set_gc_used_type): Propagate usage information to subclasses.
912         (output_mangled_typename): Use get_ultimate_base_class.
913         (walk_subclasses): Use the subclass linked list, avoiding an
914         O(N^2) when writing out all types.
915         (walk_type): Issue an error if the base class is missing a tag,
916         rather than generating bogus C code.  Add a gcc_unreachable
917         default case, in case people omit tags from concrete subclasses,
918         or get the values wrong.
919         (write_func_for_structure): Issue an error for subclasses for
920         which the base doesn't have a "desc", since otherwise the
921         autogenerated routines for the base would silently fail to visit
922         any subclass fields.
923         (write_root): Use get_ultimate_base_class, tweaking constness of
924         tp to match that function's signature.
926 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
928         * doc/gty.texi (GTY Options): Add note about inheritance to
929         description of desc and tag.
930         (Inheritance and GTY): New.
932 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
934         * gengtype-parse.c (opts_have): Drop "static" so that
935         we can use this from gengtype.c.
936         * gengtype.c (set_gc_used_type): Mark any base class as used;
937         update field traversal to visit inherited fields.
938         (output_mangled_typename):  Convert references to classes within
939         an inheritance hierarchy to reference the ultimate base class,
940         since only it will have gt_ functions.
941         (get_string_option): New.
942         (walk_subclasses): New.
943         (walk_type): Treat GTY structs that have a "desc" as being the
944         root of an inheritance hierarchy.  Generate a switch on it
945         within the marking function which walks all subclasses, adding
946         cases for them via walk_subclasses.  For subclasses, visit all
947         fields of the type (including inherited ones).
948         (write_func_for_structure): Don't write fns for subclasses, only
949         for the ultimate base class within an inheritance hierarchy.
950         Subclasses-marking will be handled by the base class marking
951         functions.
952         (write_types): Likewise.
953         (write_local_func_for_structure): Likewise.
954         (USED_BY_TYPED_GC_P): Emit allocators for subclasses that have
955         a "tag" option (and are thus concrete subclasses).
956         (write_root): Use the marker function for the ultimate base class.
957         * gengtype.h (FOR_ALL_INHERITED_FIELDS): New.
958         (opts_have): Add declaration.
960 2013-10-28  Vladimir Makarov  <vmakarov@redhat.com>
962         * lra-spills.c (lra_final_code_change): Remove useless move insns
963         originated from moves of pseudos.
965 2013-10-28  Jeff Law  <law@redhat.com>
967         * lower-subreg.c (resolve_simple_move): Fix comment typo.
969 2013-10-28  Trevor Saunders  <tsaunders@mozilla.com>
971         * df-scan.c (df_collection_rec): Adjust.
972         (copy_defs): New constant.
973         (copy_uses): Likewise.
974         (copy_eq_uses): Likewise.
975         (copy_mw): Likewise.
976         (copy_all): Likewise.
977         (df_insn_rescan): Adjust.
978         (df_notes_rescan): Likewise.
979         (df_swap_refs): Likewise.
980         (df_sort_and_compress_refs): Likewise.
981         (df_sort_and_compress_mws): Likewise.
982         (df_install_refs): Likewise.
983         (df_install_mws): Likewise.
984         (df_refs_add_to_chains): Add flags parameter controlling which vectors
985         are coppied.
986         (df_bb_refs_record): Adjust.
987         (df_record_entry_block_defs): Likewise.
988         (df_record_exit_block_defs): Likewise.
989         (df_refs_verify): Likewise.
990         (df_mws_verify): Likewise.
991         (df_insn_refs_verify): Likewise.
992         (df_bb_verify): Likewise.
993         * ipa-pure-const.c (finish_state): Remove.
994         (propagate): Adjust.
995         * tree-data-ref.c tree-ssa-alias.c tree-ssa-loop-ivcanon.c
996         tree-ssa-threadedge.c tree-vect-loop-manip.c tree-vect-slp.c
997         var-tracking.c: Adjust.
998         * vec.c (stack_vecs): Remove.
999         (register_stack_vec): Likewise.
1000         (stack_vec_register_index): Likewise.
1001         (unregister_stack_vec): Likewise.
1002         * vec.h (struct va_stack): Remove.
1003         (struct vec<T, A, vl_ptr>): Specialize as
1004         struct vec<T, va_heap, vl_ptr> instead since va_heap is the only
1005         allocation strategy compatable with the vl_ptr layout.
1006         (struct vec<T, va_gc, vl_ptr>): Remove because it now gets an empty
1007         specialization anyway.
1008         (class stack_vec): New class.
1009         (vec_stack_alloc): Remove.
1010         (vec<T, va_heap, vl_ptr>::using_auto_storage): New method.
1012 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1013             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1014             Sergey Lega  <sergey.s.lega@intel.com>
1015             Anna Tikhonova  <anna.tikhonova@intel.com>
1016             Ilya Tocar  <ilya.tocar@intel.com>
1017             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1018             Ilya Verbin  <ilya.verbin@intel.com>
1019             Kirill Yukhin  <kirill.yukhin@intel.com>
1020             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1022         * config/i386/i386.md (prefetch): Allow TARGET_AVX512PF.
1023         (*prefetch_avx512pf_<mode>): New.
1024         * config/i386/sse.md (avx512f_vmcmp<mode>3): Ditto.
1025         (avx512f_maskcmp<mode>3): Ditto.
1026         (vashrv16si3): Ditto.
1028 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1029             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1030             Sergey Lega  <sergey.s.lega@intel.com>
1031             Anna Tikhonova  <anna.tikhonova@intel.com>
1032             Ilya Tocar  <ilya.tocar@intel.com>
1033             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1034             Ilya Verbin  <ilya.verbin@intel.com>
1035             Kirill Yukhin  <kirill.yukhin@intel.com>
1036             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1038         * config/i386/i386.md (any_truncate): New.
1039         (trunsuffix): Ditto.
1040         * config/i386/predicates.md (const_8_to_9_operand): New.
1041         (const_10_to_11_operand): Ditto.
1042         (const_12_to_13_operand): Ditto.
1043         (const_14_to_15_operand): Ditto.
1044         (const_16_to_19_operand): Ditto.
1045         (const_20_to_23_operand): Ditto.
1046         (const_24_to_27_operand): Ditto.
1047         (const_28_to_31_operand): Ditto.
1048         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_FIX_NOTRUNC.
1049         (cvtusi2<ssescalarmodesuffix>32): New.
1050         (cvtusi2<ssescalarmodesuffix>64): Ditto.
1051         (ufloatv16siv16sf2): Ditto.
1052         (avx512f_fix_notruncv16sfv16si): Ditto.
1053         (avx512f_ufix_notruncv16sfv16si): Ditto.
1054         (avx512f_vcvtss2usi): Ditto.
1055         (avx512f_vcvtss2usiq): Ditto.
1056         (avx512f_vcvttss2usi): Ditto.
1057         (avx512f_vcvttss2usiq): Ditto.
1058         (avx512f_vcvtsd2usi): Ditto.
1059         (avx512f_vcvtsd2usiq): Ditto.
1060         (avx512f_vcvttsd2usi): Ditto.
1061         (avx512f_vcvttsd2usiq): Ditto.
1062         (ufloatv8siv8df): Ditto.
1063         (avx512f_cvtdq2pd512_2): Ditto.
1064         (avx512f_cvtpd2dq512): Ditto.
1065         (avx512f_ufix_notruncv8dfv8si): Ditto.
1066         (avx512f_cvtpd2ps512): Ditto.
1067         (vec_unpacks_lo_v16sf): Ditto.
1068         (vec_unpacks_hi_v16sf): Ditto.
1069         (vec_unpacks_float_hi_v16si): Ditto.
1070         (vec_unpacks_float_lo_v16si): Ditto.
1071         (avx512f_unpckhps512): Ditto.
1072         (avx512f_unpcklps512): Ditto.
1073         (avx512f_movshdup512): Ditto.
1074         (avx512f_movsldup512): Ditto.
1075         (vec_extract_lo_v32hi): Ditto.
1076         (vec_extract_hi_v32hi): Ditto.
1077         (vec_extract_lo_v64qi): Ditto.
1078         (vec_extract_hi_v64qi): Ditto.
1079         (avx512f_unpckhpd512): Ditto.
1080         (avx512f_movddup512): Ditto.
1081         (avx512f_unpcklpd512): Ditto.
1082         (*avx512f_unpcklpd512): Ditto.
1083         (avx512f_shufps512_1): Ditto.
1084         (avx512f_shufpd512_1): Ditto.
1085         (avx512f_interleave_highv8di): Ditto.
1086         (avx512f_interleave_lowv8di): Ditto.
1087         (PMOV_DST_MODE): Ditto.
1088         (pmov_src_mode): Ditto.
1089         (pmov_src_lower): Ditto.
1090         (pmov_suff): Ditto.
1091         (*avx512f_<code><pmov_src_lower><mode>2): Ditto.
1092         (*avx512f_<code>v8div16qi2): Ditto.
1093         (*avx512f_<code>v8div16qi2_store): Ditto.
1094         (vec_widen_umult_even_v16si): Ditto.
1095         (*vec_widen_umult_even_v16si): Ditto.
1096         (vec_widen_smult_even_v16si): Ditto.
1097         (*vec_widen_smult_even_v16si): Ditto.
1098         (avx512f_interleave_highv16si): Ditto.
1099         (avx512f_interleave_lowv16si): Ditto.
1100         (avx512f_<code>v16qiv16si2): Ditto.
1101         (avx512f_<code>v16hiv16si2): Ditto.
1102         (avx512f_<code>v8qiv8di2): Ditto.
1103         (avx512f_<code>v8hiv8di2): Ditto.
1104         (avx512f_<code>v8siv8di2): Ditto.
1105         (avx512cd_maskb_vec_dupv8di): Ditto.
1106         (avx512cd_maskw_vec_dupv16si): Ditto.
1107         (avx512f_vcvtph2ps512): Ditto.
1108         (avx512f_vcvtps2ph512): Ditto.
1109         (VEC_EXTRACT_MODE): Extened with wider modes.
1110         (VEC_PERM_AVX2): Ditto.
1111         (VEC_PERM_CONST): Ditto.
1114 2013-10-28  Joern Rennecke  <joern.rennecke@embecosm.com>
1116         * config/arc/arc.c (arc_ccfsm_post_advance):
1117         Add comment about TYPE_RETURN.
1119 2013-10-28  Bin Cheng  <bin.cheng@arm.com>
1121         * tree-ssa-loop-ivopts.c (strip_offset_1): Change parameter type.
1122         Count DECL_FIELD_BIT_OFFSET for COMPONENT_REF.
1123         (strip_offset): Convert offset to unsigned number.
1125 2013-10-27  Tom de Vries  <tom@codesourcery.com>
1127         * cfgexpand.c (gimple_expand_cfg): Remove test for parm_birth_insn.
1128         Don't commit insertions after NOTE_INSN_FUNCTION_BEG.
1130 2013-10-27  Oleg Endo  <olegendo@gcc.gnu.org>
1132         * config/sh/sh.c (MSW, LSW): Move and rename macros to...
1133         * config/sh/sh.h (SH_REG_MSW_OFFSET, SH_REG_LSW_OFFSET): ... here.
1134         (TARGET_BIG_ENDIAN): New macro.
1135         * config/sh/sh.md: Use it instead of !TARGET_LITTLE_ENDIAN.
1136         Use SH_REG_MSW_OFFSET and SH_REG_LSW_OFFSET.
1137         * config/sh/sh.c: Likewise.
1138         * config/sh/sh.h: Likewise.
1140 2013-10-27  Hans-Peter Nilsson  <hp@axis.com>
1142         * config/cris/cris.c (cris_emit_trap_for_misalignment): Replace the
1143         removed PRED_MUDFLAP with PRED_NORETURN.  Correct file-path in comment.
1145 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1147         * config/sh/sh.md (movmemsi): Remove empty constraints and predicates.
1148         Fix formatting.
1149         (cmpstr_t, cmpstrsi): Fix formatting.
1151 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1153         PR target/52483
1154         * config/sh/predicates.md (general_movdst_operand): Allow reg+reg
1155         addressing, do not use general_operand for memory operands.
1157 2013-10-26  Vladimir Makarov  <vmakarov@redhat.com>
1159         Revert:
1160         2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1161         * lra-spills.c (lra_final_code_change): Remove useless move insns.
1163 2013-10-26  Jeff Law  <law@redhat.com>
1165         * predict.c (PRED_MUDFLAP): Remove.
1166         * targhooks.c (build_va_arg_indirect_ref): Remove mudflap support.
1168         * Makefile.in (C_COMMON_OBJS): Remove tree-mudflap.
1169         (OBJS): Remove tree-nomudflap.o
1170         (GTFILES): Remove tree-mudflap.c
1171         * builtins.c (expand_builtin_alloc): Remove mudflap support.
1172         * gcc.c (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
1173         (mfwrap_spec, mflib_spec): Likewise.
1174         (cpp_unique_options, cc1_options, static_specs): Likewise.
1175         * gimplify (gimplify_vla_decl, build_va_arg_indirect_ref): Likewise.
1176         * passes.def: Likewise.
1177         * toplev.c (compile_file, process_options): Likewise.
1178         * tree-inline.c (copy_tree_r): Likewise.
1179         * tree-pass.,h (make_pass_mudflap_1, make_pass_mudflap_2): Likewise.
1180         * varasm.c (make_decl_rtl, make_decl_rtl_for_debug): Likewise.
1181         (build_constant_desc, output_constant_def_contents): Likewise.
1182         (categorize_decl_for_section): Likewise.
1183         * tree-mudflap.c: Removed.
1184         * tree-mudflap.h: Removed.
1185         * tree-nomudflap.c: Removed.
1186         * bfin/uclinux.h (MFWRAP_SPEC): Remove.
1187         * moxie/uclinux.h (MFWRAP_SPEC): Likewise.
1188         * rs6000/aix.h (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
1189         * config/sol2.h (MFLIB_SPEC): Likewise.
1190         * doc/install.texi: Remove mudflap references.
1191         * doc/passes.texi: Similarly.
1192         * doc/sourcebuild.texi: Similarly.
1193         * doc/invoke.texi: Remove mudlfap related options.
1195 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1197         PR rtl-optimization/58759
1198         * lra-constraints.c (lra_constraints): Remove wrong condition to
1199         remove insn setting up an equivalent pseudo.
1201 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1203         * config/rs6000/rs6000-protos.h
1204         (rs6000_secondary_memory_needed_mode): New prototype.
1205         * config/rs6000/rs6000.c: Include ira.h.
1206         (TARGET_LRA_P): Redefine.
1207         (rs6000_legitimate_offset_address_p): Call
1208         legitimate_constant_pool_address_p in strict mode for LRA.
1209         (rs6000_legitimate_address_p): Ditto.
1210         (legitimate_lo_sum_address_p): Add code for LRA.
1211         Use lra_in_progress.
1212         (rs6000_emit_move): Add LRA version of code to generate load/store
1213         of SDmode values.
1214         (rs6000_secondary_memory_needed_mode): New.
1215         (rs6000_alloc_sdmode_stack_slot): Do nothing for LRA.
1216         (rs6000_secondary_reload_class): Return NO_REGS for LRA for
1217         constants, memory, and FP registers.
1218         (rs6000_lra_p): New.
1219         * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): New
1220         macro.
1221         * config/rs6000/rs6000.opt (mlra): New option.
1222         * lra-spills.c (lra_final_code_change): Remove useless move insns.
1224 2013-10-25  Yufeng Zhang  <yufeng.zhang@arm.com>
1226         * tree-ssa-math-opts.c (convert_plusminus_to_widen): Call
1227         has_single_use () and not do the conversion if has_single_use ()
1228         returns false for the multiplication result.
1230 2013-10-25  David Malcolm  <dmalcolm@redhat.com>
1232         * tree.h (EXCEPTIONAL_CLASS_P): Rename parameter from "CODE"
1233         to "NODE", since this works on a "tree", not an
1234         "enum tree_code".
1235         (CONSTANT_CLASS_P): Likewise.
1236         (TYPE_P): Likewise.
1237         (DECL_P): Likewise.
1238         (INDIRECT_REF_P): Likewise.
1239         (REFERENCE_CLASS_P): Likewise.
1240         (COMPARISON_CLASS_P): Likewise.
1241         (UNARY_CLASS_P): Likewise.
1242         (BINARY_CLASS_P): Likewise.
1243         (STATEMENT_CLASS_P): Likewise.
1244         (VL_EXP_CLASS_P): Likewise.
1245         (EXPRESSION_CLASS_P): Likewise.
1246         (IS_TYPE_OR_DECL_P): Likewise.
1248 2013-10-25  Marc Glisse  <marc.glisse@inria.fr>
1250         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for an
1251         ADDR_EXPR in the defining statement.
1253 2013-10-25  Richard Biener  <rguenther@suse.de>
1255         PR tree-optimization/58626
1256         * tree-loop-distribution.c (enum rdg_dep_type): Remove
1257         anti_dd, output_dd and input_dd.
1258         (struct rdg_edge): Remove level and relation members.
1259         (RDGE_LEVEL, RDGE_RELATION): Remove.
1260         (dot_rdg_1): Adjust.
1261         (create_rdg_edge_for_ddr): Remove.
1262         (create_rdg_edges_for_scalar): Adjust.
1263         (create_edge_for_control_dependence): Likewise.
1264         (create_rdg_edges): Split into ...
1265         (create_rdg_flow_edges): ... this
1266         (create_rdg_cd_edges): ... and this.
1267         (free_rdg): Adjust.
1268         (build_rdg): Likewise, do not compute data dependences or
1269         add edges for them.
1270         (pg_add_dependence_edges): New function.
1271         (pgcmp): Likewise.
1272         (distribute_loop): First apply all non-dependence based
1273         partition mergings.  Then compute dependences between partitions
1274         and merge and order partitions according to them.
1276 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
1278         PR rtl-optimization/58831
1279         * alias.c (init_alias_analysis): At the beginning of each iteration, set
1280         the reg_seen[N] bit if static_reg_base_value[N] is non-null.
1282 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
1284         * recog.c (search_ofs): New static variable moved from...
1285         (peep2_find_free_register): ...here.
1286         (peephole2_optimize): Initialize it.
1288 2013-10-25  Tobias Burnus  <burnus@net-b.de>
1290         * doc/invoke.texi (fopenmp): Change supported OpenMP version to 4.0.
1292 2013-10-25  Uros Bizjak  <ubizjak@gmail.com>
1294         * config/i386/i386.h (TARGET_MPX): New define.
1295         (TARGET_MPX_P): Ditto.
1297 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
1299         * config/i386/constraints.md (B): New.
1300         (Ti): New.
1301         (Tb): New.
1302         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
1303         * config/i386/i386-modes.def (BND32): New.
1304         (BND64): New.
1305         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
1306         * config/i386/i386.c (isa_opts): Add mmpx.
1307         (regclass_map): Add bound registers.
1308         (dbx_register_map): Likewise.
1309         (dbx64_register_map): Likewise.
1310         (svr4_dbx_register_map): Likewise.
1311         (PTA_MPX): New.
1312         (ix86_option_override_internal): Support MPX ISA.
1313         (ix86_conditional_register_usage): Support bound registers.
1314         (print_reg): Likewise.
1315         (ix86_code_end): Add MPX bnd prefix.
1316         (output_set_got): Likewise.
1317         (ix86_output_call_insn): Likewise.
1318         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
1319         (ix86_print_operand_punct_valid_p): Likewise.
1320         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
1321         UNSPEC_BNDMK_ADDR.
1322         (ix86_class_likely_spilled_p): Add bound regs support.
1323         (ix86_hard_regno_mode_ok): Likewise.
1324         (x86_order_regs_for_local_alloc): Likewise.
1325         (ix86_bnd_prefixed_insn_p): New.
1326         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
1327         (FIXED_REGISTERS): Add bound registers.
1328         (CALL_USED_REGISTERS): Likewise.
1329         (REG_ALLOC_ORDER): Likewise.
1330         (HARD_REGNO_NREGS): Likewise.
1331         (TARGET_MPX): New.
1332         (VALID_BND_REG_MODE): New.
1333         (FIRST_BND_REG): New.
1334         (LAST_BND_REG): New.
1335         (reg_class): Add BND_REGS.
1336         (REG_CLASS_NAMES): Likewise.
1337         (REG_CLASS_CONTENTS): Likewise.
1338         (BND_REGNO_P): New.
1339         (ANY_BND_REG_P): New.
1340         (BNDmode): New.
1341         (HI_REGISTER_NAMES): Add bound registers.
1342         * config/i386/i386.md (UNSPEC_BNDMK): New.
1343         (UNSPEC_BNDMK_ADDR): New.
1344         (UNSPEC_BNDSTX): New.
1345         (UNSPEC_BNDLDX): New.
1346         (UNSPEC_BNDLDX_ADDR): New.
1347         (UNSPEC_BNDCL): New.
1348         (UNSPEC_BNDCU): New.
1349         (UNSPEC_BNDCN): New.
1350         (UNSPEC_MPX_FENCE): New.
1351         (BND0_REG): New.
1352         (BND1_REG): New.
1353         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
1354         (length_immediate): Likewise.
1355         (prefix_0f): Likewise.
1356         (memory): Likewise.
1357         (prefix_rep): Check for bnd prefix.
1358         (length_nobnd): New.
1359         (length): Use length_nobnd if specified.
1360         (BND): New.
1361         (bnd_ptr): New.
1362         (BNDCHECK): New.
1363         (bndcheck): New.
1364         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
1365         (*jcc_2): Likewise.
1366         (jump): Likewise.
1367         (simple_return_internal): Likewise.
1368         (simple_return_pop_internal): Likewise.
1369         (*indirect_jump): Add MPX bnd prefix.
1370         (*tablejump_1): Likewise.
1371         (simple_return_internal_long): Likewise.
1372         (simple_return_indirect_internal): Likewise.
1373         (<mode>_mk): New.
1374         (*<mode>_mk): New.
1375         (mov<mode>): New.
1376         (*mov<mode>_internal_mpx): New.
1377         (<mode>_<bndcheck>): New.
1378         (*<mode>_<bndcheck>): New.
1379         (<mode>_ldx): New.
1380         (*<mode>_ldx): New.
1381         (<mode>_stx): New.
1382         (*<mode>_stx): New.
1383         * config/i386/predicates.md (lea_address_operand): Rename to...
1384         (address_no_seg_operand): ... this.
1385         (address_mpx_no_base_operand): New.
1386         (address_mpx_no_index_operand): New.
1387         (bnd_mem_operator): New.
1388         * config/i386/i386.opt (mmpx): New.
1389         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
1390         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
1392 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
1394         * mode-classes.def (MODE_POINTER_BOUNDS): New.
1395         * tree.def (POINTER_BOUNDS_TYPE): New.
1396         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
1397         (POINTER_BOUNDS_MODE): New.
1398         (make_pointer_bounds_mode): New.
1399         * machmode.h (POINTER_BOUNDS_MODE_P): New.
1400         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
1401         (layout_type): Support POINTER_BOUNDS_TYPE.
1402         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
1403         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
1404         (type_contains_placeholder_1): Likewise.
1405         * tree.h (POINTER_BOUNDS_TYPE_P): New.
1406         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
1407         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
1409 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
1411         * expr.c (expand_expr_real_1): Use mode of memory reference rather than
1412         mode of address computation when calling memory_address_addr_space.
1414 2013-08-24  Richard Henderson  <rth@twiddle.net>
1416         PR rtl/58542
1417         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
1418         instead of create_convert_operand_to.
1419         (maybe_emit_sync_lock_test_and_set): Likewise.
1420         (expand_atomic_compare_and_swap): Likewise.
1421         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
1423 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
1425         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
1426         Change param_is to use the struct and not the pointer to the struct.
1428 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
1430         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
1431         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
1432         Move to targhooks.c.
1433         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
1434         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
1435         Relocate from gimplify.c.
1436         * targhooks.h: Add 2 prototypes.
1437         * tree.h: Delete 2 prototypes.
1439 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
1441         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
1442         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
1444 2013-10-24  Cong Hou  <congh@google.com>
1446         * convert.c (convert_to_real): Guard those unsafe math function
1447         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
1448         specially.
1450 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
1452         PR ipa/58712
1453         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
1454         as argument.
1455         (cgraph_create_edge): Use the new argument.
1456         (cgraph_create_indirect_edge): Likewise.
1458 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
1460         * gcc/config/arc/arc.c (arc_ccfsm_post_advance): Also handle
1461         TYPE_UNCOND_BRANCH.
1462         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
1463         changes statep->state.
1465 2013-10-24  Tobias Burnus  <burnus@net-b.de>
1467         PR other/33426
1468         * tree-cfg.c (replace_loop_annotate): New function.
1469         (execute_build_cfg): Call it.
1470         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
1471         * internal-fn.c (expand_ANNOTATE): New function.
1472         * internal-fn.def (ANNOTATE): Define as new internal function.
1473         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
1474         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
1475         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
1476         * doc/extend.texi (Pragmas): Document #pragma ivdep.
1477         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
1479 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
1480             Marcus Shawcroft  <marcus.shawcroft@arm.com>
1482         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
1483         Special case reload SP+C into none GENERAL_REGS.
1485 2013-10-24  Michael Matz  <matz@suse.de>
1487         * gengtype.c (is_file_equal): Check that files will be same length.
1489 2013-10-25  Christian Bruel  <christian.bruel@st.com>
1491         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
1492         * gcc/config/sh/t-sh (sh-mem.o): New rule.
1493         * gcc/config/sh/sh-mem.cc (expand_block_move): Moved here.
1494         (sh_expand_cmpstr): New function.
1495         * gcc/config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c.
1496         * gcc/config/sh/sh-protos.h (sh_expand_cmpstr): Declare.
1497         * gcc/config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
1498         (rotlhi3_8): Rename.
1500 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1502         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
1503         * Makefile.in (CXXDEPMODE): Assign and change users.
1504         (CCDEPMODE): Delete.
1505         * configure: Regenerate.
1507 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
1509         * gengtype-parse.c (require_without_advance): New.
1510         (type): For GTY-marked types that are not GTY((user)), parse any
1511         base classes, requiring them to be single-inheritance, and not
1512         be templates.  For non-GTY-marked types and GTY((user)),
1513         continue to skip over any C++ inheritance specification.
1514         * gengtype-state.c (state_writer::write_state_struct_type):
1515         Write base class of type (if any).
1516         (read_state_struct_type): Read base class of type (if any).
1517         * gengtype.c (new_structure): Add a "base_class" parameter.
1518         (create_optional_field_): Update for new parameter to new_structure.
1519         (adjust_field_rtx_def): Likewise.
1520         (adjust_field_tree_exp): Likewise.
1521         * gengtype.h (struct type): Add "base_class" field to the s
1522         union field.
1523         (new_structure): Add "base" parameter.
1525 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
1527         PR target/57756
1528         * config/i386/i386.c (ix86_option_override_internal):
1529         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
1530         (ix86_valid_target_attribute_tree):
1531         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
1532         Change TARGET_SSE to TARGET_SSE_P (opts->...)
1534 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
1536         * tree-ssa-loop.h: Remove include files.
1537         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
1538         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
1539         * graphite-clast-to-gimple.c: Likewise.
1540         * graphite-scop-detection.c: Likewise.
1541         * graphite-sese-to-poly.c: Likewise.
1542         * ipa-inline-analysis.c: Likewise.
1543         * ipa-pure-const.c: Likewise.
1544         * loop-init.c: Likewise.
1545         * passes.c: Likewise.
1546         * predict.c: Likewise.
1547         * tree-cfg.c: Likewise.
1548         * tree-cfgcleanup.c: Likewise.
1549         * tree-chrec.c: Likewise.
1550         * tree-data-ref.c: Likewise.
1551         * tree-loop-distribution.c: Likewise.
1552         * tree-parloops.c: Likewise.
1553         * tree-predcom.c: Likewise.
1554         * tree-scalar-evolution.c: Likewise.
1555         * tree-ssa-address.c: Likewise.
1556         * tree-ssa.c: Likewise.
1557         * tree-ssa-dce.c: Likewise.
1558         * tree-ssa-loop.c: Likewise.
1559         * tree-ssa-loop-im.c: Likewise.
1560         * tree-ssa-loop-ivcanon.c: Likewise.
1561         * tree-ssa-loop-ivopts.c: Likewise.
1562         * tree-ssa-loop-manip.c: Likewise.
1563         * tree-ssa-loop-niter.c: Likewise.
1564         * tree-ssa-loop-prefetch.c: Likewise.
1565         * tree-ssa-loop-unswitch.c: Likewise.
1566         * tree-ssa-reassoc.c: Likewise.
1567         * tree-vect-data-refs.c: Likewise.
1568         * tree-vect-loop.c: Likewise.
1569         * tree-vect-loop-manip.c: Likewise.
1570         * tree-vectorizer.c: Likewise.
1571         * tree-vect-stmts.c: Likewise.
1572         * tree-vrp.c: Likewise.
1574 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1576         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
1578 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
1580         PR tree-optimization/58775
1581         PR tree-optimization/58791
1582         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
1583         (insert_stmt_after): Rewritten, don't move the stmt, but really
1584         insert it.
1585         (get_stmt_uid_with_default): Remove.
1586         (build_and_add_sum): Use insert_stmt_after and
1587         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only labels.
1588         (update_range_test): Set uid on stmts added by
1589         force_gimple_operand_gsi.  Don't immediately modify statements
1590         in inter-bb optimization, just update oe->op values.
1591         (optimize_range_tests): Return bool whether any changed have been made.
1592         (update_ops): New function.
1593         (struct inter_bb_range_test_entry): New type.
1594         (maybe_optimize_range_tests): Perform statement changes here.
1595         (not_dominated_by, appears_later_in_bb, get_def_stmt,
1596         ensure_ops_are_available): Remove.
1597         (find_insert_point): Rewritten.
1598         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
1599         return LHS of the (new resp. old) stmt.  Don't call
1600         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
1601         instead of last, move new stmt at the right place.
1602         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
1603         (negate_value): Likewise.  Set uids.
1604         (break_up_subtract_bb): Initialize uids.
1605         (reassociate_bb): Adjust rewrite_expr_tree caller.
1606         (do_reassoc): Don't call renumber_gimple_stmt_uids.
1608 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
1610         PR target/58838
1611         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
1612         TARGET_32BIT final condition.
1613         (mulsi3_internal2 and splitter): Same.
1615 2013-10-23  Jeff Law  <law@redhat.com>
1617         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
1618         through joiner blocks with abnormal outgoing edges.
1620         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
1621         Add parameter JOINERS, to allow/disallow threading through joiner
1622         blocks.
1623         (thread_block): New.  Call thread_block_1.
1624         (mark_threaded_blocks): Remove code to filter out certain cases
1625         of threading through joiner blocks.
1626         (thread_through_all_blocks): Document how we can have a dangling
1627         edge AUX field and clear it.
1629 2013-10-23  Ian Lance Taylor  <iant@google.com>
1631         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
1632         (C++ Dialect Options): Likewise.
1633         (Optimize Options): Likewise.
1635 2013-10-23  Tom de Vries  <tom@codesourcery.com>
1637         PR tree-optimization/58805
1638         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
1640 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
1642         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
1643         sequence based on get_range_info returned range.
1645 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
1647         * tree-ssa.h: Remove all #include's
1648         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
1649         * alias.c: Move required includes from tree-ssa.h.
1650         * asan.c: Likewise.
1651         * builtins.c: Likewise.
1652         * calls.c: Likewise.
1653         * cfgexpand.c: Likewise.
1654         * cfghooks.c: Likewise.
1655         * cfgloop.c: Likewise.
1656         * cfgloopmanip.c: Likewise.
1657         * cgraph.c: Likewise.
1658         * cgraphbuild.c: Likewise.
1659         * cgraphclones.c: Likewise.
1660         * cgraphunit.c: Likewise.
1661         * dse.c: Likewise.
1662         * except.c: Likewise.
1663         * expr.c: Likewise.
1664         * final.c: Likewise.
1665         * fold-const.c: Likewise.
1666         * ggc-page.c: Likewise.
1667         * gimple-builder.c: Likewise.
1668         * gimple-fold.c: Likewise.
1669         * gimple-iterator.c: Likewise.
1670         * gimple-low.c: Likewise.
1671         * gimple-pretty-print.c: Likewise.
1672         * gimple-ssa-strength-reduction.c: Likewise.
1673         * gimple-streamer-in.c: Likewise.
1674         * gimple-streamer-out.c: Likewise.
1675         * gimplify.c: Likewise.
1676         * graphite-blocking.c: Likewise.
1677         * graphite-clast-to-gimple.c: Likewise.
1678         * graphite-dependences.c: Likewise.
1679         * graphite-interchange.c: Likewise.
1680         * graphite-optimize-isl.c: Likewise.
1681         * graphite-poly.c: Likewise.
1682         * graphite-scop-detection.c: Likewise.
1683         * graphite-sese-to-poly.c: Likewise.
1684         * graphite.c: Likewise.
1685         * ipa-cp.c: Likewise.
1686         * ipa-inline-analysis.c: Likewise.
1687         * ipa-inline-transform.c: Likewise.
1688         * ipa-inline.c: Likewise.
1689         * ipa-prop.c: Likewise.
1690         * ipa-pure-const.c: Likewise.
1691         * ipa-reference.c: Likewise.
1692         * ipa-split.c: Likewise.
1693         * ipa-utils.c: Likewise.
1694         * loop-init.c: Likewise.
1695         * lto-cgraph.c: Likewise.
1696         * lto-section-in.c: Likewise.
1697         * lto-section-out.c: Likewise.
1698         * lto-streamer-in.c: Likewise.
1699         * lto-streamer-out.c: Likewise.
1700         * lto-streamer.c: Likewise.
1701         * omp-low.c: Likewise.
1702         * passes.c: Likewise.
1703         * predict.c: Likewise.
1704         * print-tree.c: Likewise.
1705         * profile.c: Likewise.
1706         * sese.c: Likewise.
1707         * targhooks.c: Likewise.
1708         * tracer.c: Likewise.
1709         * trans-mem.c: Likewise.
1710         * tree-call-cdce.c: Likewise.
1711         * tree-cfg.c: Likewise.
1712         * tree-cfgcleanup.c: Likewise.
1713         * tree-chrec.c: Likewise.
1714         * tree-complex.c: Likewise.
1715         * tree-data-ref.c: Likewise.
1716         * tree-dfa.c: Likewise.
1717         * tree-eh.c: Likewise.
1718         * tree-emutls.c: Likewise.
1719         * tree-if-conv.c: Likewise.
1720         * tree-inline.c: Likewise.
1721         * tree-into-ssa.c: Likewise.
1722         * tree-loop-distribution.c: Likewise.
1723         * tree-mudflap.c: Likewise.
1724         * tree-nested.c: Likewise.
1725         * tree-nrv.c: Likewise.
1726         * tree-object-size.c: Likewise.
1727         * tree-outof-ssa.c: Likewise.
1728         * tree-parloops.c: Likewise.
1729         * tree-phinodes.c: Likewise.
1730         * tree-predcom.c: Likewise.
1731         * tree-pretty-print.c: Likewise.
1732         * tree-profile.c: Likewise.
1733         * tree-scalar-evolution.c: Likewise.
1734         * tree-sra.c: Likewise.
1735         * tree-ssa-address.c: Likewise.
1736         * tree-ssa-alias.c: Likewise.
1737         * tree-ssa-ccp.c: Likewise.
1738         * tree-ssa-coalesce.c: Likewise.
1739         * tree-ssa-copy.c: Likewise.
1740         * tree-ssa-copyrename.c: Likewise.
1741         * tree-ssa-dce.c: Likewise.
1742         * tree-ssa-dom.c: Likewise.
1743         * tree-ssa-dse.c: Likewise.
1744         * tree-ssa-forwprop.c: Likewise.
1745         * tree-ssa-ifcombine.c: Likewise.
1746         * tree-ssa-live.c: Likewise.
1747         * tree-ssa-loop-ch.c: Likewise.
1748         * tree-ssa-loop-im.c: Likewise.
1749         * tree-ssa-loop-ivcanon.c: Likewise.
1750         * tree-ssa-loop-ivopts.c: Likewise.
1751         * tree-ssa-loop-manip.c: Likewise.
1752         * tree-ssa-loop-niter.c: Likewise.
1753         * tree-ssa-loop-prefetch.c: Likewise.
1754         * tree-ssa-loop-unswitch.c: Likewise.
1755         * tree-ssa-loop.c: Likewise.
1756         * tree-ssa-math-opts.c: Likewise.
1757         * tree-ssa-operands.c: Likewise.
1758         * tree-ssa-phiopt.c: Likewise.
1759         * tree-ssa-phiprop.c: Likewise.
1760         * tree-ssa-pre.c: Likewise.
1761         * tree-ssa-propagate.c: Likewise.
1762         * tree-ssa-reassoc.c: Likewise.
1763         * tree-ssa-sccvn.c: Likewise.
1764         * tree-ssa-sink.c: Likewise.
1765         * tree-ssa-strlen.c: Likewise.
1766         * tree-ssa-structalias.c: Likewise.
1767         * tree-ssa-tail-merge.c: Likewise.
1768         * tree-ssa-ter.c: Likewise.
1769         * tree-ssa-threadedge.c: Likewise.
1770         * tree-ssa-threadupdate.c: Likewise.
1771         * tree-ssa-uncprop.c: Likewise.
1772         * tree-ssa-uninit.c: Likewise.
1773         * tree-ssa.c: Likewise.
1774         * tree-ssanames.c: Likewise.
1775         * tree-stdarg.c: Likewise.
1776         * tree-streamer-in.c: Likewise.
1777         * tree-switch-conversion.c: Likewise.
1778         * tree-tailcall.c: Likewise.
1779         * tree-vect-data-refs.c: Likewise.
1780         * tree-vect-generic.c: Likewise.
1781         * tree-vect-loop-manip.c: Likewise.
1782         * tree-vect-loop.c: Likewise.
1783         * tree-vect-patterns.c: Likewise.
1784         * tree-vect-slp.c: Likewise.
1785         * tree-vect-stmts.c: Likewise.
1786         * tree-vectorizer.c: Likewise.
1787         * tree-vrp.c: Likewise.
1788         * tree.c: Likewise.
1789         * tsan.c: Likewise.
1790         * value-prof.c: Likewise.
1791         * var-tracking.c: Likewise.
1792         * varpool.c: Likewise.
1793         * vtable-verify.c: Likewise.
1795 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1797         * config/tilegx/tilegx.c: Include "tree.h".
1799 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
1801         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
1802         the info, not after it.
1803         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
1804         dump_ssaname_info call.
1805         (pp_gimple_stmt_1): Adjust caller.
1806         (dump_phi_nodes): Likewise.  Don't print "# " here.
1808 2013-10-22  Jan Hubicka  <jh@suse.cz>
1810         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New
1811         tuning flag.
1812         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
1813         * i386.c (expand_small_movmem_or_setmem): New function.
1814         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New
1815         function.
1816         (alg_usable_p): Add support for value ranges; cleanup.
1817         (ix86_expand_set_or_movmem): Add support for misaligned moves.
1819 2013-10-22  Sterling Augustine  <saugustine@google.com>
1821         * doc/invoke.texi: Document -ggnu-pubnames.
1822         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
1823         logic.
1824         * dwarf2out.c: Include gdb/gdb-index.h.
1825         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
1826         debug_generate_pub_sections.
1827         (is_java, output_pubtables, output_pubname): New functions.
1828         (include_pubname_in_output): Handle debug_generate_pub_sections at
1829         level 2.
1830         (size_of_pubnames): Use new local space_for_flags based on
1831         debug_generate_pub_sections.
1832         (output_pubnames): Unify pubnames and pubtypes output logic.
1833         Genericize comments.  Call output_pubname.
1834         (dwarf2out_finish): Move logic to output_pubnames and call it.
1836 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
1838         PR target/58779
1839         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
1840         Remove CCCmode handling.
1841         <case LTU>: Return 'c' suffix for CCCmode.
1842         <case GEU>: Return 'nc' suffix for CCCmode.
1843         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
1844         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
1845         (*sub<mode>3_cc_overflow): Ditto.
1846         (*subsi3_zext_cc_overflow): Ditto.
1848 2013-10-22  Steve Ellcey  <sellcey@mips.com>
1850         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
1851         (AND (NOT OP1) (NOT OP2)).
1853 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
1855         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
1856         meaning of merge-high and merge-low masks for little endian; avoid
1857         use of vector-pack masks for little endian for mismatched modes.
1859 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1861         * config/tilepro/tilepro.c: Include "tree.h".
1863 2013-10-22  Andreas Schwab  <schwab@suse.de>
1865         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
1866         with PRE_DEC.
1868 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
1870         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
1871         [Ed Smith-Rowland]): New entries.
1872         ([Stephen M. Webb]): Update.
1874 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
1876         * tree-ssa-ter.h: Remove duplicate copy of file contents.
1878 2013-10-21  Marek Polacek  <polacek@redhat.com>
1880         PR middle-end/58809
1881         * fold-const.c (fold_range_test): Return 0 if the type is not
1882         an integral type.
1884 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
1886         * system.h: Move hwint.h include further down.
1887         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
1888         gcc_checking_asserts.
1889         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
1891 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1893         Fix volatile issues in optimize_bit_field_compare.
1894         * fold-const.c (optimize_bit_field_compare): Bail out if
1895         lvolatilep or rvolatilep.
1897 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1899         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
1900         and get_inner_reference returning different pmode for non-volatile
1901         bit-field members dependent on flag_strict_volatile_bitfields.
1902         * stor-layout.c (layout_decl): Remove special handling of
1903         flag_strict_volatile_bitfields.
1904         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
1905         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
1907 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
1909         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
1910         calculations.
1912 2013-10-21  Jeff Law  <law@redhat.com>
1914         * tree-ssa-threadedge.c (thread_through_normal_block): New
1915         argument VISITED.  Remove VISISTED as a local variable.  When we
1916         have a threadable jump, verify the destination of the jump has not
1917         been visised.
1918         (thread_across_edge): Allocate VISITED bitmap once at function
1919         scope and use it throughout.  Make sure to set appropriate bits in
1920         VISITED for E (start of jump thread path).
1921         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
1922         through a joiner if any edge on the path has a recorded jump thread.
1924 2013-10-21  Ian Lance Taylor  <iant@google.com>
1926         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
1927         don't imply that attributes can solve all problems.
1928         (Directory Options): Fix typo.
1930 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1932         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
1933         extend and extend_add.
1935 2013-10-21  Richard Biener  <rguenther@suse.de>
1937         PR tree-optimization/58794
1938         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
1939         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
1941 2013-10-21  Richard Biener  <rguenther@suse.de>
1943         PR middle-end/58742
1944         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
1945         to (T) X for sign-changing conversions (or no conversion).
1947 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
1949         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
1951 2013-10-20  Jan Hubicka  <jh@suse.cz>
1953         * config/i386/i386-tune.def: Add comment; organize into categories
1955 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
1957         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
1958         argument.  Update function comment.
1959         (expand_set_or_movmem_via_rep): New function combining
1960         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
1961         (expand_movmem_via_rep_mov): Remove.
1962         expand_setmem_via_rep_stos): Remove.
1963         (expand_movmem_epilogue): Update calls correspondingly.
1964         (expand_setmem_epilogue_via_loop): Likewise.
1965         (emit_memset): New.
1966         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
1967         (expand_set_or_movmem_prologue): New function combining
1968         expand_movmem_prologue and expand_setmem_prologue.
1969         (expand_movmem_prologue): Remove.
1970         (expand_setmem_prologue): Remove.
1971         (expand_set_or_movmem_constant_prologue): New function combining
1972         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
1973         (expand_constant_movmem_prologue): Remove.
1974         (expand_constant_setmem_prologue): Remove.
1975         (promote_duplicated_reg): Allow vector-const0 value.
1976         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
1977         and ix86_expand_setmem.
1978         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
1979         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
1981 2013-10-21  Diego Novillo  <dnovillo@google.com>
1983         * asan.c: Include tree.h
1984         * bb-reorder.c: Likewise.
1985         * cfgcleanup.c: Likewise.
1986         * cfgloopmanip.c: Likewise.
1987         * data-streamer-in.c: Likewise.
1988         * data-streamer-out.c: Likewise.
1989         * data-streamer.c: Likewise.
1990         * dwarf2cfi.c: Likewise.
1991         * graphite-blocking.c: Likewise.
1992         * graphite-clast-to-gimple.c: Likewise.
1993         * graphite-dependences.c: Likewise.
1994         * graphite-interchange.c: Likewise.
1995         * graphite-optimize-isl.c: Likewise.
1996         * graphite-poly.c: Likewise.
1997         * graphite-scop-detection.c: Likewise.
1998         * graphite-sese-to-poly.c: Likewise.
1999         * graphite.c: Likewise.
2000         * ipa-devirt.c: Likewise.
2001         * ipa-profile.c: Likewise.
2002         * ipa.c: Likewise.
2003         * ira.c: Likewise.
2004         * loop-init.c: Likewise.
2005         * loop-unroll.c: Likewise.
2006         * lower-subreg.c: Likewise.
2007         * lto/lto-object.c: Likewise.
2008         * recog.c: Likewise.
2009         * reginfo.c: Likewise.
2010         * tree-loop-distribution.c: Likewise.
2011         * tree-parloops.c: Likewise.
2012         * tree-ssa-strlen.c: Likewise.
2013         * tree-streamer.c: Likewise.
2014         * value-prof.c: Likewise.
2015         * target-globals.c: Likewise.
2016         * expr.h: Include tree-core.h instead of tree.h.
2017         * gimple.h: Likewise.
2018         * ipa-prop.h: Likewise.
2019         * ipa-utils.h: Likewise.
2020         * lto-streamer.h: Likewise.
2021         * streamer-hooks.h: Likewise.
2022         * ipa-reference.h: Include cgraph.h instead of tree.h.
2023         * cgraph.h: Include basic-block.h instead of tree.h.
2024         * tree-streamer.h: Do not include tree.h.
2025         * genattrtab.c (write_header): Generate inclusion of
2026         tree.h.
2027         * genautomata.c (main): Likewise.
2028         * genemit.c: Likewise.
2029         * genopinit.c: Likewise.
2030         * genoutput.c (output_prologue): Likewise.
2031         * genpeep.c: Likewise.
2033 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2035         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
2036         little endian.
2037         (vec_unpacku_hi_v8hi): Likewise.
2038         (vec_unpacku_lo_v16qi): Likewise.
2039         (vec_unpacku_lo_v8hi): Likewise.
2041 2013-10-20  Jan Hubicka  <jh@suse.cz>
2043         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
2044         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
2045         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
2047 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
2049         * config/mips/mips.h (ISA_HAS_WSBH): Define.
2050         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
2051         constants.
2052         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
2054 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
2056         PR target/58603
2057         * system.h: Undef m_slot.
2059 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2061         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
2062         all elements for both endian flavors.
2064 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2066         PR target/58792
2067         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
2068         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
2069         and SI_REG for 64bit SYSV ABI targets.
2071 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2073         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
2074         to multi_reg_return.  Clarify that we are skipping USEs of multiple
2075         return registers.  Use bool type where appropriate.
2077 2013-10-18  Jan Hubicka  <jh@suse.cz>
2079         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
2080         for cold functions.
2081         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
2082         (X86_TUNE_PUSH_MEMORY): Likewise.
2083         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
2084         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
2085         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
2086         New.
2087         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
2088         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
2089         Remove.
2090         (ix86_option_override_internal): Update to use tune features instead
2091         of variables.
2093 2013-10-18  Cong Hou  <congh@google.com>
2095         PR tree-optimization/58508
2096         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
2097         statement that contains data refs with zero-step.
2099 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2101         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
2102         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
2103         * gimple-low.c (gimple_check_call_arg,
2104         gimple_check_call_matching_types): Move to cgraph.c.
2105         * gimple-low.h: Remove prototype.
2106         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
2107         Relocate from gimple-low.c.
2108         * cgraph.h: Add prototype.  Don't include basic-block.h.
2109         * gimplify.c: Add gimple-low to include list.
2110         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
2111         * tree-eh.c: Add gimple-low to include list.
2112         * tree-nested.c: Likewise.
2113         * cfgexpand.c: Add tree-ssa-address.h to include list.
2114         * expr.c: Likewise.
2115         * gimple-fold.c: Likewise.
2116         * gimple-ssa-strength-reduction.c: Likewise.
2117         * trans-mem.c: Likewise.
2118         * tree-mudflap.c: Likewise.
2119         * tree-ssa-loop-ivopts.c: Likewise.
2120         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
2121         (degenerate_phi_result): Move to tree-phinodes.c.
2122         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
2123         * tree-ssa-threadedge.c: Likewise.
2124         * tree-vrp.c: Likewise.
2125         * tree-phinodes.c (degenerate_phi_result): Relocate here.
2126         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
2127         * tree-phinodes.h (degenerate_phi_result): Add prototype.
2128         * tree-ssa-copy.c: Include tree-ssa-dom.h.
2129         * tree-ssa-forwprop.c: Likewise.
2130         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
2131         pass_data_cleanup_cfg_post_optimizing,
2132         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
2133         * tree-optimize.c: Delete File.
2134         * graphite.c: Include tree-cfgcleanup.h.
2135         * passes.c: Likewise.
2136         * tree-cfg.c: Likewise.
2137         * tree-profile.c: Likewise.
2138         * tree-ssa-dse.c: Likewise.
2139         * tree-ssa-loop-ivcanon.c: Likewise.
2140         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
2141         * tree-outof-ssa.c: Include sbitmap.h.
2142         * tree-ssa-live.c: Likewise.
2143         * tree-ssa-propagate.c: Likewise.
2144         * tree-ssa-structalias.c: Likewise.
2145         * tree-stdarg.c: Likewise.
2146         * Makefile.in (OBJS): Delete tree-optimize.o.
2147         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
2148         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
2149         * varasm.c: Include basic-block.h.
2150         * cfgloop.h: Include function.h instead of basic-block.h
2151         (bb_loop_depth): Move to cfgloop.c.
2152         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
2154 2013-10-18  Teresa Johnson  <tejohnson@google.com>
2156         * predict.c (probably_never_executed): Compare frequency-based
2157         count to number of training runs.
2158         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
2160 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2162         * config/arm/arm.c (cortexa9_extra_costs): New table.
2163         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
2165 2013-10-18  Jeff Law  <law@redhat.com>
2167         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
2169         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
2170         (register_jump_thread): Add "registered_jump_thread" debug
2171         counter support.
2172         * dbgcnt.def (registered_jump_thread): New debug counter.
2174 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2176         * config/rs6000/rs6000.c: Include cgraph.h.
2178 2013-10-18  Teresa Johnson  <tejohnson@google.com>
2180         * tree-ssa-tail-merge.c (replace_block_by): Update edge
2181         weights during merging.
2183 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2185         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
2186         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
2187         * tree-ssa-operands.h: Remove prototype.
2188         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
2189         * gimple.c (virtual_operand_p): Relocate from gimple.c.
2190         * gimple.h: Add prototype.
2191         * gimple-ssa.h: Include tree-ssa-operands.h.
2192         * tree-dump.c: Add tree-cfg.h to include list.
2193         * tree-ssa-alias.c: Add ipa-reference.h to include list.
2194         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
2195         * config/i386/i386.c: Don't include tree-flow.h.
2196         * config/rs6000/rs6000.c: Likewise.
2198 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2200         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
2202 2013-10-18  Richard Biener  <rguenther@suse.de>
2204         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
2205         or TYPE_UNSIGNED of integral types.
2206         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
2207         NULL_TREE for zero-precision integral types.
2209 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
2211         * config/aarch64/arm_neon.h
2212         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
2214 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
2216         PR target/57756
2217         * opth-gen.awk: Define target_flags_explicit.
2219 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
2221         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
2222         fields to the reg_addr array that describes the valid addressing
2223         mode for any register, general purpose registers, floating point
2224         registers, and Altivec registers.
2225         (FIRST_RELOAD_REG_CLASS): Likewise.
2226         (LAST_RELOAD_REG_CLASS): Likewise.
2227         (struct reload_reg_map_type): Likewise.
2228         (reload_reg_map_type): Likewise.
2229         (RELOAD_REG_VALID): Likewise.
2230         (RELOAD_REG_MULTIPLE): Likewise.
2231         (RELOAD_REG_INDEXED): Likewise.
2232         (RELOAD_REG_OFFSET): Likewise.
2233         (RELOAD_REG_PRE_INCDEC): Likewise.
2234         (RELOAD_REG_PRE_MODIFY): Likewise.
2235         (reg_addr): Likewise.
2236         (mode_supports_pre_incdec_p): New helper functions to say whether
2237         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
2238         (mode_supports_pre_modify_p): Likewise.
2239         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
2240         print the valid address mode bits for each mode.
2241         (rs6000_debug_print_mode): Likewise.
2242         (rs6000_debug_reg_global): Likewise.
2243         (rs6000_setup_reg_addr_masks): New function to set up the address
2244         mask bits for each type.
2245         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
2246         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
2247         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
2248         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
2249         PRE_MODIFY are supported.
2250         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
2251         registers, instead of {src,dest}_av_p.
2252         (rs6000_print_options_internal): Tweak the debug output slightly.
2254 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
2256         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
2257         isa attribute.
2259 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
2261         * tree-flow.h (struct omp_region): Move to omp-low.c.
2262         Remove omp_ prototypes and variables.
2263         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
2264         (copy_var_decl): Relocate prototype from tree-flow.h.
2265         * gimple.c (copy_var_decl): Relocate from omp-low.c.
2266         * tree.h: Move prototype to omp-low.h.
2267         * omp-low.h: New File.  Relocate prototypes here.
2268         * omp-low.c (struct omp_region): Make local here.
2269         (root_omp_region): Make static.
2270         (copy_var_decl) Move to gimple.c.
2271         (new_omp_region): Make static.
2272         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
2273         * tree-cfg.c: Include omp-low.h.
2274         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
2275         * gimplify.c: Include omp-low.h.
2276         * tree-parloops.c: Likewise.
2278 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
2280         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
2281         operands are in memory, prefer to force non-matched operand 1 to
2282         the register.
2284 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
2286         PR target/58673
2287         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
2288         restrict TImode addresses to single indirect registers if both
2289         -mquad-memory and -mvsx-timode are used.
2290         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
2291         we should emit load/store quad.  Remove using %y for quad memory
2292         addresses.
2294         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
2295         constraints to allow load/store quad on machines where TImode is
2296         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
2297         for TImode to load integer constants.
2299 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2301         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
2303 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2305         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
2306         handling of STACK_REG.
2308 2013-10-17  Richard Biener  <rguenther@suse.de>
2310         PR tree-optimization/58143
2311         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
2312         New function.
2313         (rewrite_to_defined_overflow): Likewise.
2314         (move_computations_dom_walker::before_dom): Rewrite stmts
2315         with undefined signed overflow that are not always executed
2316         into unsigned arithmetic.
2318 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
2320         PR target/57756
2321         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
2322         explicit isa flag to be an options variable, instead of using
2323         global_options_set.  Remove define from rs6000.h.
2324         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
2326         * config/rs6000/rs6000.c (rs6000_option_override_internal):
2327         Initialize rs6000_isa_flags_explicit.
2328         (rs6000_function_specific_save): Add gcc_options* parameter, so
2329         that the powerpc builds after the 2013-10-15 changes.
2330         (rs6000_function_specific_restore): Likewise.
2332 2013-10-16  DJ Delorie  <dj@redhat.com>
2334         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
2335         op is a REG before checking REGNO.
2336         (rl78_alloc_physical_registers): Verify pattern is a SET before
2337         checking SET_SRC.
2339 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2341         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
2342         endianness.
2343         (vec_unpacks_lo_v4sf): Likewise.
2344         (vec_unpacks_float_hi_v4si): Likewise.
2345         (vec_unpacks_float_lo_v4si): Likewise.
2346         (vec_unpacku_float_hi_v4si): Likewise.
2347         (vec_unpacku_float_lo_v4si): Likewise.
2349 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2351         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
2352         (vsx_concat_v2sf): Likewise.
2354 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
2356         * config/aarch64/aarch64.md
2357         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
2359 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
2361         PR tree-optimization/58697
2362         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
2363         estimated_loop_iterations_int.
2364         (max_stmt_executions_int): Call get_max_loop_iterations_int.
2365         (get_max_loop_iterations_int): New.  HWINT version of
2366         get_max_loop_iterations.
2367         * cfgloop.h: Add prototypes.
2368         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
2369         * loop-unroll.c (decide_peel_once_rolling): Call
2370         get_estimated_loop_iterations_int.
2371         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
2372         * tree-ssa-loop-niter.h: Tweak prototypes.
2374 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
2376         * gengtype-parse.c (struct_field_seq): Ignore access-control
2377         keywords ("public:" etc).
2379 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2381         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
2382         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
2384 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
2386         * config/arm/arm.opt (mlra): New option.
2387         * config/arm/arm.c (arm_lra_p): New function.
2388         (TARGET_LRA_P): Define.
2390 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
2392         * tree-core.h (tree_code_name): Remove.
2393         * tree.h (get_tree_code_name): New prototype.
2394         * tree.c (tree_code_name): Make static.
2395         (get_tree_code_name): New function.
2396         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
2397         tree_class_check_failed, tree_range_check_failed,
2398         tree_not_class_check_failed, omp_clause_check_failed,
2399         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
2400         wrapper get_tree_code_name instead of calling tree_code_name directly.
2401         * tree-vrp.c (dump_asserts_for): Likewise.
2402         * tree-dump.c (dequeue_and_dump): Likewise.
2403         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
2404         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
2405         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
2406         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
2407         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
2408         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
2409         dump_gimple_omp_for): Likewise.
2410         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
2411         * tree-ssa-pre.c (print_pre_expr): Likewise.
2412         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2413         * print-tree.c (print_node_brief, print_node): Likewise.
2414         * gimple.c (gimple_check_failed): Likewise.
2415         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
2416         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
2417         * config/mep/mep.c (mep_validate_vliw): Likewise.
2418         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
2419         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
2421 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
2423         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
2424         for AMD bdver3.
2426 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
2428         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
2429         (MULTILIB_MATCHES): Add multilib for -march=v8.
2431 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
2433         PR target/57756
2434         * optc-save-gen.awk: Add extra parameter to the save and restore
2435         target calls.
2436         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
2437         * tree.c (build_optimization_node): New parameter.  Add extra parameter
2438         to call to cl_optimization_save.
2439         (build_target_option_node): New parameter. Add extra parameter
2440         to call to cl_target_option_save.
2441         * tree.h (build_optimization_node): New parameter.
2442         (build_target_option_node): New parameter.
2443         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
2444         build_optimization_node and build_target_option_node.
2445         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
2446         (handle_pragma_push_options): Ditto.
2447         * toplev.c (process_options): Ditto.
2448         * opts.c (init_options_struct): Check for opts_set non-null.
2449         * target.def (target_option.save): New parameter.
2450         (target_option.restore): New parameter.
2451         * tm.texi: Generate.
2452         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2453         (ix86_pragma_target_parse): Ditto.
2454         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
2455         parameters.
2456         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
2457         to  build_optimization_node and build_target_option_node.
2458         (rs6000_valid_attribute_p): Ditto.
2459         (rs6000_pragma_target_parse): Ditto.
2460         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
2461         data.
2462         * config/i386/i386.h:
2463         TARGET_64BIT_P: New Macro
2464         TARGET_MMX_P: New Macro.
2465         TARGET_3DNOW_P: New Macro.
2466         TARGET_3DNOW_A_P: New Macro.
2467         TARGET_SSE_P: New Macro.
2468         TARGET_SSE2_P: New Macro.
2469         TARGET_SSE3_P: New Macro.
2470         TARGET_SSSE3_P: New Macro.
2471         TARGET_SSE4_1_P: New Macro.
2472         TARGET_SSE4_2_P: New Macro.
2473         TARGET_AVX_P: New Macro.
2474         TARGET_AVX2_P: New Macro.
2475         TARGET_AVX512F_P: New Macro.
2476         TARGET_AVX512PF_P: New Macro.
2477         TARGET_AVX512ER_P: New Macro.
2478         TARGET_AVX512CD_P: New Macro.
2479         TARGET_FMA_P: New Macro.
2480         TARGET_SSE4A_P: New Macro.
2481         TARGET_FMA4_P: New Macro.
2482         TARGET_XOP_P: New Macro.
2483         TARGET_LWP_P: New Macro.
2484         TARGET_ABM_P: New Macro.
2485         TARGET_BMI_P: New Macro.
2486         TARGET_BMI2_P: New Macro.
2487         TARGET_LZCNT_P: New Macro.
2488         TARGET_TBM_P: New Macro.
2489         TARGET_POPCNT_P: New Macro.
2490         TARGET_SAHF_P: New Macro.
2491         TARGET_MOVBE_P: New Macro.
2492         TARGET_CRC32_P: New Macro.
2493         TARGET_AES_P: New Macro.
2494         TARGET_PCLMUL_P: New Macro.
2495         TARGET_CMPXCHG16B_P: New Macro.
2496         TARGET_FSGSBASE_P: New Macro.
2497         TARGET_RDRND_P: New Macro.
2498         TARGET_F16C_P: New Macro.
2499         TARGET_RTM_P: New Macro.
2500         TARGET_HLE_P: New Macro.
2501         TARGET_RDSEED_P: New Macro.
2502         TARGET_PRFCHW_P: New Macro.
2503         TARGET_ADX_P: New Macro.
2504         TARGET_FXSR_P: New Macro.
2505         TARGET_XSAVE_P: New Macro.
2506         TARGET_XSAVEOPT_P: New Macro.
2507         TARGET_LP64_P: New Macro.
2508         TARGET_X32_P: New Macro.
2509         TARGET_FPMATH_DEFAULT_P: New Macro.
2510         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
2511         * config/i386/i386.c (ix86_option_override_internal): New parameters.
2512         opts and opts_set.
2513         Change ix86_tune_string to access opts->x_ix86_tune_string.
2514         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
2515         Change ix86_arch_string to access opts->x_ix86_arch_string.
2516         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
2517         Change ix86_pmode to access opts->x_ix86_pmode.
2518         Change ix86_abi to access opts->x_ix86_abi.
2519         Change ix86_cmodel to access opts->x_ix86_cmodel.
2520         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
2521         Change ix86_isa_flags_explicit to access
2522         opts->x_ix86_isa_flags_explicit.
2523         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
2524         Change ix86_regparm to access opts->x_ix86_regparm.
2525         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
2526         Change ix86_preferred_stack_boundary_arg to access
2527         opts->x_ix86_preferred_stack_boundary_arg.
2528         Change ix86_force_align_arg_pointer to access
2529         opts->x_ix86_force_align_arg_pointer.
2530         Change ix86_incoming_stack_boundar_arg to access
2531         opts->x_ix86_incoming_stack_boundar_arg.
2532         Change ix86_fpmath to access opts->x_ix86_fpmath.
2533         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
2534         Change ix86_recip_name to access opts->x_ix86_recip_name.
2535         Change ix86_stack_protector_guard to access
2536         opts->x_ix86_stack_protector_guard.
2537         Change ix86_tune_memcpy_strategy to access
2538         opts->x_ix86_tune_memcpy_strategy.
2539         Change ix86_tune_memset_strategy to access
2540         opts->x_ix86_tune_memset_strategy.
2541         Change global_options to access opts.
2542         Change global_options_set to access opts_set.
2543         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
2544         Change TARGET_MMX to TARGET_MMX_P (opts->...).
2545         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
2546         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
2547         Change TARGET_SSE to TARGET_SSE_P (opts->...).
2548         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
2549         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
2550         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
2551         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
2552         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
2553         Change TARGET_AVX to TARGET_AVX_P (opts->...).
2554         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
2555         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
2556         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
2557         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
2558         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
2559         Change TARGET_FMA to TARGET_FMA_P (opts->...).
2560         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
2561         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
2562         Change TARGET_XOP to TARGET_XOP_P (opts->...).
2563         Change TARGET_LWP to TARGET_LWP_P (opts->...).
2564         Change TARGET_ABM to TARGET_ABM_P (opts->...).
2565         Change TARGET_BMI to TARGET_BMI_P (opts->...).
2566         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
2567         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
2568         Change TARGET_TBM to TARGET_TBM_P (opts->...).
2569         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
2570         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
2571         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
2572         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
2573         Change TARGET_AES to TARGET_AES_P (opts->...).
2574         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
2575         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
2576         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
2577         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
2578         Change TARGET_F16C to TARGET_F16C_P (opts->...).
2579         Change TARGET_RTM to TARGET_RTM_P (opts->...).
2580         Change TARGET_HLE to TARGET_HLE_P (opts->...).
2581         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
2582         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
2583         Change TARGET_ADX to TARGET_ADX_P (opts->...).
2584         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
2585         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
2586         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
2587         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
2588         Change TARGET_X32 to TARGET_X32_P (opts->...).
2589         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
2590         Change TARGET_FLOAT_RETURNS_IN_80387 to
2591         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
2592         (ix86_function_specific_save): New parameter. Use opts-> fields
2593         to replace global fields.
2594         (ix86_function_specific_restore): Ditto.
2595         (ix86_valid_target_attribute_inner_p): New parameters.
2596         Fix recursive call.
2597         Fix call to ix86_handle_option and set_option.
2598         (ix86_valid_target_attribute_tree): New parameters.
2599         Change global_options to access opts.
2600         Change global_options_set to access opts_set.
2601         Fix call to ix86_valid_target_attribute_inner_p.
2602         Change ix86_tune_string to access opts->x_ix86_tune_string.
2603         Change ix86_arch_string to access opts->x_ix86_arch_string.
2604         Change ix86_fpmath to access opts->x_ix86_fpmath
2605         Fix call to ix86_option_override_internal.
2606         Fix call to ix86_add_new_builtins.
2607         Fix calls to build_optimization_node and build_target_option_node.
2608         (ix86_valid_target_attribute_p): Remove access to global_options.
2609         Use new gcc_options structure func_options.
2610         Fix call to ix86_valid_target_attribute_tree.
2611         Fix call to  build_optimization_node.
2612         (get_builtin_code_for_version): Fix call to
2613         ix86_valid_target_attribute_tree.
2615 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
2617         * Makefile.in (PICFLAG): New.
2618         (enable_host_shared): New.
2619         (INTERNAL_CFLAGS): Use PICFLAG.
2620         (LIBIBERTY): Use pic build of libiberty.a if configured with
2621         --enable-host-shared.
2622         * configure.ac: Add --enable-host-shared, setting up new
2623         PICFLAG variable.
2624         * configure: Regenerate.
2625         * doc/install.texi (--enable-shared): Add note contrasting it with ...
2626         (--enable-host-shared): New option.
2628 2013-10-15  Richard Biener  <rguenther@suse.de>
2630         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
2631         for built-in functions.
2633 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2635         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
2636         (optimize_range_tests_1): New function,
2637         extracted from optimize_range_tests.
2638         (optimize_range_tests_xor): Similarly.
2639         (optimize_range_tests_diff): New function.
2640         (optimize_range_tests): Use optimize_range_tests_1.
2642 2013-10-15  Cong Hou  <congh@google.com>
2644         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
2645         requirement of the reduction pattern so that one operand of the
2646         reduction operation can come from outside of the loop.
2648 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2650         * config/arm/neon-schedgen.ml: Remove.
2651         * config/arm/cortex-a9-neon.md: Remove comment regarding
2652         neon-schedgen.ml.
2654 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2656         * config/arm/types: Remove old neon types.
2658 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2660         * config/arm/cortex-a7.md
2661         (cortex_a7_neon_type): New.
2662         (cortex_a7_neon_mul): Update for new types.
2663         (cortex_a7_neon_mla): Likewise.
2664         (cortex_a7_neon): Likewise.
2666 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2668         * config/arm/cortex-a15-neon.md
2669         (cortex_a15_neon_type): New,
2671         (cortex_a15_neon_int_1): Remove.
2672         (cortex_a15_neon_int_2): Likewise.
2673         (cortex_a15_neon_int_3): Likewise.
2674         (cortex_a15_neon_int_4): Likewise.
2675         (cortex_a15_neon_int_5): Likewise.
2676         (cortex_a15_neon_vqneg_vqabs): Likewise.
2677         (cortex_a15_neon_vmov): Likewise.
2678         (cortex_a15_neon_vaba): Likewise.
2679         (cortex_a15_neon_vaba_qqq): Likewise.
2680         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2681         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2682         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
2683         Likewise.
2684         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2685         (cortex_a15_neon_mla_qqq_8_16): Likewise.
2686         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
2687         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2688         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2689         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2690         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2691         (cortex_a15_neon_shift_1): Likewise.
2692         (cortex_a15_neon_shift_2): Likewise.
2693         (cortex_a15_neon_shift_3): Likewise.
2694         (cortex_a15_neon_vshl_ddd): Likewise.
2695         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2696         (cortex_a15_neon_vsra_vrsra): Likewise.
2697         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2698         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2699         (cortex_a15_neon_bp_3cycle): Likewise.
2700         (cortex_a15_neon_ldm_2): Likewise.
2701         (cortex_a15_neon_stm_2): Likewise.
2702         (cortex_a15_neon_mcr): Likewise.
2703         (cortex_a15_neon_mrc): Likewise.
2704         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2705         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2706         (cortex_a15_neon_fp_vmul_ddd): Likewise.
2707         (cortex_a15_neon_fp_vmul_qqd): Likewise.
2708         (cortex_a15_neon_fp_vmla_ddd): Likewise.
2709         (cortex_a15_neon_fp_vmla_qqq): Likewise.
2710         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2711         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2712         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2713         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2714         (cortex_a15_neon_bp_simple): Likewise.
2715         (cortex_a15_neon_bp_2cycle): Likewise.
2716         (cortex_a15_neon_bp_3cycle): Likewise.
2717         (cortex_a15_neon_vld1_1_2_regs): Likewise.
2718         (cortex_a15_neon_vld1_3_4_regs): Likewise.
2719         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2720         (cortex_a15_neon_vld2_4_regs): Likewise.
2721         (cortex_a15_neon_vld3_vld4): Likewise.
2722         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2723         (cortex_a15_neon_vst1_3_4_regs): Likewise.
2724         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
2725         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
2726         (cortex_a15_neon_vst3_vst4): Rename to...
2727         (cortex_a15_neon_vst4): This, update for new attributes.
2728         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
2729         (cortex_a15_neon_vld3_vld4_lane): Likewise.
2730         (cortex_a15_neon_vst1_vst2_lane): Likewise.
2731         (cortex_a15_neon_vst3_vst4_lane): Likewise.
2732         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
2733         (cortex_a15_neon_ldm_2): Likewise.
2734         (cortex_a15_neon_stm_2): Likewise.
2735         (cortex_a15_neon_mcr): Likewise.
2736         (cortex_a15_neon_mcr_2_mcrr): Likewise.
2737         (cortex_a15_neon_mrc): Likewise.
2738         (cortex_a15_neon_mrrc): Likewise.
2740         (cortex_a15_neon_abd): New.
2741         (cortex_a15_neon_abd_q): Likewise.
2742         (cortex_a15_neon_aba): Likewise.
2743         (cortex_a15_neon_aba_q): Likewise.
2744         (cortex_a15_neon_acc): Likewise.
2745         (cortex_a15_neon_acc_q): Likewise.
2746         (cortex_a15_neon_arith_basic): Likewise.
2747         (cortex_a15_neon_arith_complex): Likewise.
2748         (cortex_a15_neon_multiply): Likewise.
2749         (cortex_a15_neon_multiply_q): Likewise.
2750         (cortex_a15_neon_mla): Likewise.
2751         (cortex_a15_neon_mla_q): Likewise.
2752         (cortex_a15_neon_sat_mla_long): Likewise.
2753         (cortex_a15_neon_shift_acc): Likewise.
2754         (cortex_a15_neon_shift_imm_basic): Likewise.
2755         (cortex_a15_neon_shift_imm_complex): Likewise.
2756         (cortex_a15_neon_shift_reg_basic): Likewise.
2757         (cortex_a15_neon_shift_reg_basic_q): Likewise.
2758         (cortex_a15_neon_shift_reg_complex): Likewise.
2759         (cortex_a15_neon_shift_reg_complex_q): Likewise.
2760         (cortex_a15_neon_fp_negabs): Likewise
2761         (cortex_a15_neon_fp_arith): Likewise
2762         (cortex_a15_neon_fp_arith_q): Likewise
2763         (cortex_a15_neon_fp_cvt_int): Likewise
2764         (cortex_a15_neon_fp_cvt_int_q): Likewise
2765         (cortex_a15_neon_fp_cvt_16): Likewise
2766         (cortex_a15_neon_fp_mul): Likewise
2767         (cortex_a15_neon_fp_mul_q): Likewise
2768         (cortex_a15_neon_fp_mla): Likewise
2769         (cortex_a15_neon_fp_mla_q): Likewise
2770         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
2771         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
2772         (cortex_a15_neon_bitops): Likewise.
2773         (cortex_a15_neon_bitops_q): Likewise.
2774         (cortex_a15_neon_from_gp): Likewise.
2775         (cortex_a15_neon_from_gp_q): Likewise.
2776         (cortex_a15_neon_tbl3_tbl4): Likewise.
2777         (cortex_a15_neon_zip_q): Likewise.
2778         (cortex_a15_neon_to_gp): Likewise.
2779         (cortex_a15_neon_load_a): Likewise.
2780         (cortex_a15_neon_load_b): Likewise.
2781         (cortex_a15_neon_load_c): Likewise.
2782         (cortex_a15_neon_load_d): Likewise.
2783         (cortex_a15_neon_load_e): Likewise.
2784         (cortex_a15_neon_load_f): Likewise.
2785         (cortex_a15_neon_store_a): Likewise.
2786         (cortex_a15_neon_store_b): Likewise.
2787         (cortex_a15_neon_store_c): Likewise.
2788         (cortex_a15_neon_store_d): Likewise.
2789         (cortex_a15_neon_store_e): Likewise.
2790         (cortex_a15_neon_store_f): Likewise.
2791         (cortex_a15_neon_store_g): Likewise.
2792         (cortex_a15_neon_store_h): Likewise.
2793         (cortex_a15_vfp_to_from_gp): Likewise.
2795 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2797         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
2799         (cortex_a9_neon_vshl_ddd): Remove.
2800         (cortex_a9_neon_vst3_vst4): Likewise.
2801         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2803         (cortex_a9_neon_bit_ops_q): New.
2805         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
2806         (cortex_a9_neon_int_2): Likewise.
2807         (cortex_a9_neon_int_3): Likewise.
2808         (cortex_a9_neon_int_4): Likewise.
2809         (cortex_a9_neon_int_5): Likewise.
2810         (cortex_a9_neon_vqneg_vqabs): Likewise.
2811         (cortex_a9_neon_vmov): Likewise.
2812         (cortex_a9_neon_vaba): Likewise.
2813         (cortex_a9_neon_vaba_qqq): Likewise.
2814         (cortex_a9_neon_shift_1): Likewise.
2815         (cortex_a9_neon_shift_2): Likewise.
2816         (cortex_a9_neon_shift_3): Likewise.
2817         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2818         (cortex_a9_neon_vsra_vrsra): Likewise.
2819         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2820         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2821         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
2822         Likewise.
2823         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2824         (cortex_a9_neon_mla_qqq_8_16): Likewise.
2825         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
2826         Likewise.
2827         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2828         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2829         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
2830         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2831         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
2832         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
2833         (cortex_a9_neon_fp_vsum): Likewise.
2834         (cortex_a9_neon_fp_vmul_ddd): Likewise.
2835         (cortex_a9_neon_fp_vmul_qqd): Likewise.
2836         (cortex_a9_neon_fp_vmla_ddd): Likewise.
2837         (cortex_a9_neon_fp_vmla_qqq): Likewise.
2838         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
2839         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
2840         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2841         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2842         (cortex_a9_neon_bp_simple): Likewise.
2843         (cortex_a9_neon_bp_2cycle): Likewise.
2844         (cortex_a9_neon_bp_3cycle): Likewise.
2845         (cortex_a9_neon_ldr): Likewise.
2846         (cortex_a9_neon_str): Likewise.
2847         (cortex_a9_neon_vld1_1_2_regs): Likewise.
2848         (cortex_a9_neon_vld1_3_4_regs): Likewise.
2849         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2850         (cortex_a9_neon_vld2_4_regs): Likewise.
2851         (cortex_a9_neon_vld3_vld4): Likewise.
2852         (cortex_a9_neon_vld1_vld2_lane): Likewise.
2853         (cortex_a9_neon_vld3_vld4_lane): Likewise.
2854         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2855         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2856         (cortex_a9_neon_vst1_3_4_regs): Likewise.
2857         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
2858         (cortex_a9_neon_vst1_vst2_lane): Likewise.
2859         (cortex_a9_neon_vst3_vst4_lane): Likewise.
2860         (cortex_a9_neon_mcr): Likewise.
2861         (cortex_a9_neon_mcr_2_mcrr): Likewise.
2862         (cortex_a9_neon_mrc): Likewise.
2863         (cortex_a9_neon_mrrc): Likewise.
2865 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2867         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
2869         (cortex_a8_neon_vshl_ddd): Remove.
2870         (cortex_a8_neon_vst3_vst4): Likewise.
2871         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
2873         (cortex_a8_neon_bit_ops_q): New.
2875         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
2876         (cortex_a8_neon_int_2): Likewise..
2877         (cortex_a8_neon_int_3): Likewise.
2878         (cortex_a8_neon_int_5): Likewise.
2879         (cortex_a8_neon_vqneg_vqabs): Likewise.
2880         (cortex_a8_neon_int_4): Likewise.
2881         (cortex_a8_neon_vaba): Likewise.
2882         (cortex_a8_neon_vaba_qqq): Likewise.
2883         (cortex_a8_neon_shift_1): Likewise.
2884         (cortex_a8_neon_shift_2): Likewise.
2885         (cortex_a8_neon_shift_3): Likewise.
2886         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2887         (cortex_a8_neon_vsra_vrsra): Likewise.
2888         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2889         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2890         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
2891         Likewise.
2892         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2893         (cortex_a8_neon_mla_qqq_8_16): Likewise.
2894         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
2895         Likewise.
2896         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2897         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2898         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
2899         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2900         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
2901         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
2902         (cortex_a8_neon_fp_vsum): Likewise.
2903         (cortex_a8_neon_fp_vmul_ddd): Likewise.
2904         (cortex_a8_neon_fp_vmul_qqd): Likewise.
2905         (cortex_a8_neon_fp_vmla_ddd): Likewise.
2906         (cortex_a8_neon_fp_vmla_qqq): Likewise.
2907         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
2908         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
2909         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2910         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2911         (cortex_a8_neon_bp_simple): Likewise.
2912         (cortex_a8_neon_bp_2cycle): Likewise.
2913         (cortex_a8_neon_bp_3cycle): Likewise.
2914         (cortex_a8_neon_ldr): Likewise.
2915         (cortex_a8_neon_str): Likewise.
2916         (cortex_a8_neon_vld1_1_2_regs): Likewise.
2917         (cortex_a8_neon_vld1_3_4_regs): Likewise.
2918         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2919         (cortex_a8_neon_vld2_4_regs): Likewise.
2920         (cortex_a8_neon_vld3_vld4): Likewise.
2921         (cortex_a8_neon_vld1_vld2_lane): Likewise.
2922         (cortex_a8_neon_vld3_vld4_lane): Likewise.
2923         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2924         (cortex_a8_neon_vst1_3_4_regs): Likewise.
2925         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
2926         (cortex_a8_neon_vst1_vst2_lane): Likewise.
2927         (cortex_a8_neon_vst3_vst4_lane): Likewise.
2928         (cortex_a8_neon_mcr): Likewise.
2929         (cortex_a8_neon_mcr_2_mcrr): Likewise.
2930         (cortex_a8_neon_mrc): Likewise.
2931         (cortex_a8_neon_mrrc): Likewise.
2933 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2935         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
2936         (fp): New.
2937         * config/aarch64/aarch64-simd.md (neon_type): Remove.
2938         (aarch64_simd_dup<mode>): Add "type" attribute.
2939         (aarch64_dup_lane<mode>): Likewise.
2940         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
2941         (*aarch64_simd_mov<mode>): Likewise.
2942         (aarch64_simd_mov_from_<mode>low): Likewise.
2943         (aarch64_simd_mov_from_<mode>high): Likewise.
2944         (orn<mode>3): Likewise.
2945         (bic<mode>3): Likewise.
2946         (add<mode>3): Likewise.
2947         (sub<mode>3): Likewise.
2948         (mul<mode>3): Likewise.
2949         (*aarch64_mul3_elt<mode>): Likewise.
2950         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2951         (*aarch64_mul3_elt_to_128df): Likewise.
2952         (*aarch64_mul3_elt_to_64v2df): Likewise.
2953         (neg<mode>2): Likewise.
2954         (abs<mode>2): Likewise.
2955         (abd<mode>_3): Likewise.
2956         (aba<mode>_3): Likewise.
2957         (fabd<mode>_3): Likewise.
2958         (*fabd_scalar<mode>3): Likewise.
2959         (and<mode>3): Likewise.
2960         (ior<mode>3): Likewise.
2961         (xor<mode>3): Likewise.
2962         (one_cmpl<mode>2): Likewise.
2963         (aarch64_simd_vec_set<mode>): Likewise.
2964         (aarch64_simd_lshr<mode>): Likewise.
2965         (aarch64_simd_ashr<mode>): Likewise.
2966         (aarch64_simd_imm_shl<mode>): Likewise.
2967         (aarch64_simd_reg_sshl<mode): Likewise.
2968         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
2969         (aarch64_simd_reg_shl<mode>_signed): Likewise.
2970         (aarch64_simd_vec_setv2di): Likewise.
2971         (aarch64_simd_vec_set<mode>): Likewise.
2972         (aarch64_mla<mode>): Likewise.
2973         (*aarch64_mla_elt<mode>): Likewise.
2974         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2975         (aarch64_mls<mode>): Likewise.
2976         (*aarch64_mls_elt<mode>): Likewise.
2977         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2978         (<su><maxmin><mode>3): Likewise.
2979         (move_lo_quad_<mode>): Likewise.
2980         (aarch64_simd_move_hi_quad_<mode>): Likewise.
2981         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
2982         (vec_pack_trunc_<mode>): Likewise.
2983         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
2984         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
2985         (*aarch64_<su>mlal_lo<mode>): Likewise.
2986         (*aarch64_<su>mlal_hi<mode>): Likewise.
2987         (*aarch64_<su>mlsl_lo<mode>): Likewise.
2988         (*aarch64_<su>mlsl_hi<mode>): Likewise.
2989         (*aarch64_<su>mlal<mode>): Likewise.
2990         (*aarch64_<su>mlsl<mode>): Likewise.
2991         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
2992         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
2993         (add<mode>3): Likewise.
2994         (sub<mode>3): Likewise.
2995         (mul<mode>3): Likewise.
2996         (div<mode>3): Likewise.
2997         (neg<mode>2): Likewise.
2998         (abs<mode>2): Likewise.
2999         (fma<mode>4): Likewise.
3000         (*aarch64_fma4_elt<mode>): Likewise.
3001         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3002         (*aarch64_fma4_elt_to_128df): Likewise.
3003         (*aarch64_fma4_elt_to_64v2df): Likewise.
3004         (fnma<mode>4): Likewise.
3005         (*aarch64_fnma4_elt<mode>): Likewise.
3006         (*aarch64_fnma4_elt_<vswap_width_name><mode>
3007         (*aarch64_fnma4_elt_to_128df): Likewise.
3008         (*aarch64_fnma4_elt_to_64v2df): Likewise.
3009         (<frint_pattern><mode>2): Likewise.
3010         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
3011         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
3012         (vec_unpacks_lo_v4sf): Likewise.
3013         (aarch64_float_extend_lo_v2df): Likewise.
3014         (vec_unpacks_hi_v4sf): Likewise.
3015         (aarch64_float_truncate_lo_v2sf): Likewise.
3016         (aarch64_float_truncate_hi_v4sf): Likewise.
3017         (aarch64_vmls<mode>): Likewise.
3018         (<su><maxmin><mode>3): Likewise.
3019         (<maxmin_uns><mode>3): Likewise.
3020         (reduc_<sur>plus_<mode>): Likewise.
3021         (reduc_<sur>plus_v2di): Likewise.
3022         (reduc_<sur>plus_v2si): Likewise.
3023         (reduc_<sur>plus_<mode>): Likewise.
3024         (aarch64_addpv4sf): Likewise.
3025         (clz<mode>2): Likewise.
3026         (reduc_<maxmin_uns>_<mode>): Likewise.
3027         (reduc_<maxmin_uns>_v2di): Likewise.
3028         (reduc_<maxmin_uns>_v2si): Likewise.
3029         (reduc_<maxmin_uns>_<mode>): Likewise.
3030         (reduc_<maxmin_uns>_v4sf): Likewise.
3031         (aarch64_simd_bsl<mode>_internal): Likewise.
3032         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3033         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3034         (aarch64_get_lane<mode>): Likewise.
3035         (*aarch64_combinez<mode>): Likewise.
3036         (aarch64_combine<mode>): Likewise.
3037         (aarch64_simd_combine<mode>): Likewise.
3038         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
3039         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
3040         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
3041         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
3042         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
3043         (aarch64_<sur>h<addsub><mode>): Likewise.
3044         (aarch64_<sur><addsub>hn<mode>): Likewise.
3045         (aarch64_<sur><addsub>hn2<mode>): Likewise.
3046         (aarch64_pmul<mode>): Likewise.
3047         (aarch64_<su_optab><optab><mode>): Likewise.
3048         (aarch64_<sur>qadd<mode>): Likewise.
3049         (aarch64_sqmovun<mode>): Likewise.
3050         (aarch64_<sur>qmovn<mode>): Likewise.
3051         (aarch64_s<optab><mode>): Likewise.
3052         (aarch64_sq<r>dmulh<mode>): Likewise.
3053         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3054         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3055         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3056         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
3057         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3058         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3059         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
3060         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
3061         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3062         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
3063         (aarch64_sqdmull<mode>): Likewise.
3064         (aarch64_sqdmull_lane<mode>_internal): Likewise.
3065         (aarch64_sqdmull_n<mode>): Likewise.
3066         (aarch64_sqdmull2<mode>_internal): Likewise.
3067         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3068         (aarch64_sqdmull2_n<mode>_internal): Likewise.
3069         (aarch64_<sur>shl<mode>): Likewise.
3070         (aarch64_<sur>q<r>shl<mode>
3071         (aarch64_<sur>shll_n<mode>): Likewise.
3072         (aarch64_<sur>shll2_n<mode>): Likewise.
3073         (aarch64_<sur>shr_n<mode>): Likewise.
3074         (aarch64_<sur>sra_n<mode>): Likewise.
3075         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
3076         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
3077         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
3078         (aarch64_cm<optab><mode>): Likewise.
3079         (aarch64_cm<optab>di): Likewise.
3080         (aarch64_cm<optab><mode>): Likewise.
3081         (aarch64_cm<optab>di): Likewise.
3082         (aarch64_cmtst<mode>): Likewise.
3083         (aarch64_cmtstdi): Likewise.
3084         (aarch64_cm<optab><mode>): Likewise.
3085         (*aarch64_fac<optab><mode>): Likewise.
3086         (aarch64_addp<mode>): Likewise.
3087         (aarch64_addpdi): Likewise.
3088         (sqrt<mode>2): Likewise.
3089         (vec_load_lanesoi<mode>): Likewise.
3090         (vec_store_lanesoi<mode>): Likewise.
3091         (vec_load_lanesci<mode>): Likewise.
3092         (vec_store_lanesci<mode>): Likewise.
3093         (vec_load_lanesxi<mode>): Likewise.
3094         (vec_store_lanesxi<mode>): Likewise.
3095         (*aarch64_mov<mode>): Likewise.
3096         (aarch64_ld2<mode>_dreg): Likewise.
3097         (aarch64_ld2<mode>_dreg): Likewise.
3098         (aarch64_ld3<mode>_dreg): Likewise.
3099         (aarch64_ld3<mode>_dreg): Likewise.
3100         (aarch64_ld4<mode>_dreg): Likewise.
3101         (aarch64_ld4<mode>_dreg): Likewise.
3102         (aarch64_tbl1<mode>): Likewise.
3103         (aarch64_tbl2v16qi): Likewise.
3104         (aarch64_combinev16qi): Likewise.
3105         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
3106         (aarch64_st2<mode>_dreg): Likewise.
3107         (aarch64_st2<mode>_dreg): Likewise.
3108         (aarch64_st3<mode>_dreg): Likewise.
3109         (aarch64_st3<mode>_dreg): Likewise.
3110         (aarch64_st4<mode>_dreg): Likewise.
3111         (aarch64_st4<mode>_dreg): Likewise.
3112         (*aarch64_simd_ld1r<mode>): Likewise.
3113         (aarch64_frecpe<mode>): Likewise.
3114         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
3115         (aarch64_frecps<mode>): Likewise.
3117 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3119         * config/arm/iterators.md (V_elem_ch): New.
3120         (q): Likewise.
3121         (VQH_type): Likewise.
3122         * config/arm/arm.md (is_neon_type): New.
3123         (conds): Use is_neon_type.
3124         (anddi3_insn): Update type attribute.
3125         (xordi3_insn): Likewise.
3126         (one_cmpldi2): Likewise.
3127         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
3128         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
3129         (*movmisalign<mode>_neon_store): Likewise.
3130         (*movmisalign<mode>_neon_load): Likewise.
3131         (vec_set<mode>_internal): Likewise.
3132         (vec_set<mode>_internal): Likewise.
3133         (vec_setv2di_internal): Likewise.
3134         (vec_extract<mode>): Likewise.
3135         (vec_extract<mode>): Likewise.
3136         (vec_extractv2di): Likewise.
3137         (*add<mode>3_neon): Likewise.
3138         (adddi3_neon): Likewise.
3139         (*sub<mode>3_neon): Likewise.
3140         (subdi3_neon): Likewise.
3141         (fma<VCVTF:mode>4): Likewise.
3142         (fma<VCVTF:mode>4_intrinsic): Likewise.
3143         (*fmsub<VCVTF:mode>4): Likewise.
3144         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
3145         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
3146         (ior<mode>3): Likewise.
3147         (and<mode>3): Likewise.
3148         (orn<mode>3_neon): Likewise.
3149         (orndi3_neon): Likewise.
3150         (bic<mode>3_neon): Likewise.
3151         (bicdi3_neon): Likewise.
3152         (xor<mode>3): Likewise.
3153         (one_cmpl<mode>2): Likewise.
3154         (abs<mode>2): Likewise.
3155         (neg<mode>2): Likewise.
3156         (negdi2_neon): Likewise.
3157         (*umin<mode>3_neon): Likewise.
3158         (*umax<mode>3_neon): Likewise.
3159         (*smin<mode>3_neon): Likewise.
3160         (*smax<mode>3_neon): Likewise.
3161         (vashl<mode>3): Likewise.
3162         (vashr<mode>3_imm): Likewise.
3163         (vlshr<mode>3_imm): Likewise.
3164         (ashl<mode>3_signed): Likewise.
3165         (ashl<mode>3_unsigned): Likewise.
3166         (neon_load_count): Likewise.
3167         (ashldi3_neon_noclobber): Likewise.
3168         (ashldi3_neon): Likewise.
3169         (signed_shift_di3_neon): Likewise.
3170         (unsigned_shift_di3_neon): Likewise.
3171         (ashrdi3_neon_imm_noclobber): Likewise.
3172         (lshrdi3_neon_imm_noclobber): Likewise.
3173         (<shift>di3_neon): Likewise.
3174         (widen_ssum<mode>3): Likewise.
3175         (widen_usum<mode>3): Likewise.
3176         (quad_halves_<code>v4si): Likewise.
3177         (quad_halves_<code>v4sf): Likewise.
3178         (quad_halves_<code>v8hi): Likewise.
3179         (quad_halves_<code>v16qi): Likewise.
3180         (reduc_splus_v2di): Likewise.
3181         (neon_vpadd_internal<mode>): Likewise.
3182         (neon_vpsmin<mode>): Likewise.
3183         (neon_vpsmax<mode>): Likewise.
3184         (neon_vpumin<mode>): Likewise.
3185         (neon_vpumax<mode>): Likewise.
3186         (*ss_add<mode>_neon): Likewise.
3187         (*us_add<mode>_neon): Likewise.
3188         (*ss_sub<mode>_neon): Likewise.
3189         (*us_sub<mode>_neon): Likewise.
3190         (neon_vadd<mode>_unspec): Likewise.
3191         (neon_vaddl<mode>): Likewise.
3192         (neon_vaddw<mode>): Likewise.
3193         (neon_vhadd<mode>): Likewise.
3194         (neon_vqadd<mode>): Likewise.
3195         (neon_vaddhn<mode>): Likewise.
3196         (neon_vmul<mode>): Likewise.
3197         (neon_vfms<VCVTF:mode>): Likewise.
3198         (neon_vmlal<mode>): Likewise.
3199         (neon_vmls<mode>): Likewise.
3200         (neon_vmlsl<mode>): Likewise.
3201         (neon_vqdmulh<mode>): Likewise.
3202         (neon_vqdmlal<mode>): Likewise.
3203         (neon_vqdmlsl<mode>): Likewise.
3204         (neon_vmull<mode>): Likewise.
3205         (neon_vqdmull<mode>): Likewise.
3206         (neon_vsub<mode>_unspec): Likewise.
3207         (neon_vsubl<mode>): Likewise.
3208         (neon_vsubw<mode>): Likewise.
3209         (neon_vqsub<mode>): Likewise.
3210         (neon_vhsub<mode>): Likewise.
3211         (neon_vsubhn<mode>): Likewise.
3212         (neon_vceq<mode>): Likewise.
3213         (neon_vcge<mode>): Likewise.
3214         (neon_vcgeu<mode>): Likewise.
3215         (neon_vcgt<mode>): Likewise.
3216         (neon_vcgtu<mode>): Likewise.
3217         (neon_vcle<mode>): Likewise.
3218         (neon_vclt<mode>): Likewise.
3219         (neon_vcage<mode>): Likewise.
3220         (neon_vcagt<mode>): Likewise.
3221         (neon_vtst<mode>): Likewise.
3222         (neon_vabd<mode>): Likewise.
3223         (neon_vabdl<mode>): Likewise.
3224         (neon_vaba<mode>): Likewise.
3225         (neon_vabal<mode>): Likewise.
3226         (neon_vmax<mode>): Likewise.
3227         (neon_vmin<mode>): Likewise.
3228         (neon_vpaddl<mode>): Likewise.
3229         (neon_vpadal<mode>): Likewise.
3230         (neon_vpmax<mode>): Likewise.
3231         (neon_vpmin<mode>): Likewise.
3232         (neon_vrecps<mode>): Likewise.
3233         (neon_vrsqrts<mode>): Likewise.
3234         (neon_vqabs<mode>): Likewise.
3235         (neon_vqneg<mode>): Likewise.
3236         (neon_vcls<mode>): Likewise.
3237         (clz<mode>2): Likewise.
3238         (popcount<mode>2): Likewise.
3239         (neon_vrecpe<mode>): Likewise.
3240         (neon_vrsqrte<mode>): Likewise.
3241         (neon_vget_lane<mode>_sext_internal): Likewise.
3242         (neon_vget_lane<mode>_zext_internal): Likewise.
3243         (neon_vdup_n<mode>): Likewise.
3244         (neon_vdup_n<mode>): Likewise.
3245         (neon_vdup_nv2di): Likewise.
3246         (neon_vdup_lane<mode>_interal): Likewise.
3247         (*neon_vswp<mode>): Likewise.
3248         (neon_vcombine<mode>): Likewise.
3249         (float<mode><V_cvtto>2): Likewise.
3250         (floatuns<mode><V_cvtto>2): Likewise.
3251         (fix_trunc<mode><V_cvtto>2): Likewise.
3252         (fixuns_trunc<mode><V_cvtto>2
3253         (neon_vcvt<mode>): Likewise.
3254         (neon_vcvt<mode>): Likewise.
3255         (neon_vcvtv4sfv4hf): Likewise.
3256         (neon_vcvtv4hfv4sf): Likewise.
3257         (neon_vcvt_n<mode>): Likewise.
3258         (neon_vcvt_n<mode>): Likewise.
3259         (neon_vmovn<mode>): Likewise.
3260         (neon_vqmovn<mode>): Likewise.
3261         (neon_vqmovun<mode>): Likewise.
3262         (neon_vmovl<mode>): Likewise.
3263         (neon_vmul_lane<mode>): Likewise.
3264         (neon_vmul_lane<mode>): Likewise.
3265         (neon_vmull_lane<mode>): Likewise.
3266         (neon_vqdmull_lane<mode>): Likewise.
3267         (neon_vqdmulh_lane<mode>): Likewise.
3268         (neon_vqdmulh_lane<mode>): Likewise.
3269         (neon_vmla_lane<mode>): Likewise.
3270         (neon_vmla_lane<mode>): Likewise.
3271         (neon_vmlal_lane<mode>): Likewise.
3272         (neon_vqdmlal_lane<mode>): Likewise.
3273         (neon_vmls_lane<mode>): Likewise.
3274         (neon_vmls_lane<mode>): Likewise.
3275         (neon_vmlsl_lane<mode>): Likewise.
3276         (neon_vqdmlsl_lane<mode>): Likewise.
3277         (neon_vext<mode>): Likewise.
3278         (neon_vrev64<mode>): Likewise.
3279         (neon_vrev32<mode>): Likewise.
3280         (neon_vrev16<mode>): Likewise.
3281         (neon_vbsl<mode>_internal): Likewise.
3282         (neon_vshl<mode>): Likewise.
3283         (neon_vqshl<mode>): Likewise.
3284         (neon_vshr_n<mode>): Likewise.
3285         (neon_vshrn_n<mode>): Likewise.
3286         (neon_vqshrn_n<mode>): Likewise.
3287         (neon_vqshrun_n<mode>): Likewise.
3288         (neon_vshl_n<mode>): Likewise.
3289         (neon_vqshl_n<mode>): Likewise.
3290         (neon_vqshlu_n<mode>): Likewise.
3291         (neon_vshll_n<mode>): Likewise.
3292         (neon_vsra_n<mode>): Likewise.
3293         (neon_vsri_n<mode>): Likewise.
3294         (neon_vsli_n<mode>): Likewise.
3295         (neon_vtbl1v8qi): Likewise.
3296         (neon_vtbl2v8qi): Likewise.
3297         (neon_vtbl3v8qi): Likewise.
3298         (neon_vtbl4v8qi): Likewise.
3299         (neon_vtbl1v16qi): Likewise.
3300         (neon_vtbl2v16qi): Likewise.
3301         (neon_vcombinev16qi): Likewise.
3302         (neon_vtbx1v8qi): Likewise.
3303         (neon_vtbx2v8qi): Likewise.
3304         (neon_vtbx3v8qi): Likewise.
3305         (neon_vtbx4v8qi): Likewise.
3306         (*neon_vtrn<mode>_insn): Likewise.
3307         (*neon_vzip<mode>_insn): Likewise.
3308         (*neon_vuzp<mode>_insn): Likewise.
3309         (neon_vld1<mode>): Likewise.
3310         (neon_vld1_lane<mode>): Likewise.
3311         (neon_vld1_lane<mode>): Likewise.
3312         (neon_vld1_dup<mode>): Likewise.
3313         (neon_vld1_dup<mode>): Likewise.
3314         (neon_vld1_dupv2di): Likewise.
3315         (neon_vst1<mode>): Likewise.
3316         (neon_vst1_lane<mode>): Likewise.
3317         (neon_vst1_lane<mode>): Likewise.
3318         (neon_vld2<mode>): Likewise.
3319         (neon_vld2<mode>): Likewise.
3320         (neon_vld2_lane<mode>): Likewise.
3321         (neon_vld2_lane<mode>): Likewise.
3322         (neon_vld2_dup<mode>): Likewise.
3323         (neon_vst2<mode>): Likewise.
3324         (neon_vst2<mode>): Likewise.
3325         (neon_vst2_lane<mode>): Likewise.
3326         (neon_vst2_lane<mode>): Likewise.
3327         (neon_vld3<mode>): Likewise.
3328         (neon_vld3qa<mode>): Likewise.
3329         (neon_vld3qb<mode>): Likewise.
3330         (neon_vld3_lane<mode>): Likewise.
3331         (neon_vld3_lane<mode>): Likewise.
3332         (neon_vld3_dup<mode>): Likewise.
3333         (neon_vst3<mode>): Likewise.
3334         (neon_vst3qa<mode>): Likewise.
3335         (neon_vst3qb<mode>): Likewise.
3336         (neon_vst3_lane<mode>): Likewise.
3337         (neon_vst3_lane<mode>): Likewise.
3338         (neon_vld4<mode>): Likewise.
3339         (neon_vld4qa<mode>): Likewise.
3340         (neon_vld4qb<mode>): Likewise.
3341         (neon_vld4_lane<mode>): Likewise.
3342         (neon_vld4_lane<mode>): Likewise.
3343         (neon_vld4_dup<mode>): Likewise.
3344         (neon_vst4<mode>): Likewise.
3345         (neon_vst4qa<mode>): Likewise.
3346         (neon_vst4qb<mode>): Likewise.
3347         (neon_vst4_lane<mode>): Likewise.
3348         (neon_vst4_lane<mode>): Likewise.
3349         (neon_vec_unpack<US>_lo_<mode>): Likewise.
3350         (neon_vec_unpack<US>_hi_<mode>): Likewise.
3351         (neon_vec_<US>mult_lo_<mode>): Likewise.
3352         (neon_vec_<US>mult_hi_<mode>): Likewise.
3353         (neon_vec_<US>shiftl_<mode>): Likewise.
3354         (neon_unpack<US>_<mode>): Likewise.
3355         (neon_vec_<US>mult_<mode>): Likewise.
3356         (vec_pack_trunc_<mode>): Likewise.
3357         (neon_vec_pack_trunc_<mode>): Likewise.
3358         (neon_vabd<mode>_2): Likewise.
3359         (neon_vabd<mode>_3): Likewise.
3361 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3363         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
3364         (load_pair): Update type attribute.
3365         (store_pair): Update type attribute.
3366         * config/aarch64/iterators.md (q): New.
3368 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3370         * config/arm/types.md: Add new types for Neon insns.
3372 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3373             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3374             Sergey Lega  <sergey.s.lega@intel.com>
3375             Anna Tikhonova  <anna.tikhonova@intel.com>
3376             Ilya Tocar  <ilya.tocar@intel.com>
3377             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3378             Ilya Verbin  <ilya.verbin@intel.com>
3379             Kirill Yukhin  <kirill.yukhin@intel.com>
3380             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3382         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
3383         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
3384         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
3385         (rcp14<mode>): New.
3386         (srcp14<mode>): Ditto.
3387         (rsqrt14<mode>): Ditto.
3388         (rsqrt14<mode>): Ditto.
3389         (avx512f_vmscalef<mode>): Ditto.
3390         (avx512f_scalef<mode>): Ditto.
3391         (avx512f_getexp<mode>): Ditto.
3392         (avx512f_sgetexp<mode>): Ditto.
3393         (avx512f_fixupimm<mode>): Ditto.
3394         (avx512f_sfixupimm<mode>): Ditto.
3395         (avx512f_rndscale<mode>): Ditto.
3396         (*avx512er_exp2<mode>): Ditto.
3397         (*avx512er_rcp28<mode>): Ditto.
3398         (avx512er_rsqrt28<mode>): Ditto.
3399         (avx512f_getmant<mode>): Ditto.
3400         (avx512f_getmant<mode>): Ditto.
3401         (avx512f_rndscale<mode>): Fix formatting.
3403 2013-10-15  Martin Jambor  <mjambor@suse.cz>
3405         * ipa-utils.h (ipa_edge_within_scc): Declare.
3406         * ipa-cp.c (edge_within_scc): Moved...
3407         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
3409 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3410             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3411             Sergey Lega  <sergey.s.lega@intel.com>
3412             Anna Tikhonova  <anna.tikhonova@intel.com>
3413             Ilya Tocar  <ilya.tocar@intel.com>
3414             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3415             Ilya Verbin  <ilya.verbin@intel.com>
3416             Kirill Yukhin  <kirill.yukhin@intel.com>
3417             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3419         * config/i386/predicates.md (const_8_to_15_operand): New.
3420         (const_16_to_31_operand): Ditto.
3421         * config/i386/sse.md (V8FI): New.
3422         (V16FI): Ditto.
3423         (reduc_splus_v8df): Ditto.
3424         (reduc_splus_v16sf): Ditto.
3425         (avx512f_vextract<shuffletype>32x4_1): Ditto.
3426         (vec_extract_hi_<mode>): Ditto.
3427         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
3428         (vec_set_lo_<mode>): Ditto.
3429         (vec_set_hi_<mode>): Ditto.
3430         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
3431         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
3432         (avx512f_pshufd_1): Ditto.
3433         (avx512f_broadcast<mode>): Ditto.
3434         (avx512f_broadcast<mode>): Ditto.
3435         (define_split): Split vec_extract_lo into move.
3436         (ssequartermode): Ditto.
3437         (ssedoublemode): Extened with wider modes.
3438         (vec_extract_lo_<mode>): Ditto.
3440 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3441             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3442             Sergey Lega  <sergey.s.lega@intel.com>
3443             Anna Tikhonova  <anna.tikhonova@intel.com>
3444             Ilya Tocar  <ilya.tocar@intel.com>
3445             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3446             Ilya Verbin  <ilya.verbin@intel.com>
3447             Kirill Yukhin  <kirill.yukhin@intel.com>
3448             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3450         * config/i386/predicates.md (register_or_constm1_operand): New.
3451         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
3452         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
3453         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
3454         UNSPEC_SCATTER_PREFETCH
3455         (VI48_512): New.
3456         (avx512f_ucmp<mode>3): Ditto.
3457         (avx512f_vternlog<mode>): Ditto.
3458         (avx512f_align<mode>): Ditto.
3459         (<shift_insn><mode>3): Ditto.
3460         (avx512f_<rotate>v<mode>): Ditto.
3461         (avx512f_<rotate><mode>): Ditto.
3462         (avx512f_eq<mode>3): Ditto.
3463         (avx512f_eq<mode>3_1): Ditto.
3464         (avx512f_gt<mode>3): Ditto.
3465         (avx512f_testm<mode>3): Ditto.
3466         (avx512f_testnm<mode>3): Ditto.
3467         (avx512pf_gatherpf<mode>): Ditto.
3468         (*avx512pf_gatherpf<mode>_mask): Ditto.
3469         (*avx512pf_gatherpf<mode>): Ditto.
3470         (avx512pf_scatterpf<mode>): Ditto.
3471         (*avx512pf_scatterpf<mode>_mask): Ditto.
3472         (*avx512pf_scatterpf<mode>): Ditto.
3473         (avx512f_vec_dup_gpr<mode>): Ditto.
3474         (clz<mode>2): Ditto.
3475         (conflict<mode>): Ditto.
3476         (REDUC_SMINMAX_MODE): Extened with wider modes.
3477         (reduc_<code>_<mode>): Ditto.
3478         (vlshr<mode>3): Ditto.
3479         (vashl<mode>3): Ditto.
3481 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3482             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3483             Sergey Lega  <sergey.s.lega@intel.com>
3484             Anna Tikhonova  <anna.tikhonova@intel.com>
3485             Ilya Tocar  <ilya.tocar@intel.com>
3486             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3487             Ilya Verbin  <ilya.verbin@intel.com>
3488             Kirill Yukhin  <kirill.yukhin@intel.com>
3489             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3491         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
3492         UNSPEC_SCATTER.
3493         (VI48F_512): New.
3494         (avx512fmaskmode): Ditto.
3495         (bcstscalarsuff): Ditto.
3496         (avx512f_blendm<mode>): Ditto.
3497         (cmp_imm_predicate): Ditto.
3498         (avx512f_cmp<mode>3): Ditto.
3499         (avx512f_vec_dup<mode>): Ditto.
3500         (avx512f_vec_dup_mem<mode>): Ditto.
3501         (avx512f_vpermi2var<mode>3): Ditto.
3502         (avx512f_vpermt2var<mode>3): Ditto.
3503         (vec_init<mode>): Ditto.
3504         (avx512f_gathersi<mode>): Ditto.
3505         (*avx512f_gathersi<mode>): Ditto.
3506         (*avx512f_gathersi<mode>_2): Ditto.
3507         (avx512f_gatherdi<mode>): Ditto.
3508         (*avx512f_gatherdi<mode>): Ditto.
3509         (*avx512f_gatherdi<mode>_2): Ditto.
3510         (avx512f_scattersi<mode>): Ditto.
3511         (*avx512f_scattersi<mode>): Ditto.
3512         (avx512f_scatterdi<mode>): Ditto.
3513         (*avx512f_scatterdi<mode>): Ditto.
3514         (sseintprefix): Extened with wider modes.
3515         (VEC_GATHER_IDXSI): Ditto.
3516         (VEC_GATHER_IDXDI): Ditto.
3517         (VEC_GATHER_SRCDI): Ditto.
3519 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
3520             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3522         * config/arm/t-aprofile: New file.
3523         * config.gcc: Handle --with-multilib-list option.
3525 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
3527         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
3528         the input for r2.
3530 2013-10-15  Richard Biener  <rguenther@suse.de>
3532         * tree-loop-distribution.c (build_empty_rdg): Inline into
3533         single user.
3534         (rdg_flag_vertex): Inline into single user.
3535         (rdg_flag_vertex_and_dependent): Likewise.
3536         (build_rdg_partition_for_vertex): Remove processed bitmap.
3537         (rdg_build_partitions): Simplify.
3539 2013-10-15  Richard Biener  <rguenther@suse.de>
3541         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
3542         Restructure forwarding through conversions and copies to
3543         avoid performing copy-propagation the wrong way.  Adjust
3544         recursion invocations.
3545         (forward_propagate_addr_expr): Add argument stating if we
3546         are recursing from a single-use.
3547         (ssa_forward_propagate_and_combine): Adjust.
3549 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
3551         * dumpfile.h (gcc::dump_manager): New class, to hold state
3552         relating to dumpfile management.
3553         (get_dump_file_name): Remove in favor of method of dump_manager.
3554         (dump_initialized_p): Likewise.
3555         (dump_start): Likewise.
3556         (dump_finish): Likewise.
3557         (dump_switch_p): Likewise.
3558         (dump_register): Likewise.
3559         (get_dump_file_info): Likewise.
3560         * context.c (gcc::context::context): Construct the dump_manager
3561         instance.
3562         * context.h (gcc::context::get_dumps): New.
3563         (gcc::context::m_dumps): New.
3564         * coverage.c (coverage_init): Port to dump_manager API.
3565         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
3566         (extra_dump_files_in_use): Likewise.
3567         (extra_dump_files_alloced): Likewise.
3568         (gcc::dump_manager::dump_manager): New.
3569         (dump_register): Convert to...
3570         (gcc::dump_manager::dump_register): ...method, replacing
3571         function-static next_dump with m_next_dump field.
3572         (get_dump_file_info): Convert to...
3573         (gcc::dump_manager::get_dump_file_info): ...method.
3574         (get_dump_file_name): Convert to...
3575         (gcc::dump_manager::get_dump_file_name): ...method.
3576         (dump_start): Convert to...
3577         (gcc::dump_manager::dump_start): ...method.
3578         (dump_finish): Convert to...
3579         (gcc::dump_manager::dump_finish): ...method.
3580         (dump_begin): Replace body with...
3581         (gcc::dump_manager::dump_begin): ...new method.
3582         (dump_phase_enabled_p): Convert to...
3583         (gcc::dump_manager::dump_phase_enabled_p): ...method.
3584         (dump_phase_enabled_p): Convert to...
3585         (gcc::dump_manager::dump_phase_enabled_p): ...method.
3586         (dump_initialized_p):  Convert to...
3587         (gcc::dump_manager::dump_initialized_p): ...method.
3588         (dump_flag_name): Replace body with...
3589         (gcc::dump_manager::dump_flag_name): ...new method.
3590         (dump_enable_all): Convert to...
3591         (gcc::dump_manager::dump_enable_all): ...new method.
3592         (opt_info_enable_passes): Convert to...
3593         (gcc::dump_manager::opt_info_enable_passes): ...new method.
3594         (dump_switch_p_1): Convert to...
3595         (gcc::dump_manager::dump_switch_p_1): ...new method.
3596         (dump_switch_p):  Convert to...
3597         (gcc::dump_manager::dump_switch_p): ...new method.
3598         (opt_info_switch_p): Port to dump_manager API.
3599         (enable_rtl_dump_file): Likewise.
3600         * opts-global.c (handle_common_deferred_options): Port to new
3601         dump_manager API.
3602         * passes.c (pass_manager::finish_optimization_passes): Likewise.
3603         (pass_manager::register_one_dump_file): Likewise.
3604         (pass_manager::register_pass): Likewise.
3605         (pass_init_dump_file): Likewise.
3606         (pass_fini_dump_file): Likewise.
3607         * statistics.c (statistics_early_init): Likewise.
3609 2013-10-14  Richard Biener  <rguenther@suse.de>
3611         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
3612         iterative_hash_canonical_type, gimple_canonical_type_hash,
3613         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
3614         gimple_register_canonical_type, print_gimple_types_stats,
3615         free_gimple_type_tables): Move to lto/lto.c
3616         (gt-gimple.h): Do not include.
3617         * gimple.h (gimple_register_canonical_type,
3618         print_gimple_types_stats, free_gimple_type_tables): Remove.
3619         * Makefile.in (GTFILES): Remove gimple.c.
3621 2013-10-14  Travis Snoozy  <quandary@remstate.com>
3623         PR target/58716
3624         * config/msp430/msp430.c (msp430_option_override): Correct thinko
3625         scanning for msp430x targets.
3627 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
3629         PR bootstrap/58509
3630         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
3631         (registers_ok_for_ldd_peep): Move around.
3632         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
3633         * config/sparc/sparc.md (widening peepholes): Use it.
3635 2013-10-14  Richard Biener  <rguenther@suse.de>
3637         PR middle-end/58712
3638         PR middle-end/55358
3639         * gimple.c (iterative_hash_canonical_type): Make sure to
3640         record the hash into the correct hashtable slot.
3642 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
3644         PR rtl-optimization/58662
3645         * combine.c (try_combine): Take into account death nodes on I2 when
3646         splitting a PARALLEL of two independent SETs.  Fix dump message.
3648 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
3650         PR target/51244
3651         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
3652         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
3653         * config/sh/t-sh (sh_treg_combine.o): New entry.
3654         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
3655         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
3656         (register_sh_passes): New function.  Register sh_treg_combine pass.
3657         (sh_option_override): Invoke it.
3658         (sh_canonicalize_comparison): Handle op0_preserve_value.
3659         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
3660         opportunities.  Canonicalize branch condition.
3661         (nott): Allow only if pseudos can be created for non-SH2A.
3663 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
3665         PR target/58690
3666         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
3667         (ix86_expand_movmem): Replace copy_addr_to_reg with
3668         ix86_copy_addr_to_reg.
3669         (ix86_expand_setmem): Likewise.
3671 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
3673         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
3674         provided by ix86_fp_compare_mode instead of CCFPUmode.
3676 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
3678         * config/aarch64/arm_neon.h
3679         (vtbx<1,3>_<psu>8): Fix register constriants.
3681 2013-10-11  Jeff Law  <law@redhat.com>
3683         PR tree-optimization/58640
3684         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
3685         threading paths that cross over two loop entry points.
3687 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3689         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
3690         handle vector float as well.
3691         (*vsx_le_perm_load_v4si): Likewise.
3692         (*vsx_le_perm_store_v2di): Likewise.
3693         (*vsx_le_perm_store_v4si): Likewise.
3695 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3697         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
3698         directly to circumvent subtract from splat{31} workaround.
3699         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
3700         prototype.
3701         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
3702         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
3703         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
3704         (altivec_vperm_<mode>): Convert to define_insn_and_split to
3705         separate big and little endian logic.
3706         (*altivec_vperm_<mode>_internal): New define_insn.
3707         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
3708         separate big and little endian logic.
3709         (*altivec_vperm_<mode>_uns_internal): New define_insn.
3710         (vec_permv16qi): Add little endian logic.
3712 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
3714         * doc/extend.texi (returns_nonnull): Remove arguments.
3716 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3717             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3718             Sergey Lega  <sergey.s.lega@intel.com>
3719             Anna Tikhonova  <anna.tikhonova@intel.com>
3720             Ilya Tocar  <ilya.tocar@intel.com>
3721             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3722             Ilya Verbin  <ilya.verbin@intel.com>
3723             Kirill Yukhin  <kirill.yukhin@intel.com>
3724             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3726         * config/i386/sse.md (VI48F_256_512): New.
3727         (avx2_permvar<mode>): Change to ...
3728         (<avx2_avx512f>_permvar<mode>): This.
3730 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3731             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3732             Sergey Lega  <sergey.s.lega@intel.com>
3733             Anna Tikhonova  <anna.tikhonova@intel.com>
3734             Ilya Tocar  <ilya.tocar@intel.com>
3735             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3736             Ilya Verbin  <ilya.verbin@intel.com>
3737             Kirill Yukhin  <kirill.yukhin@intel.com>
3738             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3740         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
3741         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
3742         * config/i386/sse.md (VI4_AVX): New.
3743         (sf2simodelower): Ditto.
3744         (sse2_cvtps2dq): Change to ...
3745         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
3747 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3748             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3749             Sergey Lega  <sergey.s.lega@intel.com>
3750             Anna Tikhonova  <anna.tikhonova@intel.com>
3751             Ilya Tocar  <ilya.tocar@intel.com>
3752             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3753             Ilya Verbin  <ilya.verbin@intel.com>
3754             Kirill Yukhin  <kirill.yukhin@intel.com>
3755             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3757         * config/i386/sse.md (V_512): New.
3758         (VI_512): Ditto.
3759         (vcond<V_512:mode><VF_512:mode>): Ditto.
3760         (vcond<V_512:mode><VI_512:mode>): Ditto.
3761         (vcondu<V_512:mode><VI_512:mode>): Ditto.
3763 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3764             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3765             Sergey Lega  <sergey.s.lega@intel.com>
3766             Anna Tikhonova  <anna.tikhonova@intel.com>
3767             Ilya Tocar  <ilya.tocar@intel.com>
3768             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3769             Ilya Verbin  <ilya.verbin@intel.com>
3770             Kirill Yukhin  <kirill.yukhin@intel.com>
3771             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3773         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
3774         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
3775         (FMAMODE): Ditto.
3776         (fma<mode>4): Removed condition.
3777         (fms<mode>4): Ditto.
3778         (fnma<mode>4): Ditto.
3779         (fnms<mode>4): Ditto.
3780         (fma4i_fmadd_<mode>): Ditto.
3781         (*fma_fmadd_<mode>): Ditto.
3782         (*fma_fmsub_<mode>): Ditto.
3783         (*fma_fnmadd_<mode>): Ditto.
3784         (*fma_fnmsub_<mode>): Ditto.
3785         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
3786         (*fma_fmaddsub_<mode>): Ditto.
3787         (*fma_fmsubadd_<mode>): Ditto.
3788         (*fmai_fmadd_<mode>): Ditto.
3789         (*fmai_fmsub_<mode>): Ditto.
3790         (*fmai_fnmadd_<mode>): Ditto.
3791         (*fmai_fnmsub_<mode>): Ditto.
3793 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3794             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3795             Sergey Lega  <sergey.s.lega@intel.com>
3796             Anna Tikhonova  <anna.tikhonova@intel.com>
3797             Ilya Tocar  <ilya.tocar@intel.com>
3798             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3799             Ilya Verbin  <ilya.verbin@intel.com>
3800             Kirill Yukhin  <kirill.yukhin@intel.com>
3801             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3803         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
3804         (VI124_256): Changed to ...
3805         (VI124_256_48_512): This.
3806         (ssepackmode): Extended with wider modes.
3807         (<code><mode>3): Changed iterator.
3808         (*avx2_<code><mode>3): Ditto.
3809         (vec_pack_trunc_<mode>): Ditto.
3811 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3812             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3813             Sergey Lega  <sergey.s.lega@intel.com>
3814             Anna Tikhonova  <anna.tikhonova@intel.com>
3815             Ilya Tocar  <ilya.tocar@intel.com>
3816             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3817             Ilya Verbin  <ilya.verbin@intel.com>
3818             Kirill Yukhin  <kirill.yukhin@intel.com>
3819             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3821         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
3822         (VI8F_256_512): Ditto.
3823         (abs<mode>2): Changed iterator.
3824         (avx2_perm<mode>): Changed to ...
3825         (<avx2_avx512f>_perm<mode>): This.
3826         (avx2_perm<mode>_1): Changed to ...
3827         (<avx2_avx512f>_perm<mode>_1): This.
3829 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3830             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3831             Sergey Lega  <sergey.s.lega@intel.com>
3832             Anna Tikhonova  <anna.tikhonova@intel.com>
3833             Ilya Tocar  <ilya.tocar@intel.com>
3834             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3835             Ilya Verbin  <ilya.verbin@intel.com>
3836             Kirill Yukhin  <kirill.yukhin@intel.com>
3837             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3839         * config/i386/sse.md (VI48_AVX512F): New.
3840         (VI48_AVX2): Changed to ...
3841         (VI48_AVX2_48_AVX512F): This.
3842         (avx2_ashrv<mode>): Changed to ...
3843         (<avx2_avx512f>_ashrv<mode>): This.
3844         (avx2_<shift_insn>v<mode>): Changed to ...
3845         (<avx2_avx512f>_<shift_insn>v<mode>): This.
3847 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3848             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3849             Sergey Lega  <sergey.s.lega@intel.com>
3850             Anna Tikhonova  <anna.tikhonova@intel.com>
3851             Ilya Tocar  <ilya.tocar@intel.com>
3852             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3853             Ilya Verbin  <ilya.verbin@intel.com>
3854             Kirill Yukhin  <kirill.yukhin@intel.com>
3855             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3857         * config/i386/sse.md (VI4_AVX512F): New.
3858         (VI8_AVX2_AVX512F): Ditto.
3859         (mul<mode>3): Extended with wider modes.
3860         (*<sse4_1_avx2>_mul<mode>3): Ditto.
3861         (mul<mode>3): Ditto.
3862         (vec_widen_<s>mult_odd_<mode>): Ditto.
3864 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3865             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3866             Sergey Lega  <sergey.s.lega@intel.com>
3867             Anna Tikhonova  <anna.tikhonova@intel.com>
3868             Ilya Tocar  <ilya.tocar@intel.com>
3869             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3870             Ilya Verbin  <ilya.verbin@intel.com>
3871             Kirill Yukhin  <kirill.yukhin@intel.com>
3872             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3874         * config/i386/sse.md (VI2_AVX512F): New.
3875         (VI124_AVX512F): Ditto.
3876         (sseunpackmode): Extended with wider modes.
3877         (sseunpackfltmode): Ditto.
3878         (vec_unpacks_float_hi_<mode>): Ditto.
3879         (vec_unpacks_float_lo_<mode>): Ditto.
3880         (vec_unpacku_float_hi_<mode>): Ditto.
3881         (vec_unpacku_float_lo_<mode>): Ditto.
3882         (vec_unpacks_lo_<mode>): Ditto.
3883         (vec_unpacks_hi_<mode>): Ditto.
3884         (vec_unpacku_lo_<mode>): Ditto.
3885         (vec_unpacku_hi_<mode>): Ditto.
3887 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3888             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3889             Sergey Lega  <sergey.s.lega@intel.com>
3890             Anna Tikhonova  <anna.tikhonova@intel.com>
3891             Ilya Tocar  <ilya.tocar@intel.com>
3892             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3893             Ilya Verbin  <ilya.verbin@intel.com>
3894             Kirill Yukhin  <kirill.yukhin@intel.com>
3895             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3897         * config/i386/i386.md (multdiv): New.
3898         (multdiv_mnemonic): Ditto.
3899         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
3900         (<sse>_vm<multdiv_mnemonic><mode>3): This.
3901         (<sse>_vmdiv<mode>3): Removed.
3903 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3904             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3905             Sergey Lega  <sergey.s.lega@intel.com>
3906             Anna Tikhonova  <anna.tikhonova@intel.com>
3907             Ilya Tocar  <ilya.tocar@intel.com>
3908             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3909             Ilya Verbin  <ilya.verbin@intel.com>
3910             Kirill Yukhin  <kirill.yukhin@intel.com>
3911             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3913         * config/i386/sse.md (V): Extended with wider modes.
3914         (VF2): Ditto.
3915         (ssehalfvecmode): Ditto.
3916         (i128): Ditto.
3917         (ssepackfltmode): Ditto.
3918         (avx_vec_concat<mode>): Ditto.
3919         (V_256_512): New iterator.
3920         (VF2_512_256): Ditto.
3921         (si2dfmode): New attribute.
3922         (si2dfmodelower): Ditto.
3923         (sf2dfmode): Ditto.
3924         (concat_tg_mode): Ditto.
3925         (floatv4siv4df2): Changed to ...
3926         (float<si2dfmodelower><mode>2): This.
3927         (avx_cvtps2pd256): Changed to ...
3928         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
3929         (vec_pack_trunc_v4df): Changed to ...
3930         (vec_pack_trunc_<mode>): This.
3931         (avx_vpermil<mode>): Changed to ...
3932         (<sse2_avx_avx512f>_vpermil<mode>): This.
3933         (<fixsuffix>fix_truncv8dfv8si2): New.
3934         (vec_pack_sfix_trunc_v8df): Ditto.
3935         (avx512f_rndscale<mode>): Ditto.
3936         (avx512f_roundpd512): Ditto.
3937         (vec_pack_ufix_trunc_<mode>): Updated iterator.
3939 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3940             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3941             Sergey Lega  <sergey.s.lega@intel.com>
3942             Anna Tikhonova  <anna.tikhonova@intel.com>
3943             Ilya Tocar  <ilya.tocar@intel.com>
3944             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3945             Ilya Verbin  <ilya.verbin@intel.com>
3946             Kirill Yukhin  <kirill.yukhin@intel.com>
3947             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3949         * config/i386/i386.md (any_fix): New iterator.
3950         (fixsuffix): New attribute.
3951         * config/i386/sse.md (VF1): Extened with wider modes.
3952         (VI): Ditto.
3953         (VI_AVX2): Ditto.
3954         (VI8): Ditto.
3955         (sseintvecmodelower): Ditto.
3956         (ssescalarmode): Ditto.
3957         (ssescalarnum): Ditto.
3958         (VF1_128_256): New.
3959         (ssexmmmode): Ditto.
3960         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
3961         (<sse>_rcp<mode>2): Change iterator.
3962         (rsqrt<mode>2): Ditto.
3963         (<sse>_rsqrt<mode>2): Ditto.
3964         (avx2_vec_dup<mode>): Ditto.
3965         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
3966         (round<mode>2_sfix): Ditto.
3967         (avx2_pbroadcast<mode>): Ditto.
3968         (*andnot<mode>3): Handle XI mode.
3969         (*<code><mode>3): Ditto.
3970         (AVXTOSSEMODE): Removed.
3971         (avx_vpermil<mode>): Changed to ...
3972         (<sse2_avx_avx512f>_vpermil<mode>): This.
3974 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3975             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3976             Sergey Lega  <sergey.s.lega@intel.com>
3977             Anna Tikhonova  <anna.tikhonova@intel.com>
3978             Ilya Tocar  <ilya.tocar@intel.com>
3979             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3980             Ilya Verbin  <ilya.verbin@intel.com>
3981             Kirill Yukhin  <kirill.yukhin@intel.com>
3982             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3984         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
3985         (<sse>_comi): Ditto.
3986         (<sse>_ucomi): Ditto.
3987         (sse_cvtss2siq_2): Ditto.
3988         (sse2_cvtsd2si): Ditto.
3989         (sse2_cvtsd2siq): Ditto.
3990         (sse2_cvttsd2si): Ditto.
3991         (sse2_cvttsd2siq): Ditto.
3992         (<shift_insn><mode>3): Ditto.
3993         (sse2_cvtsi2sdq): Update constraint and prefix.
3994         (sse_cvtsi2ss): Update prefix.
3995         (sse_cvtsi2ssq): Ditto.
3997 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
3999         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
4000         ignore internal calls.
4002 2013-10-11  Richard Biener  <rguenther@suse.de>
4004         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
4005         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
4006         TREE_OVERFLOW is set.
4008 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
4010         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
4012         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
4013         * gimple.h: Likewise.
4015         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
4017         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
4018         macros.
4020         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
4022         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
4023         * configure: Regenerate.
4025 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4027         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
4028         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
4029         formatting fixes, use pp_colon instead of pp_character (..., ':'),
4030         similarly pp_right_paren.
4031         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
4032         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
4033         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
4034         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
4035         clauses.
4036         (omp_declare_simd_clauses_equal,
4037         omp_remove_redundant_declare_simd_attrs): New functions.
4038         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
4039         (walk_tree_1): Handle new OpenMP 4.0 clauses.
4040         * tree.h (OMP_LOOP_CHECK): Define.
4041         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
4042         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
4043         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
4044         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
4045         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
4046         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
4047         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
4048         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
4049         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
4050         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
4051         OMP_CLAUSE_SIMDLEN_EXPR): Define.
4052         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
4053         (omp_remove_redundant_declare_simd_attrs): New prototype.
4054         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
4055         GIMPLE_OMP_TEAMS): New codes.
4056         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
4057         * omp-low.c (struct omp_context): Add cancel_label and cancellable
4058         fields.
4059         (target_nesting_level): New variable.
4060         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
4061         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
4062         for collapse > 1 static schedule unless ordered.
4063         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
4064         (determine_parallel_type): Adjust get_ws_args_for caller.
4065         (install_var_field): Handle mask & 4 for double indirection.
4066         (scan_sharing_clauses): Ignore shared clause on teams construct.
4067         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
4068         (create_omp_child_function): If inside target or declare target
4069         constructs, set "omp declare target" attribute on the child function.
4070         (find_combined_for): New function.
4071         (scan_omp_parallel): Handle combined loops.
4072         (scan_omp_target, scan_omp_teams): New functions.
4073         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
4074         restrictions and set ctx->cancellable for cancellable constructs.
4075         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
4076         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
4077         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
4078         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
4079         (omp_clause_aligned_alignment): New function.
4080         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
4081         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
4082         on teams constructs.  Handle user defined reductions and new
4083         OpenMP 4.0 clauses.
4084         (lower_reduction_clauses): Don't set placeholder to address of ref
4085         if it has already the right type.
4086         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
4087         (expand_parallel_call): Use the new non-_start suffixed builtins,
4088         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
4089         and GOMP_parallel_end after the call.
4090         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
4091         (expand_omp_for_init_counts): Handle combined loops.
4092         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
4093         loops.
4094         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
4095         end of cancellable loops.
4096         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
4097         Likewise.  Handle collapse > 1 loops.
4098         (expand_omp_simd): Handle combined loops.
4099         (expand_omp_for): Add inner_stmt argument, adjust callers of
4100         expand_omp_for* functions, use expand_omp_for_static*chunk even
4101         for collapse > 1 unless ordered.
4102         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
4103         of cancellable sections.
4104         (expand_omp_single): Remove need_barrier variable, just rely on
4105         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
4106         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
4107         (expand_omp_atomic_load, expand_omp_atomic_store,
4108         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
4109         (expand_omp_target): New function.
4110         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
4111         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
4112         (build_omp_regions_1): Immediately close region for
4113         GF_OMP_TARGET_KIND_UPDATE.
4114         (maybe_add_implicit_barrier_cancel): New function.
4115         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
4116         cancellation.
4117         (lower_omp_single): Likewise.  Add clobber after the barrier.
4118         (lower_omp_taskgroup): New function.
4119         (lower_omp_for): Handle combined loops.  Adjust
4120         lower_rec_input_clauses caller.  Handle cancellation.
4121         (lower_depend_clauses): New function.
4122         (lower_omp_taskreg): Lower depend clauses.  Adjust
4123         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
4124         cancellation.
4125         (lower_omp_target, lower_omp_teams): New functions.
4126         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
4127         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
4128         and GOMP_cancellation_point calls.
4129         (lower_omp): Fold stmts inside of target region.
4130         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
4131         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4132         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
4133         (BT_FN_VOID_OMPFN_PTR_UINT,
4134         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
4135         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
4136         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
4137         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
4138         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
4139         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
4140         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
4141         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
4142         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
4143         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
4144         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
4145         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
4146         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
4147         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
4148         BUILT_IN_GOMP_PARALLEL_END.
4149         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
4150         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4151         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
4152         GF_OMP_FOR_KIND_DISTRIBUTE.
4153         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
4154         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
4155         (dump_gimple_omp_return): Print lhs if it has any.
4156         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
4157         gimple_omp_atomic_seq_cst_p.
4158         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
4159         and GIMPLE_OMP_TEAMS.
4160         * langhooks.c (lhd_omp_mappable_type): New function.
4161         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
4162         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
4163         hook.
4164         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
4165         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
4166         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
4167         (struct gimplify_omp_ctx): Add combined_loop field.
4168         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
4169         on stmts inside of target region.
4170         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
4171         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
4172         ORT_TARGET and ORT_TARGET_DATA.
4173         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
4174         Handle GOVD_MAP.
4175         (omp_notice_threadprivate_variable): Complain about threadprivate
4176         variables in target region.
4177         (omp_notice_variable): Complain about vars with non-mappable type
4178         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
4179         (omp_check_private): Ignore ORT_TARGET* regions.
4180         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
4181         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
4182         (find_combined_omp_for): New function.
4183         (gimplify_omp_for): Handle gimplification of combined loops.
4184         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
4185         OMP_TEAMS.
4186         (gimplify_omp_target_update): New function.
4187         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
4188         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
4189         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
4190         (gimplify_body): If fndecl has "omp declare target" attribute, add
4191         implicit ORT_TARGET context around it.
4192         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
4193         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
4194         * tree-nested.c (convert_nonlocal_reference_stmt,
4195         convert_local_reference_stmt, convert_gimple_call): Handle
4196         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4197         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
4198         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
4199         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
4200         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
4201         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
4202         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
4203         BUILT_IN_GOMP_LOOP_END_CANCEL,
4204         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
4205         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
4206         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
4207         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
4208         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
4209         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
4210         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
4211         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
4212         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
4213         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
4214         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
4215         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
4216         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
4217         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
4218         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
4219         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4220         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
4221         gimple_build_omp_teams): New functions.
4222         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
4223         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
4224         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
4225         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4226         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
4227         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
4228         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
4229         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
4230         GF_OMP_ATOMIC_SEQ_CST): New.
4231         (gimple_build_omp_taskgroup, gimple_build_omp_target,
4232         gimple_build_omp_teams): New prototypes.
4233         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
4234         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4235         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
4236         GIMPLE_OMP_SINGLE as end of range.
4237         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
4238         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
4239         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
4240         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
4241         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
4242         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
4243         gimple_omp_target_kind, gimple_omp_target_set_kind,
4244         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
4245         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
4246         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
4247         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
4248         gimple_omp_teams_set_clauses): New inlines.
4249         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
4250         and GIMPLE_OMP_TASKGROUP.
4251         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
4252         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
4253         enum omp_clause_proc_bind_kind): New.
4254         (union omp_clause_subcode): Add depend_kind, map_kind and
4255         proc_bind_kind fields.
4256         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
4257         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4258         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
4259         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
4260         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
4262 2013-10-10  Teresa Johnson  <tejohnson@google.com>
4264         * predict.c (tree_estimate_probability): Add new parameter
4265         for estimate_bb_frequencies.
4266         (estimate_bb_frequencies): Add new parameter to force estimation.
4267         (rebuild_frequencies): When max frequency in function is small,
4268         recompute counts from frequencies.
4269         * predict.h (estimate_bb_frequencies): New parameter.
4271 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
4273         * ipa-inline.c (ipa_inline): Fix leak of "order" when
4274         optimizations are disabled.
4276 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
4278         * coverage.c (coverage_finish): Fix leak of da_file_name.
4280 2013-10-10  Jan Hubicka  <jh@suse.cz>
4282         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
4283         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
4284         Bobcat and generic.
4286 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
4288         PR middle-end/58670
4289         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
4290         if any labels are in FALLTHRU_BB, use a special label emitted
4291         immediately after the asm goto insn rather than label_rtx
4292         of the LABEL_DECL.
4293         (expand_asm_stmt): Adjust caller.
4294         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
4295         edge if the last insn in predecessor is a jump with single successor,
4296         but it isn't simplejump_p.
4298 2013-10-10  Richard Biener  <rguenther@suse.de>
4300         PR tree-optimization/58656
4301         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
4303 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4305         * gimplify.c: Include expr.h and tm_p.h for targets with special
4306         va-arg padding requirements.
4308 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4310         * tree-flow.h: Move some prototypes to gimple.h.
4311         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
4312         * gimple.h: Relocate some prototypes from tree-flow.h
4313         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
4314         Move to gimplify.c.
4315         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
4316         (build_va_arg_indirect_ref): Relocate and make static.
4317         (std_gimplify_va_arg_expr): Relocate here.
4318         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
4319         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
4321 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4323         * doc/md.texi: Document the mnemonic attribute.
4325 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4327         PR target/57377
4328         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
4329         (set_attr_alternative x ...) when searching for user defined
4330         mnemonic attribute.
4332 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4334         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
4336 2013-10-09  Easwaran Raman <eraman@google.com>
4338         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
4339         * cfgexpand.c (defer_stack_allocation): ...use here
4340         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
4342 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
4344         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
4345         (operand_equal_for_value_replacement): New function, extracted from
4346         value_replacement and enhanced to catch more cases.
4347         (value_replacement): Use operand_equal_for_value_replacement.
4349 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4351         * loop-doloop.c (doloop_modify, doloop_optimize): Use
4352         get_max_loop_iterations.
4354 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4356         * config/arm/aarch-common.c (arm_early_load_addr_dep):
4357         Place comment above function.
4359 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4361         * tree-flow.h: Remove all remaining prototypes, enums and structs that
4362         are not related to tree-cfg.c.
4363         * tree-ssa-address.h: New file.  Relocate prototypes.
4364         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
4365         (addr_for_mem_ref): New.  Combine call to get_address_description and
4366         return addr_for_mem_ref.
4367         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
4368         * tree-ssa-live.h: Adjust prototypes.
4369         * passes.c: Include tree-ssa-live.h.
4370         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
4371         * graphite.c (graphite_transform_loops): Make static.
4372         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
4373         make_pass_graphite, pass_data_graphite_transforms,
4374         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
4375         * ipa-pure-const.c (warn_function_noreturn): Make static.
4376         (execute_warn_function_noreturn, gate_warn_function_noreturn,
4377         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
4378         Relocate from tree-cfg.c
4379         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
4380         static.
4381         (execute_warn_function_noreturn, gate_warn_function_noreturn,
4382         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
4383         Move to ipa-pure-const.c.
4384         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
4385         Relocate from tree-optimize.c.
4386         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
4387         make_pass_fixup_cfg): Move to tree-cfg.c.
4388         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
4389         Relocate some prototypes.
4390         * tree-data-ref.h (tree_check_data_deps) Add prototype.
4391         * tree-dump.c (dump_function_to_file): Remove prototype.
4392         Add tree-flow.h to the include file.
4393         * tree-dump.h: Remove prototype.
4394         * tree-parloops.h: New File.  Add prototypes.
4395         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
4396         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
4397         from tree-ssa-loop.c.
4398         * tree-predcom.c (run_tree_predictive_commoning,
4399         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
4400         Relocate here from tree-ssa-loop.c.
4401         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
4402         ssa_name_values.release ().
4403         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
4404         (ssa_name_values): Relocate from tree-flow.h.
4405         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
4406         * tree-ssa-loop.c (run_tree_predictive_commoning,
4407         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
4408         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
4409         make_pass_graphite, pass_data_graphite_transforms,
4410         make_pass_graphite_transforms, gate_tree_parallelize_loops,
4411         tree_parallelize_loops, pass_data_parallelize_loops,
4412         make_pass_parallelize_loops): Move to other files.
4413         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
4414         moved here.
4415         * tree.h: Remove prototypes from tree-address.c.
4417 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4419         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
4420         (struct int_tree_map): Move to tree-hasher.h
4421         (SCALE, LABEL, PERCENT): Move to gimple.h
4422         * tree-flow-inline.h: Delete.  Move functions to other files.
4423         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
4424         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
4425         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
4426         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
4427         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
4428         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
4429         * tree-hasher.h: Don't include tree-flow.h.
4430         (struct int_tree_map): Relocate from tree-flow.h.
4431         * tree-sra.c (contains_view_convert_expr_p): Relocate from
4432         tree-flow-inline.h and make static.
4433         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
4434         tree-flow-inline.h.
4435         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
4436         tree-flow-inline.h and make static.
4437         * tree.h (is_global_var, may_be_aliased): Relocate from
4438         tree-flow-inline.h.
4439         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
4440         * value-prof.c: No longer include tree-flow-inline.h.
4441         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
4443 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4445         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
4446         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
4447         (enum move_pos): Move to tree-ssa-loop-im.h
4448         * cfgloop.h: Move some prototypes.
4449         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
4450         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
4451         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
4452         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
4453         (loop_containing_stmt): Relocate from tree-flow-inline.h.
4454         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
4455         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
4456         (enum move_pos): Relocate here.
4457         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
4458         tree-ssa-loop.c.
4459         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
4460         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
4461         make_pass_lim): Relocate here from tree-ssa-loop.c.
4462         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
4463         tree-ssa-loop.c.
4464         (loop_edge_to_cancel, unloop_loops): Make static.
4465         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
4466         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
4467         (tree_complete_unroll, gate_tree_complete_unroll,
4468         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
4469         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
4470         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
4471         * tree-ssa-loop-ivopts.c: Remove local prototypes.
4472         (stmt_invariant_in_loop_p): Remove unused function.
4473         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
4474         * tree-ssa-loop-manip.h: New file.  Add prototypes.
4475         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
4476         (gcov_type_to_double_int): Move to cfgloop.h.
4477         (double_int_cmp, bound_index,
4478         estimate_numbers_of_iterations_loop): Make static.
4479         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
4480         (max_loop_iterations): Factor out get_max_loop_iterations.
4481         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
4482         cfgloop.c.
4483         * tree-ssa-loop-niter.h: New file.  Add prototypes.
4484         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
4485         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
4486         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
4487         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
4488         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
4489         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
4490         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
4491         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
4492         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
4493         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
4494         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
4495         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
4496         pass_data_complete_unroll, make_pass_complete_unroll,
4497         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
4498         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
4499         tree-ssa-loop-ivcanon.c.
4500         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
4501         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
4502         tree-ssa-loop-prefetch.c.
4503         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
4504         tree-ssa-loop-im.c.
4505         (get_lsm_tmp_name): Relocate and add suffix parameter.
4506         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
4507         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
4508         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
4509         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
4510         (get_estimated_loop_iterations): Factor out accessor from
4511         estimated_loop_iterations in tree-ssa-loop-niter.c.
4512         (get_max_loop_iterations): Factor out accessor from
4513         _max_loop_iterations in tree-ssa-niter.c.
4514         * loop-unroll.c (decide_unroll_constant_iterations,
4515         decide_unroll_runtime_iterations, decide_peel_simple,
4516         decide_unroll_stupid): Use new get_* accessors.
4518 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
4520         PR tree-optimization/20318
4521         * doc/extend.texi (returns_nonnull): New function attribute.
4522         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
4523         attribute.
4524         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
4525         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
4527 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
4529         PR middle-end/58570
4530         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
4531         false if both components are bitfields.
4533 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
4535         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
4536         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
4537         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
4538         * config/aarch64/aarch64.h
4539         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
4540         * config/aarch64/aarch64-simd-builtins.def
4541         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
4543 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
4545         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
4547 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
4549         * config/aarch64/arm_neon.h (vdiv_f64): Added.
4551 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
4553         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
4554         (vneg_f64): New intrinsic.
4555         (vneg_s8): Asm replaced with C.
4556         (vneg_s16): Likewise.
4557         (vneg_s32): Likewise.
4558         (vneg_s64): New intrinsic.
4559         (vnegq_f32): Asm replaced with C.
4560         (vnegq_f64): Likewise.
4561         (vnegq_s8): Likewise.
4562         (vnegq_s16): Likewise.
4563         (vnegq_s32): Likewise.
4564         (vnegq_s64): Likewise.
4566 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
4568         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
4570 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4572         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
4573         packed stack special handling.
4574         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
4575         back to fixed stack slots for FPRs saved due to stdarg.
4577 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4579         * config/s390/s390.c (s390_frame_info): Restructure function.
4581 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4583         * config/s390/s390.c (struct s390_frame_layout): New field
4584         gpr_save_slots.
4585         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
4586         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
4587         regs_ever_clobbered to char*.
4588         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
4589         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
4590         (s390_register_info): Enable FPR save slots.  Move/Copy some
4591         functionality into ...
4592         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
4593         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
4594         function.
4595         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
4596         not imply a stack frame.
4597         (s390_init_frame_layout): Remove variable clobbered_regs.
4598         (s390_update_register_info): Remove function.
4599         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
4600         cannot be used for register renaming.
4601         (s390_hard_regno_scratch_ok): New function.
4602         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
4603         (s390_initial_elimination_offset): Change offset calculation of
4604         the return address pointer.
4605         (save_gprs): Deal with only r6 being saved from the call-saved regs.
4606         (restore_gprs): Set frame related flag.
4607         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
4608         (s390_emit_prologue): Call s390_register_info instead of
4609         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
4610         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
4611         (s390_optimize_prologue): Call s390_optimize_register_info.
4612         Try to remove also FPR slot save/restore INSNs.  Remove frame
4613         related flags from restore INSNs.
4615 2013-10-08  DJ Delorie  <dj@redhat.com>
4617         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
4618         (movhi): Likewise.
4620         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
4621         avoid conflict with the MI use of %c.
4622         * config/rl78/rl78-real.md: change %c to %C throughout.
4623         * config/rl78/rl78-virt.md: Likewise.
4625 2013-10-08  Jan Hubicka  <jh@suse.cz>
4627         * config/i386/i386.c (ix86_option_override_internal): Switch
4628         to SSE math for -ffast-math when target ISA supports SSE2.
4630 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
4632         * tree-flow.h: Remove some prototypes.
4633         * tree.h: Remove some protypes, add a couple.
4634         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
4635         using_eh_for_cleanups_p): Add interface routines for front ends.
4636         * tree-eh.h: New file.  Add protoptyes.
4637         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
4638         (add_stmt_to_eh_lp_fn): Make static.
4639         (lower_try_finally): Use new using_eh_for_cleanups_p.
4640         * emit-rtl.c: Include tree-eh.h.
4641         * gimple.h: Include tree-eh.h.
4643 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
4645         PR tree-optimization/58480
4646         * tree-vrp.c (infer_nonnull_range): New function.
4647         (infer_value_range): Call infer_nonnull_range.
4649 2013-10-08  Dehao Chen  <dehao@google.com>
4651         PR tree-optimization/58619
4652         * tree-inline.c (copy_phis_for_bb): Combine location data
4653         only if non-null.
4655 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4657         PR target/58423
4658         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
4659         RTX_FRAME_RELATED_P on INSN.
4661 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4663         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
4664         (altivec_expand_vec_perm_const): Call it.
4666 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4668         * config/rs6000/vector.md (mov<mode>): Emit permuted move
4669         sequences for LE VSX loads and stores at expand time.
4670         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
4671         prototype.
4672         * config/rs6000/rs6000.c (rs6000_const_vec): New.
4673         (rs6000_gen_le_vsx_permute): New.
4674         (rs6000_gen_le_vsx_load): New.
4675         (rs6000_gen_le_vsx_store): New.
4676         (rs6000_gen_le_vsx_move): New.
4677         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
4678         (*vsx_le_perm_load_v4si): New.
4679         (*vsx_le_perm_load_v8hi): New.
4680         (*vsx_le_perm_load_v16qi): New.
4681         (*vsx_le_perm_store_v2di): New.
4682         (*vsx_le_perm_store_v4si): New.
4683         (*vsx_le_perm_store_v8hi): New.
4684         (*vsx_le_perm_store_v16qi): New.
4685         (*vsx_xxpermdi2_le_<mode>): New.
4686         (*vsx_xxpermdi4_le_<mode>): New.
4687         (*vsx_xxpermdi8_le_V8HI): New.
4688         (*vsx_xxpermdi16_le_V16QI): New.
4689         (*vsx_lxvd2x2_le_<mode>): New.
4690         (*vsx_lxvd2x4_le_<mode>): New.
4691         (*vsx_lxvd2x8_le_V8HI): New.
4692         (*vsx_lxvd2x16_le_V16QI): New.
4693         (*vsx_stxvd2x2_le_<mode>): New.
4694         (*vsx_stxvd2x4_le_<mode>): New.
4695         (*vsx_stxvd2x8_le_V8HI): New.
4696         (*vsx_stxvd2x16_le_V16QI): New.
4698 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
4700         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
4702 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4704         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
4705         loop to work also for 31bit ABI.
4706         Save the stack pointer for frame_size > 0.
4708 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4710         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
4711         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
4712         constraint letters from expanders.
4713         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
4714         retry count to general_operand.
4715         ("tabort"): Give operand 0 a mode.
4716         ("tabort_1"): Add mode and constraint letter for operand 0.
4717         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
4719 2013-10-04  Jeff Law  <law@redhat.com>
4721         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
4723         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
4724         out of ...
4725         (thread_across_edge): Here.  Call it.
4727 2013-10-04  Cary Coutant  <ccoutant@google.com>
4729         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
4730         discarding a location list expression (or a piece of one).
4732 2013-10-03  Jan Hubicka  <jh@suse.cz>
4734         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
4735         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
4736         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
4738 2013-10-03  Jan Hubicka  <jh@suse.cz>
4740         * config/i386/i386.c (ix86_option_override_internal): Do not enable
4741         accumulate-outgoing-args when producing unwind info.
4743 2013-10-03  Wei Mi  <wmi@google.com>
4745         * lra-constraints.c (insert_move_for_subreg): New function
4746         extracted from simplify_operand_subreg.
4747         (simplify_operand_subreg): Add reload for paradoxical subreg.
4749 2013-10-03  Rong Xu  <xur@google.com>
4751         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
4752         the candidate of builtin_expect such that we should fix the
4753         size/time estimation.
4754         (estimate_function_body_sizes): Do the acutally size/time fix-up
4755         for builtin_expect.
4757 2013-10-03  Rong Xu  <xur@google.com>
4759         * predict.c (tree_predict_by_opcode): Get the probability
4760         for builtin_expect from param builtin_expect_probability.
4761         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
4762         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
4763         * doc/invoke.texi: Add documentation for builtin-expect-probability.
4765 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
4767         PR c++/19476
4768         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
4769         * calls.c (alloca_call_p): Use get_callee_fndecl.
4770         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
4771         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
4772         Likewise.
4773         (vrp_visit_stmt): Remove duplicated code.
4775 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
4777         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
4778         ceildf2, btruncdf2, instead of vsx_* name.
4780         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
4781         iterators to only do V2DF and V4SF here.  Move the DF code to
4782         rs6000.md where it is combined with SF mode.  Replace <VSv> with
4783         just 'v' since only vector operations are handled with these insns
4784         after moving the DF support to rs6000.md.
4785         (vsx_sub<mode>3): Likewise.
4786         (vsx_mul<mode>3): Likewise.
4787         (vsx_div<mode>3): Likewise.
4788         (vsx_fre<mode>2): Likewise.
4789         (vsx_neg<mode>2): Likewise.
4790         (vsx_abs<mode>2): Likewise.
4791         (vsx_nabs<mode>2): Likewise.
4792         (vsx_smax<mode>3): Likewise.
4793         (vsx_smin<mode>3): Likewise.
4794         (vsx_sqrt<mode>2): Likewise.
4795         (vsx_rsqrte<mode>2): Likewise.
4796         (vsx_fms<mode>4): Likewise.
4797         (vsx_nfma<mode>4): Likewise.
4798         (vsx_copysign<mode>3): Likewise.
4799         (vsx_btrunc<mode>2): Likewise.
4800         (vsx_floor<mode>2): Likewise.
4801         (vsx_ceil<mode>2): Likewise.
4802         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
4803         (vsx_sminsf3): Likewise.
4804         (vsx_fmadf4): Likewise.
4805         (vsx_fmsdf4): Likewise.
4806         (vsx_nfmadf4): Likewise.
4807         (vsx_nfmsdf4): Likewise.
4808         (vsx_cmpdf_internal1): Likewise.
4810         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
4811         simpler to select whether a target has SPE or traditional floating
4812         point support in iterators.
4813         (TARGET_DF_SPE): Likewise.
4814         (TARGET_SF_FPR): Likewise.
4815         (TARGET_DF_FPR): Likewise.
4816         (TARGET_SF_INSN): Macros to say whether floating point support
4817         exists for a given operation for expanders.
4818         (TARGET_DF_INSN): Likewise.
4820         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
4821         combining of SF/DF mode operations, using both traditional and VSX
4822         registers.
4823         (Fvsx): Likewise.
4824         (Ff): Likewise.
4825         (Fv): Likewise.
4826         (Fs): Likewise.
4827         (Ffre): Likewise.
4828         (FFRE): Likewise.
4829         (abs<mode>2): Combine SF/DF modes using traditional floating point
4830         instructions.  Add support for using the upper DF registers with
4831         VSX support, and SF registers with power8-vector support.  Update
4832         expanders for operations supported by both the SPE and traditional
4833         floating point units.
4834         (abs<mode>2_fpr): Likewise.
4835         (nabs<mode>2): Likewise.
4836         (nabs<mode>2_fpr): Likewise.
4837         (neg<mode>2): Likewise.
4838         (neg<mode>2_fpr): Likewise.
4839         (add<mode>3): Likewise.
4840         (add<mode>3_fpr): Likewise.
4841         (sub<mode>3): Likewise.
4842         (sub<mode>3_fpr): Likewise.
4843         (mul<mode>3): Likewise.
4844         (mul<mode>3_fpr): Likewise.
4845         (div<mode>3): Likewise.
4846         (div<mode>3_fpr): Likewise.
4847         (sqrt<mode>3): Likewise.
4848         (sqrt<mode>3_fpr): Likewise.
4849         (fre<Fs>): Likewise.
4850         (rsqrt<mode>2): Likewise.
4851         (cmp<mode>_fpr): Likewise.
4852         (smax<mode>3): Likewise.
4853         (smin<mode>3): Likewise.
4854         (smax<mode>3_vsx): Likewise.
4855         (smin<mode>3_vsx): Likewise.
4856         (negsf2): Delete SF operations that are merged with DF.
4857         (abssf2): Likewise.
4858         (addsf3): Likewise.
4859         (subsf3): Likewise.
4860         (mulsf3): Likewise.
4861         (divsf3): Likewise.
4862         (fres): Likewise.
4863         (fmasf4_fpr): Likewise.
4864         (fmssf4_fpr): Likewise.
4865         (nfmasf4_fpr): Likewise.
4866         (nfmssf4_fpr): Likewise.
4867         (sqrtsf2): Likewise.
4868         (rsqrtsf_internal1): Likewise.
4869         (smaxsf3): Likewise.
4870         (sminsf3): Likewise.
4871         (cmpsf_internal1): Likewise.
4872         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
4873         (negdf2): Delete DF operations that are merged with SF.
4874         (absdf2): Likewise.
4875         (nabsdf2): Likewise.
4876         (adddf3): Likewise.
4877         (subdf3): Likewise.
4878         (muldf3): Likewise.
4879         (divdf3): Likewise.
4880         (fred): Likewise.
4881         (rsqrtdf_internal1): Likewise.
4882         (fmadf4_fpr): Likewise.
4883         (fmsdf4_fpr): Likewise.
4884         (nfmadf4_fpr): Likewise.
4885         (nfmsdf4_fpr): Likewise.
4886         (sqrtdf2): Likewise.
4887         (smaxdf3): Likewise.
4888         (smindf3): Likewise.
4889         (cmpdf_internal1): Likewise.
4890         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
4891         (btrunc<mode>2): Delete separate expander, and combine with the
4892         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
4893         (btrunc<mode>2_fpr): Likewise.
4894         (ceil<mode>2): Likewise.
4895         (ceil<mode>2_fpr): Likewise.
4896         (floor<mode>2): Likewise.
4897         (floor<mode>2_fpr): Likewise.
4898         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
4899         Add support for using the upper registers with VSX and
4900         power8-vector.  Move insns to be closer to the define_expands. On
4901         VSX systems, prefer the traditional form of FMA over the VSX
4902         version, since the traditional form allows the target not to
4903         overlap with the inputs.
4904         (fms<mode>4_fpr): Likewise.
4905         (nfma<mode>4_fpr): Likewise.
4906         (nfms<mode>4_fpr): Likewise.
4908 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4909             Richard Earnshaw  <richard.earnshaw@arm.com>
4911         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
4912         (struct mult_cost_table): Likewise.
4913         (struct mem_cost_table): Likewise.
4914         (struct fp_cost_table): Likewise.
4915         (struct vector_cost_table): Likewise.
4916         (cpu_cost_table): Likewise.
4917         * config/arm/arm.opt (mold-rts-costs): New option.
4918         (mnew-generic-costs): Likewise.
4919         * config/arm/arm.c (generic_extra_costs): New table.
4920         (cortexa15_extra_costs): Likewise.
4921         (arm_slowmul_tune): Use NULL as new costs.
4922         (arm_fastmul_tune): Likewise.
4923         (arm_strongarm_tune): Likewise.
4924         (arm_xscale_tune): Likewise.
4925         (arm_9e_tune): Likewise.
4926         (arm_v6t2_tune): Likewise.
4927         (arm_cortex_a5_tune): Likewise.
4928         (arm_cortex_a9_tune): Likewise.
4929         (arm_v6m_tune): Likewise.
4930         (arm_fa726te_tune): Likewise.
4931         (arm_cortex_a15_tune): Use cortex15_extra_costs.
4932         (arm_cortex_tune): Use generict_extra_costs.
4933         (shifter_op_p): New function.
4934         (arm_unspec_cost): Likewise.
4935         (LIBCALL_COST): Define.
4936         (arm_new_rtx_costs): New function.
4937         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
4938         table is available. Use old costs otherwise unless mnew-generic-costs
4939         is specified.
4940         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
4941         (cpu_cost_table): Declare.
4943 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4945         PR target/58460
4946         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
4947         (*subs_mul_imm_<mode>)
4948         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
4949         (*sub_<shift>_<mode>)
4950         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
4951         Remove k constraint.
4953 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
4955         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
4956         code.
4957         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
4959 2013-10-02  Teresa Johnson  <tejohnson@google.com>
4961         * predict.c (probably_never_executed): New function.
4962         (probably_never_executed_bb_p): Invoke probably_never_executed.
4963         (probably_never_executed_edge_p): Ditto.
4964         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4965         Treat profile insanities conservatively.
4967 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
4969         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
4971 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
4973         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
4974         use it.  Use smaller increase for scratch.  Don't increase reject
4975         for early clobber scratch.
4976         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
4977         setting eliminated regs except setting fp from hfp.
4978         (lra_eliminate): Check lra_insn_recog_data on NULL.
4980 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4982         PR target/58587
4983         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
4984         setting -mvsx-timode by default until the underlying problem is fixed.
4985         (RS6000_CPU, power7 defaults): Likewise.
4987 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
4989         * config/x-linux (host-linux.o): Remove header dependencies.
4990         Use $(COMPILE) and $(POSTCOMPILE).
4991         * config/t-linux-android (linux-android.o): Ditto.
4993 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
4995         * Makefile.in (expmed.o-warn): Remove.
4997 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
4999         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
5000         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
5002 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5004         * dojump.c (do_jump_1): Divide probability between
5005         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
5007 2013-10-02  Tom Tromey  <tromey@redhat.com>
5009         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
5011 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5013         * tree-flow.h: Remove some prototypes.
5014         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
5015         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
5016         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
5017         mark_virtual_phi_result_for_renaming): Relocate here.
5018         * tree-into-ssa.h: Add prototypes.
5019         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
5020         single_pred_before_succ_order.
5021         (blocks_in_phiopt_order): Rename and move to cfganal.c.
5022         (nonfreeing_call_p) Move to gimple.c.
5023         * cfganal.c (single_pred_before_succ_order): Move and renamed from
5024         tree-ssa-phiopt.c.
5025         * basic-block.h (single_pred_before_succ_order): Add prototype.
5026         * gimple.c (nonfreeing_call_p): Relocate here.
5027         * gimple.h: Add prototype.
5028         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
5029         * tree-ssa-dom.h: New file.  Relocate prototypes here.
5030         * tree-ssa.h: Include tree-ssa-dom.h.
5032 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5034         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
5035         Use $(COMPILE) and $(POSTCOMPILE).
5037         * config/alpha/x-alpha (driver-alpha.o): Ditto.
5039 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5041         * tree-flow.h: Remove some prototypes.
5042         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
5043         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5044         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
5045         propagate_tree_value*): Move from here to...
5046         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
5047         propagate_tree_value*): Relocate here.
5048         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5049         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
5050         * gimple-fold.c: Remove gimple-fold.h from include list.
5051         * tree-vrp.c: Remove gimple-fold.h from include list.
5052         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
5053         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
5054         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
5055         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
5056         * sese.c: Add tree-ssa-propagate.h to include list.
5058 2013-10-02  Richard Biener  <rguenther@suse.de>
5060         * tree-loop-distribution.c: Include tree-vectorizer.h for
5061         find_loop_location.
5062         (enum partition_kind): Remove PKIND_REDUCTION.
5063         (struct partition_s): Remove has_writes member, add reduction_p member.
5064         (partition_alloc): Adjust.
5065         (partition_builtin_p): Likewise.
5066         (partition_has_writes): Remove.
5067         (partition_reduction_p): New function.
5068         (partition_merge_into): Likewise.
5069         (generate_code_for_partition): Commonize builtin partition
5070         handling tail.
5071         (rdg_cannot_recompute_vertex_p): Remove.
5072         (already_processed_vertex_p): Likewise.
5073         (rdg_flag_vertex): Do not set has_writes.
5074         (classify_partition): Adjust.
5075         (rdg_build_partitions): Do not set has_writes, treat all
5076         partitions as useful.
5077         (distribute_loop): Record number of library calls generated.  Adjust.
5078         (tree_loop_distribution): Report number of loops and library
5079         calls generated as opt-info.
5081 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5083         * tree-flow.h: Include new .h files.  Move prototypes.
5084         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
5085         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
5086         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
5087         * tree-pretty-print.h: Add prototypes from tree-flow.h.
5088         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
5089         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
5090         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
5091         * tree.h (get_ref_base_and_extent): Move prototype out.
5092         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
5093         tree-dfa.h.
5094         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
5095         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
5096         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
5097         * tree-scalar-evolution.c: Include tree.h.
5098         * sese.c: Include tree.h.
5099         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
5100         * dwarf2out.c: Include tree-dfa.h.
5101         * tree-chrec.c: Include tree.h.
5102         * tree-data-ref.c: Include tree.h.
5104 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
5106         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
5107         Fix whitespace.
5109 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5111         * config/t-sol2 (sol2-c.o): Remove header dependencies.
5112         Use $(COMPILE) and $(POSTCOMPILE).
5113         (sol2-cxx.o): Likewise.
5114         (sol2-stubs.o): Likewise.
5115         (sol2.o): Likewise.
5116         * config/x-solaris (host-solaris.o): Likewise.
5118         * config/sparc/t-sparc (sparc.o): Remove.
5119         (sparc-c.o): Remove header dependencies.
5120         Use $(COMPILE) and $(POSTCOMPILE).
5121         * config/sparc/x-sparc: Likewise.
5123 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
5125         * config/arc/arc-opts.h: Add 2013 to Copyright years.
5126         * config/arc/arc700.md: Likewise.
5127         * config/arc/arc-modes.def: Likewise.
5128         * config/arc/arc-simd.h: Likewise.
5129         * config/arc/t-arc-uClibc: Likewise.
5130         * config/arc/t-arc-newlib: Likewise.
5132 2013-10-02  Renlin Li  <renlin.li@arm.com>
5134         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5135         plus_constant.
5136         (aarch64_expand_epilogue): Likewise.
5138 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5139             Yufeng Zhang  <yufeng.zhang@arm.com>
5141         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
5142         declaration.
5143         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
5144         'base_in' represent a conversion and legal_cast_p_1 holds; set
5145         'base_in' with the returned value from get_unwidened.
5147 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5149         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
5150         plus_constant is not used.
5152 2013-10-01  Wei Mi  <wmi@google.com>
5154         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
5155         * config/i386/i386.md: Add define_peephole2 to
5156         break partial reg stall for cvtss2sd/cvtsd2ss.
5158 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5160         * config/arc/arc.c (pass_arc_ifcvt::clone):
5161         Update for ctxt_ -> m_ctxt change.
5163 2013-10-01  Jeff Law  <law@redhat.com>
5165         * tree-ssa-threadupdate.c (struct redirection_data): Delete
5166         outgoing_edge and intermediate_edge fields.  Instead store the path.
5167         (redirection_data::hash): Hash on the last edge's destination index.
5168         (redirection_data::equal): Check the entire thread path.
5169         (lookup_redirectio_data): Corresponding changes.
5170         (create_edge_and_update_destination_phis): Likewise.
5171         (thread_single_edge): Likewise.
5173 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5174             Diego Novillo <dnovillo@google.com>
5176         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
5177         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
5178         (vld32wh_insn, vld32wl_insn): Delete commented-out old
5179         versions of these patterns.
5181         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
5182         (__builtin_arc_aligned): Likewise.
5184         * config/arc/arc.md: Expand adc_0 comment stating the intended
5185         purpose and why it isn't ready.
5186         Replace commented out call_value_via_label_mixed with a
5187         plain comment about bl_s.
5189         * config/arc/arc.c (stdio.h): Don't include directly.
5190         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
5191         Remove if (1) condition.
5192         (arc_encode_section_info): Fix comment.
5194 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5196         * config/arc/arc.c (arc_conditional_register_usage):
5197         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
5198         Also set reg_alloc_order for DMA config regs.
5200 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5201             Jeremy Bennett  <jeremy.bennett@embecosm.com>
5203         * doc/install.texi (--with-cpu): Mention ARC.
5204         (arc-*-elf32): New paragraph.
5205         (arc-linux-uclibc): Likewise.
5206         * doc/md.texi (Machine Constraints): Add ARC part.
5207         * doc/invoke.texi: (menu): Add ARC Options.
5208         (Machine Dependent Options) <ARC Options>: Add synopsis.
5209         (node ARC Options): Add.
5210         * doc/extend.texi (long_call / short_call attribute): Add ARC.
5211         (ARC Built-in Functions): New section defining
5212         generic ARC built-in functions.
5213         (ARC SIMD Built-in Functions): New section defining SIMD specific
5214         built-in functions.
5215         (Declaring Attributes of Functions): Extended
5216         description of short_call and long_call attributes for ARC and
5217         added index entries.
5219 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
5220             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
5221             Joern Rennecke  <joern.rennecke@embecosm.com>
5222             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
5223             Brendan Kehoe  <brendan@zen.org>
5224             Michael Eager  <eager@eagercon.com>
5225             Simon Cook  <simon.cook@embecosm.com>
5226             Jeremy Bennett  <jeremy.bennett@embecosm.com>
5228         * config/arc, common/config/arc: New directories.
5230 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5231             Brendan Kehoe  <brendan@zen.org>
5232             Simon Cook  <simon.cook@embecosm.com>
5234         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
5236 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
5238         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
5239         * tree-ssa-coalesce.h: New. Move prototype to here.
5240         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
5241         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
5242         (gimple_can_coalesce_p): Move to...
5243         * gimple.c (gimple_can_coalesce_p): Here.
5245 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
5247         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
5248         (dump_decl_set): Move to gimple.c.
5249         * gimple.h: Don't include tree-ssa-operands.h.
5250         (dump_decl_set): Add prototype.
5251         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
5252         Move to gimple-ssa.h.
5253         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
5254         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
5255         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
5256         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
5257         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
5258         than PHI_ARG_DEF.
5259         (dump_decl_set): Relocate here.
5260         * gimple-ssa.h: New file.
5261         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
5262         Relocate from gimple.h.
5263         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
5264         * tree-ssa-operands.c (swap_ssa_operands): Rename from
5265         swap_tree_operands and remove non-ssa path.
5266         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
5267         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
5268         swap_ssa_operands.
5269         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
5270         vect_is_simple_reduction_1): Use swap_ssa_operands.
5271         * tree-flow.h: Move various prototypes to tree-phinodes.h.
5272         (enum need_phi_state): Move to tree-into-ssa.c.
5273         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
5274         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
5275         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
5276         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
5277         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
5278         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
5279         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
5280         num_imm_uses): Move to ssa-iterators.h.
5281         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
5282         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
5283         tree-phinodes.h.
5284         (op_iter_done, op_iter_next_def, op_iter_next_tree,
5285         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
5286         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
5287         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
5288         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
5289         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
5290         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
5291         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
5292         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
5293         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
5294         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
5295         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
5296         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
5297         (set_phi_nodes): Move to tree-phinodes.h.
5298         * tree-ssa-operands.h (enum ssa_op_iter_type,
5299         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
5300         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
5301         (dump_decl_set): Remove prototype.
5302         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
5303         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
5304         (set_phi_nodes): Relocate from tree-flow-inline.h.
5305         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
5306         tree-flow-inline.h
5307         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
5308         include list.  Temporarily add gimple.h to include list.
5309         * ssa-iterators.h: New file.
5310         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
5311         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
5312         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
5313         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
5314         Relocate from tree-ssa-operands.h.
5315         (delink_imm_use, link_imm_use_to_list, link_imm_use,
5316         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
5317         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
5318         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
5319         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
5320         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
5321         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
5322         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
5323         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
5324         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
5325         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
5326         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
5327         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
5328         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
5329         Relocate from tree-flow-inline.h.
5330         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
5332 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
5334         * aarch64-simd.md
5335         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
5336         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
5337         Insert '\t' to output template.
5338         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
5339         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
5340         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
5341         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
5343 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
5345         * doc/install.texi (Host/target specific installation notes for GCC):
5346         Put @anchor before @heading.
5347         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
5348         Use @email for email addresses.
5350 2013-10-01  Jeff Law  <law@redhat.com>
5352         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
5353         a vec.  Only delete the path if we create one without successfully
5354         registering a jump thread.
5355         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
5356         as a pointer.
5357         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
5358         (paths): New vector of jump threading paths.
5359         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
5360         (THREAD_PATH): New accessor macro for the entire thread path.
5361         (lookup_redirection_data): Get intermediate and final outgoing edge
5362         from the thread path.
5363         (create_edge_and_update_destination_phis): Copy the threading path.
5364         (ssa_fix_duplicate_block_edges): Get edges and block types from the
5365         jump threading path.
5366         (ssa_redirect_edges): Get edges and block types from the jump threading
5367         path.  Free the path vector.
5368         (thread_block): Get edges from the jump threading path.  Look at the
5369         entire path to see if we thread to a loop exit.  If we cancel a jump
5370         thread request, then free the path vector.
5371         (thread_single_edge): Get edges and block types from the jump threading
5372         path.  Free the path vector.
5373         (thread_through_loop_header): Get edges and block types from the jump
5374         threading path.  Free the path vector.
5375         (mark_threaded_blocks): Iterate over the vector of paths and store
5376         the path on the appropriate edge.  Get edges and block types from the
5377         jump threading path.
5378         (mark_threaded_blocks): Get edges and block types from the jump
5379         threading path.  Free the path vector.
5380         (thread_through_all_blocks): Use the vector of paths rather than
5381         a vector of 3-edge sets.
5382         (register_jump_thread): Accept pointer to a path vector rather
5383         than the path vector itself.  Store the path vector for later use.
5384         Simplify.
5386 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
5387             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5389         PR target/58574
5390         * config/s390/s390.c (s390_split_branches): Modify check for table
5391         jump insns.
5392         (s390_chunkify_start): Rearrange table jump insn check in order to
5393         deal with compare and branch insns correctly.
5395 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
5397         PR target/58578
5398         Revert
5399         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
5400         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
5401         define_insn_and_split.
5402         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
5403         (shiftsi3_compare): New pattern.
5404         (rrx): New pattern.
5405         * config/arm/unspecs.md (UNSPEC_RRX): New.
5407 2013-10-01  Alan Modra  <amodra@gmail.com>
5409         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
5410         casing inout operands.
5412 2013-10-01  Richard Biener  <rguenther@suse.de>
5414         PR tree-optimization/58553
5415         * tree-loop-distribution.c (struct partition_s): Add niter member.
5416         (classify_partition): Populate niter member for the partition
5417         and properly identify whether the relevant store happens before
5418         or after the loop exit.
5419         (generate_memset_builtin): Use niter member from the partition.
5420         (generate_memcpy_builtin): Likewise.
5422 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
5424         * vec.h (vec_prefix, vec): Prefix member names with "m_".
5425         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
5427 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
5429         * basic-block.h (edge_list): Prefix member names with "m_".
5430         * context.h (context): Likewise.
5431         * domwalk.h (dom_walker): Likewise.
5432         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
5433         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
5434         * hash-table.h (hash_table): Likewise.
5435         * machmode.h (bit_field_mode_iterator): Likewise.
5436         * pass_manager.h (pass_list): Likewise.
5437         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
5438         * tree-pass.h (pass_data): Likewise.
5439         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
5440         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
5441         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
5442         * asan.c (pass_data_asan): Update accordingly.
5443         * cfganal.c (control_dependences::find_control_dependence): Likewise.
5444         (control_dependences::control_dependences): Likewise.
5445         (control_dependences::~control_dependences): Likewise.
5446         (control_dependences::~control_dependences): Likewise.
5447         (control_dependences::get_edges_dependent_on): Likewise.
5448         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
5449         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
5450         * context.c (gcc::context::context): Likewise.
5451         * cprop.c (pass_rtl_cprop::clone): Likewise.
5452         * domwalk.c (dom_walker::walk): Likewise.
5453         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
5454         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
5455         * mode-switching.c (pass_mode_switching::clone): Likewise.
5456         * passes.c (opt_pass::opt_pass): Likewise.
5457         (pass_manager::pass_manager): Likewise.
5458         * predict.c (pass_strip_predict_hints::clone): Likewise.
5459         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
5460         (pass_split_all_insns::clone): Likewise.
5461         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5462         Likewise.
5463         (bit_field_mode_iterator::next_mode): Likewise.
5464         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
5465         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
5466         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
5467         * tree-complex.c (pass_lower_complex::clone): Likewise.
5468         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
5469         * tree-object-size.c (pass_object_sizes::clone): Likewise.
5470         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
5471         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
5472         (pass_fold_builtins::clone): Likewise.
5473         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
5474         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
5475         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
5476         (pass_cd_dce::clone): Likewise.
5477         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
5478         (pass_phi_only_cprop::clone): Likewise.
5479         * tree-ssa-dse.c (pass_dse::clone): Likewise.
5480         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
5481         * tree-ssa-loop.c (pass_lim::clone): Likewise.
5482         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
5483         * tree-ssa-pre.c (pass_fre::clone): Likewise.
5484         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
5485         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
5486         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
5487         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
5488         * tree-vrp.c (pass_vrp::clone): Likewise.
5489         * tsan.c (pass_tsan::clone): Likewise.
5491 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
5493         PR middle-end/58564
5494         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
5495         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
5497         PR middle-end/58564
5498         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
5499         optimization, punt if sign_bit_p looked through any zero extension.
5501 2013-09-30  Teresa Johnson  <tejohnson@google.com>
5503         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
5504         Update redirected out edge count in joiner case.
5505         (ssa_redirect_edges): Common the joiner and non-joiner cases
5506         so that joiner case gets profile updates.
5508 2013-09-30  Richard Biener  <rguenther@suse.de>
5510         PR tree-optimization/58554
5511         * tree-loop-distribution.c (classify_partition): Require
5512         unconditionally executed stores for memcpy and memset recognition.
5513         (tree_loop_distribution): Calculate dominance info.
5515 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
5517         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
5518         (NO_PROFILE_COUNTERS): Likewise.
5519         (PROFILE_HOOK): Likewise.
5520         (FUNCTION_PROFILER): Likewise.
5521         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
5523 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
5525         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
5526         calls and defines in TARGET_MACHO conditional.
5527         (load_macho_picbase_di): Likewise.
5528         (reload_macho_picbase): Likewise.
5529         (reload_macho_picbase_si): Likewise.
5530         (reload_macho_picbase_di): Likewise.
5531         (nonlocal_goto_receiver): Likewise.
5533 2013-09-30  Nick Clifton  <nickc@redhat.com>
5535         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
5536         that use the MSP430X ISA.
5537         (msp430_option_override): Scan -mmcu command line option for any
5538         MCU name that supports the MSP430X ISA.
5539         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
5540         -mmcu options which enable the MSP430X ISA.
5542 2013-09-30  Richard Biener  <rguenther@suse.de>
5544         PR middle-end/58532
5545         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
5546         before looking for setjmp-like calls.
5548 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
5550         PR target/10901
5551         * config/darwin-protos.h (machopic_get_function_picbase): New.
5552         * config/darwin.c (machopic_get_function_picbase): New.
5553         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
5554         label for a new func.  (load_macho_picbase_di): Likewise.
5555         (reload_macho_picbase): New expand.
5556         (reload_macho_picbase_si): New insn.
5557         (reload_macho_picbase_di): New insn.
5558         (nonlocal_goto_receiver): New define and split.
5559         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
5560         (unspecv enum): Add UNSPECV_NLGR.
5562 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
5564         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
5565         that altivec registers are correctly sized on Darwin.
5567 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
5569         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
5570         darwin-driver.o): Use COMPILE and POSTCOMPILE.
5571         * config/x-darwin (host-darwin.o): Likewise.
5572         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
5573         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
5574         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
5576 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
5578         * doc/invoke.texi: Fix usage of @tie{} command.
5580 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
5582         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
5584 2013-09-28  Jan Hubicka  <jh@suse.cz>
5586         * config/i386/x86-tune.def: Add documentation for each of the options;
5587         add whitespace.
5589 2013-09-28  Jan Hubicka  <jh@suse.cz>
5591         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
5592         generic.
5593         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
5594         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
5595         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
5596         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
5597         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
5599 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
5601         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
5602         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
5603         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
5604         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
5605         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
5606         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
5607         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
5608         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
5609         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
5610         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
5611         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
5612         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
5613         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
5614         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
5615         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
5616         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
5617         gimple.h, godump.c, graphite-clast-to-gimple.c,
5618         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
5619         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
5620         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
5621         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
5622         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
5623         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
5624         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
5625         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
5626         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
5627         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
5628         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
5629         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
5630         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
5631         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
5632         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
5633         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
5634         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
5635         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
5636         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
5637         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
5638         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
5639         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
5640         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
5641         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
5642         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
5643         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
5644         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
5645         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
5646         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
5647         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
5648         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
5649         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
5650         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
5651         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
5652         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5653         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
5654         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
5655         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
5656         whitespace before "(".
5658 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
5660         * expr.h (extract_bit_field): Remove packedp parameter.
5661         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
5662         from forward declaration.
5663         (store_split_bit_field): Remove packedp arg from calls to
5664         extract_fixed_bit_field.
5665         (extract_bit_field_1): Remove packedp parameter and packedp
5666         argument from recursive calls and calls to extract_fixed_bit_field.
5667         (extract_bit_field): Remove packedp parameter and corresponding
5668         arg to extract_bit_field_1.
5669         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
5670         to issue warnings.
5671         (extract_split_bit_field): Remove packedp arg from call to
5672         extract_fixed_bit_field.
5673         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
5674         (copy_blkmode_from_reg): Likewise.
5675         (copy_blkmode_to_reg): Likewise.
5676         (read_complex_part): Likewise.
5677         (store_field): Likewise.
5678         (expand_expr_real_1): Likewise.
5679         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
5680         to extract_bit_field.
5681         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
5682         call to extract_bit_field.
5683         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
5684         call to extract_bit_field.
5685         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
5686         and special packedp behavior from -fstrict-volatile-bitfields
5687         documentation.
5689 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5691         * lra-eliminations.c (init_elim_table): Guard value_p.
5693 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
5695         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
5696         DFmode, DImode, and SFmode in the upper VSX registers based on the
5697         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
5698         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
5699         of the VSX register handling.
5701         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
5702         (f32_sv): Likewise.
5703         (zero_extendsidi2_lfiwzx): Add support for loading into the
5704         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
5705         only do VSX memory options on Altivec registers.
5706         (extendsidi2_lfiwax): Likewise.
5707         (extendsfdf2_fpr): Likewise.
5708         (mov<mode>_hardfloat, SF/SD modes): Likewise.
5709         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
5710         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
5711         (movdi_internal64): Likewise.
5713 2013-09-27  Xinliang David Li  <davidxl@google.com>
5715         * opts.c (finish_options): Adjust parameters
5716         according to vect cost model.
5717         (common_handle_option): Set dynamic vect cost
5718         model for FDO.
5719         targhooks.c (default_add_stmt_cost): Compute stmt cost
5720         unconditionally.
5721         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
5722         Use helper function.
5723         * tree-vectorizer.h (unlimited_cost_model): New function.
5724         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
5725         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
5726         function.
5727         (vect_enhance_data_refs_alignment): Ditto.
5728         * flag-types.h: New enum.
5729         * common/config/i386/i386-common.c (ix86_option_init_struct):
5730         No need to initialize vect_cost_model flag.
5731         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
5732         unconditionally.
5734 2013-09-27  Diego Novillo  <dnovillo@google.com>
5736         * gimple.h (enum ssa_mode): Remove.
5738 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
5740         * cfgloop.h (number_of_loops): Fix typo in check for null.
5742 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
5744         PR middle-end/58551
5745         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
5746         are children of outermost saved_cfun's loop, and set it up to
5747         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
5748         if loop != loop0 and SESE region contains bbs that belong to loop0.
5750 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
5752         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
5753         (binary_scale_code_p, get_base_term, get_index_term): New functions.
5754         (set_address_segment, set_address_base, set_address_index)
5755         (set_address_disp): Accept the argument unconditionally.
5756         (baseness): Remove must_be_base_p and must_be_index_p checks.
5757         (decompose_normal_address): Classify as much as possible in the
5758         main loop.
5760 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
5762         * cse.c (count_reg_usage): Handle INT_LIST.
5763         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
5764         * reginfo.c (reg_scan_mark_refs): Likewise.
5765         * reload1.c (eliminate_regs_1): Likewise.
5767 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
5769         PR middle-end/58547
5770         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
5771         signed.
5773 2013-09-27  Richard Biener  <rguenther@suse.de>
5775         PR tree-optimization/58459
5776         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
5777         restriction not propagating into loops.
5779 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
5781         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
5782         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
5783         * doc/tree-ssa.texi (Walking use-def chains): Delete.
5785 2013-09-26  Richard Biener  <rguenther@suse.de>
5787         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
5789 2013-09-26  Richard Biener  <rguenther@suse.de>
5791         * alias.h (component_uses_parent_alias_set): Rename to ...
5792         (component_uses_parent_alias_set_from): ... this.
5793         * alias.c (component_uses_parent_alias_set): Rename to ...
5794         (component_uses_parent_alias_set_from): ... this and return
5795         the desired parent.
5796         (reference_alias_ptr_type_1): Use the result from
5797         component_uses_parent_alias_set_from instead of stripping
5798         components one at a time.
5799         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
5801 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
5803         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
5804         Move prototypes to...
5805         * tree-ssa-ter.h: New File.  Move prototypes here.
5806         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
5807         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
5808         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
5809         from is_replaceable_p.
5810         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
5811         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
5812         (process_replaceable): Use ter_is_replaceable_p.
5813         (find_replaceable_in_bb): Use ter_is_replaceable_p.
5814         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
5815         newly refactored ssa_is_replaceable_p.
5816         * cfgexpand.c: Include tree-outof-ssa.h.
5817         * ssaexpand.h: Delete.
5819 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
5821         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
5822         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
5823         tree-ssa.c
5824         (create_gimple_tmp): Delete.
5825         (get_expr_type, build_assign, build_type_cast): Move to...
5826         * gimple-builder.c: New File.
5827         (get_expr_type): Relocate from gimple.c.
5828         (build_assign, build_type_cast): Change to only create ssanames.
5829         * gimple.h: Move prototypes to...
5830         * gimple-builder.h: New File. Here.
5831         * tree-ssa.h: And here.
5832         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
5833         count_uses_and_derefs): Relocate from gimple.c.
5834         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
5835         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
5836         * tree-ssa-math-opts (execute_cse_reciprocals): Use
5837         gimple_replace_ssa_lhs.
5838         * asan.c: Include gimple-builder.h.
5839         * Makefile.in: Add gimple-builder.o.
5841 2013-09-26  Richard Biener  <rguenther@suse.de>
5843         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
5844         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
5845         (loe_visit_block): Use gcc_checking_assert.
5846         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
5847         gimple_assign_ssa_name_copy_p.
5848         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
5849         change.
5851 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
5853         * config/rs6000/t-rs6000 (rs6000.o): Remove.
5854         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
5856 2013-09-26  Richard Biener  <rguenther@suse.de>
5858         PR tree-optimization/58539
5859         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
5860         the fact that debug statements are not taking part in loop-closed
5861         SSA construction.
5863 2013-09-26  Nick Clifton  <nickc@redhat.com>
5865         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
5866         time warning message.
5867         (msp430_print_operand_raw): Delete unused letter parameter.
5868         (TARGET_PRINT_OPERAND_ADDRESS): Define.
5869         (msp430_print_operand_address): New function.
5870         (msp430_print_operand): Move address printing code from here to
5871         new function.
5872         * config/msp430/msp430.md (movsipsi2): Add comment in generated
5873         assembler.
5874         (zero_extendpsisi2): Likewise.
5875         (extendpsisi2): New pattern.
5876         (andneghi3): New pattern.
5878 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
5880         * config/aarch64/aarch64.opt (mlra): New option.
5881         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
5882         (TARGET_LRA_P): Define.
5884 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
5886         * expr.c (expand_assignment): Remove obsolete comment.
5888 2013-09-25  Jeff Law  <law@redhat.com>
5890         * tree-flow.h (thread_through_all_blocks): Prototype moved into
5891         tree-ssa-threadupdate.h.
5892         (register_jump_thread): Similarly.
5893         * tree-ssa-threadupdate.h: New header file.
5894         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
5895         * tree-vrp.c: Likewise.
5896         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
5897         (thread_around_empty_blocks): Change type of path vector argument to
5898         an edge,type pair from just an edge.  Initialize both elements when
5899         appending to a jump threading path.  Tweak references to elements
5900         appropriately.
5901         (thread_across_edge): Similarly.  Release memory for the elements
5902         as needed.
5903         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
5904         (dump_jump_thread_path): New function broken out from
5905         register_jump_thread.
5906         (register_jump_thread): Use dump_jump_thread_path.  Change type of
5907         path vector entries.  Search the path for NULL edges and dump
5908         the path if one is found.  Tweak the conversion of path to 3-edge
5909         form to use the block copy type information embedded in the path.
5911 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
5913         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
5915 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
5916             Vladimir Makarov  <vmakarov@redhat.com>
5918         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
5919         from the least significant bit.
5920         (strip_address_mutations): Add bitfield operations handling.
5921         (must_be_index_p): Add shifting and rotate operations handling.
5922         (set_address_base): Use must_be_base_p predicate.
5923         (set_address_index): Use must_be_index_p predicate.
5925 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5926             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5927             Sergey Lega  <sergey.s.lega@intel.com>
5928             Anna Tikhonova  <anna.tikhonova@intel.com>
5929             Ilya Tocar  <ilya.tocar@intel.com>
5930             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5931             Ilya Verbin  <ilya.verbin@intel.com>
5932             Kirill Yukhin  <kirill.yukhin@intel.com>
5933             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5935         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
5936         Use new names.
5937         (ix86_expand_vector_move_misalign): Support new unaligned load and
5938         stores and use new names.
5939         (CODE_FOR_sse2_storedqu): Rename to ...
5940         (CODE_FOR_sse2_storedquv16qi): ... this.
5941         (CODE_FOR_sse2_loaddqu): Rename to ...
5942         (CODE_FOR_sse2_loaddquv16qi): ... this.
5943         (CODE_FOR_avx_loaddqu256): Rename to ...
5944         (CODE_FOR_avx_loaddquv32qi): ... this.
5945         (CODE_FOR_avx_storedqu256): Rename to ...
5946         (CODE_FOR_avx_storedquv32qi): ... this.
5947         * config/i386/i386.md (fpint_logic): New.
5948         * config/i386/sse.md (VMOVE): Extend for AVX512.
5949         (VF): Ditto.
5950         (VF_128_256): New.
5951         (VF_512): Ditto.
5952         (VI_UNALIGNED_LOADSTORE): Ditto.
5953         (sse2_avx_avx512f): Ditto.
5954         (sse2_avx2): Extend for AVX512.
5955         (sse4_1_avx2): Ditto.
5956         (avx2_avx512f): New.
5957         (sse): Extend for AVX512.
5958         (sse2): Ditto.
5959         (sse4_1): Ditto.
5960         (avxsizesuffix): Ditto.
5961         (sseintvecmode): Ditto.
5962         (ssePSmode): Ditto.
5963         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
5964         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
5965         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
5966         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
5967         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
5968         (<sse2_avx_avx512f>_storedqu<mode): ... this.
5969         (<sse>_movnt<mode>): Replace constraint "x" with "v".
5970         (STORENT_MODE): Extend for AVX512.
5971         (*absneg<mode>2): Replace constraint "x" with "v".
5972         (*mul<mode>3): Ditto.
5973         (*ieee_smin<mode>3): Ditto.
5974         (*ieee_smax<mode>3): Ditto.
5975         (avx_cmp<mode>3): Replace VF with VF_128_256.
5976         (*<sse>_maskcmp<mode>3_comm): Ditto.
5977         (<sse>_maskcmp<mode>3): Ditto.
5978         (<sse>_andnot<mode>3): Extend for AVX512.
5979         (<code><mode>3, anylogic): Replace VF with VF_128_256.
5980         (<code><mode>3, fpint_logic): New.
5981         (*<code><mode>3): Extend for AVX512.
5982         (avx512flogicsuff): New.
5983         (avx512f_<logic><mode>): Ditto.
5984         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
5985         VF_128_256.
5986         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
5987         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
5988         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
5989         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
5990         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
5991         (xop_vpermil2<mode>3): Ditto.
5992         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
5993         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
5994         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
5995         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
5997 2013-09-25  Tom Tromey  <tromey@redhat.com>
5999         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
6000         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
6001         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
6002         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
6003         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
6004         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
6005         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
6006         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
6007         (GRAPHITE_HTAB_H): Remove.
6009 2013-09-25  Tom Tromey  <tromey@redhat.com>
6011         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
6013 2013-09-25  Tom Tromey  <tromey@redhat.com>
6015         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
6017 2013-09-25  Tom Tromey  <tromey@redhat.com>
6019         * config/i386/t-i386 (i386.o): Remove.
6020         (i386-c.o): Use COMPILE and POSTCOMPILE.
6022 2013-09-25  Tom Tromey  <tromey@redhat.com>
6024         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
6026 2013-09-25  Tom Tromey  <tromey@redhat.com>
6028         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
6029         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
6030         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
6031         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
6032         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
6033         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
6034         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
6035         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
6036         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
6037         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
6038         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
6039         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
6040         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
6041         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
6042         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
6043         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
6044         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
6045         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
6046         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
6047         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
6048         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
6049         (tree-ssa-pre.o, tree-ssa-sccvn.o)
6050         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
6051         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
6052         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
6053         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
6054         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
6055         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
6056         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
6057         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
6058         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
6059         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
6060         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
6061         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
6062         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
6063         (graphite-clast-to-gimple.o, graphite-dependences.o)
6064         (graphite-interchange.o, graphite-poly.o)
6065         (graphite-scop-detection.o, graphite-sese-to-poly.o)
6066         (graphite-optimize-isl.o, tree-vect-loop.o)
6067         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
6068         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
6069         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
6070         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
6071         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
6072         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
6073         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
6074         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
6075         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
6076         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
6077         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
6078         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
6079         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
6080         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
6081         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
6082         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
6083         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
6084         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
6085         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
6086         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
6087         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
6088         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
6089         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
6090         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
6091         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
6092         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
6093         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
6094         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
6095         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
6096         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
6097         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
6098         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
6099         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
6100         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
6101         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
6102         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
6103         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
6104         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
6105         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
6106         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
6107         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
6108         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
6109         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
6110         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
6111         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
6112         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
6113         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
6114         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
6115         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
6116         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
6117         (gcov-dump.o): Remove.
6118         (default-c.o): Use COMPILE and POSTCOMPILE.
6119         (CFLAGS-gcc.o): New variable.
6120         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
6122 2013-09-25  Tom Tromey  <tromey@redhat.com>
6124         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
6125         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
6126         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
6127         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
6128         (c-family/c-pragma.o, c-family/c-pretty-print.o)
6129         (c-family/c-semantics.o, c-family/c-ada-spec.o)
6130         (c-family/array-notation-common.o, c-family/stub-objc.o)
6131         (c-family/c-ubsan.o): Remove.
6133 2013-09-25  Tom Tromey  <tromey@redhat.com>
6135         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
6137 2013-09-25  Tom Tromey  <tromey@redhat.com>
6139         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
6140         to add -DENABLE_SHARED_LIBGCC.
6141         (gcc.o): Don't use subshell.
6143 2013-09-25  Tom Tromey  <tromey@redhat.com>
6145         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
6146         * configure.ac: Don't invoke AM_PROG_CC_C_O.
6147         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
6148         * configure, config.in: Rebuild.
6150 2013-09-25  Tom Tromey  <tromey@redhat.com>
6152         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
6153         (COMPILE, POSTCOMPILE): New variables.
6154         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
6155         (DEPFILES): New variable.
6156         Include ".Po" files.
6157         * configure.ac: Add checks for dependency checking.
6158         * configure, aclocal.m4: Regenerate.
6160 2013-09-25  Tom Tromey  <tromey@redhat.com>
6162         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
6163         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
6165 2013-09-25  Tom Tromey  <tromey@redhat.com>
6167         * Makefile.in (generated_files): Add options.h,
6168         target-hooks-def.h, insn-opinit.h,
6169         common/common-target-hooks-def.h, pass-instances.def,
6170         c-family/c-target-hooks-def.h.
6172 2013-09-25  Jeff Law  <law@redhat.com>
6174         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
6175         than foo[foo.length () - 1] to access last member in a vec.
6176         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
6178 2013-09-25  Richard Biener  <rguenther@suse.de>
6180         PR middle-end/58521
6181         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
6183 2013-09-25  Jan Hubicka  <jh@suse.cz>
6185         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
6186         test.
6188 2013-09-25  Marek Polacek  <polacek@redhat.com>
6190         PR sanitizer/58420
6191         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
6192         when determining the type name.
6194 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
6196         * config/sh/sh.md: Fix formatting.
6198 2013-09-24  Xinliang David Li  <davidxl@google.com>
6200         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
6201         max peel iterations parameter.
6202         * param.def: New parameter.
6203         * doc/invoke.texi: Document New parameter.
6205 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
6207         * gimple-pretty-print.c: Various whitespace tweaks.
6208         * tree-core.h: Likewise.
6209         * tree-pretty-print.c: Likewise.
6210         * tree-ssa-alias.c: Likewise.
6211         * tree-ssa-copy.c: Likewise.
6212         * tree-ssanames.c: Likewise.
6213         * tree-ssanames.h: Likewise.
6214         * tree-vrp.c: Likewise.
6216 2013-09-24  Alan Modra  <amodra@gmail.com>
6218         PR middle-end/57134
6219         PR middle-end/57586
6220         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
6221         for output operands that disallow regs.  Don't use EXPAND_WRITE on
6222         inout operands.
6224 2013-09-24  Richard Biener  <rguenther@suse.de>
6226         PR middle-end/58513
6227         * tree.c (reference_alias_ptr_type): Move ...
6228         * alias.c (reference_alias_ptr_type): ... here and implement
6229         in terms of the new reference_alias_ptr_type_1.
6230         (ref_all_alias_ptr_type_p): New helper.
6231         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
6232         use ref_all_alias_ptr_type_p.
6233         (get_deref_alias_set): Add flag_strict_aliasing check here.
6234         (reference_alias_ptr_type_1): New function, split out from ...
6235         (get_alias_set): ... here.
6236         (alias_ptr_types_compatible_p): New function.
6237         * alias.h (reference_alias_ptr_type): Declare.
6238         (alias_ptr_types_compatible_p): Likewise.
6239         * tree.h (reference_alias_ptr_type): Remove.
6240         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
6241         to compare MEM_REF alias types.
6243 2013-09-24  Richard Biener  <rguenther@suse.de>
6245         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
6247 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6249         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
6250         reload helper function arrays into a single array reg_addr.
6251         (reload_fpr_gpr): Likewise.
6252         (reload_gpr_vsx): Likewise.
6253         (reload_vsx_gpr): Likewise.
6254         (struct rs6000_reg_addr): Likewise.
6255         (reg_addr): Likewise.
6256         (rs6000_debug_reg_global): Change rs6000_vector_reload,
6257         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
6258         (rs6000_init_hard_regno_mode_ok): Likewise.
6259         (rs6000_secondary_reload_direct_move): Likewise.
6260         (rs6000_secondary_reload): Likewise.
6262         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
6263         constraints: wu, ww, and wy.  Repurpose wv constraint added during
6264         power8 changes.  Put wg constraint in alphabetical order.
6266         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
6267         for future work to add ISA 2.07 VSX single precision support.
6268         (-mvsx-scalar-double): Change default from -1 to 1, update
6269         documentation comment.
6270         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
6271         (-mupper-regs-df): New debug switch to control whether DF values
6272         can go in the traditional Altivec registers.
6273         (-mupper-regs-sf): New debug switch to control whether SF values
6274         can go in the traditional Altivec registers.
6276         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
6277         and wy constraints.
6278         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
6279         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
6280         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
6281         (rs6000_debug_legitimate_address_p): Print if we are running
6282         before, during, or after reload.
6283         (rs6000_secondary_reload): Add a comment.
6284         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
6286         * config/rs6000/constraints.md (wa constraint): Sort w<x>
6287         constraints.  Update documentation string.
6288         (wd constraint): Likewise.
6289         (wf constraint): Likewise.
6290         (wg constraint): Likewise.
6291         (wn constraint): Likewise.
6292         (ws constraint): Likewise.
6293         (wt constraint): Likewise.
6294         (wx constraint): Likewise.
6295         (wz constraint): Likewise.
6296         (wu constraint): New constraint for ISA 2.07 SFmode scalar
6297         instructions.
6298         (ww constraint): Likewise.
6299         (wy constraint): Likewise.
6300         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
6301         in the previous submissions.
6302         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
6304 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
6306         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
6307         (REG_BR_PROB): Say that the probability is stored in an int_list.
6308         * reg-notes.def: Update commentary to mention INT_LIST.
6309         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
6310         (INT_LIST): New rtx.
6311         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
6312         * rtlanal.c (int_reg_note_p): New function.
6313         (alloc_reg_note): Assert that the note does not have an int argument.
6314         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
6315         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
6316         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
6317         rather than an INSN_LIST.  Handle INT_LIST.
6318         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
6319         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
6320         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
6321         Manipulate them as ints rather than rtxes.
6322         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
6323         * regmove.c (copy_src_to_dest): Likewise.
6324         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
6326         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
6327         into the cases that need it.
6328         * config/arm/arm.c (arm_unwind_emit): Likewise.
6330         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
6331         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
6332         * loop-doloop.c (add_test, doloop_modify): Likewise.
6333         * loop-unswitch.c (compare_and_jump_seq): Likewise.
6334         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
6335         * predict.c (combine_predictions_for_insn): Likewise.
6336         * print-rtl.c (print_rtx): Handle INT_LIST.
6337         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
6338         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
6339         * config/arm/arm.c (emit_unlikely_jump): Likewise.
6340         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
6341         (ix86_split_fp_branch, predict_jump): Likewise.
6342         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
6343         * config/sh/sh.c (expand_cbranchsi4): Likewise.
6344         * config/spu/spu.c (ea_load_store_inline): Likewise.
6346         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
6347         value of a REG_BR_PROB note.
6348         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
6349         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
6350         * emit-rtl.c (try_split): Likewise.
6351         * predict.c (br_prob_note_reliable_p): Likewise.
6352         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
6353         * reorg.c (mostly_true_jump): Likewise.
6354         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
6355         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
6356         * config/i386/i386.c (ix86_print_operand): Likewise.
6357         * config/ia64/ia64.c (ia64_print_operand): Likewise.
6358         * config/mmix/mmix.c (mmix_print_operand): Likewise.
6359         * config/rs6000/rs6000.c (output_cbranch): Likewise.
6360         * config/s390/s390.c (s390_expand_tbegin): Likewise.
6361         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
6362         * config/sparc/sparc.c (output_cbranch): Likewise.
6363         * config/spu/spu.c (get_branch_target): Likewise.
6364         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
6365         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
6367 2013-09-23  Jan Hubicka  <jh@suse.cz>
6369         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
6370         for ipa-devirt.
6371         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
6372         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
6373         of external calls
6374         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
6375         (gimple_fold_call): Dump inconsistent devirtualizations; add
6376         sanity check for type based devirtualizations.
6377         * ipa-prop.c: Include ipa-utils.h
6378         (ipa_intraprocedural_devirtualization): Add sanity check.
6379         (try_make_edge_direct_virtual_call): Likewise.
6381 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
6383         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
6384         assignment statements.
6386 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
6388         * gimple-pretty-print.c (dump_ssaname_info): New function.
6389         (dump_gimple_phi): Call it.
6390         (pp_gimple_stmt_1): Likewise.
6391         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
6392         (range_info_def): Declare.
6393         * tree-pretty-print.c (pp_double_int): New function.
6394         (dump_generic_node): Call it.
6395         * tree-pretty-print.h (pp_double_int): Declare.
6396         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
6397         * tree-ssanames.h (range_info_def): New structure.
6398         (value_range_type): Move definition here.
6399         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
6400         Declare.
6401         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
6402         initialization.
6403         (set_range_info): New function.
6404         (get_range_info): Likewise.
6405         (duplicate_ssa_name_range_info): Likewise.
6406         (duplicate_ssa_name_fn): Check pointer type and call
6407         duplicate_ssa_name_range_info.
6408         * tree-ssa-copy.c (fini_copy_prop): Likewise.
6409         * tree-vrp.c (value_range_type): Remove definition, now in
6410         tree-ssanames.h.
6411         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
6412         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
6413         (SSA_NAME_RANGE_INFO): New macro.
6415 2013-09-23  Richard Biener  <rguenther@suse.de>
6417         PR tree-optimization/58464
6418         * tree-ssa-pre.c (phi_trans_lookup): Remove.
6419         (phi_trans_add): Change to add conditionally on being not present.
6420         (phi_translate_1): Remove recursion detection here.
6421         (phi_translate): Pre-seed the cache with NULL to catch
6422         recursion here in a more generic way.
6423         (bitmap_find_leader): Adjust comment.
6424         (get_representative_for): Dump value-numbers.
6425         (create_expression_by_pieces): Likewise.
6426         (insert_into_preds_of_block): Likewise.
6428 2013-09-23  Christian Bruel  <christian.bruel@st.com>
6430         PR target/58475
6431         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
6432         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
6434 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
6436         Revert r202780:
6437         2013-09-20  Renlin Li  <renlin.li@arm.com>
6439         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6440         plus_constant.
6441         (aarch64_expand_epilogue): Likewise.
6442         (aarch64_legitimize_reload_address): Likewise.
6444 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
6446         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
6447         NULL_TREE before pushing them onto the vector.  Likewise for labels.
6449 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
6451         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
6452         comparison operators when -fno-trapping-math is in effect.
6453         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
6454         comparison operators in TFmode and assert that unsupported operators
6455         cannot reach here.
6456         (ia64_print_operand): Likewise.
6458 2013-09-21  Jan Hubicka  <jh@suse.cz>
6460         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
6461         (sse_typeless_stores): Enable for core
6462         (sse_load0_by_pxor): Likewise.
6463         (four_jump_limit): Disable for core.
6464         (pad_returns): Likewise.
6465         (avoid_vector_decode): Likewise.
6466         (fuse_cmp_and_branch): Enable for cores.
6467         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
6469 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
6471         PR middle-end/56791
6472         * config/pa/pa.c (pa_option_override): Disable auto increment and
6473         decrement instructions until reload is completed.
6475         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
6476         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
6477         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
6479 2013-09-20  DJ Delorie  <dj@redhat.com>
6480             Nick Clifton  <nickc@redhat.com>
6482         * config/rl78/rl78.c: Various whitespace and comment tweaks.
6483         (need_to_save): Save bank 0 on interrupts.
6484         (characterize_address): Strip far address wrappers.
6485         (rl78_as_legitimate_address): Likewise.
6486         (transcode_memory_rtx): Likewise.
6487         (rl78_peep_movhi_p): Disable this peephole after devirt.
6488         (rl78_propogate_register_origins): Forget all origins when a
6489         CLOBBER is seen.
6490         * config/rl78/rl78-virt.md: Various whitespace tweaks.
6491         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
6492         peephole2's.
6493         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
6494         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
6495         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
6496         relocating.
6497         * config/rl78/constraints.md: Various whitespace and paren tweaks.
6499 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
6501         * config/pa/pa.md: In "scc" insn patterns, change output template to
6502         handle const0_rtx in reg_or_0_operand operands.
6504 2013-09-20  Martin Husemann  <martin@NetBSD.org>
6506         PR target/56875
6507         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
6508         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
6510 2013-09-20  Richard Biener  <rguenther@suse.de>
6512         PR middle-end/58484
6513         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
6514         remembering SSA name version and block index.
6515         (new_scev_info_str): Adjust.
6516         (hash_scev_info): Likewise.  Also hash the block index.
6517         (eq_scev_info): Adjust.
6518         (find_var_scev_info): Likewise.
6519         (struct instantiate_cache_entry): Remove.
6520         (struct instantiate_cache_type): Use a htab to map name, block
6521         to chrec.
6522         (instantiate_cache_type::~instantiate_cache_type): Adjust.
6523         (get_instantiated_value_entry): Likewise.
6524         (hash_idx_scev_info, eq_idx_scev_info): New functions.
6525         (instantiate_scev_name): Adjust.
6527 2013-09-20  Jeff Law  <law@redhat.com>
6529         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
6531 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
6533         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6534         Call aarch64_simd_expand_args to update op[argc].
6536 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
6538         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
6539         plugin argument.
6541 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
6543         * gengtype.c (file_rules): Added rule for *.cc files.
6544         (get_output_file_with_visibility): Give fatal message when no
6545         rules found.
6547 2013-09-20  Renlin Li  <renlin.li@arm.com>
6549         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
6550         (aarch64_expand_epilogue): Likewise.
6551         (aarch64_legitimize_reload_address): Likewise.
6553 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
6555         PR middle-end/57748
6556         * expr.c (expand_assignment): Remove misalignp code path.
6558 2013-09-20  Marek Polacek  <polacek@redhat.com>
6560         PR sanitizer/58413
6561         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
6562         TYPE_PRECISION.  Add asserts.
6564 2013-09-20  Richard Biener  <rguenther@suse.de>
6566         PR tree-optimization/58453
6567         * tree-loop-distribution.c (distribute_loop): Apply the cost
6568         model for -ftree-loop-distribute-patterns, too.
6570 2013-09-20  Richard Biener  <rguenther@suse.de>
6572         PR middle-end/58473
6573         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
6574         make type comparison less strict.
6576 2013-09-20  Alan Modra  <amodra@gmail.com>
6578         * configure: Regenerate.
6579         * aclocal.m4: Regenerate.
6581 2013-09-20  Marek Polacek  <polacek@redhat.com>
6583         PR other/58467
6584         * doc/extend.texi: Document that attribute used is meant to be used
6585         on variables with static storage duration.
6587 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
6589         PR tree-optimization/58472
6590         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
6591         simd_lane_access set inv_p = false.
6592         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
6593         the simduid magic VAR_DECL.
6595 2013-09-19  Jan Hubicka  <jh@suse.cz>
6597         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
6599 2013-09-17  Jeff Law  <law@redhat.com>
6601         * tree-ssa-dom.c (record_temporary_equivalences): New function
6602         split out of dom_opt_dom_walker::after_dom_children.
6603         (dom_opt_dom_walker::thread_across_edge): Move common code
6604         in here from dom_opt_dom_walker::after_dom_children.
6605         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
6607 2013-09-19  Jan Hubicka  <jh@suse.cz>
6609         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
6610         (TARGET_GENERIC): Use PROCESOR_GENERIC
6611         (enum processor_type): Unify generic32 and 64.
6612         * i386.md (cpu): Likewise.
6613         * x86-tune.def (use_leave): Enable for generic32.
6614         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
6615         * athlon.md: Change generic64 to generic in all occurences.
6616         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
6617         (ix86_target_macros_internal): Likewise.
6618         * driver-i386.c (host_detect_local_cpu): Likewise.
6619         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
6620         to ..
6621         (generic_memcpy, generic_memset, generic_cost): This one.
6622         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
6623         (m_GENERIC32, m_GENERIC64): Remove.
6624         (m_GENERIC): Turn into one flag.
6625         (processor_target): Unify generic tunnings.
6626         (ix86_option_override_internal): Replace generic32/64 by generic.
6627         (ix86_issue_rate): Likewise.
6628         (ix86_adjust_cost): Likewise.
6630 2013-09-19  Jan Hubicka  <jh@suse.cz>
6632         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
6633         of speculative flag.
6635 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
6637         * omp-low.c (expand_omp_sections): Always pass len - 1 to
6638         GOMP_sections_start, even if !exit_reachable.
6640 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
6642         * lra-constraints.c (need_for_all_save_p): Use macro
6643         HARD_REGNO_CALL_PART_CLOBBERED.
6644         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
6645         set up pseudo conflict hard regs.
6647 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
6649         PR target/58452
6650         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
6651         operands.
6653 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
6655         PR rtl-optimization/58438
6656         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
6657         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
6658         from previous subpasses.
6660 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
6662         * arm.c (arm_get_frame_offsets): Validate architecture supports
6663         LDRD/STRD before accepting the tuning preference.
6664         (arm_expand_prologue): Likewise.
6665         (arm_expand_epilogue): Likewise.
6667 2013-09-18  Richard Biener  <rguenther@suse.de>
6669         PR tree-optimization/58417
6670         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
6671         have chrecs with symbols defined in the loop as operands.
6672         (chrec_fold_multiply): Likewise.
6673         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
6674         parameters before folding binary operations.
6675         (struct instantiate_cache_entry_hasher): Remove.
6676         (struct instantiate_cache_type): Use a pointer-map.
6677         (instantiate_cache_type::instantiate_cache_type): New function.
6678         (instantiate_cache_type::get): Likewise.
6679         (instantiate_cache_type::set): Likewise.
6680         (instantiate_cache_type::~instantiate_cache_type): Adjust.
6681         (get_instantiated_value_entry): Likewise.
6682         (global_cache): New global.
6683         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
6684         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
6685         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
6686         (instantiate_scev_name): Adjust.
6687         (instantiate_scev): Construct global instead of local cache.
6688         (resolve_mixers): Likewise.
6690 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
6691             Paolo Carlini  <paolo.carlini@oracle.com>
6693         PR c++/58458
6694         * doc/implement-cxx.texi: Fix references to the C++ standards.
6696 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
6698         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
6699         * tree-vect-data-refs.c (vect_analyze_data_refs): For
6700         simd_lane_access drs, update also DR_ALIGNED_TO.
6702 2013-09-18  Marek Polacek  <polacek@redhat.com>
6704         PR sanitizer/58411
6705         * doc/extend.texi: Document no_sanitize_undefined attribute.
6706         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
6707         no_sanitize_undefined attribute.
6709 2013-09-18  Nick Clifton  <nickc@redhat.com>
6711         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
6712         (ASM_DECLARE_FUNCTION_NAME): Define.
6714 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
6716         * compare-elim.c (find_comparison_dom_walker): New class
6717         (find_comparisons_in_bb): Rename to
6718         find_comparison_dom_walker::before_dom_children
6719         (find_comparisons): Adjust
6720         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
6721         adjust.
6722         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
6723         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
6724         (init_walk_dominator_tree): Remove declaration.
6725         (fini_walk_dominator_tree): Remove declaration.
6726         * fwprop.c (single_def_use_dom_walker): New class
6727         (single_def_use_enter_block): Convert to
6728         single_def_use_dom_walker::before_dom_children.
6729         (single_def_use_leave_block): Convert to
6730         single_def_use_dom_walker::after_dom_children.
6731         (build_single_def_use_links): Adjust.
6732         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
6733         class.
6734         (find_candidates_in_block): Convert to
6735         find_candidates_dom_walker::before_dom_children.
6736         (execute_strength_reduction): Adjust.
6737         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
6738         (sese_dom_walker): New class.
6739         (sese_dom_walker::sese_dom_walker): New constructor.
6740         (sese_dom_walker::~sese_dom_walker): New destructor.
6741         (build_sese_conditions_before): Convert to
6742         sese_dom_walker::before_dom_children.
6743         (build_sese_conditions_after): Convert to
6744         sese_dom_walker::after_dom_children.
6745         (build_poly_scop): Adjust
6746         * tree-into-ssa.c (rewrite_dom_walker): New class
6747         (rewrite_enter_block): Convert to
6748         rewrite_dom_walker::before_dom_children.
6749         (rewrite_leave_block): Convert to
6750         rewrite_dom_walker::after_dom_children.
6751         (rewrite_update_dom_walker): New class.
6752         (rewrite_update_enter_block): Convert to
6753         rewrite_update_dom_walker::before_dom_children.
6754         (rewrite_update_leave_block): Convert to
6755         rewrite_update_dom_walker::after_dom_children.
6756         (rewrite_blocks, rewrite_into_ssa): Adjust.
6757         (mark_def_dom_walker): New class.
6758         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
6759         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
6760         (mark_def_sites_blocks): Convert to
6761         mark_def_dom_walker::before_dom_children.
6762         (mark_def_site_blocks): Remove.
6763         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
6764         (tree_ssa_dominator_optimize): Adjust.
6765         (dom_thread_across_edge): Convert to method
6766         dom_opt_dom_walker::thread_across_edge.
6767         (dom_opt_enter_block): Convert to member function
6768         dom_opt_dom_walker::before_dom_children.
6769         (dom_opt_leave_block): Convert to member function
6770         dom_opt_dom_walker::after_dom_children.
6771         * tree-ssa-dse.c (dse_dom_walker): New class.
6772         (dse_enter_block): Convert to member function
6773         dse_dom_walker::before_dom_children.
6774         (tree_ssa_dse): Adjust.
6775         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
6776         (determine_invariantness_stmt): Convert to method
6777         invariantness_dom_walker::before_dom_children.
6778         (determine_invariantness): Remove
6779         (move_computations_dom_walker): New class.
6780         (move_computations_stmt): Convert to method
6781         move_computations_dom_walker::before_dom_children.
6782         (move_computations, tree_ssa_lim): Adjust.
6783         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
6784         (nt_init_block): Convert to method
6785         notrappping_dom_walker::before_dom_children.
6786         (nt_fini_block): Convert to method
6787         method nontrapping_dom_walker::after_dom_children.
6788         (get_non_trapping): Adjust.
6789         * tree-ssa-pre.c (eliminate_dom_walker): New class.
6790         (eliminate_bb): Convert to method
6791         eliminate_dom_walker::before_dom_children.
6792         (eliminate_leave_block): Convert to method
6793         eliminate_dom_walker::after_dom_children.
6794         (eliminate): Adjust.
6795         * tree-ssa-strlen.c (strlen_dom_walker): New class.
6796         (strlen_enter_block): Convert to method
6797         strlen_dom_walker::before_dom_children.
6798         (strlen_leave_block): Convert to method
6799         method strlen_dom_walker::after_dom_children.
6800         (tree_ssa_strlen): Adjust.
6801         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
6802         (tree_ssa_uncprop): Adjust.
6803         (uncprop_leave_block): Convert to method
6804         uncprop_dom_walker::after_dom_children.
6805         (uncprop_leave_block): Convert to method
6806         uncprop_dom_walker::before_dom_children.
6808 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
6810         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
6811         branch in same basic block.  Check both src and dest of the move insn.
6813 2013-09-17  Nick Clifton  <nickc@redhat.com>
6815         * config/rl78/rl78-real.md (bf): New pattern.
6816         (bt): New pattern.
6817         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
6818         (rl78_print_operand): Do not put a # before a %B.
6819         * config/rl78/rl78.opt: Tweak doc strings.
6821 2013-09-17  DJ Delorie  <dj@redhat.com>
6823         * config/rl78/constraints.md (Wcv): Allow up to $r31.
6824         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
6825         (rl78_option_override): Likewise, if -mallregs.
6826         (is_virtual_register): Likewise.
6827         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
6828         (REGNO_OK_FOR_BASE_P): Likewise.
6829         * config/rl78/rl78.opt (-mallregs): New.
6831 2013-09-17  Nick Clifton  <nickc@redhat.com>
6833         * config/rl78/rl78.c (need_to_save): Change return type to bool.
6834         For interrupt functions: save all call clobbered registers if the
6835         interrupt handler is not a leaf function.
6836         (rl78_expand_prologue): Always recompute the frame information.
6837         For interrupt functions: only select bank 0 if one of the bank 0
6838         registers is going to be psuhed.
6840 2013-09-17  DJ Delorie  <dj@redhat.com>
6842         * config/rl78/constraints.md: For each W* constraint, rename to C*
6843         and create a W* constraint that checks for an optional ES: prefix
6844         pattern also.
6845         * config/rl78/rl78.md (UNS_ES_ADDR): New.
6846         (es_addr): New.  Used to wrap far addresses.
6847         * config/rl78/rl78-protos.h (rl78_es_addr): New.
6848         (rl78_es_base): New.
6849         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
6850         wrapped far addresses.
6851         (rl78_print_operand_1): Unwrap far addresses before processing.
6852         (rl78_lo16): Wrap far addresses in unspecs.
6853         (rl78_es_addr): New.
6854         (rl78_es_base): New.
6855         (insn_ok_now): Check for not-yet-wrapped far addresses.
6856         (transcode_memory_rtx): Properly re-wrap far addresses.
6858 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6860         * config/sparc/t-rtems: Add leon3 multilibs.
6862 2013-09-17  Cong Hou  <congh@google.com>
6864         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
6865         when checking the dot production pattern. The type of rhs operand
6866         of multiply is now checked correctly.
6868 2013-09-17  Jeff Law  <law@redhat.com>
6870         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6871         edge implied equivalences into successor phis.
6872         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
6873         here from tree-ssa-threadedge.c.
6874         (mark_threaded_blocks): When threading through a joiner, if both
6875         successors of the joiner's clone reach the same block, verify the
6876         PHI arguments are equal.  If not, cancel the jump threading request.
6877         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
6878         tree-ssa-threadupdate.c
6879         (thread_across_edge): Don't check PHI argument equality when
6880         threading through joiner block here.
6882 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
6884         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
6885         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
6886         (warn_uninit, warn_uninitialized_vars,
6887         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
6888         Move to tree-ssa-uninit.c.
6889         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
6890         (has_undefined_value_p): New.  Pass dependant parts of
6891         ssa_undefined_value_p.
6892         (uninit_undefined_value_p): Use has_undefined_value_p.
6893         (warn_uninit, warn_uninitialized_vars,
6894         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
6895         Move from tree-ssa.c.
6896         * tree-ssa.h: Adjust prototypes.
6898 2013-09-17  Jan Hubicka  <jh@suse.cz>
6900         PR middle-end/58332
6901         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
6902         * ipa-inline.c (can_inline_edge_p): Do not downgrade
6903         FUNCTION_NOT_OPTIMIZED.
6904         * ipa-inline-analysis.c (compute_inline_parameters): Function
6905         not optimized is not inlinable unless it is alwaysinline.
6906         (inline_analyze_function): Force calls in not optimized
6907         function not inlinable.
6909 2013-09-17  Jan Hubicka  <jh@suse.cz>
6911         PR middle-end/58329
6912         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
6913         to return NULL.
6914         * ipa.c (function_and_variable_visibility): Likewise.
6915         * ipa-profile.c (ipa_profile): Likewise.
6917 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
6919         PR ipa/58398
6920         * cgraph.c (cgraph_function_body_availability): Check for ifunc
6921         attribute, and don't inline the resolver in this case.
6923 2013-09-17  Teresa Johnson  <tejohnson@google.com>
6925         * coverage.c (get_coverage_counts): Add missing newline.
6927 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6929         PR tree-optimization/58088
6930         * fold-const.c (mask_with_trailing_zeros): New function.
6931         (fold_binary_loc): Make sure we don't recurse infinitely
6932         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
6933         Use mask_with_trailing_zeros where appropriate.
6935 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
6937         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
6938         of 'prev' var to get better distance estimation.
6940 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
6942         * tree-inline.h (struct copy_body_data): Add transform_parameter.
6943         * tree-inline.c (is_parameter_of): New predicate.
6944         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
6945         a parameter has been remapped.
6946         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
6947         (optimize_inline_calls): Initialize transform_parameter.
6948         (copy_gimple_seq_and_replace_locals): Likewise.
6949         (tree_function_versioning): Likewise.
6950         (maybe_inline_call_in_expr): Likewise.
6952 2013-09-17  Nick Clifton  <nickc@redhat.com>
6954         * config/msp430/msp430-protos.h: Add prototypes for new functions.
6955         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
6956         interrupt handlers.
6957         (is_attr_func): New function.
6958         (msp430_is_interrupt_func): New function.
6959         (is_naked_func): New function.
6960         (is_reentrant_func): New function.
6961         (is_critical_func): New function.
6962         (msp430_start_function): Add annotations for function attributes.
6963         (msp430_attr): New function.
6964         (msp430_attribute_table): New.
6965         (msp430_function_section): New function.
6966         (TARGET_ASM_FUNCTION_SECTION): Define.
6967         (msp430_builtin): New enum.
6968         (msp430_init_builtins): New function.
6969         (msp430_builtin_devl): New function.
6970         (msp430_expand_builtin): New function.
6971         (TARGET_INIT_BUILTINS): Define.
6972         (TARGET_EXPAND_BUILTINS): Define.
6973         (TARGET_BUILTIN_DECL): Define.
6974         (msp430_expand_prologue): Add support for naked, interrupt,
6975         critical and reentrant functions.
6976         (msp430_expand_epilogue): Likewise.
6977         (msp430_print_operand): Handle 'O' character.
6978         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
6979         NO_TRAMPOLINES.
6980         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
6981         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
6982         (pushm): Use a 'n' rather than an 'i' constraint.
6983         (msp_return): Add generation of the interrupt return instruction.
6984         (disable_interrupts): New pattern.
6985         (enable_interrupts): New pattern.
6986         (push_intr_state): New pattern.
6987         (pop_intr_state): New pattern.
6988         (bic_SR): New pattern.
6989         (bis_SR): New pattern.
6990         * doc/extend.texi: Document MSP430 function attributes and builtin
6991         functions.
6993 2013-09-17  Richard Biener  <rguenther@suse.de>
6995         PR tree-optimization/58432
6996         * tree-loop-distribution.c (tree_loop_distribution): Also
6997         scan PHIs for outside loop uses and seed a partition from them.
6999 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
7001         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
7002         (restructure_reference): Call backtrace_base_for_ref.
7004 2013-09-17  Alan Modra  <amodra@gmail.com>
7006         PR target/57589
7007         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
7008         patch.
7010 2013-09-16  DJ Delorie  <dj@redhat.com>
7012         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
7013         vregs location for RL78/G10.
7014         (rl78_expand_prologue): Avoid SEL on G10.
7015         (rl78_expand_epilogue): Likewise.
7016         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
7017         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
7018         __RL78_G10__ when appropriate.
7019         (ASM_SPEC): Pass -mg10 along to the assembler.
7020         * config/rl78/rl78.md (sel_rb): Disable for G10.
7021         * config/rl78/rl78.opt: Add -mg10 option.
7022         * config/rl78/t-rl78: Add -mg10 multilib.
7024 2013-09-16  Xinliang David Li  <davidxl@google.com>
7026         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
7027         * omp-low.c (omp_max_vf): Ditto.
7028         (expand_omp_simd): Ditto.
7029         * tree-vectorizer.c (vectorize_loops): Ditto.
7030         (gate_vect_slp): Ditto.
7031         (gate_increase_alignment): Ditto.
7032         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
7033         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
7034         (gate_tree_vectorize): Name change.
7035         (tree_vectorize): Ditto.
7036         (pass_vectorize::gate): Call new function.
7037         (pass_vectorize::execute): Ditto.
7038         * opts.c: O3 default setting change.
7039         (finish_options): Check new flag.
7040         * doc/invoke.texi: Document new flags.
7041         * common.opt: New flags.
7043 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
7045         * doc/tm.texi.in (Cond Exec Macros): Remove node.
7046         (Condition Code): Don't reference it.
7047         * doc/tm.texi: Regenerate.
7049 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
7051         PR middle-end/58418
7052         * lra-constraints.c (undo_optional_reloads): Consider all optional
7053         reload even if it did not get a hard reg.
7055 2013-09-16  Teresa Johnson  <tejohnson@google.com>
7057         * dumpfile.c (dump_loc): Remove newline emission.
7058         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
7059         emission to dump_printf_loc calls where missing.
7060         (vect_mark_for_runtime_alias_test): Ditto.
7061         (vect_analyze_data_ref_dependence): Ditto.
7062         (vect_analyze_data_ref_dependences): Ditto.
7063         (vect_slp_analyze_data_ref_dependence): Ditto.
7064         (vect_slp_analyze_data_ref_dependences): Ditto.
7065         (vect_compute_data_ref_alignment): Ditto.
7066         (vect_update_misalignment_for_peel): Ditto.
7067         (vect_verify_datarefs_alignment): Ditto.
7068         (vector_alignment_reachable_p): Ditto.
7069         (vect_get_data_access_cost): Ditto.
7070         (vect_enhance_data_refs_alignment): Ditto.
7071         (vect_find_same_alignment_drs): Ditto.
7072         (vect_analyze_data_refs_alignment): Ditto.
7073         (vect_analyze_group_access): Ditto.
7074         (vect_analyze_data_ref_access): Ditto.
7075         (vect_analyze_data_ref_accesses): Ditto.
7076         (vect_prune_runtime_alias_test_list): Ditto.
7077         (vect_analyze_data_refs): Ditto.
7078         (vect_create_addr_base_for_vector_ref): Ditto.
7079         (vect_create_data_ref_ptr): Ditto.
7080         (vect_grouped_store_supported): Ditto.
7081         (vect_grouped_load_supported): Ditto.
7082         * value-prof.c (check_counter): Ditto.
7083         (check_ic_target): Ditto.
7084         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
7085         (vect_recog_widen_mult_pattern): Ditto.
7086         (vect_recog_widen_sum_pattern): Ditto.
7087         (vect_recog_over_widening_pattern): Ditto.
7088         (vect_recog_widen_shift_pattern): Ditto.
7089         (vect_recog_rotate_pattern): Ditto.
7090         (vect_recog_vector_vector_shift_pattern): Ditto.
7091         (vect_recog_divmod_pattern): Ditto.
7092         (vect_recog_mixed_size_cond_pattern): Ditto.
7093         (vect_recog_bool_pattern): Ditto.
7094         (vect_pattern_recog_1): Ditto.
7095         (vect_pattern_recog): Ditto.
7096         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
7097         (vect_is_simple_iv_evolution): Ditto.
7098         (vect_analyze_scalar_cycles_1): Ditto.
7099         (vect_get_loop_niters): Ditto.
7100         (vect_analyze_loop_1): Ditto.
7101         (vect_analyze_loop_form): Ditto.
7102         (vect_analyze_loop_operations): Ditto.
7103         (vect_analyze_loop_2): Ditto.
7104         (vect_analyze_loop): Ditto.
7105         (report_vect_op): Ditto.
7106         (vect_is_slp_reduction): Ditto.
7107         (vect_is_simple_reduction_1): Ditto.
7108         (vect_get_known_peeling_cost): Ditto.
7109         (vect_estimate_min_profitable_iters): Ditto.
7110         (vect_model_reduction_cost): Ditto.
7111         (vect_model_induction_cost): Ditto.
7112         (get_initial_def_for_induction): Ditto.
7113         (vect_create_epilog_for_reduction): Ditto.
7114         (vectorizable_reduction): Ditto.
7115         (vectorizable_induction): Ditto.
7116         (vectorizable_live_operation): Ditto.
7117         (vect_loop_kill_debug_uses): Ditto.
7118         (vect_transform_loop): Ditto.
7119         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
7120         (vect_stmt_relevant_p): Ditto.
7121         (process_use): Ditto.
7122         (vect_mark_stmts_to_be_vectorized): Ditto.
7123         (vect_model_simple_cost): Ditto.
7124         (vect_model_promotion_demotion_cost): Ditto.
7125         (vect_model_store_cost): Ditto.
7126         (vect_get_store_cost): Ditto.
7127         (vect_model_load_cost): Ditto.
7128         (vect_get_load_cost): Ditto.
7129         (vect_init_vector_1): Ditto.
7130         (vect_get_vec_def_for_operand): Ditto.
7131         (vect_finish_stmt_generation): Ditto.
7132         (vectorizable_call): Ditto.
7133         (vectorizable_conversion): Ditto.
7134         (vectorizable_assignment): Ditto.
7135         (vectorizable_shift): Ditto.
7136         (vectorizable_operation): Ditto.
7137         (vectorizable_store): Ditto.
7138         (vectorizable_load): Ditto.
7139         (vectorizable_condition): Ditto.
7140         (vect_analyze_stmt): Ditto.
7141         (vect_transform_stmt): Ditto.
7142         (vect_is_simple_use): Ditto.
7143         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
7144         (vect_can_advance_ivs_p): Ditto.
7145         (vect_update_ivs_after_vectorizer): Ditto.
7146         (vect_do_peeling_for_loop_bound): Ditto.
7147         (vect_gen_niters_for_prolog_loop): Ditto.
7148         (vect_update_inits_of_drs): Ditto.
7149         (vect_create_cond_for_alias_checks): Ditto.
7150         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
7151         (vect_build_slp_tree_1): Ditto.
7152         (vect_supported_load_permutation_p): Ditto.
7153         (vect_analyze_slp_instance): Ditto.
7154         (vect_analyze_slp): Ditto.
7155         (vect_make_slp_decision): Ditto.
7156         (vect_detect_hybrid_slp): Ditto.
7157         (vect_bb_vectorization_profitable_p): Ditto.
7158         (vect_slp_analyze_bb_1): Ditto.
7159         (vect_update_slp_costs_according_to_vf): Ditto.
7160         (vect_get_mask_element): Ditto.
7161         (vect_transform_slp_perm_load): Ditto.
7162         (vect_schedule_slp_instance): Ditto.
7163         (vect_schedule_slp): Ditto.
7164         (vect_slp_transform_bb): Ditto.
7165         * profile.c (read_profile_edge_counts): Ditto.
7166         (compute_branch_probabilities): Ditto.
7167         * coverage.c (get_coverage_counts): Ditto.
7169 2013-09-16  Diego Novillo  <dnovillo@google.com>
7171         * tree-core.h: Add missing comment lines from refactoring of tree.h.
7173 2013-09-16  Jan Hubicka  <jh@suse.cz>
7175         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
7176         abstract functions; for static functions check the presence of body.
7178 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7180         * config/aarch64/aarch64-simd-builtins.def (fma): New.
7181         * config/aarch64/aarch64-simd.md
7182         (aarch64_mla_elt<mode>): New.
7183         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
7184         (aarch64_mls_elt<mode>): Likewise.
7185         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
7186         (aarch64_fma4_elt<mode>): Likewise.
7187         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
7188         (aarch64_fma4_elt_to_128v2df): Likewise.
7189         (aarch64_fma4_elt_to_64df): Likewise.
7190         (fnma<mode>4): Likewise.
7191         (aarch64_fnma4_elt<mode>): Likewise.
7192         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7193         (aarch64_fnma4_elt_to_128v2df): Likewise.
7194         (aarch64_fnma4_elt_to_64df): Likewise.
7195         * config/aarch64/iterators.md (VDQSF): New.
7196         * config/aarch64/arm_neon.h
7197         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
7198         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
7200 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7202         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
7203         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7204         (aarch64_mul3_elt_to_128df): Likewise.
7205         (aarch64_mul3_elt_to_64v2df): Likewise.
7206         * config/aarch64/iterators.md (VEL): Also handle DFmode.
7207         (VMUL): New.
7208         (VMUL_CHANGE_NLANES) Likewise.
7209         (h_con): Likewise.
7210         (f): Likewise.
7211         * config/aarch64/arm_neon.h
7212         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
7214 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7216         * config/aarch64/arm_neon.h
7217         (vcvtx_high_f32_f64): Fix parameters.
7219 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7220             Uros Bizjak  <ubizjak@gmail.com>
7222         * config/alpha.c: Include tree-ssanames.h.
7224 2013-09-16  Richard Biener  <rguenther@suse.de>
7226         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
7227         (dot_rdg_1): Handle control_dd.
7228         (create_edge_for_control_dependence): New function.
7229         (create_rdg_edges): Add control dependences if asked for.
7230         (build_rdg): Likewise.
7231         (generate_loops_for_partition): If there are not necessary
7232         control stmts remove all their dependencies.
7233         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
7234         (distribute_loop): Pass on control dependences.
7235         (tree_loop_distribution): Compute control dependences and remove
7236         restriction on number of loop nodes.
7238 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
7240         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
7241         for internal calls.
7243 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
7245         * cse.c (try_const_anchors): Punt on CC modes.
7247 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7249         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
7251 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
7253         PR target/58382
7254         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
7255         calls to word_mode.
7257 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
7259         PR target/48094
7260         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
7261         seen.
7262         (darwin_objc1_section): Likewise.
7263         (darwin_file_end): Emit Image Info section when required.
7265 2013-09-14  Jan Hubicka  <jh@suse.cz>
7267         * tree-into-ssa.c (gate_into_ssa): New.
7268         (pass_data_build_ssa): Use it.
7269         * cgraph.h (expand_thunk): Update prototype.
7270         * cgraphunit.c (analyze_function): Expand thunks early.
7271         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
7272         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
7273         set lowered flag; do not add new function.
7274         (assemble_thunks_and_aliases): Update.
7275         * tree-ssa.c (gate_init_datastructures): New gate.
7276         (pass_data_init_datastructures): Use it.
7278 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
7280         PR target/58269
7281         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
7282         xmm register set described in the psABI.
7284 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
7286         * dwarf2out.c (should_emit_struct_debug): Add check
7287         for type_decl variable is not NULL.
7289 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
7291         * config.gcc: Use new winnt-c.c target hooks
7292         * config/t-winnt: New file
7293         * config/winnt-c.c: New file
7294         * doc/tm.texi.in: Document new hook
7295         * doc/tm.texi: Regenerated
7297 2013-09-13  Jan Hubicka  <jh@suse.cz>
7299         PR middle-end/58094
7300         * ipa-inline.c (check_callers): New function.
7301         (check_caller_edge): Remove.
7302         (want_inline_function_to_all_callers_p): Also permit alises that are
7303         called dirrectly.
7304         (inline_to_all_callers): Terminate the walk when devirtualization
7305         introduce new calls.
7307 2013-09-13  Jan Hubicka  <jh@suse.cz>
7309         * ipa-inline-analysis.c (struct growth_data): Add node.
7310         (do_estimate_growth_1): Fix detection of recursion.
7312 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
7314         PR tree-optimization/58392
7315         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
7316         to avoid shadowing of outer loop variable.  If
7317         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
7318         replace_by_duplicate_decl simduid of loops that have it set and
7319         set dest_cfun->has_simduid_loops and/or
7320         dest_cfun->has_force_vect_loops.
7321         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
7322         instead of maybe_lookup_decl.
7323         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
7324         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
7325         call.  Copy over force_vect and copy and remap simduid.  Set
7326         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
7327         (copy_cfg_body): Remove blocks_to_copy argument.  Use
7328         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
7329         caller.  Don't set cfun->has_simduid_loops and/or
7330         cfun->has_force_vect_loops here.
7331         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
7332         caller.
7333         (expand_call_inline, tree_function_versioning): Adjust copy_body
7334         callers.
7336 2013-09-13  Martin Jambor  <mjambor@suse.cz>
7338         PR bootstrap/58388
7339         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
7340         the assert if the edge was a speculative one.
7342 2013-09-13  Richard Biener  <rguenther@suse.de>
7344         * tree-data-ref.h (known_dependences_p): Move here ...
7345         * tree-loop-distribution.c (known_dependences_p): ... from here.
7346         (dump_rdg_component, debug_rdg_component): Remove.
7347         (dump_rdg): Adjust.
7348         (generate_loops_for_partition): Use gimple_uid instead of
7349         relying on matching stmt visit order.
7350         (rdg_build_partitions): Take starting stmt vector.
7351         (ldist_gen): Merge into ...
7352         (distribute_loop): ... this function.  Do not compute starting
7353         vertices vector.
7354         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
7356 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7358         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
7359         Set type attribute correctly. Set predicable_short_it attribute.
7360         (cmpsi_shiftsi): Remove %? from output template.
7362 2013-09-13  Richard Biener  <rguenther@suse.de>
7364         * tree-loop-distribution.c (struct rdg_component,
7365         rdg_defs_used_in_other_loops_p, free_rdg_components,
7366         rdg_build_components): Remove.
7367         (stmts_from_loop): Do not record virtual PHIs.
7368         (generate_loops_for_partition): Skip virtual PHIs.
7369         (build_rdg_partition_for_component): Rename to ...
7370         (build_rdg_partition_for_vertex): ... this and adjust.
7371         (rdg_build_partitions): Take a vector of starting vertices
7372         instead of components.  Remove unnecessary leftover handling.
7373         (ldist_gen): Do not build components or record other stores.
7374         (distribute_loop): Do not distribute loops containing stmts
7375         with side-effects.
7377 2013-09-13  Christian Bruel  <christian.bruel@st.com>
7379         PR target/58314
7380         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
7382 2013-09-13  Kai Tietz  <ktietz@redhat.com>
7384         * config.gcc: Separate cases for mingw and cygwin targets,
7385         and add 64-bit cygwin target case.
7387         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
7388         dll-export inline-functions.
7389         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
7392 2013-09-13  Jeff Law  <law@redhat.com>
7394         PR middle-end/58387
7395         Revert:
7396         2013-09-06  Jeff Law  <law@redhat.com>
7398         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7399         edge implied equivalences into successor phis.
7401 2013-09-12  DJ Delorie  <dj@redhat.com>
7403         * config/rl78/rl78-virt.md: Change from | to \; for asm line
7404         separators.
7406 2013-09-12  Brooks Moses  <bmoses@google.com>
7408         PR driver/42955
7409         * Makefile.in: Do not install driver binaries in $(target)/bin.
7411 2013-09-12  DJ Delorie  <dj@redhat.com>
7413         * config/rl78/rl78.opt (mrelax): New.
7414         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
7415         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
7417         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
7418         between SP and FP.
7419         (rl78_expand_epilogue): Likewise.
7421 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
7423         PR middle-end/58335
7424         * lra-eliminations.c (remove_reg_equal_offset_note): New.
7425         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
7426         pointer elimination with using remove_reg_equal_offset_note.
7428 2013-09-12  DJ Delorie  <dj@redhat.com>
7430         * config/msp430/: New port.
7431         * config.gcc (msp430): Added.
7432         * doc/invoke.texi: Document MSP430 options.
7433         * doc/install.texi: Document msp430-elf
7434         * doc/md.texi: Document msp430-elf
7435         * doc/contrib.texi: Document msp430-elf
7437         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
7438         PSImode.
7440 2013-09-12  Martin Jambor  <mjambor@suse.cz>
7442         PR ipa/58389
7443         * ipa-prop.c (remove_described_reference): Give up if the edge in the
7444         reference descriptor is NULL.
7445         (ipa_edge_removal_hook): If owning a reference descriptor, set its
7446         edge to NULL.
7448 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
7450         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
7451         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
7452         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
7453         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
7454         make_temp_ssa_name): move to tree-ssanames.h
7455         * tree-ssa-alias.h: Move prototype.
7456         * tree-ssa.h: Include tree-ssanames.h.
7457         * tree-ssanames.c (FREE_SSANAMES): Move to here.
7458         * tree-ssanames.h: New.  Move items from tree-flow*.h
7459         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
7461 2013-09-12  Richard Biener  <rguenther@suse.de>
7463         PR tree-optimization/58404
7464         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
7465         propagate non-invariant addresses into dereferences wrapped
7466         in component references.
7468 2013-09-12  Richard Biener  <rguenther@suse.de>
7470         PR tree-optimization/58402
7471         * passes.def: Move pass_late_warn_uninitialized later.
7473 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
7475         * tree-ssa.h: New.  Move content from tree-flow.h and
7476         tree-flow-inline.h.
7477         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
7478         Move prototypes belonging to tree-ssa.c.
7479         * tree-flow-inline.h (redirect_edge_var_map_def,
7480         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
7481         tree-ssa.h.
7482         * gimple.h: Adjust prototypes.
7483         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
7484         to...
7485         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
7486         * tree.h: Move prototype to tree-ssa.h.
7487         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
7488         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
7489         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
7490         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
7491         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
7492         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
7493         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
7494         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
7495         graphite-blocking.c, graphite-clast-to-gimple.c,
7496         graphite-dependences.c, graphite-interchange.c,
7497         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
7498         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
7499         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
7500         ipa-reference.c, ipa-split.c, ipa-utils.c,
7501         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
7502         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
7503         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
7504         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
7505         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
7506         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
7507         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
7508         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
7509         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
7510         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
7511         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
7512         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
7513         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
7514         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
7515         value-prof.c, var-tracking.c,
7516         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
7518 2013-09-12  Richard Biener  <rguenther@suse.de>
7520         PR tree-optimization/58396
7521         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
7522         (build_rdg): Take a loop-nest parameter, fix memleaks.
7523         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
7525 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
7527         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
7528         for SLM.
7530 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
7532         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
7533         The FMA instruction names should have a 'v' prefix.
7535 2013-09-12  Richard Biener  <rguenther@suse.de>
7537         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
7538         (dot_rdg): Use popen instead of system in optional code.
7539         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
7540         (already_processed_vertex_p): Adjust.
7541         (has_anti_or_output_dependence, predecessor_has_mem_write,
7542         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
7543         rdg_flag_uses): Remove.
7544         (rdg_flag_vertex): Simplify.
7545         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
7546         remove recursion.
7547         (build_rdg_partition_for_component): Process the first vertex
7548         of a component only.
7549         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
7551 2013-09-12  Alan Modra  <amodra@gmail.com>
7553         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
7555 2013-09-11  DJ Delorie  <dj@redhat.com>
7556             Nick Clifton  <nickc@redhat.com>
7558         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
7559         (rl78_stack_based_mem): New.
7560         * config/rl78/constraints.md (Iv08): New.
7561         (Iv16): New.
7562         (Iv24): New.
7563         (Is09): New.
7564         (Is17): New.
7565         (Is25): New.
7566         (ISsi): New.
7567         (IShi): New.
7568         (ISqi): New.
7569         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
7570         (movhi): Likewise.
7571         (movsi): Change from expand to insn-and-split.
7572         (ashrsi3): Clobber AX.
7573         (lshrsi3): New.
7574         (ashlsi3): New.
7575         (cbranchsi4): New.
7576         * config/rl78/rl78.md (CC_REG): Fix.
7577         (addsi3): Allow memory and immediate operands.
7578         (addsi3_internal): Split into...
7579         (addsi3_internal_virt): ...new, and ...
7580         (addsi3_internal_real): ...new.
7581         (subsi): New.
7582         (subsi3_internal_virt): New.
7583         (subsi3_internal_real): New.
7584         (mulsi3): Add memory operand.
7585         (mulsi3_rl78): Likewise.
7586         (mulsi3_g13): Likewise.
7587         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
7588         (cbranchqi4_real): Add more constraint options.
7589         (cbranchhi4_real): Expand pattern.
7590         (cbranchhi4_real_signed): New.
7591         (cbranchhi4_real_inverted): New.
7592         (cbranchsi4_real_lt): New.
7593         (cbranchsi4_real_ge): New.
7594         (cbranchsi4_real_signed): New.
7595         (cbranchsi4_real): New.
7596         (peephole2): New.
7597         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
7598         constant shifts.
7599         (lshrsi3_virt): Likewise.
7600         (ashlsi3_virt): Likewise.
7601         (cbranchqi4_virt_signed): New.
7602         (cbranchhi4_virt_signed): New.
7603         (cbranchsi4_virt): New.
7604         * config/rl78/rl78.c: Whitespace fixes throughout.
7605         (move_elim_pass): New.
7606         (pass_data_rl78_move_elim): New.
7607         (pass_rl78_move_elim): New.
7608         (make_pass_rl78_move_elim): New.
7609         (rl78_devirt_info): Run devirt earlier.
7610         (rl78_move_elim_info): New.
7611         (rl78_asm_file_start): Register it.
7612         (rl78_split_movsi): New.
7613         (rl78_as_legitimate_address): Allow virtual base registers when
7614         appropriate.
7615         (rl78_addr_space_convert): Remove spurious debug stuff.
7616         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
7617         (rl78_print_operand): More cases for not printing '#'.
7618         (rl78_expand_compare): Remove most of the logic.
7619         (content_memory): New.
7620         (clear_content_memory): New.
7621         (get_content_index): New.
7622         (get_content_name): New.
7623         (display_content_memory): New.
7624         (update_content): New.
7625         (record_content): New.
7626         (already_contains): New.
7627         (insn_ok_now): Re-recog insns with virtual registers.
7628         (add_postponed_content_update): New.
7629         (process_postponed_content_update): New.
7630         (gen_and_emit_move): New.
7631         (transcode_memory_rtx): Record new location content.
7632         Use gen_and_emit_move.
7633         (force_into_acc): New.
7634         (move_to_acc): Use gen_and_emit_move.
7635         (move_from_acc): Likewise.
7636         (move_acc_to_reg): Likewise.
7637         (move_to_x): Likewise.
7638         (move_to_hl): Likewise.
7639         (move_to_de): Likewise.
7640         (rl78_alloc_physical_registers_op1): Record location content.
7641         (has_constraint): New.
7642         (rl78_alloc_physical_registers_op2): Record location content.
7643         Optimize use of HL.
7644         (rl78_alloc_physical_registers_ro1): Likewise.
7645         (rl78_alloc_physical_registers_cmp): Likewise.
7646         (rl78_alloc_physical_registers_umul): Likewise.
7647         (rl78_alloc_address_registers_macax): New.
7648         (rl78_alloc_physical_registers): Initialize and set location
7649         content memory as needed.
7650         (rl78_reorg): Make sure split2 is called.
7651         (rl78_rtx_costs): New.
7653 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
7655         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
7656         for (not (neg ...)) and (neg (not ...)) cases.
7658 2013-09-11  Richard Biener  <rguenther@suse.de>
7660         PR middle-end/58377
7661         * passes.def: Split critical edges before late uninit warning passes.
7662         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
7664 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
7666         PR tree-optimization/58385
7667         * fold-const.c (build_range_check): If both low and high are NULL,
7668         use omit_one_operand_loc to preserve exp side-effects.
7670 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7672         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
7674 2013-09-11  Richard Biener  <rguenther@suse.de>
7676         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
7677         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
7678         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
7679         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
7680         stmts_from_loop, known_dependences_p, build_empty_rdg,
7681         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
7682         * tree-loop-distribution.c: ... here.
7683         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
7684         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
7685         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
7686         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
7687         * tree-loop-distribution.c: ... here.
7688         * tree-loop-distribution.c: Include gimple-pretty-print.h.
7689         (struct partition_s): Add loops member.
7690         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
7691         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
7692         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
7694 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7695             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
7696             Sergey Lega  <sergey.s.lega@intel.com>
7697             Anna Tikhonova  <anna.tikhonova@intel.com>
7698             Ilya Tocar  <ilya.tocar@intel.com>
7699             Andrey Turetskiy  <andrey.turetskiy@intel.com>
7700             Ilya Verbin  <ilya.verbin@intel.com>
7701             Kirill Yukhin  <kirill.yukhin@intel.com>
7702             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7704         * config/i386/constraints.md (k): New.
7705         (Yk): Ditto.
7706         * config/i386/i386.c (const regclass_map): Add new mask registers.
7707         (dbx_register_map): Ditto.
7708         (dbx64_register_map): Ditto.
7709         (svr4_dbx_register_map): Ditto.
7710         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
7711         disabled.
7712         (ix86_preferred_reload_class): Disable constants for mask registers.
7713         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
7714         (ix86_hard_regno_mode_ok): Support new mask registers.
7715         (x86_order_regs_for_local_alloc): Ditto.
7716         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
7717         (FIXED_REGISTERS): Add new mask registers.
7718         (CALL_USED_REGISTERS): Ditto.
7719         (REG_ALLOC_ORDER): Ditto.
7720         (VALID_MASK_REG_MODE): New.
7721         (FIRST_MASK_REG): Ditto.
7722         (LAST_MASK_REG): Ditto.
7723         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
7724         (MAYBE_MASK_CLASS_P): New.
7725         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
7726         (REG_CLASS_CONTENTS): Ditto.
7727         (MASK_REGNO_P): New.
7728         (ANY_MASK_REG_P): Ditto.
7729         (HI_REGISTER_NAMES): Add new mask registers.
7730         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
7731         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
7732         mask registers.
7733         (attribute "type"): Add mskmov, msklog.
7734         (attribute "length_immediate"): Support them.
7735         (attribute "memory"): Ditto.
7736         (attribute "prefix_0f"): Ditto.
7737         (*movhi_internal): Support new mask registers.
7738         (*movqi_internal): Ditto.
7739         (define_split): Split out clobber pattern is a logic
7740         insn on mask registers.
7741         (*k<logic><mode>): New.
7742         (*andhi_1): Extend to support mask regs.
7743         (*andqi_1): Extend to support mask regs.
7744         (kandn<mode>): New.
7745         (define_split): Split and-not to and and not if operands
7746         are not mask regs.
7747         (*<code><mode>_1): Separate HI mode to new pattern...
7748         (*<code>hi_1): This.
7749         (*<code>qi_1): Extend to support mask regs.
7750         (kxnor<mode>): New.
7751         (kortestzhi): Ditto.
7752         (kortestchi): Ditto.
7753         (kunpckhi): Ditto.
7754         (*one_cmpl<mode>2_1): Remove HImode and handle it...
7755         (*one_cmplhi2_1): ...Here, now with mask registers support.
7756         (*one_cmplqi2_1): Support new mask registers.
7757         (HI/QImode arithmetics splitter): Don't split if mask registers
7758         are used.
7759         (HI/QImode not splitter): Ditto.
7760         * config/i386/predicated.md (mask_reg_operand): New.
7761         (general_reg_operand): Ditto.
7763 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7765         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
7766         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
7768 2013-09-10  Jeff Law  <law@redhat.com>
7770         PR tree-optimization/58380
7771         * tree-ssa-threadupdate.c (thread_block): Recognize another case
7772         of threading through a buried loop header.
7774         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
7775         return value for single successor case.
7777 2013-09-10  Jan Hubicka  <jh@suse.cz>
7779         * ipa-devirt.c (ipa_devirt): Enable with LTO.
7781 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
7783         PR target/58361
7784         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
7785         support conditional execution.
7786         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
7788 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
7790         * lra.c (lra): Clear lra_optional_reload_pseudos before every
7791         constraint pass.
7792         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
7793         Check destination too to check move insn.
7794         (undo_optional_reloads): Add check that the original peudo did not
7795         changed its allocation and the optional reload was inherited on last
7796         inheritance pass.  Break loop after deciding to keep optional reload.
7797         (lra_undo_inheritance): Add check that inherited pseudo still in
7798         memory.
7800 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
7802         * config/aarch64/aarch64.md (generic_sched): New.
7803         * config/aarch64/aarch64-generic.md (load): Make conditional
7804         on generic_sched attribute.
7805         (nonload): Likewise.
7807 2013-09-10  Jan Hubicka  <jh@suse.cz>
7809         * lto-cgraph.c: Include ipa-utils.h.
7810         (compute_ltrans_boundary): Also add possible targets into the boundary.
7812 2013-09-10  Jan Hubicka  <jh@suse.cz>
7814         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
7815         VAR_DECL of vtable rather than full expression.
7817 2013-09-10  Jan Hubicka  <jh@suse.cz>
7818             Paolo Carlini  <paolo.carlini@oracle.com>
7820         * cgraphunit.c (analyze_functions): Save input_location, set it
7821         to UNKNOWN_LOCATION and restore it at the end.
7823 2013-09-10  Martin Jambor  <mjambor@suse.cz>
7825         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
7826         represented by a thunk.
7828 2013-09-10  Jeff Law  <law@redhat.com>
7830         PR tree-optimization/58343
7831         * tree-ssa-threadupdate.c (thread_block): Identify and disable
7832         jump threading requests through loop headers buried in the middle
7833         of a jump threading path.
7835         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
7836         in return value/type.
7838 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
7840         PR rtl-optimization/58365
7841         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
7842         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
7843         it differs.
7845 2013-09-10  Richard Biener  <rguenther@suse.de>
7847         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
7848         * tree-data-ref.c (create_rdg_vertices): Collect all data
7849         references, signal failure to the caller, use data-ref API.
7850         (build_rdg): Compute data references only once.  Maintain lifetime
7851         of data references and data dependences from within RDG.
7852         (free_rdg): Free dependence relations.
7853         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
7854         inventing extra dependences.
7855         (distribute_loop): Update for RDG API changes.
7857 2013-09-10  Kai Tietz  <ktietz@redhat.com>
7859         * doc/invoke.texi (fms-extensions): Document changed
7860         behavior for ms-abi targets.
7861         * config/i386/i386.c (ix86_option_override_internal):
7862         Set default value of option -fms-extension for ms-abi targets.
7864 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7866         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
7868 2013-09-10  Alan Modra  <amodra@gmail.com>
7870         PR target/58330
7871         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
7873 2013-09-10  Alan Modra  <amodra@gmail.com>
7875         * config/rs6000/predicates.md (add_cint_operand): New.
7876         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
7877         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
7878         using add_cint_operand.
7879         (largetoc_high_plus_aix): Likewise.
7881 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
7883         PR tree-optimization/58364
7884         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
7885         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
7886         the current range can't be an unconditional true or false.
7888 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
7890         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
7892 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
7894         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
7896 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
7898         PR c++/43452
7899         * doc/invoke.texi (-Wdelete-incomplete): Document it.
7901 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
7903         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
7904         NO_REGS for immediate that can't be moved directly into FP_REGS.
7906 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7908         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
7909         comparison with negated operand.
7910         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
7911         RTL form.
7913 2013-09-09  Richard Biener  <rguenther@suse.de>
7915         PR middle-end/58326
7916         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
7917         of a subloop record all its block as affecting loop-closed SSA form.
7919 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7921         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
7922         of an rtx/bitpos pair.
7923         (store_fixed_bit_field): Update accordingly.
7925 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7927         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
7928         GEN_INT.
7929         * builtins.c (expand_errno_check): Likewise.
7930         * dwarf2cfi.c (init_return_column_size): Likewise.
7931         * except.c (sjlj_mark_call_sites): Likewise.
7932         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
7933         * lra-constraints.c (emit_inc): Likewise.
7934         * ree.c (combine_set_extension): Likewise.
7935         * regmove.c (fixup_match_2): Likewise.
7936         * reload1.c (inc_for_reload): Likewise.
7938 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7940         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
7941         (force_to_mode, simplify_shift_const_1, simplify_comparison):
7942         Use gen_int_mode with the mode of the associated simplify_* call.
7943         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
7944         * expmed.c (expand_shift_1): Likewise.
7945         * function.c (instantiate_virtual_regs_in_insn): Likewise.
7946         * loop-iv.c (iv_number_of_iterations): Likewise.
7947         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
7948         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
7950 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7952         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
7953         of the associated expand_* call.
7954         (asan_emit_stack_protection): Likewise.
7955         * builtins.c (round_trampoline_addr): Likewise.
7956         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
7957         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
7958         (emit_store_flag): Likewise.
7959         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
7960         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
7961         Likewise.
7962         * function.c (instantiate_virtual_regs_in_insn): Likewise.
7963         * ifcvt.c (noce_try_store_flag_constants): Likewise.
7964         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
7965         * modulo-sched.c (generate_prolog_epilog): Likewise.
7966         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
7967         (expand_ctz, expand_ffs, expand_unop): Likewise.
7969 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7971         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
7972         of the associated gen_rtx_* call.
7973         * caller-save.c (init_caller_save): Likewise.
7974         * combine.c (find_split_point, make_extraction): Likewise.
7975         (make_compound_operation): Likewise.
7976         * dwarf2out.c (mem_loc_descriptor): Likewise.
7977         * explow.c (plus_constant, probe_stack_range): Likewise.
7978         * expmed.c (expand_mult_const): Likewise.
7979         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
7980         * reload1.c (init_reload): Likewise.
7981         * valtrack.c (cleanup_auto_inc_dec): Likewise.
7982         * var-tracking.c (adjust_mems): Likewise.
7983         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
7984         rather than gen_rtx_fmt_ee.
7986 2013-09-09  Jan Hubicka  <jh@suse.cz>
7988         PR middle-end/58294
7989         * value-prof.c (gimple_ic): Copy also abnormal edges.
7991 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
7993         * asan.c (asan_shadow_cst): Use gen_int_mode.
7995 2013-09-08  Jan Hubicka  <jh@suse.cz>
7997         * ipa-profile.c: Add toplevel comment.
7998         (ipa_propagate_frequency_1): Be more conservative when profile is read.
7999         (contains_hot_call_p): New function.
8000         (ipa_propagate_frequency): Set frequencies based on counts when
8001         profile is read.
8002         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
8003         profile; do not tamper with profile after inlining if it is read.
8005 2013-09-08  Jan Hubicka  <jh@suse.cz>
8007         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
8008         speculative edges.
8010 2013-09-08  Jan Hubicka  <jh@suse.cz>
8012         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
8013         summary generation.
8015 2013-09-08  Jeff Law  <law@redhat.com>
8017         PR bootstrap/58340
8018         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
8019         of 'found'.
8021 2013-09-08  Andi Kleen  <ak@linux.intel.com>
8023         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
8025 2013-09-08  Jan Hubicka  <jh@suse.cz>
8027         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
8029 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
8031         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
8032         for non-debug insns.
8033         * lra.c (new_insn_reg): Take the containing insn as a parameter.
8034         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
8035         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
8036         accordingly.
8038 2013-09-08  Jan Hubicka  <jh@suse.cz>
8040         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
8041         targets and devirtualize to BUILT_IN_UNREACHABLE.
8042         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
8043         * ipa.c (walk_polymorphic_call_targets): New function.
8044         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
8045         functions; use the new timevar.
8046         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
8047         was removed from the program.
8048         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
8049         not consider it in the walk when its vtable is dead.
8050         (possible_polymorphic_call_targets_1): Pass anonymous flag to
8051         record_binfo.
8052         (devirt_variable_node_removal_hook): New function.
8053         (possible_polymorphic_call_targets): Also register
8054         devirt_variable_node_removal_hook.
8055         (ipa_devirt): Do not do non-speculative devirtualization.
8056         (gate_ipa_devirt): One execute if devirtualizing speculatively.
8058 2013-09-08  Jan Hubicka  <jh@suse.cz>
8060         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
8061         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
8062         varpool_remove_variable_insertion_hook): Declare.
8063         * varpool.c (varpool_node_hook_list): New structure.
8064         (first_varpool_node_removal_hook,
8065         first_varpool_variable_insertion_hook): New variables.
8066         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
8067         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
8068         varpool_remove_variable_insertion_hook,
8069         varpool_call_variable_insertion_hooks): New functions.
8070         (varpool_remove_node): Use it.
8072 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
8074         PR c++/54941
8075         * diagnostic.c (diagnostic_build_prefix): When s.file is
8076         "<built-in>" don't output line and column numbers.
8078 2013-09-06  Jan Hubicka  <jh@suse.cz>
8080         * cgraphunit.c (expand_thunk): Get body before touching arguments.
8081         * lto-streamer-out.c: Stream thunks, too.
8082         * lto-streamer-in.c (input_function): Pop cfun here
8083         (lto_read_body): Instead of here.
8085 2013-09-06  Caroline Tice  <cmtice@google.com>
8087         * doc/install.texi: Add documentation for the --enable-vtable-verify
8088         and the --disable-libvtv configure options.
8090 2013-09-06  Jeff Law  <law@redhat.com>
8092         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8093         edge implied equivalences into successor phis.
8095 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
8097         * resource.c (mark_referenced_resources): Handle COND_EXEC.
8099 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
8101         * resource.c (mark_target_live_regs): Compute resources taking
8102         into account if a call is predicated or not.
8104 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
8106         * toplev.c (output_stack_usage): Be prepared for suffixes created by
8107         the compiler in the function names.
8109 2013-09-06  Jan Hubicka  <jh@suse.cz>
8111         PR middle-end/58094
8112         * ipa-inline.c (has_caller_p): New function.
8113         (want_inline_function_to_all_callers_p): Use it.
8114         (sum_callers, inline_to_all_callers): Break out from ...
8115         (ipa_inline): ... here.
8117 2013-09-06  Jan Hubicka  <jh@suse.cz>
8119         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
8120         only when AVX is enabled.
8122 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8124         * config/aarch64/aarch64.md
8125         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
8126         is fpsimd_<load/store>2.
8127         (load_pair<mode>): Likewise.
8128         (store_pair<mode>): Likewise.
8130 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8132         * config/arm/types.md (type): Add "mrs" type.
8133         * config/aarch64/aarch64.md
8134         (aarch64_load_tp_hard): Make type "mrs".
8135         * config/arm/arm.md
8136         (load_tp_hard): Make type "mrs".
8137         * config/arm/cortex-a15.md: Update with new attributes.
8138         * config/arm/cortex-a5.md: Update with new attributes.
8139         * config/arm/cortex-a53.md: Update with new attributes.
8140         * config/arm/cortex-a7.md: Update with new attributes.
8141         * config/arm/cortex-a8.md: Update with new attributes.
8142         * config/arm/cortex-a9.md: Update with new attributes.
8143         * config/arm/cortex-m4.md: Update with new attributes.
8144         * config/arm/cortex-r4.md: Update with new attributes.
8145         * config/arm/fa526.md: Update with new attributes.
8146         * config/arm/fa606te.md: Update with new attributes.
8147         * config/arm/fa626te.md: Update with new attributes.
8148         * config/arm/fa726te.md: Update with new attributes.
8150 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8152         * config/aarch64/aarch64.md
8153         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
8154         (*movtf_aarch64): Likewise.
8155         * config/arm/arm.md
8156         (thumb1_movdi_insn): Use "multiple" for type where more than one
8157         instruction is used for a move.
8158         (*arm32_movhf): Likewise.
8159         (*thumb_movdf_insn): Likewise.
8161 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8163         * config/arm/types.md (type): Rename fcpys to fmov.
8164         * config/arm/vfp.md
8165         (*arm_movsi_vfp): Rename type fcpys as fmov.
8166         (*thumb2_movsi_vfp): Likewise
8167         (*movhf_vfp_neon): Likewise
8168         (*movhf_vfp): Likewise
8169         (*movsf_vfp): Likewise
8170         (*thumb2_movsf_vfp): Likewise
8171         (*movsfcc_vfp): Likewise
8172         (*thumb2_movsfcc_vfp): Likewise
8173         * config/aarch64/aarch64-simd.md
8174         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
8175         * config/aarch64/aarch64.md
8176         (*movsi_aarch64): Replace type mov_reg with fmovs.
8177         (*movdi_aarch64): Likewise
8178         (*movsf_aarch64): Likewise
8179         (*movdf_aarch64): Likewise
8180         * config/arm/arm.c
8181         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
8182         * config/arm/iwmmxt.md
8183         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
8184         * config/arm/arm1020e.md: Update with new attributes.
8185         * config/arm/cortex-a15-neon.md: Update with new attributes.
8186         * config/arm/cortex-a5.md: Update with new attributes.
8187         * config/arm/cortex-a53.md: Update with new attributes.
8188         * config/arm/cortex-a7.md: Update with new attributes.
8189         * config/arm/cortex-a8-neon.md: Update with new attributes.
8190         * config/arm/cortex-a9.md: Update with new attributes.
8191         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8192         * config/arm/cortex-r4f.md: Update with new attributes.
8193         * config/arm/marvell-pj4.md: Update with new attributes.
8194         * config/arm/vfp11.md: Update with new attributes.
8196 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8198         * config/aarch64/aarch64.md
8199         (*madd<mode>): Fix type attribute.
8200         (*maddsi_uxtw): Likewise.
8201         (*msub<mode>): Likewise.
8202         (*msubsi_uxtw): Likewise.
8203         (<su_optab>maddsidi4): Likewise.
8204         (<su_optab>msubsidi4): Likewise.
8206 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8208         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
8209         * config/arm/arm.md (core_cycles): Remove fdiv.
8210         * config/arm/vfp.md:
8211         (*sqrtsf2_vfp): Update for attribute changes.
8212         (*sqrtdf2_vfp): Likewise.
8213         * config/aarch64/aarch64.md:
8214         (sqrt<mode>2): Update for attribute changes.
8215         * config/arm/arm1020e.md: Update with new attributes.
8216         * config/arm/cortex-a15-neon.md: Update with new attributes.
8217         * config/arm/cortex-a5.md: Update with new attributes.
8218         * config/arm/cortex-a53.md: Update with new attributes.
8219         * config/arm/cortex-a7.md: Update with new attributes.
8220         * config/arm/cortex-a8-neon.md: Update with new attributes.
8221         * config/arm/cortex-a9.md: Update with new attributes.
8222         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8223         * config/arm/cortex-r4f.md: Update with new attributes.
8224         * config/arm/marvell-pj4.md: Update with new attributes.
8225         * config/arm/vfp11.md: Update with new attributes.
8227 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8229         * config/arm/types.md
8230         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
8231         * config/aarch64/aarch64.md
8232         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
8233         new attributes.
8234         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
8235         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
8236         (float<GPI:mode><GPF:mode>2): Likewise.
8237         * config/arm/vfp.md
8238         (*truncsisf2_vfp): Update with new attributes.
8239         (*truncsidf2_vfp): Likewise.
8240         (fixuns_truncsfsi2): Likewise.
8241         (fixuns_truncdfsi2): Likewise.
8242         (*floatsisf2_vfp): Likewise.
8243         (*floatsidf2_vfp): Likewise.
8244         (floatunssisf2): Likewise.
8245         (floatunssidf2): Likewise.
8246         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
8247         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
8248         * config/arm/arm1020e.md: Update with new attributes.
8249         * config/arm/cortex-a15-neon.md: Update with new attributes.
8250         * config/arm/cortex-a5.md: Update with new attributes.
8251         * config/arm/cortex-a53.md: Update with new attributes.
8252         * config/arm/cortex-a7.md: Update with new attributes.
8253         * config/arm/cortex-a8-neon.md: Update with new attributes.
8254         * config/arm/cortex-a9.md: Update with new attributes.
8255         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8256         * config/arm/cortex-r4f.md: Update with new attributes.
8257         * config/arm/marvell-pj4.md: Update with new attributes.
8258         * config/arm/vfp11.md: Update with new attributes.
8260 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8262         * config/aarch64/arm_neon.h
8263         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
8264         (vqtbx<1,2,3,4><q>_s8): Likewise.
8266 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8268         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
8269         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
8270         (add<mode>3): Add type attribute.
8271         (add<mode>3): Likewise.
8272         (usadd<mode>3): Likewise.
8273         (ssadd<mode>3): Likewise.
8274         (sub<mode>3): Likewise.
8275         (sub<mode>3): Likewise.
8276         (ussub<mode>3): Likewise.
8277         (sssub<mode>3): Likewise.
8278         (ssmulsa3): Likewise.
8279         (usmulusa3): Likewise.
8280         (arm_usatsihi): Likewise.
8281         * config/arm/vfp.md
8282         (*movdi_vfp): Add types for all instructions.
8283         (*movdi_vfp_cortexa8): Likewise.
8284         (*movhf_vfp_neon): Likewise.
8285         (*movhf_vfp): Likewise.
8286         (*movdf_vfp): Likewise.
8287         (*thumb2_movdf_vfp): Likewise.
8288         (*thumb2_movdfcc_vfp): Likewise.
8289         * config/arm/arm.md: Add type attribute to all insn patterns.
8290         (*thumb1_adddi3): Add type attribute.
8291         (*arm_adddi3): Likewise.
8292         (*adddi_sesidi_di): Likewise.
8293         (*adddi_zesidi_di): Likewise.
8294         (*thumb1_addsi3): Likewise.
8295         (addsi3_compare0): Likewise.
8296         (*addsi3_compare0_scratch): Likewise.
8297         (*compare_negsi_si): Likewise.
8298         (cmpsi2_addneg): Likewise.
8299         (*addsi3_carryin_<optab>): Likewise.
8300         (*addsi3_carryin_alt2_<optab>): Likewise.
8301         (*addsi3_carryin_clobercc_<optab>): Likewise.
8302         (*subsi3_carryin): Likewise.
8303         (*subsi3_carryin_const): Likewise.
8304         (*subsi3_carryin_compare): Likewise.
8305         (*subsi3_carryin_compare_const): Likewise.
8306         (*arm_subdi3): Likewise.
8307         (*thumb_subdi3): Likewise.
8308         (*subdi_di_zesidi): Likewise.
8309         (*subdi_di_sesidi): Likewise.
8310         (*subdi_zesidi_di): Likewise.
8311         (*subdi_sesidi_di): Likewise.
8312         (*subdi_zesidi_ze): Likewise.
8313         (thumb1_subsi3_insn): Likewise.
8314         (*arm_subsi3_insn): Likewise.
8315         (*anddi3_insn): Likewise.
8316         (*anddi_zesidi_di): Likewise.
8317         (*anddi_sesdi_di): Likewise.
8318         (*ne_zeroextracts): Likewise.
8319         (*ne_zeroextracts): Likewise.
8320         (*ite_ne_zeroextr): Likewise.
8321         (*ite_ne_zeroextr): Likewise.
8322         (*anddi_notdi_di): Likewise.
8323         (*anddi_notzesidi): Likewise.
8324         (*anddi_notsesidi): Likewise.
8325         (andsi_notsi_si): Likewise.
8326         (thumb1_bicsi3): Likewise.
8327         (*iordi3_insn): Likewise.
8328         (*iordi_zesidi_di): Likewise.
8329         (*iordi_sesidi_di): Likewise.
8330         (*thumb1_iorsi3_insn): Likewise.
8331         (*xordi3_insn): Likewise.
8332         (*xordi_zesidi_di): Likewise.
8333         (*xordi_sesidi_di): Likewise.
8334         (*arm_xorsi3): Likewise.
8335         (*andsi_iorsi3_no): Likewise.
8336         (*smax_0): Likewise.
8337         (*smax_m1): Likewise.
8338         (*arm_smax_insn): Likewise.
8339         (*smin_0): Likewise.
8340         (*arm_smin_insn): Likewise.
8341         (*arm_umaxsi3): Likewise.
8342         (*arm_uminsi3): Likewise.
8343         (*minmax_arithsi): Likewise.
8344         (*minmax_arithsi_): Likewise.
8345         (*satsi_<SAT:code>): Likewise.
8346         (arm_ashldi3_1bit): Likewise.
8347         (arm_ashrdi3_1bit): Likewise.
8348         (arm_lshrdi3_1bit): Likewise.
8349         (*arm_negdi2): Likewise.
8350         (*thumb1_negdi2): Likewise.
8351         (*arm_negsi2): Likewise.
8352         (*thumb1_negsi2): Likewise.
8353         (*negdi_extendsid): Likewise.
8354         (*negdi_zero_extend): Likewise.
8355         (*arm_abssi2): Likewise.
8356         (*thumb1_abssi2): Likewise.
8357         (*arm_neg_abssi2): Likewise.
8358         (*thumb1_neg_abss): Likewise.
8359         (one_cmpldi2): Likewise.
8360         (extend<mode>di2): Likewise.
8361         (*compareqi_eq0): Likewise.
8362         (*arm_extendhisi2addsi): Likewise.
8363         (*arm_movdi): Likewise.
8364         (*thumb1_movdi_insn): Likewise.
8365         (*arm_movt): Likewise.
8366         (*thumb1_movsi_insn): Likewise.
8367         (pic_add_dot_plus_four): Likewise.
8368         (pic_add_dot_plus_eight): Likewise.
8369         (tls_load_dot_plus_eight): Likewise.
8370         (*thumb1_movhi_insn): Likewise.
8371         (*thumb1_movsf_insn): Likewise.
8372         (*movdf_soft_insn): Likewise.
8373         (*thumb_movdf_insn): Likewise.
8374         (cbranchsi4_insn): Likewise.
8375         (cbranchsi4_scratch): Likewise.
8376         (*negated_cbranchsi4): Likewise.
8377         (*tbit_cbranch): Likewise.
8378         (*tlobits_cbranch): Likewise.
8379         (*tstsi3_cbranch): Likewise.
8380         (*cbranchne_decr1): Likewise.
8381         (*addsi3_cbranch): Likewise.
8382         (*addsi3_cbranch_scratch): Likewise.
8383         (*arm_cmpdi_insn): Likewise.
8384         (*arm_cmpdi_unsig): Likewise.
8385         (*arm_cmpdi_zero): Likewise.
8386         (*thumb_cmpdi_zero): Likewise.
8387         (*deleted_compare): Likewise.
8388         (*mov_scc): Likewise.
8389         (*mov_negscc): Likewise.
8390         (*mov_notscc): Likewise.
8391         (*cstoresi_eq0_thumb1_insn): Likewise.
8392         (cstoresi_nltu_thumb1): Likewise.
8393         (cstoresi_ltu_thu): Likewise.
8394         (thumb1_addsi3_addgeu): Likewise.
8395         (*arm_jump): Likewise.
8396         (*thumb_jump): Likewise.
8397         (*check_arch2): Likewise.
8398         (arm_casesi_internal): Likewise.
8399         (thumb1_casesi_dispatch): Likewise.
8400         (*arm_indirect_jump): Likewise.
8401         (*thumb1_indirect_jump): Likewise.
8402         (nop): Likewise.
8403         (*and_scc): Likewise.
8404         (*ior_scc): Likewise.
8405         (*compare_scc): Likewise.
8406         (*cond_move): Likewise.
8407         (*cond_arith): Likewise.
8408         (*cond_sub): Likewise.
8409         (*cmp_ite0): Likewise.
8410         (*cmp_ite1): Likewise.
8411         (*cmp_and): Likewise.
8412         (*cmp_ior): Likewise.
8413         (*ior_scc_scc): Likewise.
8414         (*ior_scc_scc_cmp): Likewise.
8415         (*and_scc_scc): Likewise.
8416         (*and_scc_scc_cmp): Likewise.
8417         (*and_scc_scc_nod): Likewise.
8418         (*negscc): Likewise.
8419         (movcond_addsi): Likewise.
8420         (movcond): Likewise.
8421         (*ifcompare_plus_move): Likewise.
8422         (*if_plus_move): Likewise.
8423         (*ifcompare_move_plus): Likewise.
8424         (*if_move_plus): Likewise.
8425         (*ifcompare_arith_arith): Likewise.
8426         (*if_arith_arith): Likewise.
8427         (*ifcompare_arith_move): Likewise.
8428         (*if_arith_move): Likewise.
8429         (*ifcompare_move_arith): Likewise.
8430         (*if_move_arith): Likewise.
8431         (*ifcompare_move_not): Likewise.
8432         (*if_move_not): Likewise.
8433         (*ifcompare_not_move): Likewise.
8434         (*if_not_move): Likewise.
8435         (*ifcompare_shift_move): Likewise.
8436         (*if_shift_move): Likewise.
8437         (*ifcompare_move_shift): Likewise.
8438         (*if_move_shift): Likewise.
8439         (*ifcompare_shift_shift): Likewise.
8440         (*ifcompare_not_arith): Likewise.
8441         (*ifcompare_arith_not): Likewise.
8442         (*if_arith_not): Likewise.
8443         (*ifcompare_neg_move): Likewise.
8444         (*if_neg_move): Likewise.
8445         (*ifcompare_move_neg): Likewise.
8446         (*if_move_neg): Likewise.
8447         (prologue_thumb1_interwork): Likewise.
8448         (*cond_move_not): Likewise.
8449         (*sign_extract_onebit): Likewise.
8450         (*not_signextract_onebit): Likewise.
8451         (stack_tie): Likewise.
8452         (align_4): Likewise.
8453         (align_8): Likewise.
8454         (consttable_end): Likewise.
8455         (consttable_1): Likewise.
8456         (consttable_2): Likewise.
8457         (consttable_4): Likewise.
8458         (consttable_8): Likewise.
8459         (consttable_16): Likewise.
8460         (*thumb1_tablejump): Likewise.
8461         (prefetch): Likewise.
8462         (force_register_use): Likewise.
8463         (thumb_eh_return): Likewise.
8464         (load_tp_hard): Likewise.
8465         (load_tp_soft): Likewise.
8466         (tlscall): Likewise.
8467         (*arm_movtas_ze): Likewise.
8468         (*arm_rev): Likewise.
8469         (*arm_revsh): Likewise.
8470         (*arm_rev16): Likewise.
8471         * config/arm/thumb2.md
8472         (*thumb2_smaxsi3): Likewise.
8473         (*thumb2_sminsi3): Likewise.
8474         (*thumb32_umaxsi3): Likewise.
8475         (*thumb2_uminsi3): Likewise.
8476         (*thumb2_negdi2): Likewise.
8477         (*thumb2_abssi2): Likewise.
8478         (*thumb2_neg_abss): Likewise.
8479         (*thumb2_movsi_insn): Likewise.
8480         (tls_load_dot_plus_four): Likewise.
8481         (*thumb2_movhi_insn): Likewise.
8482         (*thumb2_mov_scc): Likewise.
8483         (*thumb2_mov_negs): Likewise.
8484         (*thumb2_mov_negs): Likewise.
8485         (*thumb2_mov_nots): Likewise.
8486         (*thumb2_mov_nots): Likewise.
8487         (*thumb2_movsicc_): Likewise.
8488         (*thumb2_movsfcc_soft_insn): Likewise.
8489         (*thumb2_indirect_jump): Likewise.
8490         (*thumb2_and_scc): Likewise.
8491         (*thumb2_ior_scc): Likewise.
8492         (*thumb2_ior_scc_strict_it): Likewise.
8493         (*thumb2_cond_move): Likewise.
8494         (*thumb2_cond_arith): Likewise.
8495         (*thumb2_cond_ari): Likewise.
8496         (*thumb2_cond_sub): Likewise.
8497         (*thumb2_negscc): Likewise.
8498         (*thumb2_movcond): Likewise.
8499         (thumb2_casesi_internal): Likewise.
8500         (thumb2_casesi_internal_pic): Likewise.
8501         (*thumb2_alusi3_short): Likewise.
8502         (*thumb2_mov<mode>_shortim): Likewise.
8503         (*thumb2_addsi_short): Likewise.
8504         (*thumb2_subsi_short): Likewise.
8505         (thumb2_addsi3_compare0): Likewise.
8506         (*thumb2_cbz): Likewise.
8507         (*thumb2_cbnz): Likewise.
8508         (*thumb2_one_cmplsi2_short): Likewise.
8509         (*thumb2_negsi2_short): Likewise.
8510         (*orsi_notsi_si): Likewise.
8511         * config/arm/arm1020e.md: Update with new attributes.
8512         * config/arm/arm1026ejs.md: Update with new attributes.
8513         * config/arm/arm1136jfs.md: Update with new attributes.
8514         * config/arm/arm926ejs.md: Update with new attributes.
8515         * config/arm/cortex-a15.md: Update with new attributes.
8516         * config/arm/cortex-a5.md: Update with new attributes.
8517         * config/arm/cortex-a53.md: Update with new attributes.
8518         * config/arm/cortex-a7.md: Update with new attributes.
8519         * config/arm/cortex-a8.md: Update with new attributes.
8520         * config/arm/cortex-a9.md: Update with new attributes.
8521         * config/arm/cortex-m4.md: Update with new attributes.
8522         * config/arm/cortex-r4.md: Update with new attributes.
8523         * config/arm/fa526.md: Update with new attributes.
8524         * config/arm/fa606te.md: Update with new attributes.
8525         * config/arm/fa626te.md: Update with new attributes.
8526         * config/arm/fa726te.md: Update with new attributes.
8528 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8530         * config/aarch64/aarch64-simd.md
8531         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
8532         <vwx> iterator to ensure correct register choice.
8533         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
8534         (aarch64_sqdmull_n<mode>): Likewise.
8535         (aarch64_sqdmull2_n<mode>_internal): Likewise.
8536         * config/aarch64/arm_neon.h
8537         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
8538         (vml<as><q>_n_<su>16): Likewise.
8539         (vml<as>l_high_lane<q>_<su>16): Likewise.
8540         (vml<as>l_high_n_<su>16): Likewise.
8541         (vml<as>l_lane<q>_<su>16): Likewise.
8542         (vml<as>l_n_<su>16): Likewise.
8543         (vmul<q>_lane<q>_<su>16): Likewise.
8544         (vmul<q>_n_<su>16): Likewise.
8545         (vmull_lane<q>_<su>16): Likewise.
8546         (vmull_n_<su>16): Likewise.
8547         (vmull_high_lane<q>_<su>16): Likewise.
8548         (vmull_high_n_<su>16): Likewise.
8549         (vqrdmulh<q>_n_s16): Likewise.
8551 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
8553         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
8554         have the correct lane parameter.
8556 2013-09-06  Richard Biener <rguenther@suse.de>
8558         * cfganal.c (control_dependences::~control_dependences):
8559         Properly free all of the vector.
8561 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
8563         PR target/58269
8564         * config/i386/i386.c (ix86_conditional_register_usage):
8565         Proper initialize extended SSE registers.
8567 2013-09-06  Jan Hubicka  <jh@suse.cz>
8569         PR tree-optimization/58311
8570         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
8572 2013-09-06  Jan Hubicka  <jh@suse.cz>
8574         * Makefile.in (tree-sra.o): Update dependencies.
8575         * tree-sra.c: Include ipa-utils.h
8576         (scan_function): Use recursive_call_p.
8577         (has_caller_p): New function.
8578         (cgraph_for_node_and_aliases): Count also callers of aliases.
8580 2013-09-06  Jan Hubicka  <jh@suse.cz>
8582         PR middle-end/58094
8583         * cgraph.h (symtab_semantically_equivalent_p): Declare.
8584         * tree-tailcall.c: Include ipa-utils.h.
8585         (find_tail_calls): Use it.
8586         * ipa-pure-const.c (check_call): Likewise.
8587         * ipa-utils.c (recursive_call_p): New function.
8588         * ipa-utils.h (recursive_call_p): Dclare.
8589         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
8590         (symtab_semantically_equivalent_p): New function.
8591         * Makefile.in (tree-tailcall.o): Update dependencies.
8593 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
8595         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
8596         non-inlinable part.
8598 2013-09-06  Richard Biener  <rguenther@suse.de>
8600         * lto-streamer.h (lto_global_var_decls): Remove.
8601         * Makefile.in (OBJS): Remove lto-symtab.o.
8602         (lto-symtab.o): Remove.
8603         (GTFILES): Remove lto-symtab.c
8604         * lto-symtab.c: Move to lto/
8606 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8608         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
8609         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
8610         (UNSPEC_FPINT_RINT): New constant definitions.
8611         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
8612         definition with 2 attributes.
8613         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
8614         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
8615         definitions.
8617 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8619         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
8620         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
8621         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
8622         attribute to "z196_alone".
8623         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
8624         "zEC12_simple".
8626 2013-09-06  Richard Biener  <rguenther@suse.de>
8628         * basic-block.h (class control_dependences): New.
8629         * tree-ssa-dce.c (control_dependence_map): Remove.
8630         (cd): New global.
8631         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
8632         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
8633         find_pdom, find_control_dependence, find_all_control_dependences):
8634         Move to cfganal.c.
8635         (mark_control_dependent_edges_necessary,
8636         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
8637         tree_dce_done, perform_tree_ssa_dce): Adjust.
8638         * cfganal.c (set_control_dependence_map_bit,
8639         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
8640         find_all_control_dependences): Move from tree-ssa-dce.c and
8641         implement as methods of control_dependences class.
8642         (control_dependences::control_dependences): New.
8643         (control_dependences::~control_dependences): Likewise.
8644         (control_dependences::get_edges_dependent_on): Likewise.
8645         (control_dependences::get_edge): Likewise.
8647 2013-09-04  Jan Hubicka  <jh@suse.cz>
8649         * tree.c (types_same_for_odr): Drop overactive check.
8650         * ipa-devirt.c (hash_type_name): Likewise.
8652 2013-09-04  Jan Hubicka  <jh@suse.cz>
8654         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
8655         (analyze_functions): ... here.
8657 2013-09-04  Jan Hubicka  <jh@suse.cz>
8659         PR middle-end/58201
8660         * cgraphunit.c (analyze_functions): Clear AUX fields
8661         after processing; initialize assembler name has.
8663 2013-09-05  Jeff Law  <law@redhat.com>
8665         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
8666         from thread_around_empty_block.  Record threading path into PATH.
8667         Recurse if threading through the initial block is successful.
8668         (thread_across_edge): Corresponding changes to slightly simplify.
8670 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
8672         * config/aarch64/aarch64.md
8673         (type): Remove frecpe, frecps, frecpx.
8674         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
8675         fix to be a TARGET_SIMD instruction.
8676         (aarch64_frecps): Remove.
8677         * config/aarch64/aarch64-simd.md
8678         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
8679         (aarch64_frecps<mode>): Handle all float/vector of float modes.
8681 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
8682             Sofiane Naci  <sofiane.naci@arm.com>
8684         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
8685         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
8686         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
8687         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
8688         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
8689         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
8690         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
8691         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
8692         into "clz, "rbit".  Rename "shift" to "shift_imm".
8693         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
8694         changes.  Update for attribute changes all occurrences of arlo_* and
8695         shift* types.
8696         * config/arm/arm-fixed.md: Update for attribute changes
8697         all occurrences of arlo_* types.
8698         * config/arm/thumb2.md: Update for attribute changes all occurrences
8699         of arlo_* types.
8700         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
8701         (cortexa7_older_only): Likewise.
8702         (cortexa7_younger):  Likewise.
8703         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
8704         (1020alu_shift_op): Likewise.
8705         (1020alu_shift_reg_op): Likewise.
8706         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
8707         (alu_shift_op): Likewise.
8708         (alu_shift_reg_op): Likewise.
8709         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
8710         (11_alu_shift_op): Likewise.
8711         (11_alu_shift_reg_op): Likewise.
8712         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
8713         (9_alu_shift_reg_op): Likewise.
8714         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
8715         attribute changes.
8716         (cortex_a15_alu_shift): Likewise.
8717         (cortex_a15_alu_shift_reg): Likewise.
8718         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
8719         attribute changes.
8720         (cortex_a5_alu_shift): Likewise.
8721         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
8722         attribute changes.
8723         (cortex_a53_alu_shift): Likewise.
8724         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
8725         attribute changes.
8726         (cortex_a7_alu_reg): Likewise.
8727         (cortex_a7_alu_shift): Likewise.
8728         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
8729         attribute changes.
8730         (cortex_a8_alu_shift): Likewise.
8731         (cortex_a8_alu_shift_reg): Likewise.
8732         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
8733         (cortex_a9_dp_shift): Likewise.
8734         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
8735         attribute changes.
8736         * config/arm/cortex-r4.md
8737         (cortex_r4_alu): Update for attribute changes.
8738         (cortex_r4_mov): Likewise.
8739         (cortex_r4_alu_shift_reg): Likewise.
8740         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
8741         (526_alu_shift_op): Likewise.
8742         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
8743         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
8744         (626te_alu_shift_op): Likewise.
8745         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
8746         (726te_alu_shift_op): Likewise.
8747         (726te_alu_shift_reg_op): Likewise.
8748         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
8749         (mp626_alu_shift_op): Likewise.
8750         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
8751         (pj4_alu_conds): Likewise.
8752         (pj4_shift): Likewise.
8753         (pj4_shift_conds): Likewise.
8754         (pj4_alu_shift): Likewise.
8755         (pj4_alu_shift_conds): Likewise.
8756         * config/aarch64/aarch64.md: Update for attribute change
8757         all occurrences of arlo_* and shift* types.
8759 2013-09-05  Mike Stump  <mikestump@comcast.net>
8761         * tree.h: Move documentation for tree_function_decl to tree-core.h
8762         with the declaration.
8764 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
8766         PR target/58139
8767         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
8768         looking for widest mode.
8770 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
8772         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
8774 2013-09-05  Richard Biener  <rguenther@suse.de>
8776         PR tree-optimization/58137
8777         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
8778         Do not create vectors of pointers.
8779         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
8780         types for the components of the vector initializer.
8781         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
8782         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
8784 2013-09-05  Martin Jambor  <mjambor@suse.cz>
8786         * ipa-prop.c (remove_described_reference): Accept missing references,
8787         return false if that hppens, otherwise return true.
8788         (cgraph_node_for_jfunc): New function.
8789         (try_decrement_rdesc_refcount): Likewise.
8790         (try_make_edge_direct_simple_call): Use them.
8791         (ipa_edge_removal_hook): Remove references from rdescs.
8792         (ipa_edge_duplication_hook): Clone rdescs and their references
8793         when the new edge has the same caller as the old one.
8794         * cgraph.c (cgraph_resolve_speculation): Remove speculative
8795         reference before removing any edges.
8797 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
8799         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
8800         initial store.
8801         * thumb2.md (thumb2_storewb_parisi): New pattern.
8803 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
8805         * config/aarch64/aarch64-option-extensions.def: Add
8806         AARCH64_OPT_EXTENSION of 'crc'.
8807         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
8808         (AARCH64_ISA_CRC): Ditto.
8809         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
8810         description of the CRC extension.
8812 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8814         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
8815         * config/rs6000/linux.h: Ditto.
8816         * alpha/linux.h: Ditto.
8817         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
8818         no_c99_libc_has_function.
8819         * config/c6x/uclinux-elf.h: Ditto.
8820         * config/lm32/uclinux-elf.h: Ditto.
8821         * config/m68k/uclinux.h: Ditto.
8822         * config/moxie/uclinux.h: Ditto.
8823         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
8824         (crisv32-*-linux*, cris-*-linux*): Ditto.
8825         * config/bfin/bfin.c: Include "tm_p.h".
8827 2013-09-05  Richard Biener  <rguenther@suse.de>
8829         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
8830         check for a definition without a basic-block.
8832 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
8833             Sofiane Naci <sofiane.naci@arm.com>
8835         * config/aarch64/aarch64.md
8836         (*movti_aarch64): Rename r_2_f and f_2_r.
8837         (*movsf_aarch64): Likewise.
8838         (*movdf_aarch64): Likewise.
8839         (*movtf_aarch64): Likewise.
8840         (aarch64_movdi_<mode>low): Likewise.
8841         (aarch64_movdi_<mode>high): Likewise.
8842         (aarch64_mov<mode>high_di): Likewise.
8843         (aarch64_mov<mode>low_di): Likewise.
8844         (aarch64_movtilow_tilow): Likewise.
8845         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
8846         values to config/arm/types.md
8847         (attribute "conds"): Update for attribute change.
8848         (anddi3_insn): Likewise.
8849         (iordi3_insn): Likewise.
8850         (xordi3_insn): Likewise.
8851         (one_cmpldi2): Likewise.
8852         * config/arm/types.md (type): Add Neon types.
8853         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
8854         use "type" attribute.
8855         (movmisalign<mode>_neon_store): Likewise.
8856         (movmisalign<mode>_neon_load): Likewise.
8857         (vec_set<mode>_internal): Likewise.
8858         (vec_setv2di_internal): Likewise.
8859         (vec_extract<mode>): Likewise.
8860         (vec_extractv2di): Likewise.
8861         (add<mode>3_neon): Likewise.
8862         (adddi3_neon): Likewise.
8863         (sub<mode>3_neon): Likewise.
8864         (subdi3_neon): Likewise.
8865         (mul<mode>3_neon): Likewise.
8866         (mul<mode>3add<mode>_neon): Likewise.
8867         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
8868         (fma<VCVTF:mode>4)): Likewise.
8869         (fma<VCVTF:mode>4_intrinsic): Likewise.
8870         (fmsub<VCVTF:mode>4)): Likewise.
8871         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
8872         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
8873         (ior<mode>3): Likewise.
8874         (and<mode>3): Likewise.
8875         (anddi3_neon): Likewise.
8876         (orn<mode>3_neon): Likewise.
8877         (orndi3_neon): Likewise.
8878         (bic<mode>3_neon): Likewise.
8879         (bicdi3_neon): Likewise.
8880         (xor<mode>3): Likewise.
8881         (one_cmpl<mode>2): Likewise.
8882         (abs<mode>2): Likewise.
8883         (neg<mode>2): Likewise.
8884         (umin<mode>3_neon): Likewise.
8885         (umax<mode>3_neon): Likewise.
8886         (smin<mode>3_neon): Likewise.
8887         (smax<mode>3_neon): Likewise.
8888         (vashl<mode>3): Likewise.
8889         (vashr<mode>3_imm): Likewise.
8890         (vlshr<mode>3_imm): Likewise.
8891         (ashl<mode>3_signed): Likewise.
8892         (ashl<mode>3_unsigned): Likewise.
8893         (neon_load_count): Likewise.
8894         (ashldi3_neon_noclobber): Likewise.
8895         (signed_shift_di3_neon): Likewise.
8896         (unsigned_shift_di3_neon): Likewise.
8897         (ashrdi3_neon_imm_noclobber): Likewise.
8898         (lshrdi3_neon_imm_noclobber): Likewise.
8899         (widen_ssum<mode>3): Likewise.
8900         (widen_usum<mode>3): Likewise.
8901         (quad_halves_<code>v4si): Likewise.
8902         (quad_halves_<code>v4sf): Likewise.
8903         (quad_halves_<code>v8hi): Likewise.
8904         (quad_halves_<code>v16qi): Likewise.
8905         (reduc_splus_v2di): Likewise.
8906         (neon_vpadd_internal<mode>): Likewise.
8907         (neon_vpsmin<mode>): Likewise.
8908         (neon_vpsmax<mode>): Likewise.
8909         (neon_vpumin<mode>): Likewise.
8910         (neon_vpumax<mode>): Likewise.
8911         (ss_add<mode>_neon): Likewise.
8912         (us_add<mode>_neon): Likewise.
8913         (ss_sub<mode>_neon): Likewise.
8914         (us_sub<mode>_neon): Likewise.
8915         (neon_vadd<mode>_unspec): Likewise.
8916         (neon_vaddl<mode>): Likewise.
8917         (neon_vaddw<mode>): Likewise.
8918         (neon_vhadd<mode>): Likewise.
8919         (neon_vqadd<mode>): Likewise.
8920         (neon_vaddhn<mode>): Likewise.
8921         (neon_vmul<mode>): Likewise.
8922         (neon_vmla<mode>): Likewise.
8923         (neon_vmlal<mode>): Likewise.
8924         (neon_vmls<mode>): Likewise.
8925         (neon_vmlsl<mode>): Likewise.
8926         (neon_vqdmulh<mode>): Likewise.
8927         (neon_vqdmlal<mode>): Likewise.
8928         (neon_vqdmlsl<mode>): Likewise.
8929         (neon_vmull<mode>): Likewise.
8930         (neon_vqdmull<mode>): Likewise.
8931         (neon_vsub<mode>_unspec): Likewise.
8932         (neon_vsubl<mode>): Likewise.
8933         (neon_vsubw<mode>): Likewise.
8934         (neon_vqsub<mode>): Likewise.
8935         (neon_vhsub<mode>): Likewise.
8936         (neon_vsubhn<mode>): Likewise.
8937         (neon_vceq<mode>): Likewise.
8938         (neon_vcge<mode>): Likewise.
8939         (neon_vcgeu<mode>): Likewise.
8940         (neon_vcgt<mode>): Likewise.
8941         (neon_vcgtu<mode>): Likewise.
8942         (neon_vcle<mode>): Likewise.
8943         (neon_vclt<mode>): Likewise.
8944         (neon_vcage<mode>): Likewise.
8945         (neon_vcagt<mode>): Likewise.
8946         (neon_vtst<mode>): Likewise.
8947         (neon_vabd<mode>): Likewise.
8948         (neon_vabdl<mode>): Likewise.
8949         (neon_vaba<mode>): Likewise.
8950         (neon_vabal<mode>): Likewise.
8951         (neon_vmax<mode>): Likewise.
8952         (neon_vmin<mode>): Likewise.
8953         (neon_vpaddl<mode>): Likewise.
8954         (neon_vpadal<mode>): Likewise.
8955         (neon_vpmax<mode>): Likewise.
8956         (neon_vpmin<mode>): Likewise.
8957         (neon_vrecps<mode>): Likewise.
8958         (neon_vrsqrts<mode>): Likewise.
8959         (neon_vqabs<mode>): Likewise.
8960         (neon_vqneg<mode>): Likewise.
8961         (neon_vcls<mode>): Likewise.
8962         (clz<mode>2): Likewise.
8963         (popcount<mode>2): Likewise.
8964         (neon_vrecpe): Likewise.
8965         (neon_vrsqrte): Likewise.
8966         (neon_vget_lane<mode>_sext_internal): Likewise.
8967         (neon_vget_lane<mode>_zext_internal): Likewise.
8968         (neon_vdup_n<mode>): Likewise.
8969         (neon_vdup_nv2di): Likewise.
8970         (neon_vdpu_lane<mode>_internal): Likewise.
8971         (neon_vswp<mode>): Likewise.
8972         (float<mode><V_cvtto>2): Likewise.
8973         (floatuns<mode><V_cvtto>2): Likewise.
8974         (fix_trunc<mode><V_cvtto>)2): Likewise
8975         (fixuns_trunc<mode><V_cvtto)2): Likewise.
8976         (neon_vcvt<mode>): Likewise.
8977         (neon_vcvtv4sfv4hf): Likewise.
8978         (neon_vcvtv4hfv4sf): Likewise.
8979         (neon_vcvt_n<mode>): Likewise.
8980         (neon_vmovn<mode>): Likewise.
8981         (neon_vqmovn<mode>): Likewise.
8982         (neon_vqmovun<mode>): Likewise.
8983         (neon_vmovl<mode>): Likewise.
8984         (neon_vmul_lane<mode>): Likewise.
8985         (neon_vmull_lane<mode>): Likewise.
8986         (neon_vqdmull_lane<mode>): Likewise.
8987         (neon_vqdmulh_lane<mode>): Likewise.
8988         (neon_vmla_lane<mode>): Likewise.
8989         (neon_vmlal_lane<mode>): Likewise.
8990         (neon_vqdmlal_lane<mode>): Likewise.
8991         (neon_vmls_lane<mode>): Likewise.
8992         (neon_vmlsl_lane<mode>): Likewise.
8993         (neon_vqdmlsl_lane<mode>): Likewise.
8994         (neon_vext<mode>): Likewise.
8995         (neon_vrev64<mode>): Likewise.
8996         (neon_vrev32<mode>): Likewise.
8997         (neon_vrev16<mode>): Likewise.
8998         (neon_vbsl<mode>_internal): Likewise.
8999         (neon_vshl<mode>): Likewise.
9000         (neon_vqshl<mode>): Likewise.
9001         (neon_vshr_n<mode>): Likewise.
9002         (neon_vshrn_n<mode>): Likewise.
9003         (neon_vqshrn_n<mode>): Likewise.
9004         (neon_vqshrun_n<mode>): Likewise.
9005         (neon_vshl_n<mode>): Likewise.
9006         (neon_vqshl_n<mode>): Likewise.
9007         (neon_vqshlu_n<mode>): Likewise.
9008         (neon_vshll_n<mode>): Likewise.
9009         (neon_vsra_n<mode>): Likewise.
9010         (neon_vsri_n<mode>): Likewise.
9011         (neon_vsli_n<mode>): Likewise.
9012         (neon_vtbl1v8qi): Likewise.
9013         (neon_vtbl2v8qi): Likewise.
9014         (neon_vtbl3v8qi): Likewise.
9015         (neon_vtbl4v8qi): Likewise.
9016         (neon_vtbx1v8qi): Likewise.
9017         (neon_vtbx2v8qi): Likewise.
9018         (neon_vtbx3v8qi): Likewise.
9019         (neon_vtbx4v8qi): Likewise.
9020         (neon_vtrn<mode>_internal): Likewise.
9021         (neon_vzip<mode>_internal): Likewise.
9022         (neon_vuzp<mode>_internal): Likewise.
9023         (neon_vld1<mode>): Likewise.
9024         (neon_vld1_lane<mode>): Likewise.
9025         (neon_vld1_dup<mode>): Likewise.
9026         (neon_vld1_dupv2di): Likewise.
9027         (neon_vst1<mode>): Likewise.
9028         (neon_vst1_lane<mode>): Likewise.
9029         (neon_vld2<mode>): Likewise.
9030         (neon_vld2_lane<mode>): Likewise.
9031         (neon_vld2_dup<mode>): Likewise.
9032         (neon_vst2<mode>): Likewise.
9033         (neon_vst2_lane<mode>): Likewise.
9034         (neon_vld3<mode>): Likewise.
9035         (neon_vld3qa<mode>): Likewise.
9036         (neon_vld3qb<mode>): Likewise.
9037         (neon_vld3_lane<mode>): Likewise.
9038         (neon_vld3_dup<mode>): Likewise.
9039         (neon_vst3<mode>): Likewise.
9040         (neon_vst3qa<mode>): Likewise.
9041         (neon_vst3qb<mode>): Likewise.
9042         (neon_vst3_lane<mode>): Likewise.
9043         (neon_vld4<mode>): Likewise.
9044         (neon_vld4qa<mode>): Likewise.
9045         (neon_vld4qb<mode>): Likewise.
9046         (neon_vld4_lane<mode>): Likewise.
9047         (neon_vld4_dup<mode>): Likewise.
9048         (neon_vst4<mode>): Likewise.
9049         (neon_vst4qa<mode>): Likewise.
9050         (neon_vst4qb<mode>): Likewise.
9051         (neon_vst4_lane<mode>): Likewise.
9052         (neon_vec_unpack<US>_lo_<mode>): Likewise.
9053         (neon_vec_unpack<US>_hi_<mode>): Likewise.
9054         (neon_vec_<US>mult_lo_<mode>): Likewise.
9055         (neon_vec_<US>mult_hi_<mode>): Likewise.
9056         (neon_vec_<US>shiftl_<mode>): Likewise.
9057         (neon_unpack<US>_<mode>): Likewise.
9058         (neon_vec_<US>mult_<mode>): Likewise.
9059         (vec_pack_trunc_<mode>): Likewise.
9060         (neon_vec_pack_trunk_<mode>): Likewise.
9061         (neon_vabd<mode>_2): Likewise.
9062         (neon_vabd<mode>_3): Likewise.
9063         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
9064         (thumb2_movsi_vfp): Likewise.
9065         (movdi_vfp): Likewise.
9066         (movdi_vfp_cortexa8): Likewise.
9067         (movhf_vfp_neon): Likewise.
9068         (movhf_vfp): Likewiwse.
9069         (movsf_vfp): Likewiwse.
9070         (thumb2_movsf_vfp): Likewiwse.
9071         (movdf_vfp): Likewise.
9072         (thumb2_movdf_vfp): Likewise.
9073         (movsfcc_vfp): Likewise.
9074         (thumb2_movsfcc_vfp): Likewise.
9075         (movdfcc_vfp): Likewise.
9076         (thumb2_movdfcc_vfp): Likewise.
9077         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
9078         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
9079         (v10_v2c): Likewise.
9080         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
9081         attribute change.
9082         (cortex_a15_neon_int_2): Likewise.
9083         (cortex_a15_neon_int_3): Likewise.
9084         (cortex_a15_neon_int_4): Likewise.
9085         (cortex_a15_neon_int_5): Likewise.
9086         (cortex_a15_neon_vqneg_vqabs): Likewise.
9087         (cortex_a15_neon_vmov): Likewise.
9088         (cortex_a15_neon_vaba): Likewise.
9089         (cortex_a15_neon_vaba_qqq): Likewise.
9090         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9091         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9092         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
9093         scalar_64_32_long_scalar): Likewise.
9094         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9095         (cortex_a15_neon_mla_qqq_8_16): Likewise.
9096         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9097         lotype_qdd_64_32_long): Likewise.
9098         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9099         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9100         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
9101         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9102         (cortex_a15_neon_shift_1): Likewise.
9103         (cortex_a15_neon_shift_2): Likewise.
9104         (cortex_a15_neon_shift_3): Likewise.
9105         (cortex_a15_neon_vshl_ddd): Likewise.
9106         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9107         (cortex_a15_neon_vsra_vrsra): Likewise.
9108         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
9109         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
9110         (cortex_a15_neon_fp_vmul_ddd): Likewise.
9111         (cortex_a15_neon_fp_vmul_qqd): Likewise.
9112         (cortex_a15_neon_fp_vmla_ddd): Likewise.
9113         (cortex_a15_neon_fp_vmla_qqq): Likewise.
9114         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
9115         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
9116         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9117         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9118         (cortex_a15_neon_bp_simple): Likewise.
9119         (cortex_a15_neon_bp_2cycle): Likewise.
9120         (cortex_a15_neon_bp_3cycle): Likewise.
9121         (cortex_a15_neon_vld1_1_2_regs): Likewise.
9122         (cortex_a15_neon_vld1_3_4_regs): Likewise.
9123         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9124         (cortex_a15_neon_vld2_4_regs): Likewise.
9125         (cortex_a15_neon_vld3_vld4): Likewise.
9126         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9127         (cortex_a15_neon_vst1_3_4_regs): Likewise.
9128         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
9129         (cortex_a15_neon_vst3_vst4): Likewise.
9130         (cortex_a15_neon_vld1_vld2_lane): Likewise.
9131         (cortex_a15_neon_vld3_vld4_lane" 10
9132         (cortex_a15_neon_vst1_vst2_lane): Likewise.
9133         (cortex_a15_neon_vst3_vst4_lane): Likewise.
9134         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
9135         (cortex_a15_neon_ldm_2): Likewise.0
9136         (cortex_a15_neon_stm_2): Likewise.
9137         (cortex_a15_neon_mcr): Likewise.
9138         (cortex_a15_neon_mcr_2_mcrr): Likewise.
9139         (cortex_a15_neon_mrc): Likewise.
9140         (cortex_a15_neon_mrrc): Likewise.
9141         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
9142         change.
9143         (cortex_a15_alu_shift): Likewise.
9144         (cortex_a15_alu_shift_reg): Likewise.
9145         (cortex_a15_mult32): Likewise.
9146         (cortex_a15_mult64): Likewise.
9147         (cortex_a15_block): Likewise.
9148         (cortex_a15_branch): Likewise.
9149         (cortex_a15_load1): Likewise.
9150         (cortex_a15_load3): Likewise.
9151         (cortex_a15_store1): Likewise.
9152         (cortex_a15_store3): Likewise.
9153         (cortex_a15_call): Likewise.
9154         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
9155         (cortex_a5_f2r): Likewise.
9156         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
9157         change.
9158         (cortex_a53_f2r): Likewise.
9159         * config/arm/cortex-a7.md
9160         (cortex_a7_branch): Update for attribute change.
9161         (cortex_a7_call): Likewise.
9162         (cortex_a7_alu_imm): Likewise.
9163         (cortex_a7_alu_reg): Likewise.
9164         (cortex_a7_alu_shift): Likewise.
9165         (cortex_a7_mul): Likewise.
9166         (cortex_a7_load1): Likewise.
9167         (cortex_a7_store1): Likewise.
9168         (cortex_a7_load2): Likewise.
9169         (cortex_a7_store2): Likewise.
9170         (cortex_a7_load3): Likewise.
9171         (cortex_a7_store3): Likewise.
9172         (cortex_a7_load4): Likewise.
9173         (cortex_a7_store4): Likewise.
9174         (cortex_a7_fpalu): Likewise.
9175         (cortex_a7_fconst): Likewise.
9176         (cortex_a7_fpmuls): Likewise.
9177         (cortex_a7_neon_mul): Likewise.
9178         (cortex_a7_fpmacs): Likewise.
9179         (cortex_a7_neon_mla: Likewise.
9180         (cortex_a7_fpmuld: Likewise.
9181         (cortex_a7_fpmacd: Likewise.
9182         (cortex_a7_fpfmad: Likewise.
9183         (cortex_a7_fdivs: Likewise.
9184         (cortex_a7_fdivd: Likewise.
9185         (cortex_a7_r2f: Likewise.
9186         (cortex_a7_f2r: Likewise.
9187         (cortex_a7_f_flags: Likewise.
9188         (cortex_a7_f_loads: Likewise.
9189         (cortex_a7_f_loadd: Likewise.
9190         (cortex_a7_f_stores: Likewise.
9191         (cortex_a7_f_stored: Likewise.
9192         (cortex_a7_neon): Likewise.
9193         * config/arm/cortex-a8-neon.md
9194         (cortex_a8_neon_mrc): Update for attribute change.
9195         (cortex_a8_neon_mrrc): Likewise.
9196         (cortex_a8_neon_int_1): Likewise.
9197         (cortex_a8_neon_int_2): Likewise.
9198         (cortex_a8_neon_int_3): Likewise.
9199         (cortex_a8_neon_int_4): Likewise.
9200         (cortex_a8_neon_int_5): Likewise.
9201         (cortex_a8_neon_vqneg_vqabs): Likewise.
9202         (cortex_a8_neon_vmov): Likewise.
9203         (cortex_a8_neon_vaba): Likewise.
9204         (cortex_a8_neon_vaba_qqq): Likewise.
9205         (cortex_a8_neon_vsma): Likewise.
9206         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9207         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9208         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
9209         long_scalar): Likewise.
9210         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9211         (cortex_a8_neon_mla_qqq_8_16): Likewise.
9212         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9213         long_scalar_qdd_64_32_long): Likewise.
9214         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9215         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9216         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
9217         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9218         (cortex_a8_neon_shift_1): Likewise.
9219         (cortex_a8_neon_shift_2): Likewise.
9220         (cortex_a8_neon_shift_3): Likewise.
9221         (cortex_a8_neon_vshl_ddd): Likewise.
9222         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9223         (cortex_a8_neon_vsra_vrsra): Likewise.
9224         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
9225         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
9226         (cortex_a8_neon_fp_vsum): Likewise.
9227         (cortex_a8_neon_fp_vmul_ddd): Likewise.
9228         (cortex_a8_neon_fp_vmul_qqd): Likewise.
9229         (cortex_a8_neon_fp_vmla_ddd): Likewise.
9230         (cortex_a8_neon_fp_vmla_qqq): Likewise.
9231         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
9232         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
9233         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9234         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9235         (cortex_a8_neon_bp_simple): Likewise.
9236         (cortex_a8_neon_bp_2cycle): Likewise.
9237         (cortex_a8_neon_bp_3cycle): Likewise.
9238         (cortex_a8_neon_ldr): Likewise.
9239         (cortex_a8_neon_str): Likewise.
9240         (cortex_a8_neon_vld1_1_2_regs): Likewise.
9241         (cortex_a8_neon_vld1_3_4_regs): Likewise.
9242         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9243         (cortex_a8_neon_vld2_4_regs): Likewise.
9244         (cortex_a8_neon_vld3_vld4): Likewise.
9245         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9246         (cortex_a8_neon_vst1_3_4_regs): Likewise.
9247         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
9248         (cortex_a8_neon_vst3_vst4): Likewise.
9249         (cortex_a8_neon_vld1_vld2_lane): Likewise.
9250         (cortex_a8_neon_vld3_vld4_lane): Likewise.
9251         (cortex_a8_neon_vst1_vst2_lane): Likewise.
9252         (cortex_a8_neon_vst3_vst4_lane): Likewise.
9253         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
9254         (cortex_a8_neon_mcr): Likewise.
9255         (cortex_a8_neon_mcr_2_mcrr): Likewise.
9256         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
9257         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
9258         change.
9259         (ca9_neon_mrrc): Likewise.
9260         (cortex_a9_neon_int_1): Likewise.
9261         (cortex_a9_neon_int_2): Likewise.
9262         (cortex_a9_neon_int_3): Likewise.
9263         (cortex_a9_neon_int_4): Likewise.
9264         (cortex_a9_neon_int_5): Likewise.
9265         (cortex_a9_neon_vqneg_vqabs): Likewise.
9266         (cortex_a9_neon_vmov): Likewise.
9267         (cortex_a9_neon_vaba): Likewise.
9268         (cortex_a9_neon_vaba_qqq): Likewise.
9269         (cortex_a9_neon_vsma): Likewise.
9270         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9271         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9272         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
9273         long_scalar): Likewise.
9274         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9275         (cortex_a9_neon_mla_qqq_8_16): Likewise.
9276         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9277         long_scalar_qdd_64_32_long): Likewise.
9278         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9279         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9280         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
9281         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9282         (cortex_a9_neon_shift_1): Likewise.
9283         (cortex_a9_neon_shift_2): Likewise.
9284         (cortex_a9_neon_shift_3): Likewise.
9285         (cortex_a9_neon_vshl_ddd): Likewise.
9286         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9287         (cortex_a9_neon_vsra_vrsra): Likewise.
9288         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
9289         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
9290         (cortex_a9_neon_fp_vsum): Likewise.
9291         (cortex_a9_neon_fp_vmul_ddd): Likewise.
9292         (cortex_a9_neon_fp_vmul_qqd): Likewise.
9293         (cortex_a9_neon_fp_vmla_ddd): Likewise.
9294         (cortex_a9_neon_fp_vmla_qqq): Likewise.
9295         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
9296         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
9297         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9298         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9299         (cortex_a9_neon_bp_simple): Likewise.
9300         (cortex_a9_neon_bp_2cycle): Likewise.
9301         (cortex_a9_neon_bp_3cycle): Likewise.
9302         (cortex_a9_neon_ldr): Likewise.
9303         (cortex_a9_neon_str): Likewise.
9304         (cortex_a9_neon_vld1_1_2_regs): Likewise.
9305         (cortex_a9_neon_vld1_3_4_regs): Likewise.
9306         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9307         (cortex_a9_neon_vld2_4_regs): Likewise.
9308         (cortex_a9_neon_vld3_vld4): Likewise.
9309         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9310         (cortex_a9_neon_vst1_3_4_regs): Likewise.
9311         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
9312         (cortex_a9_neon_vst3_vst4): Likewise.
9313         (cortex_a9_neon_vld1_vld2_lane): Likewise.
9314         (cortex_a9_neon_vld3_vld4_lane): Likewise.
9315         (cortex_a9_neon_vst1_vst2_lane): Likewise.
9316         (cortex_a9_neon_vst3_vst4_lane): Likewise.
9317         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
9318         (cortex_a9_neon_mcr): Likewise.
9319         (cortex_a9_neon_mcr_2_mcrr): Likewise.
9320         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
9321         (cortex_a9_fps): Likewise.
9322         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
9323         change.
9324         (cortex_m4_fmuls): Likewise.
9325         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
9326         change.
9327         (cortex_r4_mrc): Likewise.
9328         * config/arm/iterators.md: Update comment referring to neon_type.
9329         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
9330         (iwmmxt_movsi_insn): Likewise.
9331         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
9332         attribute change.
9333         (pj4_core_to_vfp): Likewise.
9334         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
9335         attribute change.
9336         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
9337         (vfp_fstore): Likewise.
9338         * doc/md.texi: Change references to neon_type to refer to type.
9340 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
9342         * tree.h (DECL_BUILT_IN): Fix typo in comment.
9344 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
9346         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
9347         lglobl if not weak.
9349 2013-09-04  Easwaran Raman  <eraman@google.com>
9351         PR middle-end/57370
9352         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
9353         (build_and_add_sum): Use it.
9354         (appears_later_in_bb): Simplify code.
9356 2013-09-04  Teresa Johnson  <tejohnson@google.com>
9358         * dumpfile.c (dump_finish): Don't close stderr/stdout.
9360 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
9362         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
9364 2013-09-04  Jan Hubicka  <jh@suse.cz>
9366         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
9367         * ipa-devirt.c: Include diganostic.h
9368         (odr_type_d): Add types and types_set.
9369         (hash_type_name): Work for types with vtables during LTO.
9370         (odr_hasher::remove): Fix comment; destroy types_set.
9371         (add_type_duplicate): New function,
9372         (get_odr_type): Use it.
9373         (dump_type_inheritance_graph): Dump type duplicates.
9374         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
9375         graph.
9376         * tree.c (types_same_for_odr): Give exact answers on types with
9377         virtual tables.
9379 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
9381         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
9382         explaining their differences.
9384 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
9386         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
9388 2013-09-03  Jeff Law  <law@redhat.com>
9390         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
9391         when not threading through a joiner block.  Pass joiner/no joiner
9392         state to register_jump_thread.
9393         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
9394         state from argument rather than implying on path length.
9395         Dump the entire jump thread path into debugging dump.
9396         * tree-flow.h (register_jump_thread): Update prototype.
9398 2013-08-29  Xinliang David Li  <davidxl@google.com>
9400         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
9401         Remove a trivial gcc_assert.
9403 2013-08-29  Xinliang David Li  <davidxl@google.com>
9405         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
9406         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
9407         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
9408         Delay base decl alignment adjustment.
9409         * tree-vectorizer.c (vect_destroy_datarefs): New function.
9410         * tree-vectorizer.h: New data structure.
9411         (set_dr_misalignment): New function.
9412         (dr_misalignment): Ditto.
9413         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
9414         (vectorizable_load): Ditto.
9415         (ensure_base_align): New function.
9416         (vectorize_loops): Add dbg_cnt support.
9417         (execute_vect_slp): Ditto.
9418         * dbgcnt.def: New debug counter.
9419         * Makefile: New dependency.
9421 2013-09-03  Meador Inge  <meadori@codesourcery.com>
9423         Revert:
9425         2013-08-30  Meador Inge  <meadori@codesourcery.com>
9427         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
9429 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
9431         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
9432         function descriptor.
9434 2013-09-03  Richard Biener  <rguenther@suse.de>
9436         * tree-affine.c (add_elt_to_tree): Fix association issue,
9437         avoid useless converts and make sure to always return a
9438         properly typed result.
9440 2013-09-03  Richard Biener  <rguenther@suse.de>
9442         PR middle-end/57656
9443         * fold-const.c (negate_expr_p): Fix division case.
9444         (negate_expr): Likewise.
9446 2013-09-03  Richard Biener  <rguenther@suse.de>
9448         PR lto/58285
9449         * tree-streamer-out.c: Include tm.h.
9450         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
9452 2013-09-03  Jan Hubicka  <jh@suse.cz>
9454         * tree-profile.c (tree_profiling): Cleanup CFG when done.
9456 2013-09-03  Alan Modra  <amodra@gmail.com>
9458         * config.gcc (powerpc*-*-linux*): Add support for little-endian
9459         multilibs to big-endian target and vice versa.
9460         * config/rs6000/t-linux64: Use := assignment on all vars.
9461         (MULTILIB_EXTRA_OPTS): Remove fPIC.
9462         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
9463         * config/rs6000/t-linux64le: New file.
9464         * config/rs6000/t-linux64bele: New file.
9465         * config/rs6000/t-linux64lebe: New file.
9467 2013-09-02  Jan Hubicka  <jh@suse.cz>
9469         * ipa-inline-transform.c (inline_transform): Do not
9470         optimize_inline_calls when not optimizing.
9472 2013-09-02  Jan Hubicka  <jh@suse.cz>
9474         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
9475         duplicated nodes for assembler names.
9476         * symtab.c (symtab_unregister_node): Do not attempt to unlink
9477         hard registers from assembler name hash.
9479 2013-09-02  Jan Hubicka  <jh@suse.cz>
9481         * ipa-split.c (execute_split_functions): Split externally visible
9482         functions called once.
9484 2013-09-02  Martin Jambor  <mjambor@suse.cz>
9486         PR ipa/58106
9487         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
9488         linked list.  When finding the correct duplicate, also consider also
9489         the caller in additon to its inlined_to node.
9491 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
9493         * config/aarch64/aarch64-simd-builtins.def
9494         (dup_lane_scalar): Remove.
9495         * config/aarch64/aarch64-simd.md
9496         (aarch64_simd_dup): Add 'w->w' alternative.
9497         (aarch64_dup_lane<mode>): Allow for VALL.
9498         (aarch64_dup_lane_scalar<mode>): Remove.
9499         (aarch64_dup_lane_<vswap_width_name><mode>): New.
9500         (aarch64_get_lane_signed<mode>): Add w->w altenative.
9501         (aarch64_get_lane_unsigned<mode>): Likewise.
9502         (aarch64_get_lane<mode>): Likewise.
9503         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
9504         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
9505         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
9506         (VCON): Change container of V2SF.
9507         (vswap_width_name): Likewise.
9508         * config/aarch64/arm_neon.h
9509         (__aarch64_vdup_lane_any): New.
9510         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
9511         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
9512         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
9514 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
9516         PR middle-end/56382
9517         * expr.c (emit_move_complex): Do not move complex FP values as parts if
9518         the source or the destination is a single hard register.
9520 2013-09-02  Richard Biener  <rguenther@suse.de>
9522         PR middle-end/57511
9523         * tree-scalar-evolution.c (instantiate_scev_name): Allow
9524         non-linear SCEVs.
9526 2013-09-02  Richard Biener  <rguenther@suse.de>
9528         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
9529         arithmetic to sizetype.
9531 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
9533         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
9534         Find auto-increment use both before and after candidate.
9536 2013-09-02  Marek Polacek  <polacek@redhat.com>
9538         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
9540 2013-09-01  Jan Hubicka  <jh@suse.cz>
9542         * Makefile.in: Add ipa-profile.o
9543         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
9544         * cgraph.c (struct cgraph_propagate_frequency_data,
9545         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
9546         ipa-profile.c; replace cgraph_ by ipa_ prefix.
9547         * cgraph.h (cgraph_propagate_frequency): Remove.
9548         * ipa-inline-analysis.c: Include ipa-utils.h;
9549         drop duplicated cfgloop.h.
9550         (inline_update_callee_summaries): Update.
9551         * ipa-profile.c: New file.
9552         * ipa-utils.h (ipa_propagate_frequency): Declare.
9553         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
9554         data-streamer.h, value-prof.h.
9555         (symtab_remove_unreachable_nodes): Update profile.
9556         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
9557         account_time_size, cmp_counts, dump_histogram,
9558         ipa_profile_generate_summary, ipa_profile_write_summary,
9559         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
9560         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
9561         Move to ipa-profile.c.
9563 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
9565         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
9567 2013-09-01  Jan Hubicka  <jh@suse.cz>
9569         * common.opt (fdevirtualize-speculatively): New function.
9570         * invoke.texi (fdevirtualize-speculatively): Document.
9571         * ipa-devirt.c: Include ipa-inline.h
9572         (likely_target_p): New function.
9573         (ipa_devirt): New function.
9574         (gate_ipa_devirt): New function.
9575         (pass_data_ipa_devirt): New static var.
9576         (pass_ipa_devirt): Likewise.
9577         (make_pass_ipa_devirt): New function.
9578         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
9579         (common_handle_option): Disable devirtualization when
9580         value range profiling is available.
9581         * passes.def (pass_ipa_devirt): Add.
9582         * timever.def (TV_IPA_DEVIRT): New timevar.
9583         * tree-pass.h (make_pass_ipa_devirt):
9585 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
9587         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
9588         include sanitize(undefined).
9590 2013-08-31  Diego Novillo  <dnovillo@google.com>
9592         * Makefile.in (TREE_CORE_H): Define.
9593         (TREE_H): Use.
9594         (GTFILES): Add tree-core.h.
9595         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
9596         size the array.
9597         * tree-core.h: New file.
9598         Move all data structures, enum, typedefs, global
9599         declarations and constants from ...
9600         * tree.h: ... here.
9602 2013-08-31  Jan Hubicka  <jh@suse.cz>
9604         * bulitins.c (expand_builtin): Do not early exit for gcov
9605         instrumented functions.
9607 2013-08-31  Marek Polacek  <polacek@redhat.com>
9609         * ubsan.c: Include tm_p.h.
9611 2013-08-31  Jan Hubicka  <jh@suse.cz>
9613         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
9614         warning.
9616         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
9617         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
9618         * tree-cfg.c (verify_gimple_label): ... here.
9619         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
9620         (ipa_merge_profiles): New function.
9621         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
9622         (lto_input_function_body): Likewise.
9623         * ipa-utils.h (ipa_merge_profiles): Declare.
9624         * lto-streamer.h (lto_input_function_body): Update prototype.
9625         (emit_label_in_global_context_p): Remove.
9626         * lto-symtab.c: Include ipa-utils.h
9627         (lto_cgraph_replace_node): Use ipa_merge_profiles.
9629 2013-08-31  Jan Hubicka  <jh@suse.cz>
9631         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
9633 2013-08-31  Jan Hubicka  <jh@suse.cz>
9635         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
9637 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
9639         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
9640         "cmp" RTX before signed_comparison_operator check to account
9641         for "code" changes.
9643 2013-08-30  Jan Hubicka  <jh@suse.cz>
9645         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
9646         (detect_type_change_1): Rename to ...
9647         (detect_type_change): ... this one; early return on non-polymorphic
9648         types.
9649         (detect_type_change_ssa): Add comp_type parameter; update
9650         use of detect_type_change.
9651         (compute_complex_assign_jump_func): Add param_type parameter;
9652         update use of detect_type_change_ssa.
9653         (compute_complex_ancestor_jump_func): Likewise.
9654         (ipa_get_callee_param_type): New function.
9655         (ipa_compute_jump_functions_for_edge): Compute parameter type;
9656         update calls to the jump function computation functions.
9658 2013-08-30  Teresa Johnson  <tejohnson@google.com>
9659             Steven Bosscher  <steven@gcc.gnu.org>
9661         * cfgrtl.c (fixup_new_cold_bb): New routine.
9662         (commit_edge_insertions): Invoke fixup_partitions.
9663         (find_partition_fixes): New routine.
9664         (fixup_partitions): Ditto.
9665         (verify_hot_cold_block_grouping): Update comments.
9666         (rtl_verify_edges): Invoke find_partition_fixes.
9667         (rtl_verify_bb_pointers): Update comments.
9668         (rtl_verify_bb_layout): Ditto.
9669         * basic-block.h (probably_never_executed_edge_p): Declare.
9670         (fixup_partitions): Ditto.
9671         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
9672         * bb-reorder.c (sanitize_hot_paths): New function.
9673         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
9674         sanitize_hot_paths.
9675         * predict.c (probably_never_executed_edge_p): New routine.
9676         * cfg.c (check_bb_profile): Add partition insanity warnings.
9678 2013-08-30  Meador Inge  <meadori@codesourcery.com>
9680         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
9682 2013-08-30  Marek Polacek  <polacek@redhat.com>
9684         * Makefile.in (ubsan.o): Add.
9685         (c-family/c-ubsan.o): Add.
9686         (builtins.o): Add ubsan.h dependency.
9687         * ubsan.h: New file.
9688         * ubsan.c: New file.
9689         * common.opt: Add -fsanitize=undefined option.
9690         (flag_sanitize): Add variable.
9691         (fsanitize=): Add option.  Add Driver.
9692         (fsanitize=thread): Remove option.
9693         (fsanitize=address): Likewise.
9694         (static-libubsan): New option.
9695         * doc/invoke.texi: Document the new flag and -static-libubsan.
9696         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
9697         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
9698         * builtin-attrs.def (ATTR_COLD): Define.
9699         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
9700         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
9701         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
9702         * flag-types.h (sanitize_code): New enum.
9703         * opts.c (common_handle_option): Parse command line arguments
9704         of -fsanitize=.  Add -fsanitize=unreachable option.
9705         * varasm.c (get_variable_section): Adjust.
9706         (assemble_noswitch_variable): Likewise.
9707         (assemble_variable): Likewise.
9708         (output_constant_def_contents): Likewise.
9709         (categorize_decl_for_section): Likewise.
9710         (place_block_symbol): Likewise.
9711         (output_object_block): Likewise.
9712         * builtins.def: Likewise.
9713         * toplev.c (compile_file): Likewise.
9714         (process_options): Likewise.
9715         * cppbuiltin.c: Likewise.
9716         * tsan.c (tsan_pass): Likewise.
9717         (tsan_gate): Likewise.
9718         (tsan_gate_O0): Likewise.
9719         * cfgexpand.c (partition_stack_vars): Likewise.
9720         (expand_stack_vars): Likewise.
9721         (defer_stack_allocation): Likewise.
9722         (expand_used_vars): Likewise.
9723         * cfgcleanup.c (old_insns_match_p): Likewise.
9724         * asan.c (asan_finish_file): Likewise.
9725         (asan_instrument): Likewise.
9726         (gate_asan): Likewise.
9727         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
9728         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
9729         (asan_global_struct): Use pointer_sized_int_node instead
9730         calling build_nonstandard_integer_type.
9731         (initialize_sanitizer_builtins): Likewise.
9732         (asan_finish_file): Likewise.
9733         * gcc.c: Document %{%:function(args):X}.
9734         (static_spec_functions): Add sanitize.
9735         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
9736         store funcval != NULL there.
9737         (do_spec_1): Adjust handle_spec_function caller.
9738         (handle_braces): Allow %:function(args) as condition.
9739         (sanitize_spec_function): New function.
9740         (ADD_STATIC_LIBUBSAN_LIBS): Define.
9741         (LIBUBSAN_SPEC): Likewise.
9742         (LIBUBSAN_EARLY_SPEC): Likewise.
9743         (SANITIZER_SPEC): Handle libubsan.
9744         (SANITIZER_EARLY_SPEC): Likewise.
9745         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
9746         instead of fsanitize=address.
9747         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
9748         instead of fsanitize=address*.
9749         * builtins.c: Include ubsan.h.
9750         (fold_builtin_0): Instrument __builtin_unreachable.
9751         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
9752         instead of flag_asan.
9753         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
9754         (pointer_sized_int_node): Define.
9755         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
9757 2013-08-30  Mike Stump  <mikestump@comcast.net>
9759         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
9760         with RE patterns.
9762 2013-08-29  Jan Hubicka  <jh@suse.cz>
9764         * cgraph.c (cgraph_function_body_availability): Handle weakref
9765         correctly.
9766         * passes.def: Remove pass_fixup_cfg.
9767         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
9768         track when we need to remove functions.
9769         (gate_ipa_inline): Execute inlining always; add comment why.
9770         (pass_data_ipa_inline): Remove TODO_remove_functions.
9771         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
9772         do not produce summaries.
9773         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
9774         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
9775         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
9776         constant pool and vtable.
9778 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
9780         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
9781         New arm_neon.h's internal macros to specify 64-bit constants.
9782         Avoid using stdint.h's macros.
9784 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
9786         * recog.c (verify_changes): Verify that changes[i].old is non-zero
9787         before applying REG_P.
9789 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
9791         PR tree-optimization/58277
9792         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
9793         after seeing too many stmts with vdef in between dombb and current
9794         bb, invalidate everything.
9796 2013-08-30  Richard Biener  <rguenther@suse.de>
9798         * fold-const.c (fold_single_bit_test): Fix overflow test.
9800 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
9802         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
9803         and which can live in a register, always retrieve the value on entry.
9804         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
9805         passed by invisible reference specially.
9806         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
9807         (vt_add_function_parameter): Correctly deal with a parameter passed by
9808         invisible reference.
9810 2013-08-30  Jan Hubicka  <jh@suse.cz>
9812         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
9814 2013-08-30  Richard Biener  <rguenther@suse.de>
9816         PR tree-optimization/58228
9817         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
9818         allow invariant loads in nested loop vectorization.
9820 2013-08-30  Richard Biener  <rguenther@suse.de>
9822         PR tree-optimization/58223
9823         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
9824         (has_anti_or_output_dependence): ... this and adjust to also
9825         look for output dependences.
9826         (mark_nodes_having_upstream_mem_writes): Adjust.
9827         (rdg_flag_uses): Likewise.
9829 2013-08-30  Richard Biener  <rguenther@suse.de>
9831         PR tree-optimization/58010
9832         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
9833         assert that we have a loop-closed PHI.
9835 2013-08-29  Jan Hubicka  <jh@suse.cz>
9837         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
9838         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
9839         * lto-section-in.c (lto_free_function_in_decl_state): New function.
9840         (lto_free_function_in_decl_state_for_node): New function.
9842 2013-08-29  Xinliang David Li  <davidxl@google.com>
9844         * loop-unroll.c (report_unroll_peel): Minor message change.
9845         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
9846         Emit alignment peeling message with default -fopt-info.
9847         (vect_loop_versioning): Emit loop version info message.
9848         * tree-vectorizer.c (vectorize_loops): Minor message change.
9849         (execute_vect_slp): Ditto.
9851 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
9853         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
9854         of the clone from the DECL_NAME of the original function.
9856 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
9858         * passes.c (register_pass): Add overload.
9859         * tree-pass.h (register_pass): Forward declare it.  Add comment.
9861 2013-08-29  Jan Hubicka  <jh@suse.cz>
9863         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
9864         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
9865         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
9866         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
9867         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
9868         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
9869         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
9870         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
9872 2013-08-29  Teresa Johnson  <tejohnson@google.com>
9874         * dumpfile.c (dump_loc): Output column number.
9875         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
9876         * doc/invoke.texi: Document optall -fopt-info flag.
9877         * profile.c (read_profile_edge_counts): Use new dump framework.
9878         (compute_branch_probabilities): Ditto.
9879         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
9880         when pass not in any opt group.
9881         * pass_manager.h (pass_manager::get_pass_profile): New method.
9882         * value-prof.c (check_counter): Use new dump framework.
9883         (check_ic_target): Ditto.
9884         * coverage.c (get_coverage_counts): Ditto.
9885         (coverage_init): Setup new dump framework.
9887 2013-08-29  Richard Biener  <rguenther@suse.de>
9889         PR tree-optimization/58246
9890         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
9891         handle the dominance check inside a basic-block.
9893 2013-08-29  Richard Biener  <rguenther@suse.de>
9895         PR middle-end/57287
9896         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
9897         of default defs that appear in abnormal PHI nodes.
9899 2013-08-29  Richard Biener  <rguenther@suse.de>
9901         PR tree-optimization/57685
9902         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
9903         single-use operands to avoid exponential complexity.
9905 2013-08-28  Dehao Chen  <dehao@google.com>
9907         * ipa-inline.c (edge_badness): Fix integer underflow.
9909 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
9911         * gtm-builtins.def (_ITM_free): Declare leaf.
9913 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
9915         PR target/58067
9916         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
9917         (*tls_local_dynamic_base_64_largepic): Likewise.
9918         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
9919         Remove predicate from call operand.
9920         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
9921         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
9923 2013-08-28  Jeff Law  <law@redhat.com>
9925         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
9926         checks for the number of predecessors and successors allowed.
9927         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
9928         which require copying a joiner block if there is a request which
9929         is a subpath that requires no joiner block copying.
9931 2013-08-28  Jan Hubicka  <jh@suse.cz>
9933         * lto-streamer-out.c (DFS_write_tree_body): Drop
9934         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
9935         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
9936         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
9937         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
9938         Do not read DECL_ERROR_ISSUED.
9939         (unpack_ts_decl_with_vis_value_fields): Do not read
9940         DECL_DEFER_OUTPUT.
9941         (lto_input_ts_binfo_tree_pointers): Do not read
9942         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
9943         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
9944         write DECL_ERROR_ISSUED..
9945         (pack_ts_decl_with_vis_value_fields): Do not write
9946         DECL_DEFER_OUTPUT.
9947         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
9948         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
9949         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
9950         * tree.h (tree_decl_common): Update comment.
9951         (DECL_ERROR_ISSUED): Remove.
9953 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
9955         PR middle-end/58257
9956         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
9958 2013-08-28  Jan Hubicka  <jh@suse.cz>
9960         * builtins.def (free): Declare leaf.
9962 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
9964         * gdbhooks.py: New.
9965         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
9966         * configure: Regenerate.
9968 2013-08-27  Martin Jambor  <mjambor@suse.cz>
9970         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
9971         (ipa_ancestor_jf_data): Likewise.
9972         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
9973         (ipa_get_jf_pass_through_type_preserved): New function.
9974         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
9975         (ipa_get_jf_ancestor_type_preserved): New function.
9976         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
9977         (ipa_get_jf_ancestor_result): Likewise.
9978         (propagate_vals_accross_pass_through): Use
9979         ipa_get_jf_pass_through_result to do all the value mappings.
9980         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
9981         type_preserved flag.
9982         (ipa_set_jf_cst_copy): New function.
9983         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
9984         (ipa_set_jf_arith_pass_through): Likewise.
9985         (ipa_set_ancestor_jf): Likewise.
9986         (compute_complex_assign_jump_func): Set type_preserved instead of
9987         punting.
9988         (ipa_compute_jump_functions_for_edge): Likewise.
9989         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
9990         (update_jump_functions_after_inlining): Update type_preserved.
9991         Explicitely create jump functions when combining one with pass_through.
9992         (ipa_write_jump_function): Stream the type_preserved flags.
9993         (ipa_read_jump_function): Likewise.
9995 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
9996             Aldy Hernandez  <aldyh@redhat.com>
9998         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
9999         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
10000         * function.h (struct function): Add has_force_vect_loops and
10001         has_simduid_loops.
10002         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
10003         * gimple.c (gimple_build_omp_critical): Add KIND argument and
10004         handle it.
10005         * gimple.def: Update CLAUSES comments.
10006         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
10007         (gimple_build_omp_for): Add argument to prototype.
10008         (gimple_omp_for_kind): New.
10009         (gimple_omp_for_set_kind): New.
10010         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
10011         GOVD_DATA_SHARE_CLASS.
10012         (enum omp_region_type): Add ORT_SIMD.
10013         (gimple_add_tmp_var): Handle ORT_SIMD.
10014         (gimplify_var_or_parm_decl): Same.
10015         (is_gimple_stmt): Same.
10016         (omp_firstprivatize_variable): Same.
10017         (omp_add_variable): Only use splay_tree_insert if lookup failed.
10018         (omp_notice_variable): Handle ORT_SIMD.
10019         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
10020         (omp_check_private): Handle ORT_SIMD.
10021         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10022         OMP_CLAUSE_SAFELEN.
10023         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
10024         Handle OMP_CLAUSE_LASTPRIVATE.
10025         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10026         OMP_CLAUSE_SAFELEN.
10027         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
10028         (gimplify_expr): Handle OMP_SIMD.
10029         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
10030         (expand_GOMP_SIMD_VF): New.
10031         (expand_GOMP_SIMD_LAST_LANE): New.
10032         * internal-fn.def (GOMP_SIMD_LANE): New.
10033         (GOMP_SIMD_VF): New.
10034         (GOMP_SIMD_LAST_LANE): New.
10035         * omp-low.c: Include target.h.
10036         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
10037         OMP_CLAUSE_SAFELEN.
10038         (check_omp_nesting_restrictions): Same.
10039         (omp_max_vf): New.
10040         (lower_rec_simd_input_clauses): New.
10041         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
10042         OMP_CLAUSE_LINEAR.
10043         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
10044         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
10045         (expand_omp_build_assign): New.
10046         (expand_omp_for_init_counts): New.
10047         (expand_omp_for_init_vars): New.
10048         (extract_omp_for_update_vars): New.
10049         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
10050         and rewrite accordingly.
10051         (expand_omp_simd): New.
10052         (expand_omp_for): Use expand_omp_simd.
10053         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
10054         (lower_omp_for): Do not lower the body.
10055         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
10056         in their own loops.
10057         * tree-flow.h (find_omp_clause): Remove prototype.
10058         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
10059         forcing vectorization of the loop, or if flag_tree_vectorize.
10060         (gate_tree_if_conversion): Similarly.
10061         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
10062         gimple_build_omp_for.
10063         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
10064         * tree-parloops (create_parallel_loop): Pass kind argument to
10065         gimple_build_omp_for.
10066         * tree-pretty-print.c (dump_omp_clause): Add cases for
10067         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
10068         OMP_CLAUSE__SIMDUID_.
10069         (dump_generic_node): Handle OMP_SIMD.
10070         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
10071         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
10072         unroll OMP_SIMD loops here.
10073         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
10074         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
10075         loop->safelen.
10076         (vect_analyze_data_refs): Handle simd loops.
10077         * tree-vect-loop.c (vectorizable_live_operation): Handle
10078         IFN_GOMP_SIMD*.
10079         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
10080         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
10081         (vectorizable_load): Same.
10082         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
10083         (struct simduid_to_vf): New.
10084         (simduid_to_vf::hash): New.
10085         (simduid_to-vf::equal): New.
10086         (struct simd_array_to_simduid): New.
10087         (simd_array_to_simduid::hash): New.
10088         (simd_array_to_simduid::equal): New.
10089         (adjust_simduid_builtins): New.
10090         (struct note_simd_array_uses_struct): New.
10091         (note_simd_array_uses_cb): New.
10092         (note_simd_array_uses): New.
10093         (vectorize_loops): Handle simd hints and adjust simd builtins
10094         accordingly.
10095         * tree-vectorizer.h (struct _stmt_vec_info): Add
10096         simd_lane_access_p field.
10097         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
10098         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
10099         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
10100         (omp_clause_code_name): Same.
10101         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
10102         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
10103         * tree.def (OMP_SIMD): New entry.
10104         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
10105         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
10106         (OMP_CLAUSE_DECL): Adjust range for new clauses.
10107         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
10108         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
10109         (OMP_CLAUSE_LINEAR_STEP): New.
10110         (OMP_CLAUSE_SAFELEN_EXPR): New.
10111         (OMP_CLAUSE__SIMDUID__DECL): New.
10112         (find_omp_clause): New prototype.
10114 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
10116         * config/i386/driver-i386.c (host_detect_local_cpu): Update
10117         Haswell processor detection.
10119 2013-08-27  Christian Widmer  <shadow@umbrox.de>
10121         PR target/57927
10122         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
10123         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
10124         AVX2 capable processors.
10126 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
10128         * config/aarch64/arm_neon.h: Replace all inline asm implementations
10129         of vget_low_* with implementations in terms of other intrinsics.
10131 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
10133         PR middle-end/57219
10134         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
10135         values to -1, 0 and 1.
10137 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
10139         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
10140         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
10141         (<optab><mode>3_insn): Remove.
10142         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
10143         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
10144         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
10145         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
10146         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
10147         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
10148         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
10149         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
10150         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
10151         (ror<mode>3_insn): Likewise.
10152         * config/aarch64/predicates.md (aarch64_simd_register): New.
10154 2013-08-27  Richard Biener  <rguenther@suse.de>
10156         PR tree-optimization/57521
10157         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
10158         one edge is non-critical.
10159         (find_phi_replacement_condition): Make sure to use a non-critical
10160         edge.  Cleanup and remove old bug workarounds.
10161         (bb_postdominates_preds): Remove.
10162         (if_convertible_loop_p_1): Do not compute post-dominators.
10163         (combine_blocks): Do not free post-dominators.
10164         (main_tree_if_conversion): Likewise.
10165         (pass_data_if_conversion): Add TODO_verify_ssa.
10167 2013-08-27  DJ Delorie  <dj@redhat.com>
10169         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
10171 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
10173         * function.c (assign_parm_find_data_types): Set passed_mode and
10174         nominal_mode to the TYPE_MODE of nominal_type for the built
10175         pointer type in case of the struct-pass-by-reference.
10177 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10179         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
10180         (UINT16_TYPE): Change default to "unsigned int".
10182         * config/avr/avr.opt (mfract-convert-truncate): New option.
10183         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
10184         is set, round negative fractional integers according to n1169
10185         when converting to integer types.
10187 2013-08-26  Jan Hubicka  <jh@suse.cz>
10189         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
10190         methods can not be called indirectly when their address is not taken.
10192 2013-08-26  Jan Hubicka  <jh@suse.cz>
10194         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
10195         ctor_for_folding.
10197 2013-08-26  Jan Hubicka  <jh@suse.cz>
10199         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
10200         can be unshared.
10202 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10204         * reload.c (find_valid_class): Allow classes that do not include
10205         FIRST_PSEUDO_REGISTER - 1.
10207 2013-08-26  Jan Hubicka  <jh@suse.cz>
10209         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
10210         fix edge count/frequency when speculation failed; fix type check
10211         for the direct call.
10213 2013-08-26  Jan Hubicka  <jh@suse.cz>
10215         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
10217 2013-08-26  Jan Hubicka  <jh@suse.cz>
10219         * ipa-inline-transform.c (inline_transform): Be ready for basic block
10220         to be changed by edge redirection.
10222 2013-08-26  Jan Hubicka  <jh@suse.cz>
10224         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
10225         formating; add sanity check.
10226         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
10227         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
10228         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
10229         (resolve_noninline_speculation): Update callee keys, too.
10231 2013-08-26  Jan Hubicka  <jh@suse.cz>
10233         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
10234         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
10236 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10238         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
10239         into proper place.
10241 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
10243         * config/i386/i386.c (ix86_debug_options): Remove prototype.
10244         (x86_64_elf_select_section): Ditto.
10245         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
10246         arguments.
10247         (ix86_pass_by_reference): Ditto.
10248         (output_set_got): Ditto.
10249         (ix86_unary_operator_ok): Ditto.
10250         (ix86_expand_builtin): Ditto.
10252 2013-08-23  Jan Hubicka  <jh@suse.cz>
10254         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
10256 2013-08-23  Jan Hubicka  <jh@suse.cz>
10258         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
10259         (tree_decl_with_vis): Add FINAL field.
10261 2013-08-23  Jeff Law  <law@redhat.com>
10263         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
10264         the debugging dump when the expression is fully redundant.
10266 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10268         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
10269         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
10270         * pretty-print.c (pp_formatted_text_data): Likewise.
10271         (pp_write_text_to_stream): Likewise.
10272         (pp_write_text_as_dot_label_to_stream): Likewise.
10273         (pp_append_r): Likewise.
10274         (pp_format): Likewise.
10275         (pp_flush): Likewise.
10276         (pp_clear_output_area): Likewise.
10277         (pp_append_text): Likewise.
10278         (pp_formatted_text): Likewise.
10279         (pp_remaining_character_count_for_line): Likewise.
10280         (pp_newline): Likewise.
10281         (pp_character): Likewise.
10282         (output_buffer::~output_buffer): Define.
10283         (pretty_printer::~pretty_printer): Destruct output buffer.
10284         * pretty-print.h (output_buffer::~output_buffer): Declare.
10285         (pretty_printer::~pretty_printer): Declare virtual.
10287 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
10289         PR other/57324
10290         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
10291         HOST_WIDE_INT_M1U): New macros.
10292         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
10293         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
10294         unsigned -1 for lshift.
10295         * cse.c (cse_insn): Likewise.
10296         * double-int.c (rshift_double, lshift_double): Likewise.
10297         * builtins.c (fold_builtin_bitop): Likewise.
10298         * combine.c (force_to_mode): Likewise.
10299         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
10300         * simplify-rtx.c (simplify_const_unary_operation,
10301         simplify_const_binary_operation): Likewise.
10302         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
10303         check_va_list_escapes): Likewise.
10304         * rtlanal.c (nonzero_bits1): Likewise.
10305         * expmed.c (expand_smod_pow2): Likewise.
10306         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
10308 2013-08-23  Jan Hubicka  <jh@suse.cz>
10310         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
10311         as having address taken.
10313 2013-08-23  Jan Hubicka  <jh@suse.cz>
10315         * ipa-utils.h (method_class_type): Declare.
10316         * ipa-devirt.c (method_class_type): Export.
10318         * cgraphunit.c (analyze_functions): Do basic devirtualization;
10319         do not walk base classes of anonymous types.
10321 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
10323         PR rtl-optimization/58220
10324         PR regression/58221
10325         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
10326         handle SEQUENCE insns properly.
10328 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10330         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
10331         definition.
10332         (pp_newline_and_indent): Likewise.
10333         (pp_separate_with): Likewise.
10334         * pretty-print.c (pp_newline_and_flush): Define.
10335         (pp_newline_and_indent): Likewise.
10336         (pp_separate_with): Likewise.
10338 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
10340         PR target/58218
10341         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
10342         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
10344 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
10346         * config/i386/predicates.md (ext_sse_reg_operand): New.
10347         * config/i386/i386.md (*movti_internal): Use
10348         predicate to determine if EVEX is needed.
10349         (*movsi_internal): Ditto.
10350         (*movdf_internal): Ditto.
10351         (*movsf_internal): Ditto.
10352         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
10354 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
10356         PR tree-optimization/58209
10357         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
10358         (find_tail_calls): Give up for pointer result types if m is non-NULL.
10359         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
10360         emit POINTER_PLUS_EXPR.
10361         (create_tailcall_accumulator): For pointer result type accumulate in
10362         sizetype type.
10364 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
10366         * configure.ac: Add backslashes missing from the last change.
10367         * configure: Regenerate.
10369 2013-08-22  Jan Hubicka  <jh@susue.cz>
10371         * ipa.c (function_and_variable_visibility): First remember function
10372         was global and then make it local.
10374 2013-08-22  Julian Brown  <julian@codesourcery.com>
10376         * configure.ac: Add aarch64 to list of arches which use "nop" in
10377         debug_line test.
10378         * configure: Regenerate.
10380 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10382         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
10383         gnu_libc_has_function.
10384         * config/s390/tpf.h: Likewise.
10386 2013-08-22  Jan Hubicka  <jh@susue.cz>
10388         * timevar.c (validate_phases): Add cast.
10390 2013-08-22  Jan Hubicka  <jh@susue.cz>
10392         * timevar.c (validate_phases): Use size_t for memory.
10393         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
10395 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10397         * pretty-print.h (output_buffer::output_buffer): Declare.
10398         (pretty_printer::pretty_printer): Likewise.
10399         (pp_construct): Remove.
10400         * pretty-print.c (output_buffer::output_buffer): Define.
10401         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
10402         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
10403         (print_gimple_expr): Likewise.
10404         (print_gimple_seq): Likewise.
10405         (gimple_dump_bb): Likewise.
10406         * sched-vis.c (dump_value_slim): Likewise.
10407         (dump_insn_slim): Likewise.
10408         (dump_rtl_slim): Likewise.
10409         (str_pattern_slim): Likewise.
10410         * tree-mudflap.c (mf_varname_tree): Likewise.
10411         * graph.c (print_graph_cfg): Likewise.
10412         (start_graph_dump): Likewise.
10413         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
10414         placement-new.
10415         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
10416         pretty printer initialization.
10417         * coretypes.h (diagnostic_context): Remove superflous type alias
10418         declaration.
10419         (pretty_printer): Likewise.  Declare directly as a class.
10420         (pretty_print_info): Remove declaration as class.
10421         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
10422         and pp_clear_output_area.
10423         (asan_add_global): Likewise.
10425 2013-08-22  Jan Hubicka  <jh@suse.cz>
10427         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
10428         * ipa-utils.h (update_type_inheritance_graph): Declare.
10429         (possible_polymorphic_call_target_p): Declare.
10430         (possible_polymorphic_call_target_p): New.
10431         * ipa-devirt.c: Update toplevel comments.
10432         (cached_polymorphic_call_targets): Move up.
10433         (odr_type_d): Move ID down.
10434         (polymorphic_type_binfo_p): Update comment.
10435         (odr_hasher::remove): Likewise;
10436         (get_odr_type): Set anonymous_namespace.
10437         (dump_odr_type): Dump it.
10438         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
10439         (maybe_record_node): Record node in cached_polymorphic_call_targets.
10440         (record_binfo): Add comment.
10441         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
10442         built.
10443         (devirt_node_removal_hook): Do not iCE when cache is freed.
10444         (possible_polymorphic_call_target_p): New predicate.
10445         (update_type_inheritance_graph): New function.
10447 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10448             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
10449             Sergey Lega  <sergey.s.lega@intel.com>
10450             Anna Tikhonova  <anna.tikhonova@intel.com>
10451             Ilya Tocar  <ilya.tocar@intel.com>
10452             Andrey Turetskiy  <andrey.turetskiy@intel.com>
10453             Ilya Verbin  <ilya.verbin@intel.com>
10454             Kirill Yukhin  <kirill.yukhin@intel.com>
10455             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
10457         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
10458         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
10459         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
10460         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
10461         (OPTION_MASK_ISA_AVX2_UNSET): Update.
10462         (OPTION_MASK_ISA_AVX512F_UNSET): New.
10463         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
10464         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
10465         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
10466         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
10467         OPT_mavx512pf, OPT_mavx512er cases.
10468         * config/i386/constraints.md (v): New constraint.
10469         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
10470         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
10471         (bit_AVX512CD): New.
10472         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
10473         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
10474         * config/i386/i386-c.c (ix86_target_macros_internal):
10475         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
10476         __AVX512PF__.
10477         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
10478         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
10479         * config/i386/i386.c (regclass_map, dbx_register_map)
10480         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
10481         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
10482         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
10483         -mavx512pf options.
10484         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
10485         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
10486         -mavx512pf options.  Fix formatting.
10487         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
10488         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
10489         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
10490         -mavx512cd, -mavx512pf options.
10491         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
10492         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
10493         (ix86_preferred_output_reload_class): Replace SSE_REGS with
10494         ALL_SSE_REGS.
10495         (ix86_hard_regno_mode_ok): Support 512-bit registers.
10496         (ix86_set_reg_reg_cost): Ditto.
10497         (x86_order_regs_for_local_alloc): Ditto.
10498         (MAX_VECT_LEN): Extend to 64-byte.
10499         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
10500         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
10501         (TARGET_AVX512ER, TARGET_AVX512CD): New.
10502         (BIGGEST_ALIGNMENT): Extend to 512-bits.
10503         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
10504         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
10505         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
10506         (SSE_REG_MODE_P): Support new modes.
10507         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
10508         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
10509         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
10510         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
10511         (REG_CLASS_CONTENTS): Add new registers.
10512         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
10513         (EXT_REX_SSE_REGNO_P): New.
10514         (HI_REGISTER_NAMES): Add new registers.
10515         * config/i386/i386.md: Define constants for new registers.
10516         (mode): Add new 512-bit modes.
10517         (prefix): Support evex prefix.
10518         (isa): Support avx512f, noavx512f, fma_avx512f.
10519         (ssemodesuffix): Add new 512-bit modes.
10520         (movxi): New.
10521         (*movxi_internal_avx512f): Ditto.
10522         (*movdi_internal): Replace constraint "x" with the new constraint "v".
10523         Support MODE_XI.
10524         (*movsi_internal): Likewise.
10525         (*movdf_internal): Likewise.
10526         (*movsf_internal): Likewise.
10527         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
10528         (<code><mode>3): Likewise.
10529         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
10530         New.
10531         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
10532         with the new constraint "v".
10533         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
10534         modes.
10535         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
10536         with the new constraint "v".
10537         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
10538         (<sse2>_storedqu<avxsizesuffix>): Likewise.
10539         (*<plusminus_insn><mode>3): Likewise.
10540         (<sse>_vm<plusminus_insn><mode>3): Likewise.
10541         (*mul<mode>3): Likewise.
10542         (<sse>_vmmul<mode>3): Likewise.
10543         (<sse>_div<mode>3): Likewise.
10544         (<sse>_vmdiv<mode>3): Likewise.
10545         (<sse>_sqrt<mode>2): Likewise.
10546         (<sse>_vmsqrt<mode>2): Likewise.
10547         (*<code><mode>3_finite): Likewise.
10548         (*<code><mode>3) <smaxmin>: Likewise.
10549         (<sse>_vm<code><mode>3): Likewise.
10550         (*<code><mode>3) <any_logic>: Likewise.
10551         (*fma_fmadd_<mode>): Likewise.
10552         (*fma_fmsub_<mode>): Likewise.
10553         (*fma_fnmadd_<mode>): Likewise.
10554         (*fma_fnmsub_<mode>): Likewise.
10555         (*fma_fmaddsub_<mode>): Likewise.
10556         (*fma_fmsubadd_<mode>): Likewise.
10557         (*fmai_fmadd_<mode>): Likewise.
10558         (*fmai_fmsub_<mode>): Likewise.
10559         (*fmai_fnmadd_<mode>): Likewise.
10560         (*fmai_fnmsub_<mode>): Likewise.
10561         (sse_cvtsi2ss): Likewise.
10562         (sse_cvtsi2ssq): Likewise.
10563         (sse_cvtss2si): Likewise.
10564         (sse_cvtss2si_2): Likewise.
10565         (sse_cvtss2siq): Likewise.
10566         (sse_cvtss2siq_2): Likewise.
10567         (sse_cvttss2si): Likewise.
10568         (sse_cvtss2siq_2): Likewise.
10569         (float<sseintvecmodelower><mode>2): Likewise.
10570         (sse2_cvtsd2si_2): Likewise.
10571         (sse2_cvtsd2siq_2): Likewise.
10572         (*<plusminus_insn><mode>3): Likewise.
10573         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
10574         (*<sse4_1_avx2>_mul<mode>3): Likewise.
10575         (ashr<mode>3): Likewise.
10576         (<shift_insn><mode>3): Likewise.
10577         (avx2_<code><mode>3): Likewise.
10578         (*avx2_<code><mode>3): Likewise.
10579         (*andnot<mode>3): Likewise.
10580         (*<code><mode>3) <any_logic>: Likewise.
10581         (abs<mode>2): Likewise.
10582         (avx2_permvar<mode>): Likewise.
10583         (avx2_perm<mode>_1): Likewise.
10584         (*avx_vpermilp<mode>): Likewise.
10585         (avx_vpermilvar<mode>3): Likewise.
10586         (avx2_ashrv<mode>): Likewise.
10587         (avx2_<shift_insn>v<mode>): Likewise.
10588         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
10589         -mavx512cd.
10590         * doc/rtl.texi: Document XImode.
10592 2013-08-21  Jeff Law  <law@redhat.com>
10594         * tree-flow.h (register_jump_thread): Pass vector of edges
10595         instead of each important edge.
10596         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
10597         thread path into a vector and pass that to register_jump_thread.
10598         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
10599         passed in edge vector to the current 3-edge form.
10601         Revert:
10602         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
10604         * dce.c (fini_dce): Call df_analyze again just in case
10605         delete_unmarked_insns removed anything.
10607 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
10609         * reload.h (struct reg_equivs): Rename to ..
10610         (struct reg_equivs_s): .. this.
10612 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
10614         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
10616 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10618         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
10620 2013-08-21  Jeff Law  <law@redhat.com>
10622         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
10623         simplify assignments too.  If the RHS collapses to a singleton
10624         range, then return the value for the range.
10626 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
10628         * config/i386/sse.md (V16): Rename to...
10629         (VMOVE): this.
10630         (mov<mode>): Update iterator name.
10631         (*mov<mode>_internal): Ditto.
10632         (push<mode>1): Ditto.
10633         (movmisalign<mode>): Ditto.
10635 2013-08-20  Jan Hubicka  <jh@suse.cz>
10637         PR bootstrap/58186
10638         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
10639         entry for direct edges.
10640         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
10642 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
10644         Revert my last two changes, r201865 and r201864:
10646         Revert r201865:
10647         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
10649         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
10650         instances can own GC refs.
10652         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
10653         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
10654         (gcc::context::gt_pch_nx): Likewise.
10655         (gcc::context::gt_pch_nx):  Likewise.
10656         * ggc.h (gt_ggc_mx <T>): New.
10657         (gt_pch_nx_with_op <T>): New.
10658         (gt_pch_nx <T>): New.
10659         * passes.c (opt_pass::gt_ggc_mx): New.
10660         (opt_pass::gt_pch_nx): New.
10661         (opt_pass::gt_pch_nx_with_op): New.
10662         (pass_manager::gt_ggc_mx): New.
10663         (pass_manager::gt_pch_nx): New.
10664         (pass_manager::gt_pch_nx_with_op): New.
10665         (pass_manager::operator new): Use
10666         ggc_internal_cleared_alloc_stat rather than xcalloc.
10667         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
10668         (pass_manager::gt_ggc_mx): New.
10669         (pass_manager::gt_pch_nx): New.
10670         (pass_manager::gt_pch_nx_with_op): New.
10671         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
10672         (opt_pass::operator new): New.
10673         (opt_pass::gt_ggc_mx): New.
10674         (opt_pass::gt_pch_nx): New.
10675         (opt_pass::gt_pch_nx_with_op): New.
10677         Revert r201864:
10678         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
10680         * Makefile.in (GTFILES): Add context.h.
10681         * context.c (gcc::context::operator new): New.
10682         (gcc::context::gt_ggc_mx): New.
10683         (gcc::context::gt_pch_nx): New.
10684         (gcc::context::gt_pch_nx): New.
10685         * context.h (gcc::context): Add GTY((user)) marking.
10686         (gcc::context::operator new): New.
10687         (gcc::context::gt_ggc_mx): New.
10688         (gcc::context::gt_pch_nx): New.
10689         (gcc::context::gt_pch_nx): New.
10690         (g): Add GTY marking.
10691         (gt_ggc_mx (gcc::context *)): New.
10692         (gt_pch_nx (gcc::context *)): New.
10693         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
10694         void *cookie)): New.
10695         * gengtype.c (open_base_files) <ifiles>: Add context.h.
10697 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
10699         * dce.c (fini_dce): Call df_analyze again just in case
10700         delete_unmarked_insns removed anything.
10702 2013-08-20  Teresa Johnson  <tejohnson@google.com>
10704         PR rtl-optimizations/57451
10705         * final.c (reemit_insn_block_notes): Prevent lexical blocks
10706         from crossing split section boundaries.
10708 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
10710         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
10711         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
10712         with MULTLIB_DEFAULTS.
10714 2013-08-20  Nick Clifton  <nickc@redhat.com>
10716         * target.def (narrow_volatile_bitfield): Note that the default
10717         value is false, not !TARGET_STRICT_ALIGN.
10718         * doc/tm.texi: Regenerate.
10720 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
10722         Fix LIB_SPEC for systems without libpthread.
10724         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
10725         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
10726         for Android.
10727         * config/i386/linux-common.h: Likewise.
10728         * config/mips/linux-common.h: Likewise.
10730 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
10732         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
10733         checks.
10735 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
10737         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
10738         instances can own GC refs.
10740         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
10741         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
10742         (gcc::context::gt_pch_nx): Likewise.
10743         (gcc::context::gt_pch_nx):  Likewise.
10744         * ggc.h (gt_ggc_mx <T>): New.
10745         (gt_pch_nx_with_op <T>): New.
10746         (gt_pch_nx <T>): New.
10747         * passes.c (opt_pass::gt_ggc_mx): New.
10748         (opt_pass::gt_pch_nx): New.
10749         (opt_pass::gt_pch_nx_with_op): New.
10750         (pass_manager::gt_ggc_mx): New.
10751         (pass_manager::gt_pch_nx): New.
10752         (pass_manager::gt_pch_nx_with_op): New.
10753         (pass_manager::operator new): Use
10754         ggc_internal_cleared_alloc_stat rather than xcalloc.
10755         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
10756         (pass_manager::gt_ggc_mx): New.
10757         (pass_manager::gt_pch_nx): New.
10758         (pass_manager::gt_pch_nx_with_op): New.
10759         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
10760         (opt_pass::operator new): New.
10761         (opt_pass::gt_ggc_mx): New.
10762         (opt_pass::gt_pch_nx): New.
10763         (opt_pass::gt_pch_nx_with_op): New.
10765 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
10767         * Makefile.in (GTFILES): Add context.h.
10768         * context.c (gcc::context::operator new): New.
10769         (gcc::context::gt_ggc_mx): New.
10770         (gcc::context::gt_pch_nx): New.
10771         (gcc::context::gt_pch_nx): New.
10772         * context.h (gcc::context): Add GTY((user)) marking.
10773         (gcc::context::operator new): New.
10774         (gcc::context::gt_ggc_mx): New.
10775         (gcc::context::gt_pch_nx): New.
10776         (gcc::context::gt_pch_nx): New.
10777         (g): Add GTY marking.
10778         (gt_ggc_mx (gcc::context *)): New.
10779         (gt_pch_nx (gcc::context *)): New.
10780         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
10781         void *cookie)): New.
10782         * gengtype.c (open_base_files) <ifiles>: Add context.h.
10784 2013-08-20  Alan Modra  <amodra@gmail.com>
10786         PR target/57865
10787         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
10788         (rs6000_emit_epilogue): Likewise.
10790 2013-08-19  Dehao Chen  <dehao@google.com>
10792         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
10794 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
10795             Jakub Jelinek  <jakub@redhat.com>
10797         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
10798         (BUILT_IN_FABSD64): Likewise.
10799         (BUILT_IN_FABSD128): Likewise.
10800         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
10801         (fold_builtin_1): Likewise.
10802         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
10803         destination and source operands.
10804         (*abstd2_fpr): Likewise.
10805         (*nabstd2_fpr): Likewise.
10807 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
10809         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
10810         JUMP_P and INSN_P.
10812 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
10814         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
10815         incomplete.
10817 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10819         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
10820         * builtins.c (default_libc_has_function): New.
10821         (gnu_libc_has_function): Ditto.
10822         (no_c99_libc_has_function): Ditto.
10823         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
10824         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
10825         (fold_builtin_sincos): Likewise.
10826         (fold_builtin_cexp): Likewise.
10827         * builtins.def (DEF_C94_BUILTIN): Likewise.
10828         (DEF_C99_BUILTIN): Likewise.
10829         (DEF_C99_C90RES_BUILTIN): Likewise.
10830         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
10831         definitions to using this define.
10832         * config/darwin-protos.h (darwin_libc_has_function): New.
10833         * config/darwin.c (darwin_libc_has_function): Ditto.
10834         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
10835         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
10836         * config/darwin.h: Ditto.
10837         * config/elfos.h: Ditto.
10838         * config/freebsd.h: Ditto.
10839         * config/i386/cygming.h: Ditto.
10840         * config/i386/djgpp.h: Ditto.
10841         * config/i386/i386-interix.h: Ditto.
10842         * config/microblaze/microblaze.h: Ditto.
10843         * config/mmix/mmix.h: Ditto.
10844         * config/gnu-user.h: Ditto.
10845         * config/ia64/hpux.h: Ditto.
10846         * config/pa/pa-hpux.h: Ditto.
10847         * config/pdp11/pdp11.h: Ditto.
10848         * config/picochip/picochip.h: Ditto.
10849         * config/linux.h: Ditto.
10850         * config/netbsd.h: Ditto.
10851         * config/openbsd.h: Ditto.
10852         * config/rs6000/aix43.h: Ditto.
10853         * config/rs6000/aix51.h: Ditto.
10854         * config/rs6000/aix52.h: Ditto.
10855         * config/rs6000/aix53.h: Ditto.
10856         * config/rs6000/aix61.h: Ditto.
10857         * config/rs6000/darwin.h: Ditto.
10858         * config/rs6000/linux.h: Ditto.
10859         * config/rs6000/linux64.h: Ditto.
10860         * config/s390/tpf.h: Ditto.
10861         * config/sol2-10.h: Ditto.
10862         * config/sol2.h: Ditto.
10863         * config/vms/vms.h: Ditto.
10864         * config/vxworks.h: Ditto.
10865         * config/linux-android.c (linux_android_libc_has_function):
10866         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
10867         * config/linux-protos.h (linux_android_libc_has_function):
10868         New declaration.
10869         * config/i386/i386.c (ix86_libc_has_function): New.
10870         * config/i386/i386-protos.h
10871         (ix86_libc_has_function): New declaration.
10872         * config/i386/i386.md
10873         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
10874         ("isinf<mode>2): Likewise.
10875         * convert.c (convert_to_integer): Using new target hook
10876         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
10877         TARGET_C99_FUNCTIONS.
10878         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
10879         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
10880         * coretypes.h (function_class): New enum for different
10881         classes of functions.
10882         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
10883         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
10884         (TARGET_HAS_SINCOS): Likewise.
10885         (TARGET_LIBC_HAS_FUNCTION): New.
10886         * doc/tm.texi: Regenerated.
10887         * targhooks.h (default_libc_has_function): New declaration.
10888         (no_c99_libc_has_function): Ditto.
10889         (gnu_libc_has_function): Ditto.
10890         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
10891         and TARGET_HAS_SINCOS.
10893 2013-08-18  Jan Hubicka  <jh@suse.cz>
10895         * Makeifle-in (ipa-devirt.o): New.
10896         (GTFILES): Add ipa-utils.h and ipa-devirt.c
10897         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
10898         (analyze_functions): Look into possible targets of polymorphic call.
10899         * dumpfile.c (dump_files): Add type-inheritance dump.
10900         * dumpfile.h (TDI_inheritance): New.
10901         * ipa-devirt.c: New file.
10902         * ipa-utils.h (odr_type_d): Forward declare.
10903         (odr_type): New type.
10904         (build_type_inheritance_graph): Declare.
10905         (possible_polymorphic_call_targets): Declare and introduce inline
10906         variant when only edge is pased.
10907         (dump_possible_polymorphic_call_targets): Likewise.
10908         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
10909         * tree.c (type_in_anonymous_namespace_p): Break out from ...
10910         (types_same_for_odr): ... here.
10911         * tree.h (type_in_anonymous_namespace_p): Declare.
10913 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
10915         PR tree-optimization/58006
10916         * tree-parloops.c (take_address_of): Don't ICE if get_name
10917         returns NULL.
10918         (eliminate_local_variables_stmt): Remove clobber stmts.
10920 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
10922         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
10923         error message is issued for an alias to undefined symbol.
10925 2013-08-18  Jan Hubicka  <jh@suse.cz>
10927         * cgraph.c (cgraph_create_indirect_edge): Discover
10928         polymorphic calls and record basic info into indirect_info.
10929         * gimple-fold.c (gimple_fold_call): When doing BINFO based
10930         devirtualization, ignore objc function calls.
10931         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
10932         call with no parm index info.
10933         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
10934         * tree.c (virtual_method_call_p): New function.
10935         * tree.h (virtual_method_call_p): Declare.
10937 2013-08-16  Jan Hubicka  <jh@suse.cz>
10939         PR middle-end/58179
10940         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
10942 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
10944         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
10945         attribute.
10947 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
10949         * gengtype.c (type_for_name): Add special-case support for
10950         locating types within the "gcc::" namespace.
10951         (open_base_files): Emit a "using namespace gcc" directive.
10953 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
10955         PR target/58160
10956         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
10957         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
10959         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
10960         array instead of each individual operand as a separate argument.
10961         (emit_fusion_gpr_load): Likewise.
10962         (expand_fusion_gpr_load): Add new function declaration.
10964         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
10965         signature to have the operands passed as an array, instead of as
10966         separate arguments.  Allow ZERO_EXTEND to be in the memory
10967         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
10968         depend on the register live/dead flags when peepholes are run.
10969         (expand_fusion_gpr_load): New function to be called from the
10970         peephole2 pass, to change the register that addis sets to be the
10971         target register.
10972         (emit_fusion_gpr_load): Change the calling signature to have the
10973         operands passed as an array, instead of as separate arguments.
10974         Allow ZERO_EXTEND to be in the memory address, and also
10975         SIGN_EXTEND if -mpower8-fusion-sign.
10977         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
10978         unspec enumeration.
10979         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
10980         adjust the register addis loads up in the peephole2 pass.  Do not
10981         depend on the register live/dead state when the peephole pass is done.
10983 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
10985         * gengtype.c (create_user_defined_type): Ensure that the kind
10986         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
10987         declaration is seen before the GTY((user)) marking.
10989 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10991         PR target/58105
10992         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
10994 2013-08-16  Jan Hubicka  <jh@suse.cz>
10996         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
10997         arugment expected_type.
10998         (gimple_fold_call): Use it.
10999         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
11000         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
11001         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
11002         (try_make_edge_direct_virtual_call): Likewise.
11003         * tree.c (obj_type_ref_class): New.
11004         * tree.h (obj_type_ref_class): Use it.
11006 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11008         * sched-vis.c (rtl_slim_pp_initialized): Remove.
11009         (rtl_slim_pp): Likewise.
11010         (init_rtl_slim_pretty_print): Likewise.
11011         (dump_value_slim):  Don't call it.  Use local pretty printer.
11012         (dump_insn_slim): Likewise.
11013         (dump_rtl_slim): Likewise.
11014         (str_pattern_slim): Likewise.
11015         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
11016         Simplify.
11018 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
11020         PR tree-optimization/58164
11021         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
11022         walk gimple_goto_dest of GIMPLE_GOTO.
11024         PR tree-optimization/58165
11025         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
11026         bi_call must be the last stmt in a bb, don't split_block, instead
11027         use fallthru edge from it and give up if there is none.
11028         Release conds vector when returning early.
11030 2013-08-14  Xinliang David Li  <davidxl@google.com>
11032         * config/i386/i386.c (ix86_option_override_internal):
11033         Remove unused variable and field.
11035 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11037         PR target/57949
11038         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
11039         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
11040         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
11041         and Linux, correct BLKmode alignment when 128-bit alignment is
11042         required and compatibility flag is not set.
11043         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
11044         for zero-size arguments when compatibility flag is not set.
11046 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11048         PR tree-optimization/58145
11049         * tree-sra.c (build_ref_for_offset): If prev_base has
11050         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
11052 2013-08-14  Xinliang David Li  <davidxl@google.com>
11054         * config/i386/i386.c (ix86_option_override_internal):
11055         Fix uninitialized variable error.
11057 2013-08-14  Xinliang David Li  <davidxl@google.com>
11059         * config/i386/i386.opt: Define two new options.
11060         * config/i386/x86-tune.def: Add arch selector field in macros.
11061         * config/i386/i386.h: Adjust macro definition.
11062         * config/i386/i386.c (ix86_option_override_internal):
11063         Refactor the code.
11064         (parse_mtune_ctrl_str): New function.
11065         (set_ix86_tune_features): New function.
11066         (ix86_function_specific_restore): Call the new helper function.
11068 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
11070         PR rtl-optimization/57662
11071         * sel-sched.c (code_motion_process_successors): When the current insn
11072         is removed after the recursive traversal, break from the loop.
11073         Add comments and debug printouts.
11075 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11076             Alexandre Oliva  <aoliva@redhat.com>
11078         PR target/58067
11079         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
11080         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
11081         there also UNSPEC_PLTOFF.
11083 2013-08-14  Marek Polacek  <polacek@redhat.com>
11085         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
11086         NUM_CONDITIONS bit positions.
11088 2013-08-13  Cary Coutant  <ccoutant@google.com>
11090         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
11091         (attr_checksum): Hash vector contents instead of pointer.
11092         (attr_checksum_ordered): Likewise.
11094 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
11096         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
11097         when Pmode != word_mode.  Add length_address attribute.
11098         (sse3_monitor_<mode>): Merge from sse3_monitor and
11099         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
11100         Pmode != word_mode.  Update insn length attribute.
11101         * config/i386/i386.c (ix86_option_override_internal): Update
11102         ix86_gen_monitor selection for merged sse3_monitor insn.
11104 2013-08-13  Julian Brown  <julian@codesourcery.com>
11106         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
11107         perform invalid legitimization on greater-than-word-size modes for
11108         TARGET_E500_DOUBLE.
11110 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
11112         * ira.c (setup_class_translate_array): Use aclass instead of cl
11113         for classes not fully covered by allocno classes.
11115 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
11117         PR tree-optimization/57661
11118         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
11119         * tree-inline.c (tree_function_versioning): Initialize it.
11120         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
11121         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
11122         that is not being copied.
11124         PR sanitizer/56417
11125         * asan.c (instrument_strlen_call): Fix typo in comment.
11126         Use char * type even for the lhs of POINTER_PLUS_EXPR.
11128 2013-08-13  Steve Ellcey  <sellcey@mips.com>
11130         * config/mips/mips.md (prefetch): Use lw instead of ld on
11131         loongson in 32bit mode.
11133 2013-08-13  Nick Clifton  <nickc@redhat.com>
11135         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
11137 2013-08-13  Jan Hubicka  <jh@suse.cz>
11139         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
11140         introduced edge; fix typo in sanity check.
11141         (cgraph_resolve_speculation): Export; improve diagnostic.
11142         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
11143         speculation at type mismatch.
11144         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
11145         (cgraph_resolve_speculation): Declare.
11146         (symtab_can_be_discarded): New function.
11147         * value-prof.c (gimple_ic_transform): Remove actual transform code.
11148         * ipa-inline-transform.c (speculation_removed): New global var.
11149         (clone_inlined_nodes): See if speculation can be removed.
11150         (inline_call): If speculations was removed, we growths may not match.
11151         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
11152         (speculation_useful_p): New function.
11153         (resolve_noninline_speculation): New function.
11154         (inline_small_functions): Resolve useless speculations.
11155         * ipa-inline.h (speculation_useful_p): Declare
11156         * ipa.c (can_replace_by_local_alias): Simplify.
11157         (ipa_profile): Produce speculative calls in non-lto, too;
11158         add simple cost model; produce local aliases.
11160 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
11162         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
11163         PASS_MANAGER_H.
11165 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
11167         * config/i386/i386.c (ix86_function_versions): Use error + inform.
11169 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
11171         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
11172         iterator instead of X87MODEF.
11174 2013-08-12  Perez Read  <netfirewall@gmail.com>
11176         PR target/58132
11177         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
11178         operand 0 for intel asm alternative.
11179         (*movabs<mode>_2): Ditto for operand 1.
11181 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
11183         * config/aarch64/arm_none.h
11184         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
11186 2013-08-12  Nick Clifton  <nickc@redhat.com>
11188         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
11190 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
11192         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
11193         expand for QI/HImode operand to produce more effictive code for
11194         unsigned char(short) --> float(double) conversion.
11196 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
11198         * doc/invoke.texi: Mention that -ftls-model does not force the final
11199         model.
11201 2013-08-12  Marek Polacek  <polacek@redhat.com>
11202             Marc Glisse  <marc.glisse@inria.fr>
11204         PR tree-optimization/57980
11205         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
11206         when creating -1 constant.
11208 2013-08-10  Jan Hubicka  <jh@suse.cz>
11210         Workaround binutils PR14342.
11211         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
11212         (gimple_init_edge_profiler): Likewise.
11213         (gimple_gen_ic_func_profiler): Likewise.
11215 2013-08-09  Jan Hubicka  <jh@suse.cz>
11217         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
11219 2013-08-09  Xinliang David Li  <davidxl@google.com>
11221         * config/i386/stringop.def: New file.
11222         * config/i386/stringop.opt: New file.
11223         * config/i386/i386-opts.h: Include stringopt.def.
11224         * config/i386/i386.opt: Include stringopt.opt.
11225         * config/i386/i386.c (ix86_option_override_internal):
11226         Override default size based stringop inline strategies with options.
11227         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
11228         New function.
11230 2013-08-09  Jan Hubicka  <jh@suse.cz>
11232         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
11234 2013-08-09  Jan Hubicka  <jh@suse.cz>
11236         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
11237         CGRAPH_FREQ_MAX.
11238         (dump_cgraph_node): Dump profile-id.
11239         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
11240         and common_target_probability.
11241         * lto-cgraph.c (lto_output_edge): Stream common targets.
11242         (lto_output_node): Stream profile ids.
11243         (input_node): Stream profile ids.
11244         (input_edge): Stream common targets.
11245         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
11246         * ipa.c: Include value-prof.h
11247         (ipa_profile_generate_summary): Turn indirect call statement histograms
11248         into common targets.
11249         (ipa_profile): Turn common targets into speculative edges.
11251 2013-08-09  Jan Hubicka  <jh@suse.cz>
11253         * cgraph.h (cgraph_node): Add profile_id.
11254         * value-prof.c (cgraph_node_map): Turn into pointer_map.
11255         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
11256         (del_node_map): Update.
11257         (find_func_by_funcdef_no): Replace by ...
11258         (find_func_by_profile_id): ... this one.
11259         (gimple_ic_transform): Do not remove useful histograms when
11260         speculation is not done; dump info when indirect call removal
11261         can happen at LTO.
11262         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
11263         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
11264         (__gcov_indirect_call_profiler_v2): .. this one.
11265         * profile.h (init_node_map): Update.
11266         * coverage.c (coverage_compute_profile_id): New function.
11267         * coverage.h (coverage_compute_profile_id): Declare.
11268         * tree-profile.c (init_ic_make_global_vars): Make
11269         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
11270         (gimple_init_edge_profiler): Update prototype of
11271         __gcov_indirect_call_profiler.
11272         (gimple_gen_ic_func_profiler): Simplify.
11273         (tree_profiling): Use init_node_map
11275 2013-08-09  Jan Hubicka  <jh@suse.cz>
11277         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
11278         non-speculative refs.
11279         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
11280         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
11281         (cgraph_set_call_stmt): Likewise.
11282         (cgraph_create_edge_1): Fix release checking compilatoin;
11283         clear lto_stmt_uid.
11284         (cgraph_free_edge): Free indirect info.
11285         (cgraph_turn_edge_to_speculative): New function.
11286         (cgraph_speculative_call_info): New function.
11287         (cgraph_make_edge_direct): Return direct edge; handle speculation.
11288         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
11289         (dump_cgraph_node): Dump speculation.
11290         (verify_edge_count_and_frequency): Accept speculative edges.
11291         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
11292         (verify_cgraph_node): Handle speculation.
11293         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
11294         (cgraph_set_call_stmt): Update prototype.
11295         (cgraph_make_edge_direct): Update prototype.
11296         (cgraph_speculative_call_info): Declare.
11297         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
11298         to change; update call of ipa_find_references.
11299         * ipa-ref.c (ipa_record_reference): Fix return value; clear
11300         lto_stmt_uid and speculative flags.
11301         (ipa_dump_references): Dump speculation.
11302         (ipa_clone_references): Clone speculative flag.
11303         (ipa_clone_referring): Likewise.
11304         (ipa_clone_ref): New function.
11305         (ipa_find_reference): Look into lto_stmt_uids
11306         (ipa_clear_stmts_in_references): Do not clear speculative calls.
11307         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
11308         (ipa_find_reference): Update declaration.
11309         (ipa_clone_ref): Declare.
11310         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
11311         stream speculative flag.
11312         (lto_output_ref): Stream statements uids and speculation.
11313         (input_ref): Likewise.
11314         (input_edge): Stream speuclation.
11315         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
11316         (cgraph_set_call_stmt_including_clones): Handle speculation.
11317         * ipa-inline.c (heap_edge_removal_hook): New function.
11318         (inline_small_functions): Register it.
11319         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
11320         also initialize refs.
11321         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
11322         edge to change.
11323         (try_make_edge_direct_simple_call): Likewise.
11324         (try_make_edge_direct_simple_call): Likewise.
11325         (update_indirect_edges_after_inlining): Likewise.
11326         (remove_described_reference): Look proper lto_stmt_uid.
11327         (propagate_controlled_uses): Likewise.
11328         (propagate_controlled_uses): Liekwise.
11329         * tree-inline.c (copy_bb): Copy speculative edges.
11330         (redirect_all_calls): New function.
11331         (copy_cfg_body): Do redirection after loop info is updated.
11332         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
11334 2013-08-09  Jan Hubicka  <jh@suse.cz>
11336         * lto-streamer-out.c (output_function): Renumber PHIs.
11337         * lto-streamer-in.c (input_function): Likewise.
11339 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
11341         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
11342         (get_lane_unsigned): Likewise.
11343         (dup_lane_scalar): Likewise.
11344         (get_lane): enable for VALL.
11345         * config/aarch64/aarch64-simd.md
11346         (aarch64_dup_lane_scalar<mode>): Remove.
11347         (aarch64_get_lane_signed<mode>): Likewise.
11348         (aarch64_get_lane_unsigned<mode>): Likewise.
11349         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
11350         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
11351         (aarch64_get_lane<mode>): Enable for all vector modes.
11352         (aarch64_get_lanedi): Remove misleading constraints.
11353         * config/aarch64/arm_neon.h
11354         (__aarch64_vget_lane_any): Define.
11355         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
11356         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
11357         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
11358         * config/aarch64/iterators.md (VDQQH): New.
11359         (VDQQHS): Likewise.
11360         (vwcore): Likewise.
11362 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
11364         * configure.ac: Add GAS check for LEON instructions on SPARC.
11365         * configure: Regenerate.
11366         * config.in: Likewise.
11367         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
11368         sparc*-*-* block.
11369         * config/sparc/sparc.opt (LEON, LEON3): New masks.
11370         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
11371         for LEON or LEON3.
11372         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
11373         (AS_LEON_FLAG): New macro.
11374         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
11375         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
11376         Deal with LEON and LEON3 for the memory model.
11377         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
11378         (atomic_compare_and_swap<mode>_1): Likewise.
11379         (*atomic_compare_and_swap<mode>_1): Likewise.
11381 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11383         * config/arm/neon.md (vcond): Fix floating-point vector
11384         comparisons against 0.
11386 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
11388         * lra-constraints.c (emit_spill_move): Remove assert.
11389         (process_alt_operands): Add more debugging
11390         output.  Increase reject for spilling into memory.  Decrease
11391         reject for reloading scratch.
11392         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
11394 2013-08-08  Steve Ellcey  <sellcey@mips.com>
11396         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
11397         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
11398         micromips incompatible.  Add nan2008.
11399         (MULTILIB_DIRNAMES): Add nan2008.
11400         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
11401         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
11402         and micromips incompatible.  Add nan2008.
11403         (MULTILIB_DIRNAMES): Add nan2008.
11404         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
11406 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
11408         PR rtl-optimization/58079
11409         * combine.c (combine_simplify_rtx): Avoid using SUBST if
11410         simplify_comparison has widened a comparison with an integer.
11412 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11414         * config/arm/neon.md (movmisalign<mode>): Disable when we
11415         don't allow unaligned accesses.
11416         (*movmisalign<mode>_neon_store): Likewise.
11417         (*movmisalign<mode>_neon_load): Likewise.
11418         (*movmisalign<mode>_neon_store): Likewise.
11419         (*movmisalign<mode>_neon_load): Likewise.
11421 2013-08-08  Jan Hubicka  <jh@suse.cz>
11423         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
11424         (make_pass_rebuild_cgraph_edges): Also clear references.
11425         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
11426         * ipa-inline-transform.c (inline_transform): Remove all references
11427         after inlining.
11428         * cgraphunit.c (expand_function): Remove all references after
11429         expansion.
11430         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
11431         (ipa_find_reference): Rewrite to iterator.
11432         (remove_stmt_references): Likewise.
11433         (ipa_clear_stmts_in_references): New function.
11434         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
11435         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
11436         clear references.
11437         * ipa-split.c (split_function): Remove references in split function.
11439 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
11441         PR target/57431
11442         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
11443         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
11445 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
11447         PR target/56979
11448         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
11449         suggested mode for the assignment isn't compatible with the
11450         registers required.
11452 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
11454         PR target/58065
11455         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
11457 2013-08-07  Xinliang David Li  <davidxl@google.com>
11459         * config/i386/i386.opt: New option -mtune-ctrl=.
11460         * config/i386/x86-tune.def: New file.
11461         * config/i386/i386.h: include x86-tune.def.
11462         * config/i386/i386.c (ix86_option_override_internal):
11463         Parsing -mtune-ctrl= option and set tune features.
11465 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
11467         PR other/12081
11468         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
11469         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
11470         to gen_2arg_fn_t.
11472 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
11474         * rtl.h (update_alignments): Declare.
11475         * final.c (grow_label_align): New function extracted from...
11476         (shorten_branches): ...here.  Call it.
11477         (update_alignments): New function.
11478         * reorg.c (sibling_labels): New variable.
11479         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
11480         the new label along with it onto the sibling_labels vector.
11481         (fill_simple_delay_slots): Adjust call to get_label_before.
11482         (fill_slots_from_thread): Likewise.
11483         (relax_delay_slots): Likewise.
11484         (make_return_insns): Likewise.
11485         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
11487 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
11489         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
11490         document its semantics.
11491         (diagnostic_report_diagnostic): Adjust accordingly.
11493 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
11495         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
11496         (sparc_option_override): ...and port to new C++ pass API.
11497         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
11499 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
11501         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
11503 2013-08-06  Caroline Tice  <cmtice@google.com>
11505         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
11506         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
11507         * tree-pass.h: Add pass_vtable_verify.
11508         * varasm.c (assemble_variable): Add code to properly set the comdat
11509         section and name for the .vtable_map_vars section.
11510         (assemble_vtyv_preinit_initializer): New function.
11511         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
11512         LINK_ONCE flag.
11513         * output.h: Add function decl for assemble_vtv_preinit_initializer.
11514         * vtable-verify.c: New file.
11515         * vtable-verify.h: New file.
11516         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
11517         initialiation levels.
11518         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
11519         * passes.def: Insert pass_vtable_verify.
11520         * aclocal.m4: Reorder includes.
11521         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
11522         -fvtv-counts options.
11523         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
11524         as appropriate, if -fvtable-verify=... is used.
11525         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
11526         -fvtable-verify=... is used.
11527         * Makefile.in (OBJS):  Add vtable-verify.o to list.
11528         (vtable-verify.o): Add new build rule.
11529         (GTFILES): Add vtable-verify.c to list.
11530         * common.opt (fvtable-verify=): New flag.
11531         (vtv_priority): Values for fvtable-verify= flag.
11532         (fvtv-counts): New flag.
11533         (fvtv-debug): New flag.
11534         * tree.h (save_vtable_map_decl): New extern function decl.
11536 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
11538         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
11539         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
11540         (pass_data_rl78_devirt): ...new pass_data instance and...
11541         (make_pass_rl78_devirt): ...new function.
11542         (rl78_asm_file_start): Port pass registration to new C++ API.
11544 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
11546         * coretypes.h (rtl_opt_pass): Add.
11547         (gcc::context): Add.
11548         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
11549         (epiphany_init): Port to new C++ pass API.
11550         (epiphany_optimize_mode_switching): Likewise.
11551         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
11552         (pass_manager::get_pass_mode_switching): New.
11553         (pass_manager::get_pass_peephole2): New.
11554         * mode-switching.c (pass_mode_switching): Add clone method.
11555         * recog.c (pass_peephole2): Add clone method.
11556         (pass_split_all_insns): Add clone method.
11558 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
11560         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
11561         (mips_option_override): ...here, porting to new C++ API for passes.
11563 2013-08-06  Jan Hubicka  <jh@suse.cz>
11565         * cgraph.c (cgraph_get_body): New function based on lto.c
11566         implementation.
11567         * cgraph.h (cgraph_get_body): Declare.
11568         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
11569         LTO paths.
11570         * cgraphunit.c (expand_function): Get body prior expanding.
11571         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
11572         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
11573         really need.
11574         * passes.c (do_per_function_toporder): Get body.
11575         * tree-inline.c (expand_call_inline): Get body prior inlining it.
11576         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
11578 2013-08-06  Martin Jambor  <mjambor@suse.cz>
11580         PR fortran/57987
11581         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
11582         is not re-finalized.  Rename second parameter to no_collect.
11584 2013-08-06  Martin Jambor  <mjambor@suse.cz>
11586         PR middle-end/58041
11587         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
11588         MEM_REF has proper alignment information.
11590 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
11592         PR other/12081
11593         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
11594         class insn_gen_fn.
11595         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
11596         rtx (*) (rtx, ...) with insn_gen_fn.
11597         * genoutput.c (output_insn_data): Cast gen_? function pointers to
11598         insn_gen_fn::stored_funcptr.  Add initializer braces.
11600 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11602         Rewrite how instances of passes are cloned to remove assumptions
11603         about their sizes (thus allowing pass subclasses to have
11604         additional data fields, albeit non-GC-managed ones at this point).
11606         * passes.c (make_pass_instance): Now that passes have clone
11607         methods, rewrite this function to eliminate XNEW and memcpy
11608         calls that used hardcoded sizes.  Since this function no longer
11609         creates pass instances, rename it to...
11610         (add_pass_instance): ...this.  Document the old way that passes were
11611         numbered and flagged, and rework this function to continue using it.
11612         (next_pass_1): Add an initial_pass argument for use by
11613         add_pass_instance.
11614         (position_pass): When adding multiple instances of a pass, use
11615         the pass's clone method, rather than relying on the XNEW/memcpy
11616         within the former make_pass_instance (now add_pass_instance).
11617         (pass_manager::pass_manager): When invoking next_pass_1, also supply
11618         the initial instance of the current pass within the pass manager.
11620 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11622         This is the automated part of the conversion of passes from C
11623         structs to C++ classes.
11625         Patch autogenerated by refactor_passes.py from
11626         https://github.com/davidmalcolm/gcc-refactoring-scripts
11627         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
11629         * asan.c (pass_asan): Convert from a global struct to a subclass of
11630         gimple_opt_pass along with...
11631         (pass_data_asan): ...new pass_data instance and...
11632         (make_pass_asan): ...new function.
11633         (pass_asan_O0): Convert from a global struct to a subclass of
11634         gimple_opt_pass along with...
11635         (pass_data_asan_O0): ...new pass_data instance and...
11636         (make_pass_asan_O0): ...new function.
11637         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
11638         subclass of rtl_opt_pass along with...
11639         (pass_data_inc_dec): ...new pass_data instance and...
11640         (make_pass_inc_dec): ...new function.
11641         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
11642         a subclass of rtl_opt_pass along with...
11643         (pass_data_reorder_blocks): ...new pass_data instance and...
11644         (make_pass_reorder_blocks): ...new function.
11645         (pass_duplicate_computed_gotos): Convert from a global struct to a
11646         subclass of rtl_opt_pass along with...
11647         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
11648         (make_pass_duplicate_computed_gotos): ...new function.
11649         (pass_partition_blocks): Convert from a global struct to a subclass of
11650         rtl_opt_pass along with...
11651         (pass_data_partition_blocks): ...new pass_data instance and...
11652         (make_pass_partition_blocks): ...new function.
11653         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
11654         struct to a subclass of rtl_opt_pass along with...
11655         (pass_data_branch_target_load_optimize1): ...new pass_data instance
11656         and...
11657         (make_pass_branch_target_load_optimize1): ...new function.
11658         (pass_branch_target_load_optimize2): Convert from a global struct to a
11659         subclass of rtl_opt_pass along with...
11660         (pass_data_branch_target_load_optimize2): ...new pass_data instance
11661         and...
11662         (make_pass_branch_target_load_optimize2): ...new function.
11663         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
11664         of rtl_opt_pass along with...
11665         (pass_data_jump): ...new pass_data instance and...
11666         (make_pass_jump): ...new function.
11667         (pass_jump2): Convert from a global struct to a subclass of
11668         rtl_opt_pass along with...
11669         (pass_data_jump2): ...new pass_data instance and...
11670         (make_pass_jump2): ...new function.
11671         * cfgexpand.c (pass_expand): Convert from a global struct to a
11672         subclass of rtl_opt_pass along with...
11673         (pass_data_expand): ...new pass_data instance and...
11674         (make_pass_expand): ...new function.
11675         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
11676         of rtl_opt_pass along with...
11677         (pass_data_free_cfg): ...new pass_data instance and...
11678         (make_pass_free_cfg): ...new function.
11679         (pass_into_cfg_layout_mode): Convert from a global struct to a
11680         subclass of rtl_opt_pass along with...
11681         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
11682         (make_pass_into_cfg_layout_mode): ...new function.
11683         (pass_outof_cfg_layout_mode): Convert from a global struct to a
11684         subclass of rtl_opt_pass along with...
11685         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
11686         (make_pass_outof_cfg_layout_mode): ...new function.
11687         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
11688         struct to a subclass of gimple_opt_pass along with...
11689         (pass_data_build_cgraph_edges): ...new pass_data instance and...
11690         (make_pass_build_cgraph_edges): ...new function.
11691         (pass_rebuild_cgraph_edges): Convert from a global struct to a
11692         subclass of gimple_opt_pass along with...
11693         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
11694         (make_pass_rebuild_cgraph_edges): ...new function.
11695         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
11696         subclass of gimple_opt_pass along with...
11697         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
11698         and...
11699         (make_pass_remove_cgraph_callee_edges): ...new function.
11700         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
11701         struct to a subclass of rtl_opt_pass along with...
11702         (pass_data_stack_adjustments): ...new pass_data instance and...
11703         (make_pass_stack_adjustments): ...new function.
11704         * combine.c (pass_combine): Convert from a global struct to a subclass
11705         of rtl_opt_pass along with...
11706         (pass_data_combine): ...new pass_data instance and...
11707         (make_pass_combine): ...new function.
11708         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
11709         global struct to a subclass of rtl_opt_pass along with...
11710         (pass_data_compare_elim_after_reload): ...new pass_data instance
11711         and...
11712         (make_pass_compare_elim_after_reload): ...new function.
11713         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
11714         of rtl_opt_pass along with...
11715         (pass_data_rtl_cprop): ...new pass_data instance and...
11716         (make_pass_rtl_cprop): ...new function.
11717         * cse.c (pass_cse): Convert from a global struct to a subclass of
11718         rtl_opt_pass along with...
11719         (pass_data_cse): ...new pass_data instance and...
11720         (make_pass_cse): ...new function.
11721         (pass_cse2): Convert from a global struct to a subclass of
11722         rtl_opt_pass along with...
11723         (pass_data_cse2): ...new pass_data instance and...
11724         (make_pass_cse2): ...new function.
11725         (pass_cse_after_global_opts): Convert from a global struct to a
11726         subclass of rtl_opt_pass along with...
11727         (pass_data_cse_after_global_opts): ...new pass_data instance and...
11728         (make_pass_cse_after_global_opts): ...new function.
11729         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
11730         of rtl_opt_pass along with...
11731         (pass_data_ud_rtl_dce): ...new pass_data instance and...
11732         (make_pass_ud_rtl_dce): ...new function.
11733         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
11734         rtl_opt_pass along with...
11735         (pass_data_fast_rtl_dce): ...new pass_data instance and...
11736         (make_pass_fast_rtl_dce): ...new function.
11737         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
11738         a subclass of rtl_opt_pass along with...
11739         (pass_data_df_initialize_opt): ...new pass_data instance and...
11740         (make_pass_df_initialize_opt): ...new function.
11741         (pass_df_initialize_no_opt): Convert from a global struct to a
11742         subclass of rtl_opt_pass along with...
11743         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
11744         (make_pass_df_initialize_no_opt): ...new function.
11745         (pass_df_finish): Convert from a global struct to a subclass of
11746         rtl_opt_pass along with...
11747         (pass_data_df_finish): ...new pass_data instance and...
11748         (make_pass_df_finish): ...new function.
11749         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
11750         rtl_opt_pass along with...
11751         (pass_data_rtl_dse1): ...new pass_data instance and...
11752         (make_pass_rtl_dse1): ...new function.
11753         (pass_rtl_dse2): Convert from a global struct to a subclass of
11754         rtl_opt_pass along with...
11755         (pass_data_rtl_dse2): ...new pass_data instance and...
11756         (make_pass_rtl_dse2): ...new function.
11757         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
11758         subclass of rtl_opt_pass along with...
11759         (pass_data_dwarf2_frame): ...new pass_data instance and...
11760         (make_pass_dwarf2_frame): ...new function.
11761         * except.c (pass_set_nothrow_function_flags): Convert from a global
11762         struct to a subclass of rtl_opt_pass along with...
11763         (pass_data_set_nothrow_function_flags): ...new pass_data instance
11764         and...
11765         (make_pass_set_nothrow_function_flags): ...new function.
11766         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
11767         subclass of rtl_opt_pass along with...
11768         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
11769         and...
11770         (make_pass_convert_to_eh_region_ranges): ...new function.
11771         * final.c (pass_compute_alignments): Convert from a global struct to a
11772         subclass of rtl_opt_pass along with...
11773         (pass_data_compute_alignments): ...new pass_data instance and...
11774         (make_pass_compute_alignments): ...new function.
11775         (pass_final): Convert from a global struct to a subclass of
11776         rtl_opt_pass along with...
11777         (pass_data_final): ...new pass_data instance and...
11778         (make_pass_final): ...new function.
11779         (pass_shorten_branches): Convert from a global struct to a subclass of
11780         rtl_opt_pass along with...
11781         (pass_data_shorten_branches): ...new pass_data instance and...
11782         (make_pass_shorten_branches): ...new function.
11783         (pass_clean_state): Convert from a global struct to a subclass of
11784         rtl_opt_pass along with...
11785         (pass_data_clean_state): ...new pass_data instance and...
11786         (make_pass_clean_state): ...new function.
11787         * function.c (pass_instantiate_virtual_regs): Convert from a global
11788         struct to a subclass of rtl_opt_pass along with...
11789         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
11790         (make_pass_instantiate_virtual_regs): ...new function.
11791         (pass_leaf_regs): Convert from a global struct to a subclass of
11792         rtl_opt_pass along with...
11793         (pass_data_leaf_regs): ...new pass_data instance and...
11794         (make_pass_leaf_regs): ...new function.
11795         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
11796         subclass of rtl_opt_pass along with...
11797         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
11798         and...
11799         (make_pass_thread_prologue_and_epilogue): ...new function.
11800         (pass_match_asm_constraints): Convert from a global struct to a
11801         subclass of rtl_opt_pass along with...
11802         (pass_data_match_asm_constraints): ...new pass_data instance and...
11803         (make_pass_match_asm_constraints): ...new function.
11804         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
11805         subclass of rtl_opt_pass along with...
11806         (pass_data_rtl_fwprop): ...new pass_data instance and...
11807         (make_pass_rtl_fwprop): ...new function.
11808         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
11809         rtl_opt_pass along with...
11810         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
11811         (make_pass_rtl_fwprop_addr): ...new function.
11812         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
11813         rtl_opt_pass along with...
11814         (pass_data_rtl_pre): ...new pass_data instance and...
11815         (make_pass_rtl_pre): ...new function.
11816         (pass_rtl_hoist): Convert from a global struct to a subclass of
11817         rtl_opt_pass along with...
11818         (pass_data_rtl_hoist): ...new pass_data instance and...
11819         (make_pass_rtl_hoist): ...new function.
11820         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
11821         subclass of gimple_opt_pass along with...
11822         (pass_data_lower_cf): ...new pass_data instance and...
11823         (make_pass_lower_cf): ...new function.
11824         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
11825         from a global struct to a subclass of gimple_opt_pass along with...
11826         (pass_data_strength_reduction): ...new pass_data instance and...
11827         (make_pass_strength_reduction): ...new function.
11828         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
11829         of rtl_opt_pass along with...
11830         (pass_data_rtl_ifcvt): ...new pass_data instance and...
11831         (make_pass_rtl_ifcvt): ...new function.
11832         (pass_if_after_combine): Convert from a global struct to a subclass of
11833         rtl_opt_pass along with...
11834         (pass_data_if_after_combine): ...new pass_data instance and...
11835         (make_pass_if_after_combine): ...new function.
11836         (pass_if_after_reload): Convert from a global struct to a subclass of
11837         rtl_opt_pass along with...
11838         (pass_data_if_after_reload): ...new pass_data instance and...
11839         (make_pass_if_after_reload): ...new function.
11840         * init-regs.c (pass_initialize_regs): Convert from a global struct to
11841         a subclass of rtl_opt_pass along with...
11842         (pass_data_initialize_regs): ...new pass_data instance and...
11843         (make_pass_initialize_regs): ...new function.
11844         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
11845         of ipa_opt_pass_d along with...
11846         (pass_data_ipa_cp): ...new pass_data instance and...
11847         (make_pass_ipa_cp): ...new function.
11848         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
11849         global struct to a subclass of gimple_opt_pass along with...
11850         (pass_data_inline_parameters): ...new pass_data instance and...
11851         (make_pass_inline_parameters): ...new function.
11852         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
11853         subclass of gimple_opt_pass along with...
11854         (pass_data_early_inline): ...new pass_data instance and...
11855         (make_pass_early_inline): ...new function.
11856         (pass_ipa_inline): Convert from a global struct to a subclass of
11857         ipa_opt_pass_d along with...
11858         (pass_data_ipa_inline): ...new pass_data instance and...
11859         (make_pass_ipa_inline): ...new function.
11860         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
11861         struct to a subclass of gimple_opt_pass along with...
11862         (pass_data_local_pure_const): ...new pass_data instance and...
11863         (make_pass_local_pure_const): ...new function.
11864         (pass_ipa_pure_const): Convert from a global struct to a subclass of
11865         ipa_opt_pass_d along with...
11866         (pass_data_ipa_pure_const): ...new pass_data instance and...
11867         (make_pass_ipa_pure_const): ...new function.
11868         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
11869         to a subclass of ipa_opt_pass_d along with...
11870         (pass_data_ipa_reference): ...new pass_data instance and...
11871         (make_pass_ipa_reference): ...new function.
11872         * ipa-split.c (pass_split_functions): Convert from a global struct to
11873         a subclass of gimple_opt_pass along with...
11874         (pass_data_split_functions): ...new pass_data instance and...
11875         (make_pass_split_functions): ...new function.
11876         (pass_feedback_split_functions): Convert from a global struct to a
11877         subclass of gimple_opt_pass along with...
11878         (pass_data_feedback_split_functions): ...new pass_data instance and...
11879         (make_pass_feedback_split_functions): ...new function.
11880         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
11881         global struct to a subclass of simple_ipa_opt_pass along with...
11882         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
11883         instance and...
11884         (make_pass_ipa_function_and_variable_visibility): ...new function.
11885         (pass_ipa_free_inline_summary): Convert from a global struct to a
11886         subclass of simple_ipa_opt_pass along with...
11887         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
11888         (make_pass_ipa_free_inline_summary): ...new function.
11889         (pass_ipa_whole_program_visibility): Convert from a global struct to a
11890         subclass of ipa_opt_pass_d along with...
11891         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
11892         and...
11893         (make_pass_ipa_whole_program_visibility): ...new function.
11894         (pass_ipa_profile): Convert from a global struct to a subclass of
11895         ipa_opt_pass_d along with...
11896         (pass_data_ipa_profile): ...new pass_data instance and...
11897         (make_pass_ipa_profile): ...new function.
11898         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
11899         ipa_opt_pass_d along with...
11900         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
11901         (make_pass_ipa_cdtor_merge): ...new function.
11902         * ira.c (pass_ira): Convert from a global struct to a subclass of
11903         rtl_opt_pass along with...
11904         (pass_data_ira): ...new pass_data instance and...
11905         (make_pass_ira): ...new function.
11906         (pass_reload): Convert from a global struct to a subclass of
11907         rtl_opt_pass along with...
11908         (pass_data_reload): ...new pass_data instance and...
11909         (make_pass_reload): ...new function.
11910         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
11911         subclass of rtl_opt_pass along with...
11912         (pass_data_cleanup_barriers): ...new pass_data instance and...
11913         (make_pass_cleanup_barriers): ...new function.
11914         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
11915         of rtl_opt_pass along with...
11916         (pass_data_loop2): ...new pass_data instance and...
11917         (make_pass_loop2): ...new function.
11918         (pass_rtl_loop_init): Convert from a global struct to a subclass of
11919         rtl_opt_pass along with...
11920         (pass_data_rtl_loop_init): ...new pass_data instance and...
11921         (make_pass_rtl_loop_init): ...new function.
11922         (pass_rtl_loop_done): Convert from a global struct to a subclass of
11923         rtl_opt_pass along with...
11924         (pass_data_rtl_loop_done): ...new pass_data instance and...
11925         (make_pass_rtl_loop_done): ...new function.
11926         (pass_rtl_move_loop_invariants): Convert from a global struct to a
11927         subclass of rtl_opt_pass along with...
11928         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
11929         (make_pass_rtl_move_loop_invariants): ...new function.
11930         (pass_rtl_unswitch): Convert from a global struct to a subclass of
11931         rtl_opt_pass along with...
11932         (pass_data_rtl_unswitch): ...new pass_data instance and...
11933         (make_pass_rtl_unswitch): ...new function.
11934         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
11935         subclass of rtl_opt_pass along with...
11936         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
11937         and...
11938         (make_pass_rtl_unroll_and_peel_loops): ...new function.
11939         (pass_rtl_doloop): Convert from a global struct to a subclass of
11940         rtl_opt_pass along with...
11941         (pass_data_rtl_doloop): ...new pass_data instance and...
11942         (make_pass_rtl_doloop): ...new function.
11943         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
11944         a subclass of rtl_opt_pass along with...
11945         (pass_data_lower_subreg): ...new pass_data instance and...
11946         (make_pass_lower_subreg): ...new function.
11947         (pass_lower_subreg2): Convert from a global struct to a subclass of
11948         rtl_opt_pass along with...
11949         (pass_data_lower_subreg2): ...new pass_data instance and...
11950         (make_pass_lower_subreg2): ...new function.
11951         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
11952         struct to a subclass of ipa_opt_pass_d along with...
11953         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
11954         (make_pass_ipa_lto_gimple_out): ...new function.
11955         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
11956         of ipa_opt_pass_d along with...
11957         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
11958         (make_pass_ipa_lto_finish_out): ...new function.
11959         * mode-switching.c (pass_mode_switching): Convert from a global struct
11960         to a subclass of rtl_opt_pass along with...
11961         (pass_data_mode_switching): ...new pass_data instance and...
11962         (make_pass_mode_switching): ...new function.
11963         * modulo-sched.c (pass_sms): Convert from a global struct to a
11964         subclass of rtl_opt_pass along with...
11965         (pass_data_sms): ...new pass_data instance and...
11966         (make_pass_sms): ...new function.
11967         * omp-low.c (pass_expand_omp): Convert from a global struct to a
11968         subclass of gimple_opt_pass along with...
11969         (pass_data_expand_omp): ...new pass_data instance and...
11970         (make_pass_expand_omp): ...new function.
11971         (pass_lower_omp): Convert from a global struct to a subclass of
11972         gimple_opt_pass along with...
11973         (pass_data_lower_omp): ...new pass_data instance and...
11974         (make_pass_lower_omp): ...new function.
11975         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
11976         of gimple_opt_pass along with...
11977         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
11978         (make_pass_diagnose_omp_blocks): ...new function.
11979         * passes.c (pass_early_local_passes): Convert from a global struct to
11980         a subclass of simple_ipa_opt_pass along with...
11981         (pass_data_early_local_passes): ...new pass_data instance and...
11982         (make_pass_early_local_passes): ...new function.
11983         (pass_all_early_optimizations): Convert from a global struct to a
11984         subclass of gimple_opt_pass along with...
11985         (pass_data_all_early_optimizations): ...new pass_data instance and...
11986         (make_pass_all_early_optimizations): ...new function.
11987         (pass_all_optimizations): Convert from a global struct to a subclass
11988         of gimple_opt_pass along with...
11989         (pass_data_all_optimizations): ...new pass_data instance and...
11990         (make_pass_all_optimizations): ...new function.
11991         (pass_all_optimizations_g): Convert from a global struct to a subclass
11992         of gimple_opt_pass along with...
11993         (pass_data_all_optimizations_g): ...new pass_data instance and...
11994         (make_pass_all_optimizations_g): ...new function.
11995         (pass_rest_of_compilation): Convert from a global struct to a subclass
11996         of rtl_opt_pass along with...
11997         (pass_data_rest_of_compilation): ...new pass_data instance and...
11998         (make_pass_rest_of_compilation): ...new function.
11999         (pass_postreload): Convert from a global struct to a subclass of
12000         rtl_opt_pass along with...
12001         (pass_data_postreload): ...new pass_data instance and...
12002         (make_pass_postreload): ...new function.
12003         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
12004         subclass of rtl_opt_pass along with...
12005         (pass_data_gcse2): ...new pass_data instance and...
12006         (make_pass_gcse2): ...new function.
12007         * postreload.c (pass_postreload_cse): Convert from a global struct to
12008         a subclass of rtl_opt_pass along with...
12009         (pass_data_postreload_cse): ...new pass_data instance and...
12010         (make_pass_postreload_cse): ...new function.
12011         * predict.c (pass_profile): Convert from a global struct to a subclass
12012         of gimple_opt_pass along with...
12013         (pass_data_profile): ...new pass_data instance and...
12014         (make_pass_profile): ...new function.
12015         (pass_strip_predict_hints): Convert from a global struct to a subclass
12016         of gimple_opt_pass along with...
12017         (pass_data_strip_predict_hints): ...new pass_data instance and...
12018         (make_pass_strip_predict_hints): ...new function.
12019         * recog.c (pass_peephole2): Convert from a global struct to a subclass
12020         of rtl_opt_pass along with...
12021         (pass_data_peephole2): ...new pass_data instance and...
12022         (make_pass_peephole2): ...new function.
12023         (pass_split_all_insns): Convert from a global struct to a subclass of
12024         rtl_opt_pass along with...
12025         (pass_data_split_all_insns): ...new pass_data instance and...
12026         (make_pass_split_all_insns): ...new function.
12027         (pass_split_after_reload): Convert from a global struct to a subclass
12028         of rtl_opt_pass along with...
12029         (pass_data_split_after_reload): ...new pass_data instance and...
12030         (make_pass_split_after_reload): ...new function.
12031         (pass_split_before_regstack): Convert from a global struct to a
12032         subclass of rtl_opt_pass along with...
12033         (pass_data_split_before_regstack): ...new pass_data instance and...
12034         (make_pass_split_before_regstack): ...new function.
12035         (pass_split_before_sched2): Convert from a global struct to a subclass
12036         of rtl_opt_pass along with...
12037         (pass_data_split_before_sched2): ...new pass_data instance and...
12038         (make_pass_split_before_sched2): ...new function.
12039         (pass_split_for_shorten_branches): Convert from a global struct to a
12040         subclass of rtl_opt_pass along with...
12041         (pass_data_split_for_shorten_branches): ...new pass_data instance
12042         and...
12043         (make_pass_split_for_shorten_branches): ...new function.
12044         * ree.c (pass_ree): Convert from a global struct to a subclass of
12045         rtl_opt_pass along with...
12046         (pass_data_ree): ...new pass_data instance and...
12047         (make_pass_ree): ...new function.
12048         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
12049         subclass of rtl_opt_pass along with...
12050         (pass_data_stack_regs): ...new pass_data instance and...
12051         (make_pass_stack_regs): ...new function.
12052         (pass_stack_regs_run): Convert from a global struct to a subclass of
12053         rtl_opt_pass along with...
12054         (pass_data_stack_regs_run): ...new pass_data instance and...
12055         (make_pass_stack_regs_run): ...new function.
12056         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
12057         subclass of rtl_opt_pass along with...
12058         (pass_data_cprop_hardreg): ...new pass_data instance and...
12059         (make_pass_cprop_hardreg): ...new function.
12060         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
12061         subclass of rtl_opt_pass along with...
12062         (pass_data_reginfo_init): ...new pass_data instance and...
12063         (make_pass_reginfo_init): ...new function.
12064         * regmove.c (pass_regmove): Convert from a global struct to a subclass
12065         of rtl_opt_pass along with...
12066         (pass_data_regmove): ...new pass_data instance and...
12067         (make_pass_regmove): ...new function.
12068         * regrename.c (pass_regrename): Convert from a global struct to a
12069         subclass of rtl_opt_pass along with...
12070         (pass_data_regrename): ...new pass_data instance and...
12071         (make_pass_regrename): ...new function.
12072         * reorg.c (pass_delay_slots): Convert from a global struct to a
12073         subclass of rtl_opt_pass along with...
12074         (pass_data_delay_slots): ...new pass_data instance and...
12075         (make_pass_delay_slots): ...new function.
12076         (pass_machine_reorg): Convert from a global struct to a subclass of
12077         rtl_opt_pass along with...
12078         (pass_data_machine_reorg): ...new pass_data instance and...
12079         (make_pass_machine_reorg): ...new function.
12080         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
12081         of rtl_opt_pass along with...
12082         (pass_data_sched): ...new pass_data instance and...
12083         (make_pass_sched): ...new function.
12084         (pass_sched2): Convert from a global struct to a subclass of
12085         rtl_opt_pass along with...
12086         (pass_data_sched2): ...new pass_data instance and...
12087         (make_pass_sched2): ...new function.
12088         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
12089         to a subclass of rtl_opt_pass along with...
12090         (pass_data_stack_ptr_mod): ...new pass_data instance and...
12091         (make_pass_stack_ptr_mod): ...new function.
12092         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
12093         to a subclass of rtl_opt_pass along with...
12094         (pass_data_rtl_store_motion): ...new pass_data instance and...
12095         (make_pass_rtl_store_motion): ...new function.
12096         * tracer.c (pass_tracer): Convert from a global struct to a subclass
12097         of gimple_opt_pass along with...
12098         (pass_data_tracer): ...new pass_data instance and...
12099         (make_pass_tracer): ...new function.
12100         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
12101         to a subclass of gimple_opt_pass along with...
12102         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
12103         (make_pass_diagnose_tm_blocks): ...new function.
12104         (pass_lower_tm): Convert from a global struct to a subclass of
12105         gimple_opt_pass along with...
12106         (pass_data_lower_tm): ...new pass_data instance and...
12107         (make_pass_lower_tm): ...new function.
12108         (pass_tm_init): Convert from a global struct to a subclass of
12109         gimple_opt_pass along with...
12110         (pass_data_tm_init): ...new pass_data instance and...
12111         (make_pass_tm_init): ...new function.
12112         (pass_tm_mark): Convert from a global struct to a subclass of
12113         gimple_opt_pass along with...
12114         (pass_data_tm_mark): ...new pass_data instance and...
12115         (make_pass_tm_mark): ...new function.
12116         (pass_tm_edges): Convert from a global struct to a subclass of
12117         gimple_opt_pass along with...
12118         (pass_data_tm_edges): ...new pass_data instance and...
12119         (make_pass_tm_edges): ...new function.
12120         (pass_tm_memopt): Convert from a global struct to a subclass of
12121         gimple_opt_pass along with...
12122         (pass_data_tm_memopt): ...new pass_data instance and...
12123         (make_pass_tm_memopt): ...new function.
12124         (pass_ipa_tm): Convert from a global struct to a subclass of
12125         simple_ipa_opt_pass along with...
12126         (pass_data_ipa_tm): ...new pass_data instance and...
12127         (make_pass_ipa_tm): ...new function.
12128         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
12129         subclass of gimple_opt_pass along with...
12130         (pass_data_call_cdce): ...new pass_data instance and...
12131         (make_pass_call_cdce): ...new function.
12132         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
12133         subclass of gimple_opt_pass along with...
12134         (pass_data_build_cfg): ...new pass_data instance and...
12135         (make_pass_build_cfg): ...new function.
12136         (pass_split_crit_edges): Convert from a global struct to a subclass of
12137         gimple_opt_pass along with...
12138         (pass_data_split_crit_edges): ...new pass_data instance and...
12139         (make_pass_split_crit_edges): ...new function.
12140         (pass_warn_function_return): Convert from a global struct to a
12141         subclass of gimple_opt_pass along with...
12142         (pass_data_warn_function_return): ...new pass_data instance and...
12143         (make_pass_warn_function_return): ...new function.
12144         (pass_warn_function_noreturn): Convert from a global struct to a
12145         subclass of gimple_opt_pass along with...
12146         (pass_data_warn_function_noreturn): ...new pass_data instance and...
12147         (make_pass_warn_function_noreturn): ...new function.
12148         (pass_warn_unused_result): Convert from a global struct to a subclass
12149         of gimple_opt_pass along with...
12150         (pass_data_warn_unused_result): ...new pass_data instance and...
12151         (make_pass_warn_unused_result): ...new function.
12152         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
12153         a subclass of gimple_opt_pass along with...
12154         (pass_data_merge_phi): ...new pass_data instance and...
12155         (make_pass_merge_phi): ...new function.
12156         * tree-complex.c (pass_lower_complex): Convert from a global struct to
12157         a subclass of gimple_opt_pass along with...
12158         (pass_data_lower_complex): ...new pass_data instance and...
12159         (make_pass_lower_complex): ...new function.
12160         (pass_lower_complex_O0): Convert from a global struct to a subclass of
12161         gimple_opt_pass along with...
12162         (pass_data_lower_complex_O0): ...new pass_data instance and...
12163         (make_pass_lower_complex_O0): ...new function.
12164         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
12165         subclass of gimple_opt_pass along with...
12166         (pass_data_lower_eh): ...new pass_data instance and...
12167         (make_pass_lower_eh): ...new function.
12168         (pass_refactor_eh): Convert from a global struct to a subclass of
12169         gimple_opt_pass along with...
12170         (pass_data_refactor_eh): ...new pass_data instance and...
12171         (make_pass_refactor_eh): ...new function.
12172         (pass_lower_resx): Convert from a global struct to a subclass of
12173         gimple_opt_pass along with...
12174         (pass_data_lower_resx): ...new pass_data instance and...
12175         (make_pass_lower_resx): ...new function.
12176         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
12177         of gimple_opt_pass along with...
12178         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
12179         (make_pass_lower_eh_dispatch): ...new function.
12180         (pass_cleanup_eh): Convert from a global struct to a subclass of
12181         gimple_opt_pass along with...
12182         (pass_data_cleanup_eh): ...new pass_data instance and...
12183         (make_pass_cleanup_eh): ...new function.
12184         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
12185         to a subclass of simple_ipa_opt_pass along with...
12186         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
12187         (make_pass_ipa_lower_emutls): ...new function.
12188         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
12189         a subclass of gimple_opt_pass along with...
12190         (pass_data_if_conversion): ...new pass_data instance and...
12191         (make_pass_if_conversion): ...new function.
12192         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
12193         subclass of gimple_opt_pass along with...
12194         (pass_data_build_ssa): ...new pass_data instance and...
12195         (make_pass_build_ssa): ...new function.
12196         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
12197         global struct to a subclass of gimple_opt_pass along with...
12198         (pass_data_loop_distribution): ...new pass_data instance and...
12199         (make_pass_loop_distribution): ...new function.
12200         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
12201         subclass of gimple_opt_pass along with...
12202         (pass_data_mudflap_1): ...new pass_data instance and...
12203         (make_pass_mudflap_1): ...new function.
12204         (pass_mudflap_2): Convert from a global struct to a subclass of
12205         gimple_opt_pass along with...
12206         (pass_data_mudflap_2): ...new pass_data instance and...
12207         (make_pass_mudflap_2): ...new function.
12208         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
12209         subclass of gimple_opt_pass along with...
12210         (pass_data_mudflap_1): ...new pass_data instance and...
12211         (make_pass_mudflap_1): ...new function.
12212         (pass_mudflap_2): Convert from a global struct to a subclass of
12213         gimple_opt_pass along with...
12214         (pass_data_mudflap_2): ...new pass_data instance and...
12215         (make_pass_mudflap_2): ...new function.
12216         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
12217         gimple_opt_pass along with...
12218         (pass_data_nrv): ...new pass_data instance and...
12219         (make_pass_nrv): ...new function.
12220         (pass_return_slot): Convert from a global struct to a subclass of
12221         gimple_opt_pass along with...
12222         (pass_data_return_slot): ...new pass_data instance and...
12223         (make_pass_return_slot): ...new function.
12224         * tree-object-size.c (pass_object_sizes): Convert from a global struct
12225         to a subclass of gimple_opt_pass along with...
12226         (pass_data_object_sizes): ...new pass_data instance and...
12227         (make_pass_object_sizes): ...new function.
12228         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
12229         global struct to a subclass of gimple_opt_pass along with...
12230         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
12231         and...
12232         (make_pass_cleanup_cfg_post_optimizing): ...new function.
12233         (pass_fixup_cfg): Convert from a global struct to a subclass of
12234         gimple_opt_pass along with...
12235         (pass_data_fixup_cfg): ...new pass_data instance and...
12236         (make_pass_fixup_cfg): ...new function.
12237         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
12238         (make_pass_mudflap_1): ...new function.
12239         (pass_mudflap_2): Replace declaration with that of...
12240         (make_pass_mudflap_2): ...new function.
12241         (pass_asan): Replace declaration with that of...
12242         (make_pass_asan): ...new function.
12243         (pass_asan_O0): Replace declaration with that of...
12244         (make_pass_asan_O0): ...new function.
12245         (pass_tsan): Replace declaration with that of...
12246         (make_pass_tsan): ...new function.
12247         (pass_tsan_O0): Replace declaration with that of...
12248         (make_pass_tsan_O0): ...new function.
12249         (pass_lower_cf): Replace declaration with that of...
12250         (make_pass_lower_cf): ...new function.
12251         (pass_refactor_eh): Replace declaration with that of...
12252         (make_pass_refactor_eh): ...new function.
12253         (pass_lower_eh): Replace declaration with that of...
12254         (make_pass_lower_eh): ...new function.
12255         (pass_lower_eh_dispatch): Replace declaration with that of...
12256         (make_pass_lower_eh_dispatch): ...new function.
12257         (pass_lower_resx): Replace declaration with that of...
12258         (make_pass_lower_resx): ...new function.
12259         (pass_build_cfg): Replace declaration with that of...
12260         (make_pass_build_cfg): ...new function.
12261         (pass_early_tree_profile): Replace declaration with that of...
12262         (make_pass_early_tree_profile): ...new function.
12263         (pass_cleanup_eh): Replace declaration with that of...
12264         (make_pass_cleanup_eh): ...new function.
12265         (pass_sra): Replace declaration with that of...
12266         (make_pass_sra): ...new function.
12267         (pass_sra_early): Replace declaration with that of...
12268         (make_pass_sra_early): ...new function.
12269         (pass_early_ipa_sra): Replace declaration with that of...
12270         (make_pass_early_ipa_sra): ...new function.
12271         (pass_tail_recursion): Replace declaration with that of...
12272         (make_pass_tail_recursion): ...new function.
12273         (pass_tail_calls): Replace declaration with that of...
12274         (make_pass_tail_calls): ...new function.
12275         (pass_tree_loop): Replace declaration with that of...
12276         (make_pass_tree_loop): ...new function.
12277         (pass_tree_loop_init): Replace declaration with that of...
12278         (make_pass_tree_loop_init): ...new function.
12279         (pass_lim): Replace declaration with that of...
12280         (make_pass_lim): ...new function.
12281         (pass_tree_unswitch): Replace declaration with that of...
12282         (make_pass_tree_unswitch): ...new function.
12283         (pass_predcom): Replace declaration with that of...
12284         (make_pass_predcom): ...new function.
12285         (pass_iv_canon): Replace declaration with that of...
12286         (make_pass_iv_canon): ...new function.
12287         (pass_scev_cprop): Replace declaration with that of...
12288         (make_pass_scev_cprop): ...new function.
12289         (pass_empty_loop): Replace declaration with that of...
12290         (make_pass_empty_loop): ...new function.
12291         (pass_record_bounds): Replace declaration with that of...
12292         (make_pass_record_bounds): ...new function.
12293         (pass_graphite): Replace declaration with that of...
12294         (make_pass_graphite): ...new function.
12295         (pass_graphite_transforms): Replace declaration with that of...
12296         (make_pass_graphite_transforms): ...new function.
12297         (pass_if_conversion): Replace declaration with that of...
12298         (make_pass_if_conversion): ...new function.
12299         (pass_loop_distribution): Replace declaration with that of...
12300         (make_pass_loop_distribution): ...new function.
12301         (pass_vectorize): Replace declaration with that of...
12302         (make_pass_vectorize): ...new function.
12303         (pass_slp_vectorize): Replace declaration with that of...
12304         (make_pass_slp_vectorize): ...new function.
12305         (pass_complete_unroll): Replace declaration with that of...
12306         (make_pass_complete_unroll): ...new function.
12307         (pass_complete_unrolli): Replace declaration with that of...
12308         (make_pass_complete_unrolli): ...new function.
12309         (pass_parallelize_loops): Replace declaration with that of...
12310         (make_pass_parallelize_loops): ...new function.
12311         (pass_loop_prefetch): Replace declaration with that of...
12312         (make_pass_loop_prefetch): ...new function.
12313         (pass_iv_optimize): Replace declaration with that of...
12314         (make_pass_iv_optimize): ...new function.
12315         (pass_tree_loop_done): Replace declaration with that of...
12316         (make_pass_tree_loop_done): ...new function.
12317         (pass_ch): Replace declaration with that of...
12318         (make_pass_ch): ...new function.
12319         (pass_ccp): Replace declaration with that of...
12320         (make_pass_ccp): ...new function.
12321         (pass_phi_only_cprop): Replace declaration with that of...
12322         (make_pass_phi_only_cprop): ...new function.
12323         (pass_build_ssa): Replace declaration with that of...
12324         (make_pass_build_ssa): ...new function.
12325         (pass_build_alias): Replace declaration with that of...
12326         (make_pass_build_alias): ...new function.
12327         (pass_build_ealias): Replace declaration with that of...
12328         (make_pass_build_ealias): ...new function.
12329         (pass_dominator): Replace declaration with that of...
12330         (make_pass_dominator): ...new function.
12331         (pass_dce): Replace declaration with that of...
12332         (make_pass_dce): ...new function.
12333         (pass_dce_loop): Replace declaration with that of...
12334         (make_pass_dce_loop): ...new function.
12335         (pass_cd_dce): Replace declaration with that of...
12336         (make_pass_cd_dce): ...new function.
12337         (pass_call_cdce): Replace declaration with that of...
12338         (make_pass_call_cdce): ...new function.
12339         (pass_merge_phi): Replace declaration with that of...
12340         (make_pass_merge_phi): ...new function.
12341         (pass_split_crit_edges): Replace declaration with that of...
12342         (make_pass_split_crit_edges): ...new function.
12343         (pass_pre): Replace declaration with that of...
12344         (make_pass_pre): ...new function.
12345         (pass_profile): Replace declaration with that of...
12346         (make_pass_profile): ...new function.
12347         (pass_strip_predict_hints): Replace declaration with that of...
12348         (make_pass_strip_predict_hints): ...new function.
12349         (pass_lower_complex_O0): Replace declaration with that of...
12350         (make_pass_lower_complex_O0): ...new function.
12351         (pass_lower_complex): Replace declaration with that of...
12352         (make_pass_lower_complex): ...new function.
12353         (pass_lower_vector): Replace declaration with that of...
12354         (make_pass_lower_vector): ...new function.
12355         (pass_lower_vector_ssa): Replace declaration with that of...
12356         (make_pass_lower_vector_ssa): ...new function.
12357         (pass_lower_omp): Replace declaration with that of...
12358         (make_pass_lower_omp): ...new function.
12359         (pass_diagnose_omp_blocks): Replace declaration with that of...
12360         (make_pass_diagnose_omp_blocks): ...new function.
12361         (pass_expand_omp): Replace declaration with that of...
12362         (make_pass_expand_omp): ...new function.
12363         (pass_expand_omp_ssa): Replace declaration with that of...
12364         (make_pass_expand_omp_ssa): ...new function.
12365         (pass_object_sizes): Replace declaration with that of...
12366         (make_pass_object_sizes): ...new function.
12367         (pass_strlen): Replace declaration with that of...
12368         (make_pass_strlen): ...new function.
12369         (pass_fold_builtins): Replace declaration with that of...
12370         (make_pass_fold_builtins): ...new function.
12371         (pass_stdarg): Replace declaration with that of...
12372         (make_pass_stdarg): ...new function.
12373         (pass_early_warn_uninitialized): Replace declaration with that of...
12374         (make_pass_early_warn_uninitialized): ...new function.
12375         (pass_late_warn_uninitialized): Replace declaration with that of...
12376         (make_pass_late_warn_uninitialized): ...new function.
12377         (pass_cse_reciprocals): Replace declaration with that of...
12378         (make_pass_cse_reciprocals): ...new function.
12379         (pass_cse_sincos): Replace declaration with that of...
12380         (make_pass_cse_sincos): ...new function.
12381         (pass_optimize_bswap): Replace declaration with that of...
12382         (make_pass_optimize_bswap): ...new function.
12383         (pass_optimize_widening_mul): Replace declaration with that of...
12384         (make_pass_optimize_widening_mul): ...new function.
12385         (pass_warn_function_return): Replace declaration with that of...
12386         (make_pass_warn_function_return): ...new function.
12387         (pass_warn_function_noreturn): Replace declaration with that of...
12388         (make_pass_warn_function_noreturn): ...new function.
12389         (pass_cselim): Replace declaration with that of...
12390         (make_pass_cselim): ...new function.
12391         (pass_phiopt): Replace declaration with that of...
12392         (make_pass_phiopt): ...new function.
12393         (pass_forwprop): Replace declaration with that of...
12394         (make_pass_forwprop): ...new function.
12395         (pass_phiprop): Replace declaration with that of...
12396         (make_pass_phiprop): ...new function.
12397         (pass_tree_ifcombine): Replace declaration with that of...
12398         (make_pass_tree_ifcombine): ...new function.
12399         (pass_dse): Replace declaration with that of...
12400         (make_pass_dse): ...new function.
12401         (pass_nrv): Replace declaration with that of...
12402         (make_pass_nrv): ...new function.
12403         (pass_rename_ssa_copies): Replace declaration with that of...
12404         (make_pass_rename_ssa_copies): ...new function.
12405         (pass_sink_code): Replace declaration with that of...
12406         (make_pass_sink_code): ...new function.
12407         (pass_fre): Replace declaration with that of...
12408         (make_pass_fre): ...new function.
12409         (pass_check_data_deps): Replace declaration with that of...
12410         (make_pass_check_data_deps): ...new function.
12411         (pass_copy_prop): Replace declaration with that of...
12412         (make_pass_copy_prop): ...new function.
12413         (pass_vrp): Replace declaration with that of...
12414         (make_pass_vrp): ...new function.
12415         (pass_uncprop): Replace declaration with that of...
12416         (make_pass_uncprop): ...new function.
12417         (pass_return_slot): Replace declaration with that of...
12418         (make_pass_return_slot): ...new function.
12419         (pass_reassoc): Replace declaration with that of...
12420         (make_pass_reassoc): ...new function.
12421         (pass_rebuild_cgraph_edges): Replace declaration with that of...
12422         (make_pass_rebuild_cgraph_edges): ...new function.
12423         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
12424         (make_pass_remove_cgraph_callee_edges): ...new function.
12425         (pass_build_cgraph_edges): Replace declaration with that of...
12426         (make_pass_build_cgraph_edges): ...new function.
12427         (pass_local_pure_const): Replace declaration with that of...
12428         (make_pass_local_pure_const): ...new function.
12429         (pass_tracer): Replace declaration with that of...
12430         (make_pass_tracer): ...new function.
12431         (pass_warn_unused_result): Replace declaration with that of...
12432         (make_pass_warn_unused_result): ...new function.
12433         (pass_diagnose_tm_blocks): Replace declaration with that of...
12434         (make_pass_diagnose_tm_blocks): ...new function.
12435         (pass_lower_tm): Replace declaration with that of...
12436         (make_pass_lower_tm): ...new function.
12437         (pass_tm_init): Replace declaration with that of...
12438         (make_pass_tm_init): ...new function.
12439         (pass_tm_mark): Replace declaration with that of...
12440         (make_pass_tm_mark): ...new function.
12441         (pass_tm_memopt): Replace declaration with that of...
12442         (make_pass_tm_memopt): ...new function.
12443         (pass_tm_edges): Replace declaration with that of...
12444         (make_pass_tm_edges): ...new function.
12445         (pass_split_functions): Replace declaration with that of...
12446         (make_pass_split_functions): ...new function.
12447         (pass_feedback_split_functions): Replace declaration with that of...
12448         (make_pass_feedback_split_functions): ...new function.
12449         (pass_strength_reduction): Replace declaration with that of...
12450         (make_pass_strength_reduction): ...new function.
12451         (pass_ipa_lower_emutls): Replace declaration with that of...
12452         (make_pass_ipa_lower_emutls): ...new function.
12453         (pass_ipa_function_and_variable_visibility): Replace declaration with
12454         that of...
12455         (make_pass_ipa_function_and_variable_visibility): ...new function.
12456         (pass_ipa_tree_profile): Replace declaration with that of...
12457         (make_pass_ipa_tree_profile): ...new function.
12458         (pass_early_local_passes): Replace declaration with that of...
12459         (make_pass_early_local_passes): ...new function.
12460         (pass_ipa_whole_program_visibility): Replace declaration with that
12461         of...
12462         (make_pass_ipa_whole_program_visibility): ...new function.
12463         (pass_ipa_lto_gimple_out): Replace declaration with that of...
12464         (make_pass_ipa_lto_gimple_out): ...new function.
12465         (pass_ipa_increase_alignment): Replace declaration with that of...
12466         (make_pass_ipa_increase_alignment): ...new function.
12467         (pass_ipa_inline): Replace declaration with that of...
12468         (make_pass_ipa_inline): ...new function.
12469         (pass_ipa_free_lang_data): Replace declaration with that of...
12470         (make_pass_ipa_free_lang_data): ...new function.
12471         (pass_ipa_free_inline_summary): Replace declaration with that of...
12472         (make_pass_ipa_free_inline_summary): ...new function.
12473         (pass_ipa_cp): Replace declaration with that of...
12474         (make_pass_ipa_cp): ...new function.
12475         (pass_ipa_reference): Replace declaration with that of...
12476         (make_pass_ipa_reference): ...new function.
12477         (pass_ipa_pure_const): Replace declaration with that of...
12478         (make_pass_ipa_pure_const): ...new function.
12479         (pass_ipa_pta): Replace declaration with that of...
12480         (make_pass_ipa_pta): ...new function.
12481         (pass_ipa_lto_finish_out): Replace declaration with that of...
12482         (make_pass_ipa_lto_finish_out): ...new function.
12483         (pass_ipa_tm): Replace declaration with that of...
12484         (make_pass_ipa_tm): ...new function.
12485         (pass_ipa_profile): Replace declaration with that of...
12486         (make_pass_ipa_profile): ...new function.
12487         (pass_ipa_cdtor_merge): Replace declaration with that of...
12488         (make_pass_ipa_cdtor_merge): ...new function.
12489         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
12490         of...
12491         (make_pass_cleanup_cfg_post_optimizing): ...new function.
12492         (pass_init_datastructures): Replace declaration with that of...
12493         (make_pass_init_datastructures): ...new function.
12494         (pass_fixup_cfg): Replace declaration with that of...
12495         (make_pass_fixup_cfg): ...new function.
12496         (pass_expand): Replace declaration with that of...
12497         (make_pass_expand): ...new function.
12498         (pass_instantiate_virtual_regs): Replace declaration with that of...
12499         (make_pass_instantiate_virtual_regs): ...new function.
12500         (pass_rtl_fwprop): Replace declaration with that of...
12501         (make_pass_rtl_fwprop): ...new function.
12502         (pass_rtl_fwprop_addr): Replace declaration with that of...
12503         (make_pass_rtl_fwprop_addr): ...new function.
12504         (pass_jump): Replace declaration with that of...
12505         (make_pass_jump): ...new function.
12506         (pass_jump2): Replace declaration with that of...
12507         (make_pass_jump2): ...new function.
12508         (pass_lower_subreg): Replace declaration with that of...
12509         (make_pass_lower_subreg): ...new function.
12510         (pass_cse): Replace declaration with that of...
12511         (make_pass_cse): ...new function.
12512         (pass_fast_rtl_dce): Replace declaration with that of...
12513         (make_pass_fast_rtl_dce): ...new function.
12514         (pass_ud_rtl_dce): Replace declaration with that of...
12515         (make_pass_ud_rtl_dce): ...new function.
12516         (pass_rtl_dce): Replace declaration with that of...
12517         (make_pass_rtl_dce): ...new function.
12518         (pass_rtl_dse1): Replace declaration with that of...
12519         (make_pass_rtl_dse1): ...new function.
12520         (pass_rtl_dse2): Replace declaration with that of...
12521         (make_pass_rtl_dse2): ...new function.
12522         (pass_rtl_dse3): Replace declaration with that of...
12523         (make_pass_rtl_dse3): ...new function.
12524         (pass_rtl_cprop): Replace declaration with that of...
12525         (make_pass_rtl_cprop): ...new function.
12526         (pass_rtl_pre): Replace declaration with that of...
12527         (make_pass_rtl_pre): ...new function.
12528         (pass_rtl_hoist): Replace declaration with that of...
12529         (make_pass_rtl_hoist): ...new function.
12530         (pass_rtl_store_motion): Replace declaration with that of...
12531         (make_pass_rtl_store_motion): ...new function.
12532         (pass_cse_after_global_opts): Replace declaration with that of...
12533         (make_pass_cse_after_global_opts): ...new function.
12534         (pass_rtl_ifcvt): Replace declaration with that of...
12535         (make_pass_rtl_ifcvt): ...new function.
12536         (pass_into_cfg_layout_mode): Replace declaration with that of...
12537         (make_pass_into_cfg_layout_mode): ...new function.
12538         (pass_outof_cfg_layout_mode): Replace declaration with that of...
12539         (make_pass_outof_cfg_layout_mode): ...new function.
12540         (pass_loop2): Replace declaration with that of...
12541         (make_pass_loop2): ...new function.
12542         (pass_rtl_loop_init): Replace declaration with that of...
12543         (make_pass_rtl_loop_init): ...new function.
12544         (pass_rtl_move_loop_invariants): Replace declaration with that of...
12545         (make_pass_rtl_move_loop_invariants): ...new function.
12546         (pass_rtl_unswitch): Replace declaration with that of...
12547         (make_pass_rtl_unswitch): ...new function.
12548         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
12549         (make_pass_rtl_unroll_and_peel_loops): ...new function.
12550         (pass_rtl_doloop): Replace declaration with that of...
12551         (make_pass_rtl_doloop): ...new function.
12552         (pass_rtl_loop_done): Replace declaration with that of...
12553         (make_pass_rtl_loop_done): ...new function.
12554         (pass_web): Replace declaration with that of...
12555         (make_pass_web): ...new function.
12556         (pass_cse2): Replace declaration with that of...
12557         (make_pass_cse2): ...new function.
12558         (pass_df_initialize_opt): Replace declaration with that of...
12559         (make_pass_df_initialize_opt): ...new function.
12560         (pass_df_initialize_no_opt): Replace declaration with that of...
12561         (make_pass_df_initialize_no_opt): ...new function.
12562         (pass_reginfo_init): Replace declaration with that of...
12563         (make_pass_reginfo_init): ...new function.
12564         (pass_inc_dec): Replace declaration with that of...
12565         (make_pass_inc_dec): ...new function.
12566         (pass_stack_ptr_mod): Replace declaration with that of...
12567         (make_pass_stack_ptr_mod): ...new function.
12568         (pass_initialize_regs): Replace declaration with that of...
12569         (make_pass_initialize_regs): ...new function.
12570         (pass_combine): Replace declaration with that of...
12571         (make_pass_combine): ...new function.
12572         (pass_if_after_combine): Replace declaration with that of...
12573         (make_pass_if_after_combine): ...new function.
12574         (pass_ree): Replace declaration with that of...
12575         (make_pass_ree): ...new function.
12576         (pass_partition_blocks): Replace declaration with that of...
12577         (make_pass_partition_blocks): ...new function.
12578         (pass_match_asm_constraints): Replace declaration with that of...
12579         (make_pass_match_asm_constraints): ...new function.
12580         (pass_regmove): Replace declaration with that of...
12581         (make_pass_regmove): ...new function.
12582         (pass_split_all_insns): Replace declaration with that of...
12583         (make_pass_split_all_insns): ...new function.
12584         (pass_fast_rtl_byte_dce): Replace declaration with that of...
12585         (make_pass_fast_rtl_byte_dce): ...new function.
12586         (pass_lower_subreg2): Replace declaration with that of...
12587         (make_pass_lower_subreg2): ...new function.
12588         (pass_mode_switching): Replace declaration with that of...
12589         (make_pass_mode_switching): ...new function.
12590         (pass_sms): Replace declaration with that of...
12591         (make_pass_sms): ...new function.
12592         (pass_sched): Replace declaration with that of...
12593         (make_pass_sched): ...new function.
12594         (pass_ira): Replace declaration with that of...
12595         (make_pass_ira): ...new function.
12596         (pass_reload): Replace declaration with that of...
12597         (make_pass_reload): ...new function.
12598         (pass_clean_state): Replace declaration with that of...
12599         (make_pass_clean_state): ...new function.
12600         (pass_branch_prob): Replace declaration with that of...
12601         (make_pass_branch_prob): ...new function.
12602         (pass_value_profile_transformations): Replace declaration with that
12603         of...
12604         (make_pass_value_profile_transformations): ...new function.
12605         (pass_postreload_cse): Replace declaration with that of...
12606         (make_pass_postreload_cse): ...new function.
12607         (pass_gcse2): Replace declaration with that of...
12608         (make_pass_gcse2): ...new function.
12609         (pass_split_after_reload): Replace declaration with that of...
12610         (make_pass_split_after_reload): ...new function.
12611         (pass_branch_target_load_optimize1): Replace declaration with that
12612         of...
12613         (make_pass_branch_target_load_optimize1): ...new function.
12614         (pass_thread_prologue_and_epilogue): Replace declaration with that
12615         of...
12616         (make_pass_thread_prologue_and_epilogue): ...new function.
12617         (pass_stack_adjustments): Replace declaration with that of...
12618         (make_pass_stack_adjustments): ...new function.
12619         (pass_peephole2): Replace declaration with that of...
12620         (make_pass_peephole2): ...new function.
12621         (pass_if_after_reload): Replace declaration with that of...
12622         (make_pass_if_after_reload): ...new function.
12623         (pass_regrename): Replace declaration with that of...
12624         (make_pass_regrename): ...new function.
12625         (pass_cprop_hardreg): Replace declaration with that of...
12626         (make_pass_cprop_hardreg): ...new function.
12627         (pass_reorder_blocks): Replace declaration with that of...
12628         (make_pass_reorder_blocks): ...new function.
12629         (pass_branch_target_load_optimize2): Replace declaration with that
12630         of...
12631         (make_pass_branch_target_load_optimize2): ...new function.
12632         (pass_leaf_regs): Replace declaration with that of...
12633         (make_pass_leaf_regs): ...new function.
12634         (pass_split_before_sched2): Replace declaration with that of...
12635         (make_pass_split_before_sched2): ...new function.
12636         (pass_compare_elim_after_reload): Replace declaration with that of...
12637         (make_pass_compare_elim_after_reload): ...new function.
12638         (pass_sched2): Replace declaration with that of...
12639         (make_pass_sched2): ...new function.
12640         (pass_stack_regs): Replace declaration with that of...
12641         (make_pass_stack_regs): ...new function.
12642         (pass_stack_regs_run): Replace declaration with that of...
12643         (make_pass_stack_regs_run): ...new function.
12644         (pass_df_finish): Replace declaration with that of...
12645         (make_pass_df_finish): ...new function.
12646         (pass_compute_alignments): Replace declaration with that of...
12647         (make_pass_compute_alignments): ...new function.
12648         (pass_duplicate_computed_gotos): Replace declaration with that of...
12649         (make_pass_duplicate_computed_gotos): ...new function.
12650         (pass_variable_tracking): Replace declaration with that of...
12651         (make_pass_variable_tracking): ...new function.
12652         (pass_free_cfg): Replace declaration with that of...
12653         (make_pass_free_cfg): ...new function.
12654         (pass_machine_reorg): Replace declaration with that of...
12655         (make_pass_machine_reorg): ...new function.
12656         (pass_cleanup_barriers): Replace declaration with that of...
12657         (make_pass_cleanup_barriers): ...new function.
12658         (pass_delay_slots): Replace declaration with that of...
12659         (make_pass_delay_slots): ...new function.
12660         (pass_split_for_shorten_branches): Replace declaration with that of...
12661         (make_pass_split_for_shorten_branches): ...new function.
12662         (pass_split_before_regstack): Replace declaration with that of...
12663         (make_pass_split_before_regstack): ...new function.
12664         (pass_convert_to_eh_region_ranges): Replace declaration with that
12665         of...
12666         (make_pass_convert_to_eh_region_ranges): ...new function.
12667         (pass_shorten_branches): Replace declaration with that of...
12668         (make_pass_shorten_branches): ...new function.
12669         (pass_set_nothrow_function_flags): Replace declaration with that of...
12670         (make_pass_set_nothrow_function_flags): ...new function.
12671         (pass_dwarf2_frame): Replace declaration with that of...
12672         (make_pass_dwarf2_frame): ...new function.
12673         (pass_final): Replace declaration with that of...
12674         (make_pass_final): ...new function.
12675         (pass_rtl_seqabstr): Replace declaration with that of...
12676         (make_pass_rtl_seqabstr): ...new function.
12677         (pass_release_ssa_names): Replace declaration with that of...
12678         (make_pass_release_ssa_names): ...new function.
12679         (pass_early_inline): Replace declaration with that of...
12680         (make_pass_early_inline): ...new function.
12681         (pass_inline_parameters): Replace declaration with that of...
12682         (make_pass_inline_parameters): ...new function.
12683         (pass_update_address_taken): Replace declaration with that of...
12684         (make_pass_update_address_taken): ...new function.
12685         (pass_convert_switch): Replace declaration with that of...
12686         (make_pass_convert_switch): ...new function.
12687         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
12688         to a subclass of simple_ipa_opt_pass along with...
12689         (pass_data_ipa_tree_profile): ...new pass_data instance and...
12690         (make_pass_ipa_tree_profile): ...new function.
12691         * tree-sra.c (pass_sra_early): Convert from a global struct to a
12692         subclass of gimple_opt_pass along with...
12693         (pass_data_sra_early): ...new pass_data instance and...
12694         (make_pass_sra_early): ...new function.
12695         (pass_sra): Convert from a global struct to a subclass of
12696         gimple_opt_pass along with...
12697         (pass_data_sra): ...new pass_data instance and...
12698         (make_pass_sra): ...new function.
12699         (pass_early_ipa_sra): Convert from a global struct to a subclass of
12700         gimple_opt_pass along with...
12701         (pass_data_early_ipa_sra): ...new pass_data instance and...
12702         (make_pass_early_ipa_sra): ...new function.
12703         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
12704         subclass of gimple_opt_pass along with...
12705         (pass_data_ccp): ...new pass_data instance and...
12706         (make_pass_ccp): ...new function.
12707         (pass_fold_builtins): Convert from a global struct to a subclass of
12708         gimple_opt_pass along with...
12709         (pass_data_fold_builtins): ...new pass_data instance and...
12710         (make_pass_fold_builtins): ...new function.
12711         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
12712         subclass of gimple_opt_pass along with...
12713         (pass_data_copy_prop): ...new pass_data instance and...
12714         (make_pass_copy_prop): ...new function.
12715         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
12716         global struct to a subclass of gimple_opt_pass along with...
12717         (pass_data_rename_ssa_copies): ...new pass_data instance and...
12718         (make_pass_rename_ssa_copies): ...new function.
12719         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
12720         subclass of gimple_opt_pass along with...
12721         (pass_data_dce): ...new pass_data instance and...
12722         (make_pass_dce): ...new function.
12723         (pass_dce_loop): Convert from a global struct to a subclass of
12724         gimple_opt_pass along with...
12725         (pass_data_dce_loop): ...new pass_data instance and...
12726         (make_pass_dce_loop): ...new function.
12727         (pass_cd_dce): Convert from a global struct to a subclass of
12728         gimple_opt_pass along with...
12729         (pass_data_cd_dce): ...new pass_data instance and...
12730         (make_pass_cd_dce): ...new function.
12731         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
12732         subclass of gimple_opt_pass along with...
12733         (pass_data_dominator): ...new pass_data instance and...
12734         (make_pass_dominator): ...new function.
12735         (pass_phi_only_cprop): Convert from a global struct to a subclass of
12736         gimple_opt_pass along with...
12737         (pass_data_phi_only_cprop): ...new pass_data instance and...
12738         (make_pass_phi_only_cprop): ...new function.
12739         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
12740         subclass of gimple_opt_pass along with...
12741         (pass_data_dse): ...new pass_data instance and...
12742         (make_pass_dse): ...new function.
12743         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
12744         a subclass of gimple_opt_pass along with...
12745         (pass_data_forwprop): ...new pass_data instance and...
12746         (make_pass_forwprop): ...new function.
12747         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
12748         struct to a subclass of gimple_opt_pass along with...
12749         (pass_data_tree_ifcombine): ...new pass_data instance and...
12750         (make_pass_tree_ifcombine): ...new function.
12751         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
12752         subclass of gimple_opt_pass along with...
12753         (pass_data_ch): ...new pass_data instance and...
12754         (make_pass_ch): ...new function.
12755         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
12756         subclass of gimple_opt_pass along with...
12757         (pass_data_tree_loop): ...new pass_data instance and...
12758         (make_pass_tree_loop): ...new function.
12759         (pass_tree_loop_init): Convert from a global struct to a subclass of
12760         gimple_opt_pass along with...
12761         (pass_data_tree_loop_init): ...new pass_data instance and...
12762         (make_pass_tree_loop_init): ...new function.
12763         (pass_lim): Convert from a global struct to a subclass of
12764         gimple_opt_pass along with...
12765         (pass_data_lim): ...new pass_data instance and...
12766         (make_pass_lim): ...new function.
12767         (pass_tree_unswitch): Convert from a global struct to a subclass of
12768         gimple_opt_pass along with...
12769         (pass_data_tree_unswitch): ...new pass_data instance and...
12770         (make_pass_tree_unswitch): ...new function.
12771         (pass_predcom): Convert from a global struct to a subclass of
12772         gimple_opt_pass along with...
12773         (pass_data_predcom): ...new pass_data instance and...
12774         (make_pass_predcom): ...new function.
12775         (pass_vectorize): Convert from a global struct to a subclass of
12776         gimple_opt_pass along with...
12777         (pass_data_vectorize): ...new pass_data instance and...
12778         (make_pass_vectorize): ...new function.
12779         (pass_graphite): Convert from a global struct to a subclass of
12780         gimple_opt_pass along with...
12781         (pass_data_graphite): ...new pass_data instance and...
12782         (make_pass_graphite): ...new function.
12783         (pass_graphite_transforms): Convert from a global struct to a subclass
12784         of gimple_opt_pass along with...
12785         (pass_data_graphite_transforms): ...new pass_data instance and...
12786         (make_pass_graphite_transforms): ...new function.
12787         (pass_check_data_deps): Convert from a global struct to a subclass of
12788         gimple_opt_pass along with...
12789         (pass_data_check_data_deps): ...new pass_data instance and...
12790         (make_pass_check_data_deps): ...new function.
12791         (pass_iv_canon): Convert from a global struct to a subclass of
12792         gimple_opt_pass along with...
12793         (pass_data_iv_canon): ...new pass_data instance and...
12794         (make_pass_iv_canon): ...new function.
12795         (pass_scev_cprop): Convert from a global struct to a subclass of
12796         gimple_opt_pass along with...
12797         (pass_data_scev_cprop): ...new pass_data instance and...
12798         (make_pass_scev_cprop): ...new function.
12799         (pass_record_bounds): Convert from a global struct to a subclass of
12800         gimple_opt_pass along with...
12801         (pass_data_record_bounds): ...new pass_data instance and...
12802         (make_pass_record_bounds): ...new function.
12803         (pass_complete_unroll): Convert from a global struct to a subclass of
12804         gimple_opt_pass along with...
12805         (pass_data_complete_unroll): ...new pass_data instance and...
12806         (make_pass_complete_unroll): ...new function.
12807         (pass_complete_unrolli): Convert from a global struct to a subclass of
12808         gimple_opt_pass along with...
12809         (pass_data_complete_unrolli): ...new pass_data instance and...
12810         (make_pass_complete_unrolli): ...new function.
12811         (pass_parallelize_loops): Convert from a global struct to a subclass
12812         of gimple_opt_pass along with...
12813         (pass_data_parallelize_loops): ...new pass_data instance and...
12814         (make_pass_parallelize_loops): ...new function.
12815         (pass_loop_prefetch): Convert from a global struct to a subclass of
12816         gimple_opt_pass along with...
12817         (pass_data_loop_prefetch): ...new pass_data instance and...
12818         (make_pass_loop_prefetch): ...new function.
12819         (pass_iv_optimize): Convert from a global struct to a subclass of
12820         gimple_opt_pass along with...
12821         (pass_data_iv_optimize): ...new pass_data instance and...
12822         (make_pass_iv_optimize): ...new function.
12823         (pass_tree_loop_done): Convert from a global struct to a subclass of
12824         gimple_opt_pass along with...
12825         (pass_data_tree_loop_done): ...new pass_data instance and...
12826         (make_pass_tree_loop_done): ...new function.
12827         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
12828         struct to a subclass of gimple_opt_pass along with...
12829         (pass_data_cse_reciprocals): ...new pass_data instance and...
12830         (make_pass_cse_reciprocals): ...new function.
12831         (pass_cse_sincos): Convert from a global struct to a subclass of
12832         gimple_opt_pass along with...
12833         (pass_data_cse_sincos): ...new pass_data instance and...
12834         (make_pass_cse_sincos): ...new function.
12835         (pass_optimize_bswap): Convert from a global struct to a subclass of
12836         gimple_opt_pass along with...
12837         (pass_data_optimize_bswap): ...new pass_data instance and...
12838         (make_pass_optimize_bswap): ...new function.
12839         (pass_optimize_widening_mul): Convert from a global struct to a
12840         subclass of gimple_opt_pass along with...
12841         (pass_data_optimize_widening_mul): ...new pass_data instance and...
12842         (make_pass_optimize_widening_mul): ...new function.
12843         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
12844         subclass of gimple_opt_pass along with...
12845         (pass_data_phiopt): ...new pass_data instance and...
12846         (make_pass_phiopt): ...new function.
12847         (pass_cselim): Convert from a global struct to a subclass of
12848         gimple_opt_pass along with...
12849         (pass_data_cselim): ...new pass_data instance and...
12850         (make_pass_cselim): ...new function.
12851         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
12852         subclass of gimple_opt_pass along with...
12853         (pass_data_phiprop): ...new pass_data instance and...
12854         (make_pass_phiprop): ...new function.
12855         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
12856         subclass of gimple_opt_pass along with...
12857         (pass_data_pre): ...new pass_data instance and...
12858         (make_pass_pre): ...new function.
12859         (pass_fre): Convert from a global struct to a subclass of
12860         gimple_opt_pass along with...
12861         (pass_data_fre): ...new pass_data instance and...
12862         (make_pass_fre): ...new function.
12863         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
12864         subclass of gimple_opt_pass along with...
12865         (pass_data_reassoc): ...new pass_data instance and...
12866         (make_pass_reassoc): ...new function.
12867         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
12868         subclass of gimple_opt_pass along with...
12869         (pass_data_sink_code): ...new pass_data instance and...
12870         (make_pass_sink_code): ...new function.
12871         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
12872         subclass of gimple_opt_pass along with...
12873         (pass_data_strlen): ...new pass_data instance and...
12874         (make_pass_strlen): ...new function.
12875         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
12876         struct to a subclass of gimple_opt_pass along with...
12877         (pass_data_build_alias): ...new pass_data instance and...
12878         (make_pass_build_alias): ...new function.
12879         (pass_build_ealias): Convert from a global struct to a subclass of
12880         gimple_opt_pass along with...
12881         (pass_data_build_ealias): ...new pass_data instance and...
12882         (make_pass_build_ealias): ...new function.
12883         (pass_ipa_pta): Convert from a global struct to a subclass of
12884         simple_ipa_opt_pass along with...
12885         (pass_data_ipa_pta): ...new pass_data instance and...
12886         (make_pass_ipa_pta): ...new function.
12887         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
12888         subclass of gimple_opt_pass along with...
12889         (pass_data_uncprop): ...new pass_data instance and...
12890         (make_pass_uncprop): ...new function.
12891         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
12892         global struct to a subclass of gimple_opt_pass along with...
12893         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
12894         (make_pass_late_warn_uninitialized): ...new function.
12895         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
12896         to a subclass of gimple_opt_pass along with...
12897         (pass_data_init_datastructures): ...new pass_data instance and...
12898         (make_pass_init_datastructures): ...new function.
12899         (pass_early_warn_uninitialized): Convert from a global struct to a
12900         subclass of gimple_opt_pass along with...
12901         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
12902         (make_pass_early_warn_uninitialized): ...new function.
12903         (pass_update_address_taken): Convert from a global struct to a
12904         subclass of gimple_opt_pass along with...
12905         (pass_data_update_address_taken): ...new pass_data instance and...
12906         (make_pass_update_address_taken): ...new function.
12907         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
12908         struct to a subclass of gimple_opt_pass along with...
12909         (pass_data_release_ssa_names): ...new pass_data instance and...
12910         (make_pass_release_ssa_names): ...new function.
12911         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
12912         subclass of gimple_opt_pass along with...
12913         (pass_data_stdarg): ...new pass_data instance and...
12914         (make_pass_stdarg): ...new function.
12915         * tree-switch-conversion.c (pass_convert_switch): Convert from a
12916         global struct to a subclass of gimple_opt_pass along with...
12917         (pass_data_convert_switch): ...new pass_data instance and...
12918         (make_pass_convert_switch): ...new function.
12919         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
12920         to a subclass of gimple_opt_pass along with...
12921         (pass_data_tail_recursion): ...new pass_data instance and...
12922         (make_pass_tail_recursion): ...new function.
12923         (pass_tail_calls): Convert from a global struct to a subclass of
12924         gimple_opt_pass along with...
12925         (pass_data_tail_calls): ...new pass_data instance and...
12926         (make_pass_tail_calls): ...new function.
12927         * tree-vect-generic.c (pass_lower_vector): Convert from a global
12928         struct to a subclass of gimple_opt_pass along with...
12929         (pass_data_lower_vector): ...new pass_data instance and...
12930         (make_pass_lower_vector): ...new function.
12931         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
12932         gimple_opt_pass along with...
12933         (pass_data_lower_vector_ssa): ...new pass_data instance and...
12934         (make_pass_lower_vector_ssa): ...new function.
12935         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
12936         to a subclass of gimple_opt_pass along with...
12937         (pass_data_slp_vectorize): ...new pass_data instance and...
12938         (make_pass_slp_vectorize): ...new function.
12939         (pass_ipa_increase_alignment): Convert from a global struct to a
12940         subclass of simple_ipa_opt_pass along with...
12941         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
12942         (make_pass_ipa_increase_alignment): ...new function.
12943         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
12944         gimple_opt_pass along with...
12945         (pass_data_vrp): ...new pass_data instance and...
12946         (make_pass_vrp): ...new function.
12947         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
12948         subclass of simple_ipa_opt_pass along with...
12949         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
12950         (make_pass_ipa_free_lang_data): ...new function.
12951         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
12952         gimple_opt_pass along with...
12953         (pass_data_tsan): ...new pass_data instance and...
12954         (make_pass_tsan): ...new function.
12955         (pass_tsan_O0): Convert from a global struct to a subclass of
12956         gimple_opt_pass along with...
12957         (pass_data_tsan_O0): ...new pass_data instance and...
12958         (make_pass_tsan_O0): ...new function.
12959         * var-tracking.c (pass_variable_tracking): Convert from a global
12960         struct to a subclass of rtl_opt_pass along with...
12961         (pass_data_variable_tracking): ...new pass_data instance and...
12962         (make_pass_variable_tracking): ...new function.
12963         * web.c (pass_web): Convert from a global struct to a subclass of
12964         rtl_opt_pass along with...
12965         (pass_data_web): ...new pass_data instance and...
12966         (make_pass_web): ...new function.
12967         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
12968         declaration with that of...
12969         (make_pass_mode_switch_use): ...new function.
12970         (pass_resolve_sw_modes): Replace declaration with that of...
12971         (make_pass_resolve_sw_modes): ...new function.
12972         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
12973         from a global struct to a subclass of rtl_opt_pass along with...
12974         (pass_data_mode_switch_use): ...new pass_data instance and...
12975         (make_pass_mode_switch_use): ...new function.
12976         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
12977         from a global struct to a subclass of rtl_opt_pass along with...
12978         (pass_data_resolve_sw_modes): ...new pass_data instance and...
12979         (make_pass_resolve_sw_modes): ...new function.
12980         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
12981         struct to a subclass of rtl_opt_pass along with...
12982         (pass_data_insert_vzeroupper): ...new pass_data instance and...
12983         (make_pass_insert_vzeroupper): ...new function.
12984         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
12985         global struct to a subclass of rtl_opt_pass along with...
12986         (pass_data_work_around_errata): ...new pass_data instance and...
12987         (make_pass_work_around_errata): ...new function.
12988         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
12989         struct to a subclass of rtl_opt_pass along with...
12990         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
12991         (make_pass_mips_machine_reorg2): ...new function.
12993 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12995         * passes.c (pass_manager::operator new): New.
12997 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12999         Handwritten part of conversion of passes to C++ classes.
13001         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
13002         (toplev.o): Add dep on PASS_MANAGER_H.
13003         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
13004         of early local pases to reflect this moving from a global to a
13005         member of gcc::pass_manager.
13006         (cgraph_add_new_function): Likewise.
13007         * lto-cgraph.c (lto_output_node): Update for conversion of
13008         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
13009         * passes.c (opt_pass::clone): New.
13010         (opt_pass::gate): New.
13011         (opt_pass::execute): New.
13012         (opt_pass::opt_pass): New.
13013         (pass_manager::execute_early_local_passes): New.
13014         (pass_manager::execute_pass_mode_switching): new.
13015         (finish_optimization_passes): Convert to...
13016         (pass_manager::finish_optimization_passes): ...this.
13017         (finish_optimization_passes): Update for conversion of passes to
13018         C++ classes.
13019         (register_dump_files_1): Use has_gate since we cannot portably
13020         check a vtable entry against NULL.
13021         (dump_one_pass): Likewise.
13022         (ipa_write_summaries_2): Likewise.
13023         (ipa_write_optimization_summaries_1): Likewise.
13024         (ipa_read_summaries_1): Likewise.
13025         (ipa_read_optimization_summaries_1): Likewise.
13026         (execute_ipa_stmt_fixups): Likewise.
13027         (pass_manager::pass_manager): Rewrite pass-creation, invoking
13028         pass-creation functions rather than wiring up globals, and
13029         storing the results in fields of pass_manager generated using
13030         pass-instances.def.
13031         (pass_manager::dump_profile_report): Update for conversion of
13032         passes to C++ classes.
13033         (pass_manager::execute_ipa_summary_passes): Likewise.
13034         (execute_one_ipa_transform_pass): Likewise.
13035         (execute_one_pass): Use has_gate and has_execute since we cannot
13036         portably check a vtable entry against NULL.
13037         * pass_manager.h (pass_manager::finish_optimization_passes): New.
13038         (pass_manager): Use pass-instances.def to add fields for the
13039         various pass instances.
13040         * toplev.c (finalize): Update for move of
13041         finish_optimization_passes to a method of gcc::pass_manager.
13042         * toplev.h (finish_optimization_passes): Move to method of class
13043         pass_manager.
13044         * tree-pass.h (struct pass_data): New.
13045         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
13046         (opt_pass::gate): Convert to virtual function.
13047         (opt_pass::~opt_pass): New.
13048         (opt_pass::clone): New.
13049         (opt_pass::execute): Convert to virtual function.
13050         (opt_pass::opt_pass): New.
13051         (opt_pass::ctxt_): new.
13052         (gimple_opt_pass): Convert to subclass of opt_pass.
13053         (gimple_opt_pass::gimple_opt_pass): New.
13054         (rtl_opt_pass): Convert to subclass of opt_pass.
13055         (rtl_opt_pass::rtl_opt_pass): New.
13056         (ipa_opt_pass_d): Convert to subclass of opt_pass.
13057         (ipa_opt_pass_d::ipa_opt_pass_d): New.
13058         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
13059         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
13060         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
13061         invocation of pass_mode_switching to reflect this moving from a
13062         global to a member of gcc::pass_manager.
13063         (ix86_option_override): Rework how pass_insert_vzeroupper is
13064         added to the pass_manager to reflect autogenerated changes.
13065         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
13067 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
13069         PR rtl-optimization/57708
13070         * recog.c (peep2_find_free_register): Validate all regs in a
13071         multi-reg mode.
13073 2013-08-05  Jan Hubicka  <jh@suse.cz>
13075         PR lto/57602
13076         * cgraph.c (verify_cgraph_node): Accept local flags from other
13077         partitions.
13078         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
13079         (function_and_variable_visibility): Likewise.
13080         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
13082 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13084         * graph.c (init_graph_slim_pretty_print): Remove.
13085         (print_graph_cfg): Do not call it.  Use local pretty printer.
13086         (start_graph_dump): Likewise.
13088 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13090         * gimple-pretty-print.c (buffer): Remove.
13091         (initialized): Likewise.
13092         (maybe_init_pretty_print): Likewise.
13093         (print_gimple_stmt): Do not call it.  Use non-static local
13094         pretty_printer variable.
13095         (print_gimple_expr): Likewise.
13096         (print_gimple_seq): Likewise.
13097         (gimple_dump_bb): Likewise.
13099 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13101         * asan.c (asan_pp): Remove.
13102         (asan_pp_initialized): Likewise.
13103         (asan_pp_initialize): Likewise.
13104         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
13105         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
13106         (asan_add_global): Likewise.
13108 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13110         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
13111         * diagnostic.h (diagnostic_flush_buffer): Adjust.
13112         * pretty-print.c (pp_formatted_text_data): Likewise.
13113         (pp_indent): Rename from pp_base_indent.
13114         (pp_format): Rename from pp_base_format.
13115         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
13116         (pp_format_verbatim): Rename from pp_base_format_verbatim.
13117         (pp_flush): Rename from pp_base_flush.
13118         (pp_set_line_maximum_length): Rename from
13119         pp_base_set_line_maximum_length.
13120         (pp_clear_output_area): Rename from pp_base_clear_output_area.
13121         (pp_set_prefix): Rename from pp_base_set_prefix.
13122         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
13123         (pp_emit_prefix): Rename from pp_base_emit_prefix.
13124         (pp_append_text): Rename from pp_base_append_text.
13125         (pp_formatted_text): Rename from pp_base_formatted_text.
13126         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
13127         (pp_remaining_character_count_for_line): Rename from
13128         pp_base_remaining_character_count_for_line.
13129         (pp_newline): Rename from pp_base_newline.
13130         (pp_character): Rename from pp_base_character.
13131         (pp_string): Rename from pp_base_string.
13132         (pp_maybe_space): Rename from pp_base_maybe_space.
13133         * asan.c (asan_pp_string): Adjust.
13134         (asan_emit_stack_protection): Likewise.
13135         (asan_add_global): Likewise.
13136         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
13137         * tree-mudflap.c (mf_varname_tree): Likewise.
13138         * tree-pretty-print.c (pp_tree_identifier): Rename from
13139         pp_base_tree_identifier.
13140         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
13141         Declare as function.
13143 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13145         * pretty-print.h (pp_bar_bar): New.
13146         (pp_ampersand_ampersand): Likewise.
13147         (pp_less_equal): Likewise.
13148         (pp_greater_equal): Likewise.
13149         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
13150         printer functions instead of pp_string or operators and punctuators.
13151         (dump_gimple_call): Likewise.
13152         (dump_gimple_omp_for): Likewise.
13153         (dump_gimple_transaction): Likewise.
13154         (dump_gimple_phi): Likewise.
13155         (pp_gimple_stmt_1): Likewise.
13156         * sched-vis.c (print_insn): Likewise.
13157         * tree-mudflap.c (mf_varname_tree): Likewise.
13158         * tree-pretty-print.c (dump_block_node): Likewise.
13159         (dump_generic_node): Likewise.
13161 2013-08-02  Jan Hubicka  <jh@suse.cz>
13163         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
13164         boundaries.
13165         * lto-streamer-out.c (tree_is_indexable): Results decls and
13166         parm decls are not indexable.
13167         (DFS_write_tree_body): Do not follow args and results.
13168         (hash_tree): Likewise.
13169         (output_functions): Rearrange so struct function is needed
13170         only when real body is output; be able to also ouptut abstract
13171         functions; output DECL_ARGUMENTS and DECL_RESULT.
13172         (lto_output): When not in WPA, ale store abstract functions.
13173         (write_symbol): Do not care about RESULT_DECL.
13174         (output_symbol_p): Handle correctly sbtract decls.
13175         * lto-streamer-in.c (input_function): Rearrange so struct
13176         function can be NULL at entry; allow streaming of
13177         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
13178         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
13179         sanity check during LTO.
13180         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
13181         RESULT_DECl and DECL_ARGUMENTS.
13182         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
13183         Likewise.
13185 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13187         * pretty-print.h (pp_underscore): New.
13188         (pp_comma): Tidy.
13189         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
13190         printer functions instead of pp_character.
13191         (dump_binary_rhs): Likewise.
13192         (dump_ternary_rhs): Likewise.
13193         (dump_gimple_call_args): Likewise.
13194         (pp_points_to_solution): Likewise.
13195         (dump_gimple_call): Likewise.
13196         (dump_gimple_switch): Likewise.
13197         (dump_gimple_cond): Likewise.
13198         (dump_gimple_bind): Likewise.
13199         (dump_gimple_try): Likewise.
13200         (dump_gimple_omp_for): Likewise.
13201         (dump_gimple_omp_continue): Likewise.
13202         (dump_gimple_omp_single): Likewise.
13203         (dump_gimple_omp_sections): Likewise.
13204         (dump_gimple_omp_block): Likewise.
13205         (dump_gimple_omp_critical): Likewise.
13206         (dump_gimple_transaction): Likewise.
13207         (dump_gimple_asm): Likewise.
13208         (dump_gimple_phi): Likewise.
13209         (dump_gimple_omp_parallel): Likewise.
13210         (dump_gimple_omp_task): Likewise.
13211         (dump_gimple_omp_atomic_load): Likewise.
13212         (dump_gimple_omp_atomic_store): Likewise.
13213         (dump_gimple_mem_ops): Likewise.
13214         (pp_gimple_stmt_1): Likewise.
13215         (pp_cfg_jump): Likewise.
13216         (dump_implicit_edges): Likewise.
13217         (gimple_dump_bb_for_graph): Likewise.
13218         * graph.c (draw_cfg_node): Likewise.
13219         * langhooks.c (lhd_print_error_function): Likewise.
13220         * sched-vis.c (print_exp): Likewise.
13221         (print_value): Likewise.
13222         (print_pattern): Likewise.
13223         (print_insn): Likewise.
13224         (rtl_dump_bb_for_graph): Likewise.
13225         * tree-pretty-print.c (dump_function_declaration): Likewise.
13226         (dump_array_domain): Likewise.
13227         (dump_omp_clause): Likewise.
13228         (dump_location): Likewise.
13229         (dump_generic_node): Likewise.
13230         (print_struct_decl): Likewise.
13231         * diagnostic.c (diagnostic_show_locus): Use pp_space.
13233 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
13235         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
13236         candidate table when replacing a candidate statement.
13237         (replace_rhs_if_not_dup): Likewise.
13238         (replace_one_candidate): Likewise.
13240 2013-08-02  Jan Hubicka  <jh@suse.cz>
13241             Martin Liska  <marxin.liska@gmail.com>
13243         * cgraphunit.c (add_new_function): Fix logic when adding from
13244         late IPA pass.
13245         (assemble_thunk): Rename to ...
13246         (expand_thunk); .. this one; export; get it working with
13247         general functions; make produced gimple valid.
13248         * cgraph.h (expand_thunk): Declare.
13250 2013-08-02  Jan Hubicka  <jh@suse.cz>
13252         * ipa-cp.c (gather_context_independent_values): Use
13253         ipa_get_param_move_cost.
13254         (get_replacement_map): Remove PARAM; move parameter folding
13255         into tree-inline.c
13256         (create_specialized_node): Update.
13257         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
13258         assert that we have gimple body; update move_cost.
13259         (count_formal_params): Assert that we have gimple body.
13260         (ipa_dump_param): New function.
13261         (ipa_alloc_node_params): Break out from ...
13262         (ipa_initialize_node_params): ... here.
13263         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
13264         (ipa_write_node_info): Stream move costs.
13265         (ipa_read_node_info): Read move costs.
13266         (ipa_update_after_lto_read): Do not recompute node params.
13267         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
13268         (ipa_get_param): Check we are not in WPA.
13269         (ipa_get_param_move_cost): New.
13270         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
13271         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
13272         parm numbers to be present.
13274 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
13276         PR rtl-optimization/58048
13277         * lra-constraints.c (process_alt_operands): Don't check asm
13278         operand on register.
13280 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
13282         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
13283         the implied StoreLoad barrier for atomic operations if before.
13285 2013-08-02  Jan Hubicka  <jh@suse.cz>
13286             Martin Liska  <marxin.liska@gmail.com>
13288         * cgraph.c (cgraph_function_body_availability): Do not check
13289         cgraph flags.
13290         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
13291         symtab_node_availability): Declare.
13292         * ipa.c (can_replace_by_local_alias): New.
13293         (function_and_variable_visibility): Use it.
13294         * symtab.c (symtab_for_node_and_aliases,
13295         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
13297 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
13299         PR rtl-optimization/57963
13300         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
13301         (lra_constraints): Use them.
13303 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
13305         * config/arm/types.md (define_attr "type"): Add "load_acq"
13306         and "store_rel".
13307         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
13308         changes.
13309         (cortex_a53_store1): Likewise.
13311 2013-08-01  Jan Hubicka  <jh@suse.cz>
13313         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
13314         partitions are not needed.
13316 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
13318         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
13319         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
13320         MAYBE_NON_Q_CLASS_P where appropriate.
13322 2013-08-01  Jan Hubicka  <jh@suse.cz>
13324         * cgraph.h (release_function_body): Declare.
13325         * tree.c (free_lang_data_in_decl): Free, parameters and return values
13326         of unused delcarations.
13328 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13330         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
13331         RTL form when subtracting a constant.
13333 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13335         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
13336         Generate canonical plus rtx with negated immediate instead of minus
13337         where appropriate.
13338         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
13340 2013-08-01  Jan Hubicka  <jh@suse.cz>
13342         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
13343         (cgraph_release_function_body): Likewise.
13344         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
13345         * cgraph.h (cgrpah_node): Rename abstract_and_needed
13346         to used_as_abstract_origin.
13347         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
13348         symbols used as abstract origins.
13349         * cgraphunit.c (analyze_functions): Update.
13350         * ipa.c (symtab_remove_unreachable_nodes): Recompute
13351         used_as_abstract_origin.
13352         * tree-inline.c (tree_function_versioning): Update
13353         used_as_abstract_origin; be ready for DECL_RESULT and
13354         DECL_ARGUMENTS to be NULL.
13356         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
13357         for abstract functions.
13358         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
13359         real symbols.
13361 2013-08-01  Jan Hubicka  <jh@suse.cz>
13363         * profile.c (compute_value_histograms): Fix thinko.
13365 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
13367         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
13368         aarch-common-protos.h to extra_headers.
13369         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
13370         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
13371         * config/aarch64/t-aarch64 (aarch-common.o): Define.
13373 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
13375         * config/aarch64/aarch64.md (define_attr "type"): Delete.
13376         Include "../arm/types.md".  Define "type" attribute for all patterns.
13377         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
13378         attribute changes.
13380 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
13382         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
13383         to support power8 load fusion.
13384         (fusion_gpr_mem_load): Likewise.
13386         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
13388         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
13389         declarations for power8 load fusion.
13390         (emit_fusion_gpr_load): Likewise.
13392         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
13393         tuning for power8, turn on fusion mode by default.  Turn on sign
13394         extending fusion mode if normal fusion mode is on, and we are at
13395         -O2 or -O3.
13396         (fusion_gpr_load_p): New function, return true if we can fuse an
13397         addis instruction with a dependent load to a GPR.
13398         (emit_fusion_gpr_load): Emit the instructions for power8 load
13399         fusion to GPRs.
13401         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
13402         (VSX load fusion peepholes): New peepholes to fuse together an
13403         addi instruction with a VSX load instruction.
13405         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
13406         peepholes to fuse an addis instruction with a load to a GPR base
13407         register.  If we are supporting sign extending fusions, convert
13408         sign extending loads to zero extending loads and add an explicit
13409         sign extension.
13411 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
13413         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
13414         aarch-common-protos.h to extra_headers.
13415         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
13416         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
13417         (arm_early_store_addr_dep): Likewise.
13418         (arm_no_early_alu_shift_dep): Likewise.
13419         (arm_no_early_alu_shift_value_dep): Likewise.
13420         (arm_no_early_mul_dep): Likewise.
13421         (arm_no_early_store_addr_dep): Likewise.
13422         (arm_mac_accumulator_is_mul_result): Likewise.
13423         (arm_mac_accumulator_is_result): Likewise.
13424         * config/arm/aarch-common.c: ... here.  New file.
13425         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
13426         here to ...
13427         (arm_early_store_addr_dep): Likewise.
13428         (arm_no_early_alu_shift_dep): Likewise.
13429         (arm_no_early_alu_shift_value_dep): Likewise.
13430         (arm_no_early_mul_dep): Likewise.
13431         (arm_no_early_store_addr_dep): Likewise.
13432         (arm_mac_accumulator_is_mul_result): Likewise.
13433         (arm_mac_accumulator_is_result): Likewise.
13434         * config/arm/aarch-common-protos.h: ... here.  New file.
13435         * config/arm/t-arm (aarch-common.o): Define.
13437 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
13439         * config/arm/arm.md: Include new file "types.md".
13440         (define_attr "type"): Move from here to ...
13441         (define_attr "mul32"): Likewise.
13442         (define_attr "mul64"): Likewise.
13443         * config/arm/types.md: ... here.  New file.
13445 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13447         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
13448         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
13450 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
13452         * gen-pass-instances.awk: Fix offset of substr().
13454 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
13456         * Makefile.in (pass-instances.def): New.
13457         (passes.o): Replace dependency on passes.def with one on
13458         pass-instances.def
13460         * gen-pass-instances.awk: New.
13462         * passes.c (pass_manager::pass_manager): Use pass-instances.def
13463         rather than passes.def, updating local definition of NEXT_PASS
13464         macro to add an extra NUM parameter (currently unused).
13466 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
13468         * Makefile.in (PASS_MANAGER_H): New.
13469         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
13470         (passes.o): Likewise.
13471         (statistics.o): Likewise.
13472         (cgraphunit.o): Likewise.
13473         (context.o): Depend on PASS_MANAGER_H.
13475         * pass_manager.h: New.
13477         * cgraphunit.c (cgraph_add_new_function): Update for moves
13478         of globals to fields of pass_manager.
13479         (analyze_function): Likewise.
13480         (expand_function): Likewise.
13481         (ipa_passes): Likewise.
13482         (compile): Likewise.
13484         * context.c (context::context): New.
13485         * context.h  (context::context): New.
13486         (context::get_passes): New.
13487         (context::passes_): New.
13489         * lto-cgraph.c (input_node): Update for moves of globals to
13490         fields of pass_manager.
13492         * passes.c (all_passes): Remove, in favor of a field of the
13493         same name within the new class pass_manager.
13494         (all_small_ipa_passes): Likewise.
13495         (all_lowering_passes): Likewise.
13496         (all_regular_ipa_passes): Likewise.
13497         (all_late_ipa_passes): Likewise.
13498         (all_lto_gen_passes): Likewise.
13499         (passes_by_id): Likewise.
13500         (passes_by_id_size): Likewise.
13501         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
13502         the new class pass_manager.
13503         (set_pass_for_id): Convert to...
13504         (pass_manager::set_pass_for_id): ...method.
13505         (get_pass_for_id): Convert to...
13506         (pass_manager::get_pass_for_id): ...method.
13507         (register_one_dump_file): Move body of implementation into...
13508         (pass_manager::register_one_dump_file): ...here.
13509         (register_dump_files_1): Convert to...
13510         (pass_manager::register_dump_files_1): ...method.
13511         (register_dump_files): Convert to...
13512         (pass_manager::register_dump_files): ...method.
13513         (create_pass_tab): Update for moves of globals to fields of
13514         pass_manager.
13515         (dump_passes): Move body of implementation into...
13516         (pass_manager::dump_passes): ...here.
13517         (register_pass): Move body of implementation into...
13518         (pass_manager::register_pass): ...here.
13519         (init_optimization_passes): Convert into...
13520         (pass_manager::pass_manager): ...constructor for new
13521         pass_manager class, and initialize the pass_lists array.
13522         (check_profile_consistency): Update for moves of globals to
13523         fields of pass_manager.
13524         (dump_profile_report): Move body of implementation into...
13525         (pass_manager::dump_profile_report): ...here.
13526         (ipa_write_summaries_1): Update for moves of pass lists from
13527         being globals to fields of pass_manager.
13528         (ipa_write_optimization_summaries): Likewise.
13529         (ipa_read_summaries):  Likewise.
13530         (ipa_read_optimization_summaries): Likewise.
13531         (execute_all_ipa_stmt_fixups): Likewise.
13533         * statistics.c (statistics_fini): Update for moves of globals to
13534         fields of pass_manager.
13536         * toplev.c (general_init): Replace call to
13537         init_optimization_passes with construction of the pass_manager
13538         instance.
13540         * tree-pass.h (all_passes): Remove, in favor of a field of the
13541         same name within the new class pass_manager.
13542         (all_small_ipa_passes): Likewise.
13543         (all_lowering_passes): Likewise.
13544         (all_regular_ipa_passes): Likewise.
13545         (all_lto_gen_passes): Likewise.
13546         (all_late_ipa_passes): Likewise.
13547         (passes_by_id): Likewise.
13548         (passes_by_id_size): Likewise.
13549         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
13550         the new class pass_manager.
13551         (get_pass_for_id): Remove.
13553 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
13555         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
13556         configs.
13558 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
13560         * arm.md (mulhi3): New expand pattern.
13562 2013-07-30  Jan Hubicka  <jh@suse.cz>
13563             Martin Liska  <marxin.liska@gmail.com>
13565         * profile.c (compute_value_histograms): Do not ICE when
13566         there is mismatch only on some counters.
13568 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
13570         PR rtl-optimization/57637
13571         * function.c (move_insn_for_shrink_wrap): Also check the
13572         GEN set of the LIVE problem for the liveness analysis
13573         if it exists, otherwise give up.
13575 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
13577         PR tree-optimization/57993
13578         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
13579         replaced statement in the candidate table.
13580         (phi_add_costs): Return infinite cost when the hidden basis does
13581         not dominate all phis on which the candidate is dependent.
13582         (replace_one_candidate): Record replaced statement in the
13583         candidate table.
13585 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
13587         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
13588         (ashlv2si3): New expander.
13589         (*ashlv2si3_i): New define_insn_and_split.
13590         * predicates.md (float_operation): Allow patterns with three
13591         basic sub-patterns.
13593         PR rtl-optimization/58021
13594         * mode-switching.c (create_pre_exit): Always split off preceding
13595         insns if we are not at the basic block head.
13597 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
13599         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
13600         (UCLIBC_DYNAMIC_LINKER): New macro.
13601         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
13602         `-mnan=2008'.
13603         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
13604         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
13605         `-mnan=2008'.
13606         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
13607         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
13608         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
13609         for SF and DF modes.  Use ieee_quad_format for TF mode.
13610         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
13611         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
13612         (mips_option_override): Handle `-mnan=legacy'.
13613         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
13614         `-mabs=2008' and `-mnan=2008'.
13615         (OPTION_DEFAULT_SPECS): Add "nan" default.
13616         (ASM_SPEC): Handle `-mnan='.
13617         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
13618         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
13619         comment accordingly.
13620         (neg<mode>2): Likewise.
13621         * config/mips/mips.opt (mabs, mnan): New options.
13622         * doc/install.texi (Configuration): Document `--with-nan=' option.
13623         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
13624         `-mnan=' options.
13625         (MIPS Options): Document them.
13626         * config.gcc <mips*-*-*>: Handle `--with-nan='.
13627         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
13628         * configure: Regenerate.
13629         * config.in: Regenerate.
13631 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
13633         * config/i386/i386.md (float post-reload splitters): Do not check
13634         for subregs of SSE registers.
13636 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
13637             H.J. Lu  <hongjiu.lu@intel.com>
13639         PR target/57954
13640         PR target/57988
13641         * config/i386/i386.md (post-reload splitter
13642         to avoid partial SSE reg dependency stalls): New pattern.
13644 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
13646         * config/s390/s390.md ("movcc"): Swap load and store instructions.
13648 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
13650         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13651         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
13653 2013-07-26  Cary Coutant  <ccoutant@google.com>
13655         * dwarf2out.c (die_checksum_ordered): Don't include template
13656         instantiations in signature.
13657         (is_template_parameter): New function.
13658         (is_template_instantiation): New function.
13659         (generate_skeleton_bottom_up): Don't include template instantiations
13660         in type unit DIE.
13661         (generate_skeleton): Likewise.
13662         (break_out_comdat_types): Move recursive call to break out nested
13663         types earlier.
13664         (prune_unused_types_mark_generic_parms_dies): Call
13665         is_template_parameter.
13667 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
13669         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
13670         uses vector registers.
13671         * config/aarch64/iterators.md: Add attributes rtn and vas.
13673 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13674             Richard Earnshaw  <richard.earnshaw@arm.com>
13676         * combine.c (simplify_comparison): Re-canonicalize operands
13677         where appropriate.
13678         * config/arm/arm.md (movcond_addsi): New splitter.
13680 2013-07-25  Sterling Augustine  <saugustine@google.com>
13682         * dwarf2out.c (size_of_pubnames): Move code to...
13683         (include_pubname_in_output): ...here.  New.
13684         (want_pubnames): Rearrange.
13685         (output_pubnames): Call include_pubname_in_output.  Move assertion.
13687 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
13689         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
13691 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
13693         PR target/38836
13694         * doc/extend.texi: Remove obsolete builtins. Fix
13695         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
13697 2013-07-25  Jan Hubicka  <jh@suse.cz>
13699         * cgraph.c (release_function_body): Break out from ...
13700         (cgraph_release_function_body): ... this one; also release DECL_RESULT
13701         and DECL_ARGUMENTS.
13702         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
13703         old_tree in the map.
13704         (create_specialized_node): Update.
13705         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
13706         into index.
13707         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
13708         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
13709         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
13710         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
13711         DECL_RESULT.
13713 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13715         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
13716         addsi3_carryin_alt2_<optab>): Correct output template.
13718 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13720         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
13721         Adjust for arm_restrict_it.
13722         Remove trailing whitespace.
13724 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
13726         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
13727         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
13729         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
13731 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
13733         PR rtl-optimization/57960
13734         * lra-constraints.c (process_alt_operands): Use the right mode
13735         when checking strict_low.
13737 2013-07-25  Jan Hubicka  <jh@suse.cz>
13739         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
13740         * cgraph.c (cgraph_remove_node): Do not release function body
13741         when in cgraph streaming.
13742         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
13743         in other partitions are not considered reachable; fix handling of
13744         clones.
13746 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13748         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
13750 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13752         PR target/19599
13753         PR target/57731
13754         PR target/57837
13755         * config/arm/arm.md ("*sibcall_insn): Replace use of
13756         Ss with US. Adjust output for v5 and v4t.
13757         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
13759         * config/arm/constraints.md ("Ss"): Rename to US.
13761 2013-07-25  Terry Guo  <terry.guo@arm.com>
13763         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
13764         shift_add/shift_sub0/shift_sub1 RTXs.
13766 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
13767             Anton Blanchard  <anton@au1.ibm.com>
13769         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
13770         (altivec_vpks<VI_char>ss): Likewise.
13771         (altivec_vpks<VI_char>us): Likewise.
13772         (altivec_vpku<VI_char>us): Likewise.
13773         (altivec_vpku<VI_char>um): Likewise.
13775 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
13777         Introduce context class.
13779         * Makefile.in (CONTEXT_H): New.
13780         (OBJS): Add context.o.
13781         (toplev.o): Add CONTEXT_H to dependencies.
13782         (context.o): New.
13784         * toplev.c (general_init): Create the singleton gcc::context instance.
13786         * context.c: New.
13788         * context.h: New.
13790 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
13792         PR rtl-optimization/57968
13793         * mode-switching.c (create_pre_exit): Allow instructions that
13794         don't set a return register to need a non-exit mode.
13796 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
13797             Anton Blanchard  <anton@au1.ibm.com>
13799         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
13800         operands to vperm for little endian.
13801         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
13802         of lvsl to create the control mask for a vperm for little endian.
13804 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13805             Anton Blanchard  <anton@au1.ibm.com>
13807         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
13808         two operands for little-endian.
13810 2013-07-23  Steve Ellcey  <sellcey@mips.com>
13812         * config/mips/mips.c (mips_case_values_threshold): New.
13813         (TARGET_CASE_VALUES_THRESHOLD): Define.
13815 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13816             Anton Blanchard  <anton@au1.ibm.com>
13818         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
13819         selection of field for vector splat in little endian mode.
13821 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
13823         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
13824         expanders to rs6000.md.
13825         (ior<mode>3): Likewise.
13826         (and<mode>3): Likewise.
13827         (one_cmpl<mode>2): Likewise.
13828         (nor<mode>3): Likewise.
13829         (andc<mode>3): Likewise.
13830         (eqv<mode>3): Likewise.
13831         (nand<mode>3): Likewise.
13832         (orc<mode>3): Likewise.
13834         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
13835         declaration.
13837         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
13838         to split multi-word logical operations.
13839         (rs6000_split_logical_di): Likewise.
13840         (rs6000_split_logical): Likewise.
13842         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
13843         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
13844         and allow TImode operations in 32-bit.
13845         (vsx_and<mode>3_64bit): Likewise.
13846         (vsx_ior<mode>3_32bit): Likewise.
13847         (vsx_ior<mode>3_64bit): Likewise.
13848         (vsx_xor<mode>3_32bit): Likewise.
13849         (vsx_xor<mode>3_64bit): Likewise.
13850         (vsx_one_cmpl<mode>2_32bit): Likewise.
13851         (vsx_one_cmpl<mode>2_64bit): Likewise.
13852         (vsx_nor<mode>3_32bit): Likewise.
13853         (vsx_nor<mode>3_64bit): Likewise.
13854         (vsx_andc<mode>3_32bit): Likewise.
13855         (vsx_andc<mode>3_64bit): Likewise.
13856         (vsx_eqv<mode>3_32bit): Likewise.
13857         (vsx_eqv<mode>3_64bit): Likewise.
13858         (vsx_nand<mode>3_32bit): Likewise.
13859         (vsx_nand<mode>3_64bit): Likewise.
13860         (vsx_orc<mode>3_32bit): Likewise.
13861         (vsx_orc<mode>3_64bit): Likewise.
13863         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
13864         logical types in GPRs.
13866         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
13867         logical insns to rs6000.md, and allow TImode operations in
13868         32-bit.
13869         (altivec_ior<mode>3): Likewise.
13870         (altivec_xor<mode>3): Likewise.
13871         (altivec_one_cmpl<mode>2): Likewise.
13872         (altivec_nor<mode>3): Likewise.
13873         (altivec_andc<mode>3): Likewise.
13875         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
13876         attributes for moving the 128-bit logical operations into
13877         rs6000.md.
13878         (BOOL_REGS_OUTPUT): Likewise.
13879         (BOOL_REGS_OP1): Likewise.
13880         (BOOL_REGS_OP2): Likewise.
13881         (BOOL_REGS_UNARY): Likewise.
13882         (BOOL_REGS_AND_CR0): Likewise.
13883         (one_cmpl<mode>2): Add support for DI logical operations on
13884         32-bit, splitting the operations to 32-bit.
13885         (anddi3): Likewise.
13886         (iordi3): Likewise.
13887         (xordi3): Likewise.
13888         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
13889         changes to combine the 32/64-bit code, allow logical operations on
13890         TI mode in 32-bit, and to use similar match_operator patterns like
13891         scalar mode uses.  Combine the Altivec and VSX code for logical
13892         operations, and move it here.
13893         (ior<mode>3, 128-bit types): Likewise.
13894         (xor<mode>3, 128-bit types): Likewise.
13895         (one_cmpl<mode>3, 128-bit types): Likewise.
13896         (nor<mode>3, 128-bit types): Likewise.
13897         (andc<mode>3, 128-bit types): Likewise.
13898         (eqv<mode>3, 128-bit types): Likewise.
13899         (nand<mode>3, 128-bit types): Likewise.
13900         (orc<mode>3, 128-bit types): Likewise.
13901         (and<mode>3_internal): Likewise.
13902         (bool<mode>3_internal): Likewise.
13903         (boolc<mode>3_internal1): Likewise.
13904         (boolc<mode>3_internal2): Likewise.
13905         (boolcc<mode>3_internal1): Likewise.
13906         (boolcc<mode>3_internal2): Likewise.
13907         (eqv<mode>3_internal1): Likewise.
13908         (eqv<mode>3_internal2): Likewise.
13909         (one_cmpl1<mode>3_internal): Likewise.
13911 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
13913         * config/microblaze/microblaze.c (microblaze_expand_prologue):
13914         Rename flag_stack_usage to flag_stack_usage_info.
13916 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
13918         * config/microblaze/sync.md: New file.
13919         * config/microblaze/microblaze.md: Include sync.md
13920         * config/microblaze/microblaze.c: Add print_operand 'y'.
13921         * config/microblaze/constraints.md: Add memory_contraint
13922         'Q' which is a single register.
13924 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
13926         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
13928 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
13930         * reload.c (find_reloads): Exit loop once we find this operand
13931         cannot be reloaded somehow for this alternative.
13933         * reload.c (find_reloads): Exit loop once we find a hard register.
13935         * rtlanal.c (computed_jump_p): Exit loop once we find label
13936         reference is used.
13938         * i386.c (ix86_pad_returns): Exit loop after setting replace.
13940         * cfgloopmanip.c (remove_path): Exit loop after setting
13941         irred_invalidated.
13943         * gensupport.c (subst_dup): Avoid loop if code is not
13944         MATCH_DUP nor MATCH_OP_DUP.
13946 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
13948         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
13950 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
13952         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
13953         true for SP_REGNUM if mode == ptr_mode.
13954         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
13955         with value R0_REGNUM + 31.
13957 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
13959         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
13960         pad pointer-typed argument downward.
13962 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
13964         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
13965         and __ILP32__ when the ILP32 model is in use.
13967 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
13969         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
13970         (aarch64_load_symref_appropriately): In the case of
13971         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
13972         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
13973         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
13974         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
13975         if 'mode' doesn't equal to 'ptr_mode'.
13976         (aarch64_output_mi_thunk): Add an assertion on the alignment of
13977         'vcall_offset'; change to call aarch64_emit_move differently depending
13978         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
13979         to calculate the upper bound of 'vcall_offset'.
13980         (aarch64_cannot_force_const_mem): Change to also return true if
13981         mode != ptr_mode.
13982         (aarch64_legitimize_reload_address): In the case of large
13983         displacements, add new local variable 'xmode' and an assertion
13984         based on it; change to use 'xmode' to generate the new rtx and
13985         reload.
13986         (aarch64_asm_trampoline_template): Change to generate the template
13987         differently depending on TARGET_ILP32 or not; change to use
13988         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
13989         (aarch64_trampoline_size): Removed.
13990         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
13991         and replace immediate literals with it.  Change to use 'ptr_mode'
13992         instead of 'DImode' and call convert_memory_address if the mode
13993         of 'fnaddr' doesn't equal to 'ptr_mode'.
13994         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
13995         to output symbol.
13996         (aarch64_elf_asm_destructor): Likewise.
13997         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
13998         on TARGET_ILP32 instead of aarch64_trampoline_size.
13999         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
14000         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
14001         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
14002         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
14003         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
14004         (storewb_pair<GPI:mode>_<P:mode>): ... this.
14005         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
14006         depending on the value of 'mode'.
14007         (add_losym_<mode>): New.
14008         (ldr_got_small_<mode>): New, based on ldr_got_small.
14009         (ldr_got_small): Remove.
14010         (ldr_got_small_sidi): New.
14011         * config/aarch64/iterators.md (P): New.
14012         (PTR): Change to 'ptr_mode' in the condition.
14014 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14016         * config.gcc (aarch64*-*-*): Support --with-abi.
14017         (aarch64*-*-elf): Support --with-multilib-list.
14018         (aarch64*-*-linux*): Likewise.
14019         (supported_defaults): Add abi to aarch64*-*-*.
14020         * configure.ac: Mention AArch64 for --with-multilib-list.
14021         * configure: Re-generated.
14022         * config/aarch64/biarchilp32.h: New file.
14023         * config/aarch64/biarchlp64.h: New file.
14024         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
14025         (ABI_SPEC): Ditto.
14026         (MULTILIB_DEFAULTS): Ditto.
14027         (DRIVER_SELF_SPECS): Ditto.
14028         (ASM_SPEC): Update to also substitute -mabi.
14029         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
14030         file whose name depends on -mabi= and -mbig-endian.
14031         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
14032         TARGET_ILP32.
14033         (POINTER_SIZE): New define.
14034         (POINTERS_EXTEND_UNSIGNED): Ditto.
14035         (enum aarch64_abi_type): New enumeration tag.
14036         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
14037         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
14038         (TARGET_ILP32): New define.
14039         * config/aarch64/aarch64.opt (mabi): New.
14040         (aarch64_abi): New.
14041         (ilp32, lp64): New values for -mabi.
14042         * config/aarch64/t-aarch64 (comma): New define.
14043         (MULTILIB_OPTIONS): Ditto.
14044         (MULTILIB_DIRNAMES): Ditto.
14045         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
14046         * doc/invoke.texi: Document -mabi for AArch64.
14048 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
14050         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
14052 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14053             Anton Blanchard  <anton@au1.ibm.com>
14055         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
14056         endianness when selecting field to splat.
14058 2013-07-22  Eric Christopher  <echristo@gmail.com>
14060         * dwarf2out.c (die_odr_checksum): New function to use
14061         CHECKSUM_ macros and ULEB128 for DIE tag.
14062         (generate_type_signature): Use.
14064 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
14066         * config.gcc (sparc*-*-*): Accept leon3 processor.
14067         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
14068         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
14069         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
14070         * config/sparc/sparc.opt (enum processor_type): Add leon3.
14071         (mfix-ut699): Adjust comment.
14072         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
14073         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
14074         (CPP_CPU_SPEC): Likewise.
14075         (ASM_CPU_SPEC): Likewise.
14076         * config/sparc/sparc.c (leon3_cost): New constant.
14077         (sparc_option_override): Add leon3 support.
14078         (mem_ref): New function.
14079         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
14080         (sparc_do_work_around_errata): Look into the instruction in the delay
14081         slot and adjust accordingly.  Add fix for the data cache nullify issues
14082         of the UT699.  Change insertion position for the NOP.
14083         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
14084         (leon3_load): New reservation.
14085         (leon_store): Bump latency to 2.
14086         (grfpu): New automaton.
14087         (grfpu_alu): New unit.
14088         (grfpu_ds): Likewise.
14089         (leon_fp_alu): Adjust.
14090         (leon_fp_mult): Delete.
14091         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
14092         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
14093         * config/sparc/sparc.md (cpu): Add leon3.
14094         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
14095         (swapsi): Likewise.
14096         (atomic_test_and_set): Likewise.
14097         (ldstub): Likewise.
14099 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
14101         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
14102         default for R5900 targets.
14103         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
14104         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
14105         * config/mips/mips.c (mips_option_override): Report an error for
14106         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
14107         for -march=r5900 -mhard-float.
14109 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14111         * df-problems.c (can_move_insns_across): Exit loop once we
14112         find a non-fixed, non-global register.
14114         * ipa-pure-const.c (propagate_nothrow): Exit loop after
14115         setting can_throw.
14117         * omega.c (omega_eliminate_red): Break after setting red_found.
14118         (omega_problem_has_red_equations): Similarly after setting found.
14119         (omega_query_variable): Similarly after setting coupled.
14121 2013-07-22  Marek Polacek  <polacek@redhat.com>
14123         * gimplify.c: Don't include gimple.h twice.
14125 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14127         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
14128         instead of TARGET_THUMB1.
14129         (Pz): New constraint.
14130         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
14131         encodings.
14132         (compare_negsi_si): Likewise.
14133         (compare_addsi2_op0): Likewise.
14134         (compare_addsi2_op1): Likewise.
14135         (addsi3_carryin_<optab>): Likewise.
14136         (addsi3_carryin_alt2_<optab>): Likewise.
14137         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
14138         for arm_restrict_it.
14139         (subsi3_carryin): Likewise.
14140         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
14141         (minmax_arithsi): Disable for arm_restrict_it.
14142         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
14143         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
14144         (satsi_<SAT:code>_shift): Likewise.
14145         (arm_shiftsi3): Add alternative for 16-bit encoding.
14146         (arm32_movhf): Disable for arm_restrict_it.
14147         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
14148         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
14150 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
14152         * config/arm/arm.md (attribute "insn"): Delete.
14153         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
14154         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
14155         (not_shiftsi): Update for attribute change.
14156         (not_shiftsi_compare0): Likewise.
14157         (not_shiftsi_compare0_scratch): Likewise.
14158         (arm_one_cmplsi2): Likewise.
14159         (thumb1_one_cmplsi2): Likewise.
14160         (notsi_compare0): Likewise.
14161         (notsi_compare0_scratch): Likewise.
14162         (thumb1_movdi_insn): Likewise.
14163         (arm_movsi_insn): Likewise.
14164         (movhi_insn_arch4): Likewise.
14165         (movhi_bytes): Likewise.
14166         (arm_movqi_insn): Likewise.
14167         (thumb1_movqi_insn): Likewise.
14168         (arm32_movhf): Likewise.
14169         (thumb1_movhf): Likewise.
14170         (arm_movsf_soft_insn): Likewise.
14171         (thumb1_movsf_insn): Likewise.
14172         (thumb_movdf_insn): Likewise.
14173         (movsicc_insn): Likewise.
14174         (movsfcc_soft_insn): Likewise.
14175         (and_scc): Likewise.
14176         (cond_move): Likewise.
14177         (if_move_not): Likewise.
14178         (if_not_move): Likewise.
14179         (if_shift_move): Likewise.
14180         (if_move_shift): Likewise.
14181         (if_shift_shift): Likewise.
14182         (if_not_arith): Likewise.
14183         (if_arith_not): Likewise.
14184         (cond_move_not): Likewise.
14185         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
14186         (neon_mov<mode>): Likewise.
14187         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
14188         (thumb2_movsi_vfp): Likewise.
14189         (movsf_vfp): Likewise.
14190         (thumb2_movsf_vfp): Likewise.
14191         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
14192         change.
14193         (cortexa7_older_only): Likewise.
14194         (cortexa7_younger): Likewise.
14195         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
14196         (1020alu_shift_op): Likewise.
14197         (1020alu_shift_reg_op): Likewise.
14198         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
14199         (alu_shift_op): Likewise.
14200         (alu_shift_reg_op): Likewise.
14201         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
14202         (11_alu_shift_op): Likewise.
14203         (11_alu_shift_reg_op): Likewise.
14204         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
14205         (9_alu_shift_reg_op): Likewise.
14206         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
14207         change.
14208         (cortex_a15_alu_shift): Likewise.
14209         (cortex_a15_alu_shift_reg): Likewise.
14210         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
14211         (cortex_a5_alu_shift): Likewise.
14212         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
14213         change.
14214         (cortex_a53_alu_shift): Likewise.
14215         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
14216         change.
14217         (cortex_a7_alu_reg): Likewise.
14218         (cortex_a7_alu_shift): Likewise.
14219         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
14220         (cortex_a8_alu_shift): Likewise.
14221         (cortex_a8_alu_shift_reg): Likewise.
14222         (cortex_a8_mov): Likewise.
14223         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
14224         (cortex_a9_dp_shift): Likewise.
14225         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
14226         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
14227         (cortex_r4_mov): Likewise.
14228         (cortex_r4_alu_shift): Likewise.
14229         (cortex_r4_alu_shift_reg): Likewise.
14230         * config/arm/fa526.md (526_alu_op): Update for attribute change.
14231         (526_alu_shift_op): Likewise.
14232         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
14233         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
14234         (626te_alu_shift_op): Likewise.
14235         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
14236         (726te_alu_op): Likewise.
14237         (726te_alu_shift_op): Likewise.
14238         (726te_alu_shift_reg_op): Likewise.
14239         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
14240         (mp626_alu_shift_op): Likewise.
14241         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
14242         (pj4_alu_e1_conds): Likewise.
14243         (pj4_alu): Likewise.
14244         (pj4_alu_conds): Likewise.
14245         (pj4_shift): Likewise.
14246         (pj4_shift_conds): Likewise.
14247         (pj4_alu_shift): Likewise.
14248         (pj4_alu_shift_conds): Likewise.
14250 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14252         * config/arm/predicates.md (shiftable_operator_strict_it):
14253         New predicate.
14254         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
14255         Disable cond_exec version for arm_restrict_it.
14256         (thumb2_smaxsi3): Convert to generate cond_exec.
14257         (thumb2_sminsi3): Likewise.
14258         (thumb32_umaxsi3): Likewise.
14259         (thumb2_uminsi3): Likewise.
14260         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
14261         (thumb2_neg_abssi2): Likewise.
14262         (thumb2_mov_scc): Add alternative for 16-bit encoding.
14263         (thumb2_movsicc_insn): Adjust alternatives.
14264         (thumb2_mov_negscc): Disable for arm_restrict_it.
14265         (thumb2_mov_negscc_strict_it): New pattern.
14266         (thumb2_mov_notscc_strict_it): New pattern.
14267         (thumb2_mov_notscc): Disable for arm_restrict_it.
14268         (thumb2_ior_scc): Likewise.
14269         (thumb2_ior_scc_strict_it): New pattern.
14270         (thumb2_cond_move): Adjust for arm_restrict_it.
14271         (thumb2_cond_arith): Disable for arm_restrict_it.
14272         (thumb2_cond_arith_strict_it): New pattern.
14273         (thumb2_cond_sub): Adjust for arm_restrict_it.
14274         (thumb2_movcond): Likewise.
14275         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
14276         (thumb2_zero_extendhisi2_v6): Likewise.
14277         (thumb2_zero_extendqisi2_v6): Likewise.
14278         (orsi_notsi_si): Likewise.
14279         (orsi_not_shiftsi_si): Likewise.
14281 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
14283         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
14284         instruction sequence is 1 byte shorter.
14286 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
14288         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
14289         it is not needed after split.
14291 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
14293         PR target/51784
14294         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
14295         second label for nonlocal goto receivers. Don't output pic base labels
14296         unless we're producing PIC; mark that action unreachable().
14297         (ix86_save_reg): If the function contains a nonlocal label, save the
14298         PIC base reg.
14299         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
14300         * config/darwin.c (emitted_pic_label_num): New GTY.
14301         (update_pic_label_number_if_needed): New.
14302         (machopic_output_function_base_name): Adjust for nonlocal receiver
14303         case.
14304         (machopic_should_output_picbase_label): New.
14305         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
14306         (nonlocal_goto_receiver): New insn and split.
14308 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
14310         * config/aarch64/aarch64-builtins.c
14311         (aarch64_fold_builtin): Fold abs in all modes.
14312         * config/aarch64/aarch64-simd-builtins.def
14313         (abs): Enable for all modes.
14314         * config/aarch64/arm_neon.h
14315         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
14316         (vabs_f64): Add missing intrinsic.
14318 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
14320         * config/aarch64/arm_neon.h (vabs_s64): New function
14322 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
14324         PR target/57516
14325         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
14326         * config/avr/avr.md (adjust_len): Add `round'.
14327         * config/avr/avr-protos.h (avr_out_round): New prototype.
14328         (avr_out_plus): Add `out_label' argument.
14329         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
14330         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
14331         Handle the case where `insn' is just a pattern.
14332         (avr_out_bitop): Handle the case where `insn' is just a pattern.
14333         (avr_out_round): New function.
14334         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
14336 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
14338         * config/microblaze/microblaze.c (microblaze_expand_prologue):
14339         Add check for flag_stack_usage to handle -fstack-usage support
14341 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
14343         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
14344         interaction for new Power8 flags and VSX.
14346 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
14348         PR middle-end/57698
14349         * tree-inline.c (expand_call_inline): Emit errors during
14350         early_inlining only if optimization is not turned on.
14352 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14354         * passes.def: New.
14356         * passes.c (init_optimization_passes): Move the construction of
14357         the pass hierarchy into a new passes.def file.
14359         * Makefile.in (passes.o): Add dependency on passes.def.
14361 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14363         * passes.c (init_optimization_passes): Introduce macros for
14364         constructing the tree of passes (INSERT_PASSES_AFTER,
14365         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
14366         TERMINATE_PASS_LIST).
14368 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
14369             Wei Mi  <wmi@google.com>
14371         PR rtl-optimization/57878
14372         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
14373         top.
14374         (reload_pseudo_compare_func): Check nregs first for reload
14375         pseudos.
14377 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14379         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
14381 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
14383         * read-rtl.c (validate_const_int): Once an invalid character is
14384         seen, quit the loop.
14386         * gengtype.c (write_roots): Similarly once we find the "deletable"
14387         or "if_marked" option.
14389 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
14391         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
14392         "xtab" and "sat".  Move value "clz" from here to ...
14393         (attriubte "type"): ... here.
14394         (satsi_<SAT:code>): Delete "insn" attribute.
14395         (satsi_<SAT:code>_shift): Likewise.
14396         (arm_zero_extendqisi2addsi): Likewise.
14397         (arm_extendqisi2addsi): Likewise.
14398         (clzsi2): Update for attribute changes.
14399         (rbitsi2): Likewise.
14400         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
14401         attribute.
14402         (arm_usatsihi): Likewise.
14403         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
14405 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
14407         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
14408         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
14409         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
14410         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
14411         in alphabetical order.
14412         (attribute "core_cycles"): Update for attribute changes.
14413         (arm_addsi3): Likewise.
14414         (addsi3_compare0): Likewise.
14415         (addsi3_compare0_scratch): Likewise.
14416         (addsi3_compare_op1): Likewise.
14417         (addsi3_compare_op2): Likewise.
14418         (compare_addsi2_op0): Likewise.
14419         (compare_addsi2_op1): Likewise.
14420         (addsi3_carryin_shift_<optab>): Likewise.
14421         (subsi3_carryin_shift): Likewise.
14422         (rsbsi3_carryin_shift): Likewise.
14423         (arm_subsi3_insn): Likewise.
14424         (subsi3_compare0): Likewise.
14425         (subsi3_compare): Likewise.
14426         (arm_andsi3_insn): Likewise.
14427         (thumb1_andsi3_insn): Likewise.
14428         (andsi3_compare0): Likewise.
14429         (andsi3_compare0_scratch): Likewise.
14430         (zeroextractsi_compare0_scratch
14431         (andsi_not_shiftsi_si): Likewise.
14432         (iorsi3_insn): Likewise.
14433         (iorsi3_compare0): Likewise.
14434         (iorsi3_compare0_scratch): Likewise.
14435         (arm_xorsi3): Likewise.
14436         (thumb1_xorsi3_insn): Likewise.
14437         (xorsi3_compare0): Likewise.
14438         (xorsi3_compare0_scratch): Likewise.
14439         (satsi_<SAT:code>_shift): Likewise.
14440         (rrx): Likewise.
14441         (arm_shiftsi3): Likewise.
14442         (shiftsi3_compare0): Likewise.
14443         (not_shiftsi): Likewise.
14444         (not_shiftsi_compare0): Likewise.
14445         (not_shiftsi_compare0_scratch): Likewise.
14446         (arm_one_cmplsi2): Likewise.
14447         (thumb_one_complsi2): Likewise.
14448         (notsi_compare0): Likewise.
14449         (notsi_compare0_scratch): Likewise.
14450         (thumb1_zero_extendhisi2): Likewise.
14451         (arm_zero_extendhisi2): Likewise.
14452         (arm_zero_extendhisi2_v6): Likewise.
14453         (arm_zero_extendhisi2addsi): Likewise.
14454         (thumb1_zero_extendqisi2): Likewise.
14455         (thumb1_zero_extendqisi2_v6): Likewise.
14456         (arm_zero_extendqisi2): Likewise.
14457         (arm_zero_extendqisi2_v6): Likewise.
14458         (arm_zero_extendqisi2addsi): Likewise.
14459         (thumb1_extendhisi2): Likewise.
14460         (arm_extendhisi2): Likewise.
14461         (arm_extendhisi2_v6): Likewise.
14462         (arm_extendqisi): Likewise.
14463         (arm_extendqisi_v6): Likewise.
14464         (arm_extendqisi2addsi): Likewise.
14465         (thumb1_extendqisi2): Likewise.
14466         (thumb1_movdi_insn): Likewise.
14467         (arm_movsi_insn): Likewise.
14468         (movsi_compare0): Likewise.
14469         (movhi_insn_arch4): Likewise.
14470         (movhi_bytes): Likewise.
14471         (arm_movqi_insn): Likewise.
14472         (thumb1_movqi_insn): Likewise.
14473         (arm32_movhf): Likewise.
14474         (thumb1_movhf): Likewise.
14475         (arm_movsf_soft_insn): Likewise.
14476         (thumb1_movsf_insn): Likewise.
14477         (movdf_soft_insn): Likewise.
14478         (thumb_movdf_insn): Likewise.
14479         (arm_cmpsi_insn): Likewise.
14480         (cmpsi_shiftsi): Likewise.
14481         (cmpsi_shiftsi_swp): Likewise.
14482         (arm_cmpsi_negshiftsi_si): Likewise.
14483         (movsicc_insn): Likewise.
14484         (movsfcc_soft_insn): Likewise.
14485         (arith_shiftsi): Likewise.
14486         (arith_shiftsi_compare0
14487         (arith_shiftsi_compare0_scratch
14488         (sub_shiftsi): Likewise.
14489         (sub_shiftsi_compare0
14490         (sub_shiftsi_compare0_scratch
14491         (and_scc): Likewise.
14492         (cond_move): Likewise.
14493         (if_plus_move): Likewise.
14494         (if_move_plus): Likewise.
14495         (if_move_not): Likewise.
14496         (if_not_move): Likewise.
14497         (if_shift_move): Likewise.
14498         (if_move_shift): Likewise.
14499         (if_shift_shift): Likewise.
14500         (if_not_arith): Likewise.
14501         (if_arith_not): Likewise.
14502         (cond_move_not): Likewise.
14503         (thumb1_ashlsi3): Set type attribute.
14504         (thumb1_ashrsi3): Likewise.
14505         (thumb1_lshrsi3): Likewise.
14506         (thumb1_rotrsi3): Likewise.
14507         (shiftsi3_compare0_scratch): Likewise.
14508         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
14509         (neon_mov<mode>): Likewise.
14510         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
14511         attribute changes.
14512         (thumb2_movsi_insn): Likewise.
14513         (thumb2_cmpsi_neg_shiftsi): Likewise.
14514         (thumb2_extendqisi_v6): Likewise.
14515         (thumb2_zero_extendhisi2_v6): Likewise.
14516         (thumb2_zero_extendqisi2_v6): Likewise.
14517         (thumb2_shiftsi3_short): Likewise.
14518         (thumb2_addsi3_compare0_scratch): Likewise.
14519         (orsi_not_shiftsi_si): Likewise.
14520         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
14521         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
14522         changes.
14523         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
14524         (1020alu_shift_op): Likewise.
14525         (1020alu_shift_reg_op): Likewise.
14526         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
14527         (alu_shift_op): Likewise.
14528         (alu_shift_reg_op): Likewise.
14529         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
14530         (11_alu_shift_op): Likewise.
14531         (11_alu_shift_reg_op): Likewise.
14532         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
14533         (9_alu_shift_reg_op): Likewise.
14534         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
14535         changes.
14536         (cortex_a15_alu_shift): Likewise.
14537         (cortex_a15_alu_shift_reg): Likewise.
14538         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
14539         changes.
14540         (cortex_a5_alu_shift): Likewise.
14541         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
14542         changes.
14543         (cortex_a53_alu_shift): Likewise.
14544         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
14545         changes.
14546         (cortex_a7_alu_reg): Likewise.
14547         (cortex_a7_alu_shift): Likewise.
14548         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
14549         changes.
14550         (cortex_a8_alu_shift): Likewise.
14551         (cortex_a8_alu_shift_reg): Likewise.
14552         (cortex_a8_mov): Likewise.
14553         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
14554         (cortex_a9_dp_shift): Likewise.
14555         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
14556         changes.
14557         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
14558         changes.
14559         (cortex_r4_mov): Likewise.
14560         (cortex_r4_alu_shift): Likewise.
14561         (cortex_r4_alu_shift_reg): Likewise.
14562         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
14563         (526_alu_shift_op): Likewise.
14564         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
14565         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
14566         (626te_alu_shift_op): Likewise.
14567         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
14568         (726te_alu_op): Likewise.
14569         (726te_alu_shift_op): Likewise.
14570         (726te_alu_shift_reg_op): Likewise.
14571         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
14572         (mp626_alu_shift_op): Likewise.
14573         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
14574         (pj4_alu_e1_conds): Likewise.
14575         (pj4_alu): Likewise.
14576         (pj4_alu_conds): Likewise.
14577         (pj4_shift): Likewise.
14578         (pj4_shift_conds): Likewise.
14579         (pj4_alu_shift): Likewise.
14580         (pj4_alu_shift_conds): Likewise.
14581         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
14582         changes.
14583         (cortexa7_older_only): Likewise.
14584         (cortexa7_younger): Likewise.
14586 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14588         * ipa-pure-const.c (generate_summary): Rename to...
14589         (pure_const_generate_summary): ... this.
14591 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
14593         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
14595 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
14597         PR target/57909
14598         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
14599         usage in HI mode.
14601 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14603         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
14604         enabled without -march=zEC12.
14605         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
14606         flags to be set.
14608 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
14610         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
14611         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
14612         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
14613         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
14614         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
14615         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
14616         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
14617         ISA_HAS_FP4.
14618         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
14619         and ISA_HAS_NMADD3_NMSUB3.
14620         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
14621         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
14622         (nmsub4<mode>, nmsub3<mode>): Likewise.
14623         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
14625 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
14627         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
14628         TARGET_MIPS5400 checking.
14630 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
14631             Peter Bergner  <bergner@vnet.ibm.com>
14633         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
14634         registers in the comment.
14635         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
14636         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
14637         rather than FIRST_PSEUDO_REGISTERS.
14639 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
14641         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
14642         enable extra ISA flags with TARGET_HTM.
14644 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
14646         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
14647         Fix comment typos.
14649 2013-07-15  Cong Hou  <congh@google.com>
14651         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
14652         in compare function for sorting.
14654 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
14656         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
14657         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
14658         * config/rs6000/rs6000.opt: Add -mhtm option.
14659         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
14660         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
14661         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
14662         __HTM__ if the HTM instructions are available.
14663         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
14664         htm_spr_reg_operand): New define_predicates.
14665         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
14666         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
14667         Include htm.md.
14668         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
14669         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
14670         HTM builtin functions.
14671         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
14672         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
14673         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
14674         (rs6000_builtin_mask_calculate): Likewise.
14675         (rs6000_option_override_internal): Likewise.
14676         (bdesc_htm): Add new HTM builtin support.
14677         (htm_spr_num): New function.
14678         (htm_spr_regno): Likewise.
14679         (rs6000_htm_spr_icode): Likewise.
14680         (htm_expand_builtin): Likewise.
14681         (htm_init_builtins): Likewise.
14682         (rs6000_expand_builtin): Add support for HTM builtin functions.
14683         (rs6000_init_builtins): Likewise.
14684         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
14685         option.
14686         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
14687         (TARGET_HTM, MASK_HTM): Define macros.
14688         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
14689         (FIXED_REGISTERS): Likewise.
14690         (CALL_USED_REGISTERS): Likewise.
14691         (CALL_REALLY_USED_REGISTERS): Likewise.
14692         (REG_ALLOC_ORDER): Likewise.
14693         (enum reg_class): Likewise.
14694         (REG_CLASS_NAMES): Likewise.
14695         (REG_CLASS_CONTENTS): Likewise.
14696         (REGISTER_NAMES): Likewise.
14697         (ADDITIONAL_REGISTER_NAMES): Likewise.
14698         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
14699         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
14700         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
14701         * config/rs6000/htm.md: New file.
14702         * config/rs6000/htmintrin.h: New file.
14703         * config/rs6000/htmxlintrin.h: New file.
14705 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
14707         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
14708         Define SYMBOL_TINY_GOT, update comment.
14709         * config/aarch64/aarch64.c
14710         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
14711         (aarch64_expand_mov_immediate): Likewise.
14712         (aarch64_print_operand): Likewise.
14713         (aarch64_classify_symbol): Likewise.
14714         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
14715         (ldr_got_tiny): Define.
14717 2013-07-13  Tobias Grosser  <tobias@grosser.es>
14719         PR tree-optimization/54094
14720         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
14721           scheduling dimension for the parallelism check from the polyhedral
14722           information in the AST.
14723         * graphite-dependences.c (carries_deps): Do not assume the schedule is
14724           in 2D + 1 form.
14726 2013-07-13  Jason Merrill  <jason@redhat.com>
14728         * print-tree.c (debug_vec_tree): Use debug_raw.
14729         (debug_raw (vec<tree, va_gc> &)): New.
14730         (debug_raw (vec<tree, va_gc> *)): New.
14731         * tree.h: Declare them.
14733 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
14735         * ifcvt.c (ifcvt_after_combine): New static variable.
14736         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
14737         for size.
14738         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
14739         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
14740         rest_of_handle_if_after_reload): Pass new argument for if_convert.
14742 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
14744         * config/mips/mips.c (mips_expand_call): Remove empty statement.
14746 2013-07-12  Michael Matz  <matz@suse.de>
14748         PR middle-end/55771
14749         * convert.c (convert_to_real): Reject non-float inner types.
14751 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
14753         * config/aarch64/aarch64-protos.h
14754         (aarch64_simd_immediate_valid_for_move): Remove.
14755         * config/aarch64/aarch64.c (simd_immediate_info): New member.
14756         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
14757         cases.
14758         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
14760 2013-07-11  Steve Ellcey  <sellcey@mips.com>
14762         * config/mips/mips.c (mips_conditional_register_usage): Do not
14763         use t[0-7] registers in MIPS16 mode when optimizing for size.
14765 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
14767         * config/i386/i386.c (dispatch_function_versions): Fix array
14768         indexing of function_version_info to match actual_versions.
14770 2013-07-11  Teresa Johnson  <tejohnson@google.com>
14772         * vec.h (struct va_gc): Move release out-of-line.
14773         (va_gc::release): Call ggc_free on released vec.
14775 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
14777         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
14778         Require GOT register as additional operand in UNSPEC.
14779         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
14780         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
14781         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
14782         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
14783         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
14784         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
14785         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
14787 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
14789         PR target/57631
14790         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
14791         name seen by assembler/linker rather if available.
14793 2013-07-11  Andreas Schwab  <schwab@suse.de>
14795         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
14797 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
14799         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
14801 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
14803         * doc/tm.texi.in: Move hook documentation to ....
14804         * target.def: ... here.
14806         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
14807         text on @hook line.
14808         * doc/tm.texi: Regenerate.
14810 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
14812         PR c++/57869
14813         * doc/invoke.texi: Document -Wconditionally-supported.
14815 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
14817         PR target/57844
14818         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
14819         of my_fp.
14821 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
14823         PR target/57506
14824         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
14825         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
14826         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
14827         Remove duplicate devices.
14828         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
14829         * config/avr/t-multilib: Regenerate.
14830         * config/avr/avr-tables.opt: Regenerate.
14831         * doc/avr-mmcu.texi: Regenerate.
14833 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
14835         PR target/56987
14836         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
14838 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
14840         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
14841         the cost of MULT when optimizing for size.
14843 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
14845         * config/cr16/cr16-protos.h: Don't include target.h.
14847 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
14849         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
14850         adjust register size for TDmode and TFmode for VSX registers.
14852 2013-07-08  Kai Tietz  <ktietz@redhat.com>
14854         PR target/56892
14855         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
14856         hook_bool_const_tree_true.
14858 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14860         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
14861         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
14862         * config/s390/s390.md: Define FPR*_REGNUM constants.
14863         Fix FPR2_REGNUM constant (18 -> 17).
14864         ("*trunc<BFP:mode><DFP_ALL:mode>2")
14865         ("*trunc<DFP_ALL:mode><BFP:mode>2")
14866         ("trunc<BFP:mode><DFP_ALL:mode>2")
14867         ("trunc<DFP_ALL:mode><BFP:mode>2")
14868         ("*extend<BFP:mode><DFP_ALL:mode>2")
14869         ("*extend<DFP_ALL:mode><BFP:mode>2")
14870         ("extend<BFP:mode><DFP_ALL:mode>2")
14871         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
14872         FPR4_REGNUM.
14874 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
14876         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
14878 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14880         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
14881         and cfun_fpr_bit_p to cfun_fpr_save_p.
14882         (s390_frame_area, s390_register_info, s390_frame_info)
14883         (s390_emit_prologue, s390_emit_epilogue)
14884         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
14885         register numbers.
14886         * config/s390/s390.h: Define *_REGNUM macros for floating point
14887         register numbers.
14889 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
14891         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
14893 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
14895         PR rtl-optimization/57786
14896         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
14897         and break out of the loop when it is set to false.
14899 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
14901         PR target/57819
14902         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
14903         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
14904         (const_int 63)) 0)).
14905         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
14906         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
14907         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
14909         PR rtl-optimization/57829
14910         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
14911         mask bits outside of mode are just sign-extension from mode to HWI.
14913 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
14915         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
14916         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
14917         adjust_address instead of change_address to keep info about alignment.
14918         (emit_strmov): Remove.
14919         (emit_memmov): New function.
14920         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
14921         (expand_movmem_epilogue): Likewise and return updated rtx for
14922         destination.
14923         (expand_constant_movmem_prologue): Likewise and return updated rtx for
14924         destination and source.
14925         (decide_alignment): Refactor, handle vector_loop.
14926         (ix86_expand_movmem): Likewise.
14927         (ix86_expand_setmem): Likewise.
14928         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
14930 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
14932         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
14933         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
14935 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
14937         * config/i386/sse.md (sse_movlhps): Change alternative 3
14938         of operand 2 to "m".
14940 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
14942         PR target/57807
14943         * config/i386/sse.md (iptr): New mode attribute.
14944         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
14945         (<sse>_vm<plusminus_insn><mode>3): Ditto.
14946         (<sse>_vmmul<mode>3): Ditto.
14947         (<sse>_vmdiv<mode>3): Ditto.
14948         (sse_vmrcpv4sf2): Ditto.
14949         (<sse>_vmsqrt<mode>2): Ditto.
14950         (sse_vmrsqrtv4sf2): Ditto.
14951         (<sse>_vm<code><mode>3): Ditto.
14952         (avx_vmcmp<mode>3): Ditto.
14953         (<sse>_vmmaskcmp<mode>3): Ditto.
14954         (<sse>_comi): Ditto.
14955         (<sse>_ucomi): Ditto.
14956         (*xop_vmfrcz_<mode>): Ditto.
14957         (*fmai_fmadd_<mode>): Ditto.
14958         (*fmai_fmsub_<mode>): Ditto.
14959         (*fmai_fnmadd_<mode>): Ditto.
14960         (*fmai_fnmsub_<mode>): Ditto.
14961         (*fma4i_vmfmadd_<mode>): Ditto.
14962         (*fma4i_vmfmsub_<mode>): Ditto.
14963         (*fma4i_vmfnmadd_<mode>): Ditto.
14964         (*fma4i_vmfnmsub_<mode>): Ditto.
14965         (*xop_vmfrcz_<mode>): Ditto.
14966         (sse_cvtps2pi): Ditto.
14967         (sse_cvttps2pi): Ditto.
14968         (sse_cvtss2si): Ditto.
14969         (sse_cvtss2si_2): Ditto.
14970         (sse_cvtss2siq_2): Ditto.
14971         (sse_cvttss2si): Ditto.
14972         (sse_cvttss2siq): Ditto.
14973         (sse_cvtsd2si): Ditto.
14974         (sse_cvtsd2si_2): Ditto.
14975         (sse_cvtsd2siq_2): Ditto.
14976         (sse_cvttsd2si): Ditto.
14977         (sse_cvttsd2siq): Ditto.
14978         (sse_cvtsd2ss): Ditto.
14979         (sse_cvtss2sd): Ditto.
14980         (avx2_pbroadcast<mode>): Ditto.
14981         (avx2_pbroadcast<mode>_1): Ditto.
14982         (*avx_vperm_broadcast_v4sf): Ditto.
14984         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
14985         (sse_movlhps): Ditto.
14986         (sse_storehps): Ditto.
14987         (sse_loadhps): Ditto.
14988         (sse_storelps): Ditto.
14989         (sse_loadlps): Ditto.
14990         (*vec_concatv4sf): Ditto.
14991         (*vec_interleave_highv2df): Ditto.
14992         (*vec_interleave_lowv2df): Ditto.
14993         (*vec_extractv2df_1_sse): Ditto.
14994         (*vec_extractv2df_0_sse): Ditto.
14995         (sse2_storelpd): Ditto.
14996         (sse2_loadlpd): Ditto.
14997         (sse2_movsd): Ditto.
14998         (*vec_concatv4si): Ditto.
14999         (vec_concatv2di): Ditto.
15001         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
15002         for Intel asm dialect.
15003         (mmx_punpcklwd): Ditto.
15004         (mmx_punpckldq): Ditto.
15006         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
15007         for intel assembler dialect.
15009 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
15011         PR target/29776
15012         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
15013         for BUILT_IN_C{LZ,LRSB}*.
15014         * tree.h (CASE_INT_FN): Add FN##IMAX case.
15015         * tree-vrp.c (extract_range_basic): Handle
15016         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
15017         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
15018         fall thru to code calling set_value*.
15019         * builtins.c (expand_builtin): Remove *IMAX cases.
15020         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
15021         if width is bigger than 2*HWI.
15023 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
15025         PR rtl-optimization/55342
15026         * lra-int.h (lra_subreg_reload_pseudos): New.
15027         * lra.c: Add undoing optional reloads to the block diagram.
15028         (lra_subreg_reload_pseudos): New.
15029         (lra_optional_reload_pseudos): Change comments.
15030         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
15031         lra_optional_reload_pseudos after undo transformations.
15032         * lra-assigns.c (pseudo_prefix_title): New.
15033         (lra_setup_reg_renumber): Use it.
15034         (spill_for): Ditto.  Check subreg reload pseudos too.
15035         (assign_by_spills): Consider subreg reload pseudos too.
15036         * lra-constraints.c (simplify_operand_subreg): Use
15037         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
15038         (curr_insn_transform): Recognize and do optional reloads.
15039         (undo_optional_reloads): New.
15040         (lra_undo_inheritance): Call undo_optional_reloads.
15042 2013-07-05  Thomas Quinot  <quinot@adacore.com>
15044         * tree-complex.c (expand_complex_operations_1): Fix typo.
15046 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
15048         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
15049         (tune_params): New member 'const vec_costs'.
15050         * config/aarch64/aarch64.c (generic_vector_cost): New.
15051         (generic_tunings): New member 'generic_vector_cost'.
15052         (aarch64_builtin_vectorization_cost): New.
15053         (aarch64_add_stmt_cost): New.
15054         (TARGET_VECTORIZE_ADD_STMT_COST): New.
15055         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
15057 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
15059         PR target/57777
15060         * config/i386/predicates.md (vsib_address_operand): Disallow
15061         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
15063 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
15065         PR middle-end/55030
15066         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
15067         expand_builtin_setjmp_receiver.
15068         (expand_label): Adjust, call expand_builtin_setjmp_receiver
15069         with NULL for the label parameter.
15070         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
15071         the frame-pointer.  Adjust comments.
15072         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
15073         only if LABEL is non-NULL.
15075 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
15077         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
15078         (ARM_ABI_AAPCS64): Ditto.
15079         (arm_abi): Ditto.
15080         (ARM_DEFAULT_ABI): Ditto.
15082 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
15084         * config/aarch64/aarch64-builtins.c
15085         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
15086         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
15087         (st1): Likewise.
15088         * config/aarch64/aarch64-simd.md
15089         (aarch64_ld1<VALL:mode>): New.
15090         (aarch64_st1<VALL:mode>): Likewise.
15091         * config/aarch64/arm_neon.h
15092         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
15094 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
15096         * config/i386/i386.c (gate_insert_vzeroupper): Check if
15097         target ISA is AVX.
15098         (ix86_option_override_internal):Turn on all -mavx target flags by
15099         default as they are dependent on AVX anyway.
15101 2013-07-02  Cary Coutant  <ccoutant@google.com>
15103         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
15104         deterministic hash.
15105         (loc_checksum_ordered): Likewise.
15106         (hash_loc_operands): Remove inline keyword.
15108 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
15110         PR tree-optimization/57741
15111         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
15112         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
15113         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
15114         Allow REAL_CST step_exprs if flag_associative_math.
15115         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
15117 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15119         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
15121 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15123         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
15125 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15127         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
15128         encoding.
15129         (iorsi3_insn): Likewise.
15130         (arm_xorsi3): Likewise.
15132 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
15134         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
15135         to ...
15136         (attribute "type"): ... here, and prefix with "wmmx_".
15137         (attribute "core_cycles"): Update for attribute changes.
15138         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
15139         (tbcstv4hi): Likewise.
15140         (tbcstv2si): Likewise.
15141         (iwmmxt_iordi3): Likewise.
15142         (iwmmxt_xordi3): Likewise.
15143         (iwmmxt_anddi3): Likewise.
15144         (iwmmxt_nanddi3): Likewise.
15145         (iwmmxt_arm_movdi): Likewise.
15146         (iwmmxt_movsi_insn): Likewise.
15147         (mov<mode>_internal): Likewise.
15148         (and<mode>3_iwmmxt): Likewise.
15149         (ior<mode>3_iwmmxt): Likewise.
15150         (xor<mode>3_iwmmxt): Likewise.
15151         (add<mode>3_iwmmxt): Likewise.
15152         (ssaddv8qi3): Likewise.
15153         (ssaddv4hi3): Likewise.
15154         (ssaddv2si3): Likewise.
15155         (usaddv8qi3): Likewise.
15156         (usaddv4hi3): Likewise.
15157         (usaddv2si3): Likewise.
15158         (sub<mode>3_iwmmxt): Likewise.
15159         (sssubv8qi3): Likewise.
15160         (sssubv4hi3): Likewise.
15161         (sssubv2si3): Likewise.
15162         (ussubv8qi3): Likewise.
15163         (ussubv4hi3): Likewise.
15164         (ussubv2si3): Likewise.
15165         (mulv4hi3_iwmmxt): Likewise.
15166         (smulv4hi3_highpart): Likewise.
15167         (umulv4hi3_highpart): Likewise.
15168         (iwmmxt_wmacs): Likewise.
15169         (iwmmxt_wmacsz): Likewise.
15170         (iwmmxt_wmacu): Likewise.
15171         (iwmmxt_wmacuz): Likewise.
15172         (iwmmxt_clrdi): Likewise.
15173         (iwmmxt_clrv8qi): Likewise.
15174         (iwmmxt_clr4hi): Likewise.
15175         (iwmmxt_clr2si): Likewise.
15176         (iwmmxt_uavgrndv8qi3): Likewise.
15177         (iwmmxt_uavgrndv4hi3): Likewise.
15178         (iwmmxt_uavgv8qi3): Likewise.
15179         (iwmmxt_uavgv4hi3): Likewise.
15180         (iwmmxt_tinsrb): Likewise.
15181         (iwmmxt_tinsrh): Likewise.
15182         (iwmmxt_tinsrw): Likewise.
15183         (iwmmxt_textrmub): Likewise.
15184         (iwmmxt_textrmsb): Likewise.
15185         (iwmmxt_textrmuh): Likewise.
15186         (iwmmxt_textrmsh): Likewise.
15187         (iwmmxt_textrmw): Likewise.
15188         (iwmxxt_wshufh): Likewise.
15189         (eqv8qi3): Likewise.
15190         (eqv4hi3): Likewise.
15191         (eqv2si3): Likewise.
15192         (gtuv8qi3): Likewise.
15193         (gtuv4hi3): Likewise.
15194         (gtuv2si3): Likewise.
15195         (gtv8qi3): Likewise.
15196         (gtv4hi3): Likewise.
15197         (gtv2si3): Likewise.
15198         (smax<mode>3_iwmmxt): Likewise.
15199         (umax<mode>3_iwmmxt): Likewise.
15200         (smin<mode>3_iwmmxt): Likewise.
15201         (umin<mode>3_iwmmxt): Likewise.
15202         (iwmmxt_wpackhss): Likewise.
15203         (iwmmxt_wpackwss): Likewise.
15204         (iwmmxt_wpackdss): Likewise.
15205         (iwmmxt_wpackhus): Likewise.
15206         (iwmmxt_wpackwus): Likewise.
15207         (iwmmxt_wpackdus): Likewise.
15208         (iwmmxt_wunpckihb): Likewise.
15209         (iwmmxt_wunpckihh): Likewise.
15210         (iwmmxt_wunpckihw): Likewise.
15211         (iwmmxt_wunpckilb): Likewise.
15212         (iwmmxt_wunpckilh): Likewise.
15213         (iwmmxt_wunpckilw): Likewise.
15214         (iwmmxt_wunpckehub): Likewise.
15215         (iwmmxt_wunpckehuh): Likewise.
15216         (iwmmxt_wunpckehuw): Likewise.
15217         (iwmmxt_wunpckehsb): Likewise.
15218         (iwmmxt_wunpckehsh): Likewise.
15219         (iwmmxt_wunpckehsw): Likewise.
15220         (iwmmxt_wunpckelub): Likewise.
15221         (iwmmxt_wunpckeluh): Likewise.
15222         (iwmmxt_wunpckeluw): Likewise.
15223         (iwmmxt_wunpckelsb): Likewise.
15224         (iwmmxt_wunpckelsh): Likewise.
15225         (iwmmxt_wunpckelsw): Likewise.
15226         (ror<mode>3): Likewise.
15227         (ashr<mode>3_iwmmxt): Likewise.
15228         (lshr<mode>3_iwmmxt): Likewise.
15229         (ashl<mode>3_iwmmxt): Likewise.
15230         (ror<mode>3_di): Likewise.
15231         (ashr<mode>3_di): Likewise.
15232         (lshr<mode>3_di): Likewise.
15233         (ashl<mode>3_di): Likewise.
15234         (iwmmxt_wmadds): Likewise.
15235         (iwmmxt_wmaddu): Likewise.
15236         (iwmmxt_tmia): Likewise.
15237         (iwmmxt_tmiaph): Likewise.
15238         (iwmmxt_tmiabb): Likewise.
15239         (iwmmxt_tmiatb): Likewise.
15240         (iwmmxt_tmiabt): Likewise.
15241         (iwmmxt_tmiatt): Likewise.
15242         (iwmmxt_tmovmskb): Likewise.
15243         (iwmmxt_tmovmskh): Likewise.
15244         (iwmmxt_tmovmskw): Likewise.
15245         (iwmmxt_waccb): Likewise.
15246         (iwmmxt_wacch): Likewise.
15247         (iwmmxt_waccw): Likewise.
15248         (iwmmxt_waligni): Likewise.
15249         (iwmmxt_walignr): Likewise.
15250         (iwmmxt_walignr0): Likewise.
15251         (iwmmxt_walignr1): Likewise.
15252         (iwmmxt_walignr2): Likewise.
15253         (iwmmxt_walignr3): Likewise.
15254         (iwmmxt_wsadb): Likewise.
15255         (iwmmxt_wsadh): Likewise.
15256         (iwmmxt_wsadbz): Likewise.
15257         (iwmmxt_wsadhz): Likewise.
15258         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
15259         (iwmmxt_wabsdiffb): Likewise.
15260         (iwmmxt_wabsdiffh): Likewise.
15261         (iwmmxt_wabsdiffw): Likewise.
15262         (iwmmxt_waddsubhx): Likewise
15263         (iwmmxt_wsubaddhx): Likewise.
15264         (addc<mode>3): Likewise.
15265         (iwmmxt_avg4): Likewise.
15266         (iwmmxt_avg4r): Likewise.
15267         (iwmmxt_wmaddsx): Likewise.
15268         (iwmmxt_wmaddux): Likewise.
15269         (iwmmxt_wmaddsn): Likewise.
15270         (iwmmxt_wmaddun): Likewise.
15271         (iwmmxt_wmulwsm): Likewise.
15272         (iwmmxt_wmulwum): Likewise.
15273         (iwmmxt_wmulsmr): Likewise.
15274         (iwmmxt_wmulumr): Likewise.
15275         (iwmmxt_wmulwsmr): Likewise.
15276         (iwmmxt_wmulwumr): Likewise.
15277         (iwmmxt_wmulwl): Likewise.
15278         (iwmmxt_wqmulm): Likewise.
15279         (iwmmxt_wqmulwm): Likewise.
15280         (iwmmxt_wqmulmr): Likewise.
15281         (iwmmxt_wqmulwmr): Likewise.
15282         (iwmmxt_waddbhusm): Likewise.
15283         (iwmmxt_waddbhusl): Likewise.
15284         (iwmmxt_wqmiabb): Likewise.
15285         (iwmmxt_wqmiabt): Likewise.
15286         (iwmmxt_wqmiatb): Likewise.
15287         (iwmmxt_wqmiatt): Likewise.
15288         (iwmmxt_wqmiabbn): Likewise.
15289         (iwmmxt_wqmiabtn): Likewise.
15290         (iwmmxt_wqmiatbn): Likewise.
15291         (iwmmxt_wqmiattn): Likewise.
15292         (iwmmxt_wmiabb): Likewise.
15293         (iwmmxt_wmiabt): Likewise.
15294         (iwmmxt_wmiatb): Likewise.
15295         (iwmmxt_wmiatt): Likewise.
15296         (iwmmxt_wmiabbn): Likewise.
15297         (iwmmxt_wmiabtn): Likewise.
15298         (iwmmxt_wmiatbn): Likewise.
15299         (iwmmxt_wmiattn): Likewise.
15300         (iwmmxt_wmiawbb): Likewise.
15301         (iwmmxt_wmiawbt): Likewise.
15302         (iwmmxt_wmiawtb): Likewise.
15303         (iwmmxt_wmiawtt): Likewise.
15304         (iwmmxt_wmiawbbn): Likewise.
15305         (iwmmxt_wmiawbtn): Likewise.
15306         (iwmmxt_wmiawtbn): Likewise.
15307         (iwmmxt_wmiawttn): Likewise.
15308         (iwmmxt_wmerge): Likewise.
15309         (iwmmxt_tandc<mode>3): Likewise.
15310         (iwmmxt_torc<mode>3): Likewise.
15311         (iwmmxt_torvsc<mode>3): Likewise.
15312         (iwmmxt_textrc<mode>3): Likewise.
15313         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
15314         (wmmxt_pack): Likewise.
15315         (wmmxt_mult_c1): Likewise.
15316         (wmmxt_mult_c2): Likewise.
15317         (wmmxt_alu_c1): Likewise.
15318         (wmmxt_alu_c2): Likewise.
15319         (wmmxt_alu_c3): Likewise.
15320         (wmmxt_transfer_c1): Likewise.
15321         (wmmxt_transfer_c2): Likewise.
15322         (wmmxt_transfer_c3): Likewise.
15323         (marvell_f_iwmmxt_wstr): Likewise.
15324         (marvell_f_iwmmxt_wldr): Likewise.
15326 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
15328         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
15330 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15332         Revert:
15333         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15334         * lra-constraints.c (need_for_split_p): Check call used hard regs
15335         living through calls.
15337         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
15338         call used regs for call insn.
15340 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
15342         PR target/57736
15343         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
15344         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
15346 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
15348         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
15350 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15352         * lra-constraints.c (need_for_split_p): Check call used hard regs
15353         living through calls.
15355 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
15357         PR target/57744
15358         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
15359         to tie with any other modes.  Eliminate Altivec vector mode tests,
15360         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
15361         code, to return 0 if testing MODE2 for a condition, if we've
15362         already tested MODE1 for the same condition.
15364 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15366         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
15367         layout.
15369 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15371         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15372         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
15374 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15376         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
15377         Define.
15378         (aarch64_symbolic_constant_p): Remove.
15379         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
15380         static.  Fix line length and white space.
15381         (aarch64_symbolic_constant_p): Remove.
15382         * config/aarch64/predicates.md (aarch64_valid_symref):
15383         Use aarch64_classify_symbol_expression.
15385 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15387         * config/arm/constraints.md (Ts): New constraint.
15388         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
15389         16-bit encodings.
15390         (compare_scc): Use "Ts" constraint for operand 0.
15391         (ior_scc_scc): Likewise.
15392         (and_scc_scc): Likewise.
15393         (and_scc_scc_nodom): Likewise.
15394         (ior_scc_scc_cmp): Likewise for operand 7.
15395         (and_scc_scc_cmp): Likewise.
15396         * config/arm/thumb2.md (thumb2_movsi_insn):
15397         Add alternatives for 16-bit encodings.
15398         (thumb2_movhi_insn): Likewise.
15399         (thumb2_movsicc_insn): Likewise.
15400         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
15401         (thumb2_negscc): Use "Ts" constraint.
15402         Move mvn instruction outside cond_exec block.
15403         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
15404         for 16-bit encodings.
15406 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15408         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
15409         encoding.
15410         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
15411         (mulsi3subsi): Likewise.
15412         (mulsidi3adddi): Likewise.
15413         (mulsidi3_v6): Likewise.
15414         (umulsidi3_v6): Likewise.
15415         (umulsidi3adddi_v6): Likewise.
15416         (smulsi3_highpart_v6): Likewise.
15417         (umulsi3_highpart_v6): Likewise.
15418         (mulhisi3tb): Likewise.
15419         (mulhisi3bt): Likewise.
15420         (mulhisi3tt): Likewise.
15421         (maddhisi4): Likewise.
15422         (maddhisi4tb): Likewise.
15423         (maddhisi4tt): Likewise.
15424         (maddhidi4): Likewise.
15425         (maddhidi4tb): Likewise.
15426         (maddhidi4tt): Likewise.
15427         (zeroextractsi_compare0_scratch): Likewise.
15428         (insv_zero): Likewise.
15429         (insv_t2): Likewise.
15430         (anddi_notzesidi_di): Likewise.
15431         (anddi_notsesidi_di): Likewise.
15432         (andsi_notsi_si): Likewise.
15433         (iordi_zesidi_di): Likewise.
15434         (xordi_zesidi_di): Likewise.
15435         (andsi_iorsi3_notsi): Likewise.
15436         (smax_0): Likewise.
15437         (smax_m1): Likewise.
15438         (smin_0): Likewise.
15439         (not_shiftsi): Likewise.
15440         (unaligned_loadsi): Likewise.
15441         (unaligned_loadhis): Likewise.
15442         (unaligned_loadhiu): Likewise.
15443         (unaligned_storesi): Likewise.
15444         (unaligned_storehi): Likewise.
15445         (extv_reg): Likewise.
15446         (extzv_t2): Likewise.
15447         (divsi3): Likewise.
15448         (udivsi3): Likewise.
15449         (arm_zero_extendhisi2addsi): Likewise.
15450         (arm_zero_extendqisi2addsi): Likewise.
15451         (compareqi_eq0): Likewise.
15452         (arm_extendhisi2_v6): Likewise.
15453         (arm_extendqisi2addsi): Likewise.
15454         (arm_movt): Likewise.
15455         (thumb2_ldrd): Likewise.
15456         (thumb2_ldrd_base): Likewise.
15457         (thumb2_ldrd_base_neg): Likewise.
15458         (thumb2_strd): Likewise.
15459         (thumb2_strd_base): Likewise.
15460         (thumb2_strd_base_neg): Likewise.
15461         (arm_negsi2): Add alternative for 16-bit encoding.
15462         (arm_one_cmplsi2): Likewise.
15464 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15466         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
15467         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
15468         (movdfcc): Likewise.
15469         * config/arm/vfp.md (*thumb2_movsf_vfp):
15470         Disable predication for arm_restrict_it.
15471         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
15472         (*thumb2_movdfcc_vfp): Likewise.
15473         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
15474         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
15475         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
15476         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
15477         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
15478         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
15479         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
15480         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
15481         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
15482         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
15483         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
15484         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
15485         Disable predication for arm_restrict_it.
15487 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
15489         * config/i386/bmiintrin.h (_bextr_u32): New.
15490         (_bextr_u64): Ditto.
15492 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
15494         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
15495         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
15496         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
15497         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
15498         * config/mips/n32-elf.h: ...this new file.
15500 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
15502         PR target/57224
15503         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
15504         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
15506 2013-06-27  Catherine Moore  <clm@codesourcery.com>
15508         * config/mips/mips-tables.opt: Regenerate.
15509         * config/mips/mips-cpus.def: Add m14ke and m14kec.
15510         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
15511         * doc/invoke.texi: Add -m14kc.
15513 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
15515         PR target/57623
15516         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
15517         constraints of operand 1 and 2.
15519         PR target/57623
15520         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
15521         to match RTL canonicalization.  Swap predicates and
15522         constraints of operand 1 and 2.
15524 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
15526         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
15527         Process OP_INOUT regs for splitting too.
15529 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
15531         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
15532         decl before the loop, initialize to NULL.
15533         (vectorizable_load): Initialize ptr_incr to NULL.
15535 2013-06-27  Martin Jambor  <mjambor@suse.cz>
15537         PR lto/57208
15538         * ipa-ref.h (ipa_maybe_record_reference): Declare.
15539         * ipa-ref.c (ipa_maybe_record_reference): New function.
15540         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
15541         * ipa-cp.c (create_specialized_node): Record potential references from
15542         aggvals.
15543         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
15545 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
15547         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
15548         parameter 'mode' of type 'enum machine_mode mode'; change to pass
15549         'mode' to force_reg.
15550         (aarch64_add_offset): Update calls to aarch64_force_temporary.
15551         (aarch64_expand_mov_immediate): Likewise.
15553 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
15555         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
15556         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
15558 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15560         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
15561         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
15562         (struct machine_function): Add tbegin_p.
15563         (s390_canonicalize_comparison): Fold CC mode compares to
15564         conditional jump if possible.
15565         (s390_emit_jump): Return the emitted jump.
15566         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
15567         Handle CCRAWmode compares.
15568         (s390_option_override): Default to -mhtm if available.
15569         (s390_reg_clobbered_rtx): Handle floating point regs as well.
15570         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
15571         FPRs instead of df_regs_ever_live_p.
15572         (s390_optimize_nonescaping_tx): New function.
15573         (s390_init_frame_layout): Extend clobbered_regs array to cover
15574         FPRs as well.
15575         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
15576         (s390_expand_tbegin): New function.
15577         (enum s390_builtin): New enum definition.
15578         (code_for_builtin): New array definition.
15579         (s390_init_builtins): New function.
15580         (s390_expand_builtin): New function.
15581         (TARGET_INIT_BUILTINS): Define.
15582         (TARGET_EXPAND_BUILTIN): Define.
15583         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
15584         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
15585         (s390_alc_comparison): Likewise.
15586         * config/s390/s390-modes.def: Add CCRAWmode.
15587         * config/s390/s390.h (processor_flags): Add PF_TX.
15588         (TARGET_CPU_HTM): Define macro.
15589         (TARGET_HTM): Define macro.
15590         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
15591         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
15592         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
15593         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
15594         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
15595         (TBEGIN_MASK, TBEGINC_MASK): New constants.
15596         ("*cc_to_int"): Move up.
15597         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
15598         constants other than 0.
15599         ("*ccraw_to_int"): New insn and splitter definition.
15600         ("tbegin", "tbegin_nofloat", "tbegin_retry")
15601         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
15602         ("tx_assist"): New expander.
15603         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
15604         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
15605         * config/s390/s390.opt: Add -mhtm option.
15606         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
15607         * config/s390/htmxlintrin.h: New file.
15608         * config/s390/htmintrin.h: New file.
15609         * config/s390/s390intrin.h: New file.
15610         * doc/extend.texi: Document htm builtins.
15611         * config.gcc: Add the new header files to extra_headers.
15613 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
15615         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
15616         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
15618 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
15619             Pat Haugen  <pthaugen@us.ibm.com>
15620             Peter Bergner  <bergner@vnet.ibm.com>
15622         * config/rs6000/power8.md: New.
15623         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
15624         setting for power8 entry.
15625         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
15626         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
15627         test for Power4/Power5 only.
15628         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
15629         support.
15630         (force_new_group): Adjust comment.
15631         * config/rs6000/rs6000.md: Include power8.md.
15633 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
15635         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
15636         * config/arm/arm-protos.h (arm_max_conditional_execute): New
15637         declaration.
15638         (tune_params): Update comment.
15639         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
15640         (arm_max_conditional_execute): New function.
15641         (thumb2_final_prescan_insn): Use max_insn_skipped and
15642         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
15644 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
15646         PR tree-optimization/57705
15647         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
15648         SSA_NAME step, provided that it is not defined inside the loop.
15649         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
15650         (get_initial_def_for_induction): Handle SSA_NAME IV step.
15652 2013-06-25  Martin Jambor  <mjambor@suse.cz>
15654         PR middle-end/57670
15655         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
15656         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
15657         calls in the dump.
15658         (ipa_note_param_call): Initialize member_ptr flag.
15659         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
15660         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
15661         (ipa_write_indirect_edge_info): Stream member_ptr flag.
15662         (ipa_read_indirect_edge_info): Likewise.
15664 2013-06-25  Richard Biener  <rguenther@suse.de>
15666         PR middle-end/56977
15667         * passes.c (init_optimization_passes): Move pass_fold_builtins
15668         and pass_dce earlier with -Og.
15670 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
15672         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
15673         <BIT_FIELD_REF>: Remove trailing TAB.
15674         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
15675         remove blank line.
15677 2013-06-24  Martin Jambor  <mjambor@suse.cz>
15679         PR tree-optimization/57358
15680         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
15681         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
15682         (ipa_analyze_params_uses): Generate pessimistic info when true.
15684 2013-06-24  Martin Jambor  <mjambor@suse.cz>
15686         PR tree-optimization/57539
15687         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
15688         global.inlined_to of the new node to it.  All callers changed.
15689         * ipa-inline-transform.c (clone_inlined_nodes): New variable
15690         inlining_into, pass it to cgraph_clone_node.
15691         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
15692         ipa_free_edge_args_substructures.
15693         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
15694         rdesc linked list.  Do not assert rdesc edges have inlined caller.
15695         Assert we have found an rdesc in the rdesc list.
15697 2013-06-24  Richard Biener  <rguenther@suse.de>
15699         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
15700         (pointer_set_lookup): Declare.
15701         (class pointer_map): New template class implementing a
15702         generic pointer to T map.
15703         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
15704         pointer_map<T>::contains, pointer_map<T>::insert,
15705         pointer_map<T>::traverse): New functions.
15706         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
15707         (pointer_set_lookup): New function.
15708         (pointer_set_contains): Use pointer_set_lookup.
15709         (pointer_set_insert): Likewise.
15710         (insert_aux): Remove.
15711         (struct pointer_map_t): Embed a pointer_set_t.
15712         (pointer_map_create): Adjust.
15713         (pointer_map_destroy): Likewise.
15714         (pointer_map_contains): Likewise.
15715         (pointer_map_insert): Likewise.
15716         (pointer_map_traverse): Likewise.
15717         * tree-streamer.h (struct streamer_tree_cache_d): Use a
15718         pointer_map<unsigned> instead of a pointer_map_t.
15719         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
15720         (streamer_tree_cache_lookup): Likewise.
15721         (streamer_tree_cache_create): Likewise.
15722         (streamer_tree_cache_delete): Likewise.
15723         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
15724         pointer_map<unsigned> instead of a pointer_map_t.
15725         (lto_init_tree_ref_encoder): Adjust.
15726         (lto_destroy_tree_ref_encoder): Likewise.
15727         * lto-section-out.c (lto_output_decl_index): Likewise.
15728         (lto_record_function_out_decl_state): Likewise.
15729         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
15731 2013-06-24  Richard Biener  <rguenther@suse.de>
15733         PR tree-optimization/57488
15734         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
15736 2013-06-24  Alan Modra  <amodra@gmail.com>
15738         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
15739         (gen_easy_altivec_constant): Likewise.
15740         * config/rs6000/predicates.md (easy_vector_constant_add_self,
15741         easy_vector_constant_msb): Likewise.
15743 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
15745         PR target/57688
15746         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
15747         add missing return true.
15749 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
15751         PR target/52483
15752         * config/sh/predicates.md (general_extend_operand): Invoke
15753         general_movsrc_operand for memory operands.
15754         (general_movsrc_operand): Allow reg+reg addressing, do not use
15755         general_operand for memory operands.
15757 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
15759         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
15760         when current target options does not apply.
15761         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
15762         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
15763         * config/i386/bmiintrin.h: Pass appropriate target
15764         attributes to header.
15765         * config/i386/mmintrin.h: Ditto.
15766         * config/i386/nmmintrin.h: Ditto.
15767         * config/i386/avx2intrin.h: Ditto.
15768         * config/i386/fxsrintrin.h: Ditto.
15769         * config/i386/tbmintrin.h: Ditto.
15770         * config/i386/xsaveintrin.h: Ditto.
15771         * config/i386/f16cintrin.h: Ditto.
15772         * config/i386/xtestintrin.h: Ditto.
15773         * config/i386/xsaveoptintrin.h: Ditto.
15774         * config/i386/bmi2intrin.h: Ditto.
15775         * config/i386/lzcntintrin.h: Ditto.
15776         * config/i386/smmintrin.h: Ditto.
15777         * config/i386/wmmintrin.h: Ditto.
15778         * config/i386/x86intrin.h: Remove all header include guards.
15779         * config/i386/prfchwintrin.h: Ditto.
15780         * config/i386/pmmintrin.h: Ditto.
15781         * config/i386/tmmintrin.h: Ditto.
15782         * config/i386/xmmintrin.h: Ditto.
15783         * config/i386/popcntintrin.h: Ditto.
15784         * config/i386/rdseedintrin.h: Ditto.
15785         * config/i386/ammintrin.h: Ditto.
15786         * config/i386/emmintrin.h: Ditto.
15787         * config/i386/immintrin.h: Remove all header include guards.
15788         * config/i386/fma4intrin.h: Ditto.
15789         * config/i386/lwpintrin.h: Ditto.
15790         * config/i386/xopintrin.h: Ditto.
15791         * config/i386/ia32intrin.h: Ditto.
15792         * config/i386/avxintrin.h: Ditto.
15793         * config/i386/rtmintrin.h: Ditto.
15794         * config/i386/fmaintrin.h: Ditto.
15795         * config/i386/mm3dnow.h: Ditto.
15797 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
15799         * common/config/i386/i386-common.c: Handle LZCNT.
15801 2013-06-22  Andi Kleen  <ak@linux.intel.com>
15803         * doc/extend.texi: Use __atomic_store_n instead of
15804         __atomic_store in HLE example.
15806 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
15808         * config/sh/sh.c: Remove <cstdlib> workaround.
15810 2013-06-21  Andi Kleen  <ak@linux.intel.com>
15812         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
15814 2013-06-21  Andi Kleen  <ak@linux.intel.com>
15816         * doc/extend.texi: Document that __atomic_clear and
15817           __atomic_test_and_set should only be used with bool.
15819 2013-06-20  Jan Hubicka  <jh@suse.cz>
15821         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
15822         types_same_for_odr.
15823         * tree.c (decls_same_for_odr): New function.
15824         (same_for_edr): New function.
15825         (types_same_for_odr): New function.
15826         (get_binfo_at_offset): Use it.
15827         * tree.h (types_same_for_odr): Declare.
15829 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
15830             Jason Merrill  <jason@redhat.com>
15832         * system.h: Include <cstdlib> as well as <stdlib.h>.
15834 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
15836         PR target/57655
15837         * config/i386/i386.c (construct_container): Report error if
15838         long double is used with disabled x87 float returns.
15840 2013-06-20  Jan Hubicka  <jh@suse.cz>
15842         * lto-cgraph.c (input_symtab): Do not set cgraph state.
15844 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
15846         PR rtl-optimization/57425
15847         PR rtl-optimization/57569
15848         * alias.c (write_dependence_p): Remove parameters mem_mode and
15849         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
15850         Changed all callers.
15851         (canon_anti_dependence): Get comments and semantics in sync.
15852         Add parameter mem_canonicalized.  Changed all callers.
15853         * rtl.h (canon_anti_dependence): Update prototype.
15855 2013-06-20  Richard Biener  <rguenther@suse.de>
15857         * data-streamer-in.c (streamer_read_uhwi): Optimize single
15858         byte case, inline streamer_read_uchar and defer section
15859         overrun check.
15861 2013-06-20  Richard Biener  <rguenther@suse.de>
15863         PR tree-optimization/57584
15864         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
15865         SSA names into the expanded expression that take part in
15866         abnormal coalescing.
15868 2013-06-19  Sharad Singhai  <singhai@google.com>
15870         * gcov.c (print_usage): Handle new option.
15871         (process_args): Ditto.
15872         (get_gcov_intermediate_filename): New function.
15873         (output_intermediate_file): New function.
15874         (output_gcov_file): New function
15875         (generate_results): Handle new option.
15876         (release_function): Relase demangled name.
15877         (read_graph_file): Handle demangled name.
15878         (output_lines): Ditto.
15879         * doc/gcov.texi: Document gcov intermediate format.
15881 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
15883         PR bootstrap/57604
15884         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
15885         (lra_emit_add): Use the functions.  Add comment about Y as an
15886         address segment.
15888 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
15890         PR driver/57652
15891         * collect2.c (collect_atexit): New.
15892         (collect_exit): Delete.
15893         (main): Register collect_atexit with atexit.
15894         (collect_wait): Change collect_exit to exit.
15895         (do_wait): Same.
15896         * collect2.h (collect_exit): Delete.
15897         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
15899 2013-06-19  Wei Mi  <wmi@google.com>
15901         PR rtl-optimization/57518
15902         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
15903         if regno is used in paradoxical subreg.
15904         (update_equiv_regs): Check pdx_subregs[regno] before
15905         set a reg to be equivalent with a mem.
15907 2013-06-19  Matthias Klose  <doko@ubuntu.com>
15909         PR driver/57651
15910         * file-find.h (find_a_file): Add a mode parameter.
15911         * file-find.c (find_a_file): Likewise.
15912         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
15913         with X_OK for the executables.
15914         * collect2.c (main): Call find_a_file with X_OK.
15916 2013-06-19  Steve Ellcey  <sellcey@mips.com>
15918         PR target/56942
15919         * config/mips/mips.md (casesi_internal_mips16_<mode>):
15920         Use NEXT_INSN instead of next_real_insn.
15922 2013-06-19  Jan Hubicka  <jh@suse.cz>
15924         * cgraph.h (const_value_known_p): Replace by ...
15925         (ctor_for_folding): .. this one.
15926         * cgraphunit.c (process_function_and_variable_attributes): Use it.
15927         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
15928         * expr.c (expand_expr_real_1): Likewise.
15929         (string_constant): Likewise.
15930         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
15931         * ipa.c (process_references): Likewise.
15932         (symtab_remove_unreachable_nodes): Likewise.
15933         * ipa-inline-analysis.c (param_change_prob): Likewise.
15934         * gimple-fold.c (canonicalize_constructor_val): Likewise.
15935         (get_base_constructor): Likwise.
15936         * varpool.c (varpool_remove_node): Likewise.
15937         (varpool_remove_initializer): LIkewise.
15938         (dump_varpool_node): LIkwise.
15939         (const_value_known_p): Rewrite to ...
15940         (ctor_for_folding): ... this one.
15942 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
15944         PR driver/57651
15945         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
15946         PERSONALITY in $PATH derived prefixes.
15948 2013-06-19  Jeff Law  <law@redhat.com>
15950         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
15951         in comment.
15953         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
15954         (simplify_bitwise_binary): Use it to simpify certain binary ops on
15955         booleans.
15957 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
15959         * config/arm/vfp.md: Move VFP instruction classification documentation
15960         to ...
15961         * config/arm/arm.md: ... here.  Update instruction classification
15962         documentation.
15964 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
15966         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
15967         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
15968         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
15970 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
15972         PR target/57609
15973         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
15974         with NEXT_INSN.  Use tablejump_p to check for jump table data
15975         insns.
15977 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
15979         PR c++/56544
15980         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
15981         that now in C++ the value is correct per the C++ standards.
15983 2013-06-19  Richard Biener  <rguenther@suse.de>
15985         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
15986         for global context.
15988 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15990         Revert:
15991         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15993         PR target/57609
15994         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
15995         with next_active_insn.
15997 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
15999         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
16000         functions are inlined during failures to flag an error.
16001         * tree-inline.c (expand_call_inline): Allow the error to be flagged
16002         in early inline pass.
16004 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
16006         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
16007         in comments.
16009 2013-06-18  Julian Brown  <julian@codesourcery.com>
16011         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
16012         Permit virtual register pre-reload if !strict.
16013         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
16014         change.
16015         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
16016         prototype.
16017         * config/arm/neon.md (movmisalign<mode>): Use
16018         neon_perm_struct_or_reg_operand instead of
16019         neon_struct_or_register_operand.
16020         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
16021         neon_permissive_struct_operand instead of neon_struct_operand.
16022         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
16023         neon_vector_mem_operand.
16024         * config/arm/predicates.md (neon_struct_operand): Adjust call to
16025         neon_vector_mem_operand.
16026         (neon_permissive_struct_operand): New.
16027         (neon_struct_or_register_operand): Rename to...
16028         (neon_perm_struct_or_reg_operand): This. Adjust call to
16029         neon_vector_mem_operand.
16031 2013-06-18  Richard Biener  <rguenther@suse.de>
16033         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
16034         * lto-streamer.h: Include pointer-set.h.
16035         (struct lto_decl_slot): Remove.
16036         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
16037         Remove next_index entry.
16038         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16039         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16040         (lto_init_tree_ref_encoder): Adjust.
16041         (lto_destroy_tree_ref_encoder): Likewise.
16042         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16043         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16044         (lto_output_decl_index): Adjust.
16045         (lto_new_out_decl_state): Likewise.
16046         (lto_record_function_out_decl_state): Likewise.
16047         * lto-streamer-out.c (copy_function): Likewise.
16049 2013-06-18  Richard Biener  <rguenther@suse.de>
16051         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
16052         * cgraphunit.c: Include cfgloop.h.
16053         (init_lowered_empty_function): Initialize the loop tree.
16054         (assemble_thunk): Insert new BBs into loops.
16056 2013-06-18  Richard Biener  <rguenther@suse.de>
16058         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
16059         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
16060         the map from cache entry to cache index optional.
16061         (streamer_tree_cache_replace_tree): Adjust accordingly.
16062         (streamer_tree_cache_append): Likewise.
16063         (streamer_tree_cache_delete): Likewise.
16064         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
16065         streamer cache map from cache entry to cache index.
16066         * lto-streamer-out.c (create_output_block): Adjust.
16068 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
16070         * config/arm/arm.md (attribute "insn"): Move multiplication and
16071         division attributes to...
16072         (attribute "type"): ... here.  Remove mult.
16073         (attribute "mul32"): New attribute.
16074         (attribute "mul64"): Add umaal.
16075         (*arm_mulsi3): Update attributes.
16076         (*arm_mulsi3_v6): Likewise.
16077         (*thumb_mulsi3): Likewise.
16078         (*thumb_mulsi3_v6): Likewise.
16079         (*mulsi3_compare0): Likewise.
16080         (*mulsi3_compare0_v6): Likewise.
16081         (*mulsi_compare0_scratch): Likewise.
16082         (*mulsi_compare0_scratch_v6): Likewise.
16083         (*mulsi3addsi): Likewise.
16084         (*mulsi3addsi_v6): Likewise.
16085         (*mulsi3addsi_compare0): Likewise.
16086         (*mulsi3addsi_compare0_v6): Likewise.
16087         (*mulsi3addsi_compare0_scratch): Likewise.
16088         (*mulsi3addsi_compare0_scratch_v6): Likewise.
16089         (*mulsi3subsi): Likewise.
16090         (*mulsidi3adddi): Likewise.
16091         (*mulsi3addsi_v6): Likewise.
16092         (*mulsidi3adddi_v6): Likewise.
16093         (*mulsidi3_nov6): Likewise.
16094         (*mulsidi3_v6): Likewise.
16095         (*umulsidi3_nov6): Likewise.
16096         (*umulsidi3_v6): Likewise.
16097         (*umulsidi3adddi): Likewise.
16098         (*umulsidi3adddi_v6): Likewise.
16099         (*smulsi3_highpart_nov6): Likewise.
16100         (*smulsi3_highpart_v6): Likewise.
16101         (*umulsi3_highpart_nov6): Likewise.
16102         (*umulsi3_highpart_v6): Likewise.
16103         (mulhisi3): Likewise.
16104         (*mulhisi3tb): Likewise.
16105         (*mulhisi3bt): Likewise.
16106         (*mulhisi3tt): Likewise.
16107         (maddhisi4): Likewise.
16108         (*maddhisi4tb): Likewise.
16109         (*maddhisi4tt): Likewise.
16110         (maddhidi4): Likewise.
16111         (*maddhidi4tb): Likewise.
16112         (*maddhidi4tt): Likewise.
16113         (divsi3): Likewise.
16114         (udivsi3): Likewise.
16115         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
16116         (thumb2_mulsi_short_compare0): Likewise.
16117         (thumb2_mulsi_short_compare0_scratch): Likewise.
16118         * config/arm/arm1020e.md (1020mult1): Update attribute change.
16119         (1020mult2): Likewise.
16120         (1020mult3): Likewise.
16121         (1020mult4): Likewise.
16122         (1020mult5): Likewise.
16123         (1020mult6): Likewise.
16124         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
16125         change.
16126         (cortex_a15_mult64): Likewise.
16127         (cortex_a15_sdiv): Likewise.
16128         (cortex_a15_udiv): Likewise.
16129         * config/arm/arm1026ejs.md (mult1): Update attribute change.
16130         (mult2): Likewise.
16131         (mult3): Likewise.
16132         (mult4): Likewise.
16133         (mult5): Likewise.
16134         (mult6): Likewise.
16135         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
16136         (pj4_ir_div): Likewise.
16137         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
16138         (11_mult2): Likewise.
16139         (11_mult3): Likewise.
16140         (11_mult4): Likewise.
16141         (11_mult5): Likewise.
16142         (11_mult6): Likewise.
16143         (11_mult7): Likewise.
16144         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
16145         (cortex_a8_mla): Likewise.
16146         (cortex_a8_mull): Likewise.
16147         (cortex_a8_smulwy): Likewise.
16148         (cortex_a8_smlald): Likewise.
16149         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
16150         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
16151         (cortex_r4_mul_3): Likewise.
16152         (cortex_r4_mla_4): Likewise.
16153         (cortex_r4_mla_3): Likewise.
16154         (cortex_r4_smlald): Likewise.
16155         (cortex_r4_mull): Likewise.
16156         (cortex_r4_sdiv): Likewise.
16157         (cortex_r4_udiv): Likewise.
16158         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
16159         (cortex_a7_idiv): Likewise.
16160         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
16161         (9_mult2): Likewise.
16162         (9_mult3): Likewise.
16163         (9_mult4): Likewise.
16164         (9_mult5): Likewise.
16165         (9_mult6): Likewise.
16166         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
16167         (cortex_a53_sdiv): Likewise.
16168         (cortex_a53_udiv): Likewise.
16169         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
16170         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
16171         (mp626_mult2): Likewise.
16172         (mp626_mult3): Likewise.
16173         (mp626_mult4): Likewise.
16174         * config/arm/fa526.md (526_mult1): Update attribute change.
16175         (526_mult2): Likewise.
16176         * config/arm/arm-generic.md (mult): Update attribute change.
16177         (mult_ldsched_strongarm): Likewise.
16178         (mult_ldsched): Likewise.
16179         (multi_cycle): Likewise.
16180         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
16181         * config/arm/fa606te.md (606te_mult1): Update attribute change.
16182         (606te_mult2): Likewise.
16183         (606te_mult3): Likewise.
16184         (606te_mult4): Likewise.
16185         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
16186         (cortex_a9_mac16): Likewise.
16187         (cortex_a9_multiply): Likewise.
16188         (cortex_a9_mac): Likewise.
16189         (cortex_a9_multiply_long): Likewise.
16190         * config/arm/fa626te.md (626te_mult1): Update attribute change.
16191         (626te_mult2): Likewise.
16192         (626te_mult3): Likewise.
16193         (626te_mult4): Likewise.
16195 2013-06-18  Richard Biener  <rguenther@suse.de>
16197         PR lto/57334
16198         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
16200 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16202         PR target/57609
16203         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16204         with next_active_insn.
16206 2013-06-18  Alan Modra  <amodra@gmail.com>
16208         * config/rs6000/rs6000.h (enum data_align): New.
16209         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
16210         (DATA_ABI_ALIGNMENT): Define.
16211         (CONSTANT_ALIGNMENT): Correct comment.
16212         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
16213         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
16215 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
16217         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
16218         ATTRIBUTE_UNUSED marking.
16220 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
16222         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
16223         alternative and update.
16224         (aarch64_dup_lanedi): Delete.
16225         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
16226         * config/aarch64/aarch64-simd-builtins.def: Update.
16228 2013-06-17  Richard Biener  <rguenther@suse.de>
16230         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
16231         (lto_input_scc): Declare.
16232         (lto_input_tree_1): Likewise.
16233         (struct lto_stats_d): Add num_tree_bodies_output and
16234         num_pickle_refs_output.
16235         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
16236         (lto_read_tree_1): Split out from ...
16237         (lto_read_tree): ... this.
16238         (lto_input_scc): New function.
16239         (lto_input_tree_1): Split out from ...
16240         (lto_input_tree): ... this.  Handle LTO_tree_scc.
16241         (lto_data_in_create): Create the streamer cache without hashes.
16242         * lto-streamer-out.c (create_output_block): Create the streamer
16243         cache with hashes when not doing WPA.
16244         (lto_write_tree_1): Split out from ...
16245         (lto_write_tree): ... this.
16246         (get_symbol_initial_value): New function.
16247         (lto_output_tree_1): Split out from ...
16248         (lto_output_tree): ... this.  Write trees as series of SCCs
16249         using a DFS walk via DFS_write_tree.
16250         (struct sccs, struct scc_entry): New types.
16251         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
16252         (DFS_write_tree_body): New function.
16253         (DFS_write_tree): Likewise.
16254         (hash_tree): Likewise.
16255         (scc_entry_compare): Likewise.
16256         (hash_scc): Likewise.
16257         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
16258         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
16259         TREE_CHAIN as regular reference.
16260         (streamer_read_integer_cst): Remove.
16261         (streamer_get_pickled_tree): Adjust.
16262         * tree-streamer-out.c (streamer_write_chain): Disable streaming
16263         of DECL_EXTERNALs in BLOCK_VARS for now.
16264         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
16265         reference.
16266         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
16267         Add hash value argument and record that if hashes are recorded
16268         in the cache.
16269         (streamer_tree_cache_insert_1): Adjust.
16270         (streamer_tree_cache_insert): Likewise.
16271         (streamer_tree_cache_insert_at): Rename to ...
16272         (streamer_tree_cache_replace_tree): ... this and adjust.
16273         (streamer_tree_cache_append): Adjust.
16274         (record_common_node): Likewise.
16275         (streamer_tree_cache_create): Add argument whether to
16276         record hash values together with trees.
16277         (streamer_tree_cache_delete): Adjust.
16278         * tree-streamer.h (struct streamer_tree_cache_d): Add
16279         vector of hashes.
16280         (streamer_read_integer_cst): Remove.
16281         (streamer_tree_cache_insert): Adjust.
16282         (streamer_tree_cache_append): Likewise.
16283         (streamer_tree_cache_insert_at): Rename to ...
16284         (streamer_tree_cache_replace_tree): ... this and adjust.
16285         (streamer_tree_cache_create): Add argument whether to record hashes.
16286         (streamer_tree_cache_get): Rename to ...
16287         (streamer_tree_cache_get_tree): ... this.
16288         (streamer_tree_cache_get_hash): New function.
16289         * tree.c (cache_integer_cst): New function.
16290         * tree.h (cache_integer_cst): Declare.
16291         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
16292         * lto-symtab.c (lto_varpool_replace_node): Only release
16293         DECL_INITIAL of non-prevailing decls.
16294         * varpool.c (varpool_remove_initializer): Do not release
16295         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
16297 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
16299         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
16300         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
16301         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
16302         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
16303         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
16304         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
16305         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
16306         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
16307         instead of TARGET_64BIT.
16308         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
16309         Require ISA_HAS_<D>DIV.
16311 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
16313         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
16314         (mips*-*-linux*): Move default with_llsc setting to where other
16315         defaults are set.
16316         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
16317         with_arch block.
16318         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
16319         Likewise.  Remove default with_tune setting.  Move default float
16320         setting to its own block.  Handle with_llsc in the same block as above.
16322 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
16324         PR rtl-optimization/57425
16325         PR rtl-optimization/57569
16326         * alias.c (write_dependence_p): Add new parameters mem_mode,
16327         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
16328         Changed all callers.
16329         (canon_anti_dependence): New function.
16330         * cse.c (check_dependence): Use canon_anti_dependence.
16331         * cselib.c (cselib_invalidate_mem): Likewise.
16332         * rtl.h (canon_anti_dependence): Declare.
16334 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
16336         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
16337         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
16338         ".set mips3" for 64-bit targets.
16340 2013-06-15  Dehao Chen  <dehao@google.com>
16342         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
16343         * gimple-low.c (gimple_check_call_matching_types): Likewise.
16344         (gimple_check_call_args): Likewise.
16345         * value-prof.c (check_ic_target): Likewise.
16346         * ipa-inline.c (early_inliner): Likewise.
16347         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
16348         * cgraph.c (cgraph_create_edge_1): Likewise.
16349         (cgraph_make_edge_direct): Likewise.
16351 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
16353         PR target/57615
16354         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
16355         rs6000_output_move_128bit to handle emitting quad memory
16356         operations.  Set attribute length to 8 bytes.
16358 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
16360         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
16361         New pattern.
16362         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
16363         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
16364         (aarch64_<su>mlsl<mode>): Likewise.
16366 2013-06-14  Mike Stump  <mikestump@comcast.net>
16368         * Makefile.in (TARGET_H): Add insn-codes.h.
16370 2013-06-14  Alan Modra  <amodra@gmail.com>
16372         PR middle-end/57134
16373         PR middle-end/57586
16374         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
16375         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
16376         bitfield expansion when EXPAND_MEMORY.
16377         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
16379 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
16381         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
16382         test for clearing quad memory on 32-bit later.
16384 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
16386         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
16387         (fold_negate_expr): Likewise.
16388         (fold_real_zero_addition_p): Handle vectors.
16389         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
16391 2013-06-14  Alan Modra  <amodra@gmail.com>
16393         * varasm.c (force_const_mem): Revert 2013-06-07 change.
16395 2013-06-13  Jan Hubicka  <jh@suse.cz>
16397         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
16398         Local comdats are not externally visible.
16399         * symtab.c (dump_symtab_base): Dump externally visible.
16400         (verify_symtab_base): Verify back links in the symtab hash.
16402 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
16404         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
16405         CONVERT_EXPR as equal nodes.
16407 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
16409         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
16411 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
16413         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
16414         Generalize to complex and vector.
16415         * tree.c (build_all_ones_cst): New function.
16416         * tree.h (build_all_ones_cst): Declare it.
16418 2013-06-13  Alan Modra  <amodra@gmail.com>
16420         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
16421         * config/rs6000/rs6000.md (signbittf2): New insn.
16422         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
16423         (abstf2_internal, cmptf_internal2): Likewise.
16424         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
16426 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
16427             Pat Haugen  <pthaugen@us.ibm.com>
16428             Peter Bergner  <bergner@vnet.ibm.com>
16430         * config/rs6000/rs6000.c (emit_load_locked): Add support for
16431         power8 byte, half-word, and quad-word atomic instructions.
16432         (emit_store_conditional): Likewise.
16433         (rs6000_expand_atomic_compare_and_swap): Likewise.
16434         (rs6000_expand_atomic_op): Likewise.
16436         * config/rs6000/sync.md (larx): Add new modes for power8.
16437         (stcx): Likewise.
16438         (AINT): New mode iterator to include TImode as well as normal
16439         integer modes on power8.
16440         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
16441         that VSX registers are not considered.  Use AINT mode iterator
16442         instead of INT1 to allow inclusion of quad word atomic operations
16443         on power8.
16444         (load_locked<mode>): Likewise.
16445         (store_conditional<mode>): Likewise.
16446         (atomic_compare_and_swap<mode>): Likewise.
16447         (atomic_exchange<mode>): Likewise.
16448         (atomic_nand<mode>): Likewise.
16449         (atomic_fetch_<fetchop_name><mode>): Likewise.
16450         (atomic_nand_fetch<mode>): Likewise.
16451         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
16452         each type.
16453         (ATOMIC): On power8, add QImode, HImode modes.
16454         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
16455         modes that promote to SImode.
16456         (load_lockedti): Convert TImode arguments to PTImode, so that we
16457         get a guaranteed even/odd register pair.
16458         (load_lockedpti): Likewise.
16459         (store_conditionalti): Likewise.
16460         (store_conditionalpti): Likewise.
16462         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
16463         atomic load/store instructions.
16464         (HSI): Likewise.
16466 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
16468         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
16469         loads.
16470         (insn_count): New attribute, with most cases extracted from...
16471         (length): ...here.  Redefine most cases in terms of insn_count.
16472         (single_insn): Delete.
16473         (can_delay): Use insn_count to check for single instructions.
16474         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
16475         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
16476         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
16477         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
16478         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
16479         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
16480         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
16481         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
16482         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
16483         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
16484         rather than "length".
16485         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
16486         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
16487         Use "insn_count" rather than "length".
16488         * config/mips/mips-dsp.md
16489         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
16490         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
16491         length attributes.
16493 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
16495         PR tree-optimization/57361
16496         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
16498 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
16500         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
16501         to split.
16502         (aarch64_simd_combine<mode>): New instruction expansion.
16503         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
16504         function prototype.
16505         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
16506         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
16508 2013-06-12  Jan Hubicka  <jh@suse.cz>
16510         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
16511         decl has when in streaming stage.
16512         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
16513         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
16515 2013-06-12  Roland Stigge  <stigge@antcom.de>
16517         PR target/57578
16518         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
16520 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
16522         PR tree-optimization/57537
16523         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
16524         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
16526 2013-06-12  Richard Biener  <rguenther@suse.de>
16528         * data-streamer.h (streamer_write_char_stream): CSE
16529         obs->current_pointer.
16530         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
16531         streamer_write_char_stream manually and optimize the resulting loop.
16532         (streamer_write_hwi_stream): Likewise.
16534 2013-06-12  Jan Hubicka  <jh@suse.cz>
16536         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
16537         * cgraph.h (varpool_create_empty_node): Declare.
16538         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
16539         duplicated nodes.
16540         * symtab.c (symtab_unregister_node): Be lax about missin entries
16541         in node hash.
16542         (symtab_get_node): Update comment.
16543         * varpool.c (varpool_create_empty_node): Break out from ...
16544         (varpool_node_for_decl): ... here.
16545         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
16547 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
16549         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
16550         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
16551         part.  Use straight-line flow at the end.
16552         <COMPONENT_REF>: Remove superfluous else.
16553         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
16555 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
16557         PR target/56564
16558         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
16559         target hook even for !TREE_PUBLIC decls.  If no resolution info
16560         is available, return false for common and external decls.
16562 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
16564         * config/rl78/constraints.md (U): New constraint.
16565         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
16566         valloc attribute.
16568 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
16570         PR target/57589
16571         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
16572         to allow returning address to AT_PLATFORM name.
16574 2013-06-11  Jan Hubicka  <jh@suse.cz>
16576         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
16577         * cgraph.h (symtab_node_base): Add weakref flag.
16578         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
16579         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
16580         (output_weakrefs): Use weakref flag.
16581         * fold-const.c (simple_operand_p): Handle WEAK.
16582         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
16583         * ipa.c (varpool_externally_visible_p): Drop weakref.
16584         (function_and_variable_visibility): Update comment; fix weakref
16585         sanity checks; do not clear DECL_WEAK on them.
16586         * lto-cgraph.c (lto_output_node): update.
16587         (lto_output_varpool_node): Update.
16588         (input_overwrite_node): Update.
16589         (input_node): Update.
16590         (input_varpool_node): Update.
16591         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
16592         (lto_symtab_merge_symbols): Add sanity check.
16593         (lto_symtab_prevailing_decl): Do not special case weakrefs.
16594         * passes.c (rest_of_decl_compilation): Set static flag, too.
16595         * symtab.c (dump_symtab_base): Dump weakref.
16596         (verify_symtab_base): Sanity check weakrefs.
16597         (symtab_make_decl_local): Remove duplicated code.
16598         (symtab_alias_ultimate_target): Simplify.
16599         * varpool.c (varpool_create_variable_alias): Set weakref flag.
16601 2013-06-11  Tom de Vries  <tom@codesourcery.com>
16603         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
16604         sequence_vect == NULL.
16606 2013-06-11  DJ Delorie  <dj@redhat.com>
16608         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
16609         (rl78_unwind_word_mode): New.
16611 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
16613         * final.c (debug_prefix_maps): Make static.
16615 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
16617         * function.c (initial_trampoline): Remove stray copy.
16619 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
16621         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
16623 2013-06-11  Martin Jambor  <mjambor@suse.cz>
16625         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
16626         within bounds at the beginning of the function.
16628 2013-06-11  Alan Modra  <amodra@gmail.com>
16630         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
16631         reporting.
16632         (get_named_section): Don't NULL !DECL_P decl.
16634 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
16636         * doc/invoke.texi (core-avx2): Document.
16637         (slm): Likewise.
16638         (atom): Updated with MOVBE.
16640 2013-06-11  Richard Biener  <rguenther@suse.de>
16642         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
16644 2013-06-11  Anton Blanchard  <anton@samba.org>
16646         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
16647         correct shift value in little-endian mode.
16649 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
16651         PR target/56564
16652         * varasm.c (get_variable_align): Move #endif to the right place.
16654 2013-06-10  Cary Coutant  <ccoutant@google.com>
16656         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
16657         for hash so that hash table traversal order is deterministic.
16659 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
16660             Pat Haugen  <pthaugen@us.ibm.com>
16661             Peter Bergner  <bergner@vnet.ibm.com>
16663         * config/rs6000/vector.md (GPR move splitter): Do not split moves
16664         of vectors in GPRS if they are direct moves or quad word load or
16665         store moves.
16667         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
16668         declaration.
16669         (direct_move_p): Likewise.
16670         (quad_load_store_p): Likewise.
16672         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
16673         classes into bins based on the physical register type.
16674         (reg_class_to_reg_type): Likewise.
16675         (IS_STD_REG_TYPE): Likewise.
16676         (IS_FP_VECT_REG_TYPE): Likewise.
16677         (reload_fpr_gpr): Arrays to determine what insn to use if we can
16678         use direct move instructions.
16679         (reload_gpr_vsx): Likewise.
16680         (reload_vsx_gpr): Likewise.
16681         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
16682         information that is a simplification of register classes.  Also
16683         precalculate direct move reload helpers.
16684         (direct_move_p): New function to return true if the operation can
16685         be done as a direct move instruciton.
16686         (quad_load_store_p): New function to return true if the operation
16687         is a quad memory operation.
16688         (rs6000_legitimize_address): If quad memory, only allow register
16689         indirect for TImode addresses.
16690         (rs6000_legitimate_address_p): Likewise.
16691         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
16692         (rs6000_reload_register_type): Likewise.
16693         (register_to_reg_type): Return register type.
16694         (rs6000_secondary_reload_simple_move): New helper function for
16695         secondary reload and secondary memory needed to identify anything
16696         that is a simple move, and does not need reloading.
16697         (rs6000_secondary_reload_direct_move): New helper function for
16698         secondary reload to identify cases that can be done with several
16699         instructions via the direct move instructions.
16700         (rs6000_secondary_reload_move): New helper function for secondary
16701         reload to identify moves between register types that can be done.
16702         (rs6000_secondary_reload): Add support for quad memory operations
16703         and for direct move.
16704         (rs6000_secondary_memory_needed): Likewise.
16705         (rs6000_debug_secondary_memory_needed): Change argument names.
16706         (rs6000_output_move_128bit): New function to return the move to
16707         use for 128-bit moves, including knowing about the various
16708         limitations of quad memory operations.
16710         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
16711         memory operations.  call rs6000_output_move_128bit for the actual
16712         instruciton(s) to generate.
16713         (vsx_movti_64bit): Likewise.
16715         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
16716         (UNSPEC_P8V_MTVSRWZ): Likewise.
16717         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
16718         (UNSPEC_P8V_MTVSRD): Likewise.
16719         (UNSPEC_P8V_XXPERMDI): Likewise.
16720         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
16721         (UNSPEC_FUSION_GPR): Likewise.
16722         (FMOVE128_GPR): New iterator for direct move.
16723         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
16724         (f32_sv): Likewise.
16725         (f32_dm): Likewise.
16726         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
16727         loads and direct move instructions.
16728         (zero_extendsidi2_lfiwzx): Likewise.
16729         (extendsidi2_lfiwax): Likewise.
16730         (extendsidi2_nocell): Likewise.
16731         (floatsi<mode>2_lfiwax): Likewise.
16732         (lfiwax): Likewise.
16733         (floatunssi<mode>2_lfiwzx): Likewise.
16734         (lfiwzx): Likewise.
16735         (fix_trunc<mode>_stfiwx): Likewise.
16736         (fixuns_trunc<mode>_stfiwx): Likewise.
16737         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
16738         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
16739         (parity<mode>2_cmpb): Set length/type attr.
16740         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
16741         for 'mr.' to fast_compare.
16742         (bpermd_<mode>): Change type attr to popcnt.
16743         (p8_fmrgow_<mode>): New insns for power8 direct move support.
16744         (p8_mtvsrwz_1): Likewise.
16745         (p8_mtvsrwz_2): Likewise.
16746         (reload_fpr_from_gpr<mode>): Likewise.
16747         (p8_mtvsrd_1): Likewise.
16748         (p8_mtvsrd_2): Likewise.
16749         (p8_xxpermdi_<mode>): Likewise.
16750         (reload_vsx_from_gpr<mode>): Likewise.
16751         (reload_vsx_from_gprsf): Likewise.
16752         (p8_mfvsrd_3_<mode>): LIkewise.
16753         (reload_gpr_from_vsx<mode>): Likewise.
16754         (reload_gpr_from_vsxsf): Likewise.
16755         (p8_mfvsrd_4_disf): Likewise.
16756         (multi-word GPR splits): Do not split direct moves or quad memory
16757         operations.
16759 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
16761         * tree-into-ssa.c (interesting_blocks): Make static.
16763 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
16765         PR target/56564
16766         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
16767         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
16768         Use DATA_ABI_ALIGNMENT for that case instead if defined.
16769         (get_variable_align): New function.
16770         (get_variable_section, emit_bss, emit_common,
16771         assemble_variable_contents, place_block_symbol): Use
16772         get_variable_align instead of DECL_ALIGN.
16773         (assemble_noswitch_variable): Add align argument, use it
16774         instead of DECL_ALIGN.
16775         (assemble_variable): Adjust caller.  Use get_variable_align
16776         instead of DECL_ALIGN.
16777         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
16778         caller.
16779         (DATA_ABI_ALIGNMENT): Define.
16780         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
16781         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
16782         opt is false, only return the psABI mandated alignment increase.
16783         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
16784         (DATA_ABI_ALIGNMENT): ... this.
16785         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
16786         (DATA_ABI_ALIGNMENT): ... this.
16787         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
16788         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
16789         (DATA_ABI_ALIGNMENT): ... this.
16790         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
16791         * doc/tm.texi: Regenerated.
16793 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
16795         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
16796         cmp_code to construct REG_EQUAL note.
16798 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
16800         PR target/57568
16801         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
16802         that operands[2] doesn't overlap with operands[0].
16804 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
16805             Jan Hubicka  <jh@suse.cz>
16807         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
16808         hack to mark symbols as used.
16810 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
16812         PR rtl-optimization/57559
16813         * lra-constraints.c (process_alt_operands): Don't discourage
16814         memory with known offset for offsetable memory constraint.
16815         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
16817 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
16819         * varasm.c (struct oc_local_state): Reorder fields.
16820         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
16821         and adjust accordingly.
16822         (output_constructor): Reorder initialization code and adjust call to
16823         output_constructor_bitfield.
16825 2013-06-07  Jan Hubicka  <jh@suse.cz>
16827         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
16829 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
16831         * tree-object-size.c (unknown): Make const.
16833 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16835         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
16836         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
16837         for last alternative in the cpu_facility attribute.
16839 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16841         PR target/56315
16842         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
16843         (xordi3): Change operand 2 constraint to arm_xordi_operand.
16844         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
16845         * config/arm/constraints.md (Dg): New constraint.
16846         * config/arm/neon.md (xordi3_neon): Remove.
16847         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
16848         * config/arm/predicates.md (arm_xordi_operand): New predicate.
16850 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16852         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
16853         Clean up alternatives.
16855 2013-06-07  Alan Modra  <amodra@gmail.com>
16857         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
16858         va_list_gpr_size.
16860 2013-06-07  Alan Modra  <amodra@gmail.com>
16862         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
16864 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16866         * config/arm/constraints.md (Df): New constraint.
16867         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
16868         Correct length attribute for last two alternatives.
16870 2013-06-07  Alan Modra  <amodra@gmail.com>
16872         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
16873         override user -mfp-in-toc.
16874         (offsettable_ok_by_alignment): Consider just the current access
16875         rather than the whole object, unless BLKmode.  Handle
16876         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
16877         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
16878         for -mcmodel=medium.
16879         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
16880         override user -mfp-in-toc or -msum-in-toc.  Default to
16881         -mno-fp-in-toc for -mcmodel=medium.
16883 2013-06-06  DJ Delorie  <dj@redhat.com>
16885         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
16886         TARGET_VALID_POINTER_MODE.
16888 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
16889             Pat Haugen  <pthaugen@us.ibm.com>
16890             Peter Bergner  <bergner@vnet.ibm.com>
16892         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
16893         Document new power8 builtins.
16895         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
16896         condition code register, to allow 128-bit logical operations to be
16897         done in the VSX or GPR registers.
16898         (nor<mode>3): Use the canonical form for nor.
16899         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
16900         vclz*, and vpopcnt* vector instructions.
16901         (nand<mode>3): Likewise.
16902         (orc<mode>3): Likewise.
16903         (clz<mode>2): LIkewise.
16904         (popcount<mode>2): Likewise.
16906         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
16907         that only the GPRs are recognized.
16909         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16910         support for new power8 builtins.
16912         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
16913         builtin functions.
16914         (xscvdpspn): Likewise.
16915         (vclz): Likewise.
16916         (vclzb): Likewise.
16917         (vclzh): Likewise.
16918         (vclzw): Likewise.
16919         (vclzd): Likewise.
16920         (vpopcnt): Likewise.
16921         (vpopcntb): Likewise.
16922         (vpopcnth): Likewise.
16923         (vpopcntw): Likewise.
16924         (vpopcntd): Likewise.
16925         (vgbbd): Likewise.
16926         (vmrgew): Likewise.
16927         (vmrgow): Likewise.
16928         (eqv): Likewise.
16929         (eqv_v16qi3): Likewise.
16930         (eqv_v8hi3): Likewise.
16931         (eqv_v4si3): Likewise.
16932         (eqv_v2di3): Likewise.
16933         (eqv_v4sf3): Likewise.
16934         (eqv_v2df3): Likewise.
16935         (nand): Likewise.
16936         (nand_v16qi3): Likewise.
16937         (nand_v8hi3): Likewise.
16938         (nand_v4si3): Likewise.
16939         (nand_v2di3): Likewise.
16940         (nand_v4sf3): Likewise.
16941         (nand_v2df3): Likewise.
16942         (orc): Likewise.
16943         (orc_v16qi3): Likewise.
16944         (orc_v8hi3): Likewise.
16945         (orc_v4si3): Likewise.
16946         (orc_v2di3): Likewise.
16947         (orc_v4sf3): Likewise.
16948         (orc_v2df3): Likewise.
16950         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
16951         allow power8 quad mode in 64-bit.
16952         (rs6000_builtin_vectorized_function): Add support to vectorize
16953         ISA 2.07 count leading zeros, population count builtins.
16954         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
16955         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
16956         (builtin_function_type): Add vgbbd builtin function which takes an
16957         unsigned argument.
16958         (altivec_expand_vec_perm_const): Add support for new power8 merge
16959         instructions.
16961         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
16962         that does not include TImdoe for use with 32-bit.
16963         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
16964         instructions.
16965         (UNSPEC_VSX_CVDPSPN): Likewise.
16966         (vsx_xscvdpspn): Likewise.
16967         (vsx_xscvspdpn): Likewise.
16968         (vsx_xscvdpspn_scalar): Likewise.
16969         (vsx_xscvspdpn_directmove): Likewise.
16970         (vsx_and<mode>3): Split logical operations into 32-bit and
16971         64-bit. Add support to do logical operations on TImode as well as
16972         VSX vector types.  Allow logical operations to be done in either
16973         VSX registers or in general purpose registers in 64-bit mode.  Add
16974         splitters if GPRs were used. For AND, add clobber of CCmode to
16975         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
16976         encoding.
16977         (vsx_and<mode>3_32bit): Likewise.
16978         (vsx_and<mode>3_64bit): Likewise.
16979         (vsx_ior<mode>3): Likewise.
16980         (vsx_ior<mode>3_32bit): Likewise.
16981         (vsx_ior<mode>3_64bit): Likewise.
16982         (vsx_xor<mode>3): Likewise.
16983         (vsx_xor<mode>3_32bit): Likewise.
16984         (vsx_xor<mode>3_64bit): Likewise.
16985         (vsx_one_cmpl<mode>2): Likewise.
16986         (vsx_one_cmpl<mode>2_32bit): Likewise.
16987         (vsx_one_cmpl<mode>2_64bit): Likewise.
16988         (vsx_nor<mode>3): Likewise.
16989         (vsx_nor<mode>3_32bit): Likewise.
16990         (vsx_nor<mode>3_64bit): Likewise.
16991         (vsx_andc<mode>3): Likewise.
16992         (vsx_andc<mode>3_32bit): Likewise.
16993         (vsx_andc<mode>3_64bit): Likewise.
16994         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
16995         and xxlorc instructions.
16996         (vsx_eqv<mode>3_64bit): Likewise.
16997         (vsx_nand<mode>3_32bit): Likewise.
16998         (vsx_nand<mode>3_64bit): Likewise.
16999         (vsx_orc<mode>3_32bit): Likewise.
17000         (vsx_orc<mode>3_64bit): Likewise.
17002         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
17004         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
17005         instruction.
17006         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
17007         (p8_vmrgow): Likewise.
17008         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
17009         GPRs to be split under VSX.
17010         (p8v_clz<mode>2): Add power8 count leading zero support.
17011         (p8v_popcount<mode>2): Add power8 population count support.
17012         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
17013         support.
17015         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
17016         instruction.
17018         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
17019         builtin functions.
17020         (vec_nand): Likewise.
17021         (vec_vclz): Likewise.
17022         (vec_vclzb): Likewise.
17023         (vec_vclzd): Likewise.
17024         (vec_vclzh): Likewise.
17025         (vec_vclzw): Likewise.
17026         (vec_vgbbd): Likewise.
17027         (vec_vmrgew): Likewise.
17028         (vec_vmrgow): Likewise.
17029         (vec_vpopcnt): Likewise.
17030         (vec_vpopcntb): Likewise.
17031         (vec_vpopcntd): Likewise.
17032         (vec_vpopcnth): Likewise.
17033         (vec_vpopcntw): Likewise.
17035 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17037         PR rtl-optimization/57468
17038         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
17039         spilled pseudos.
17041 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17043         PR rtl-optimization/57459
17044         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
17045         type when setting live regs.
17047 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17049         * config/s390/s390.opt (mlra): New option.
17050         * config/s390/s390.c (s390_decompose_address): Check displacement
17051         for all registers for LRA.
17052         (s390_secondary_reload): Don't used secondary reloads for LRA.
17053         (s390_lra_p): New function.
17054         (TARGET_LRA_P): Define.
17055         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
17056         of attribute cpu_facility to zarch for the last alternative.
17057         (*cmpmem_short): Ditto.
17059 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
17061         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
17062         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
17063         (arm_expand_prologue): Likewise.
17065 2013-06-06  Teresa Johnson  <tejohnson@google.com>
17067         PR c++/53743
17068         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
17069         as this is now done by redirect_edge_and_branch_force.
17070         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
17071         barriers, and fix interaction with splitting.
17072         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
17073         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
17074         reflect changes made in the routine.
17075         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
17076         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
17077         since this is called in cfglayout mode, and replace partition fixup
17078         with assert as that is now done by force_nonfallthru_and_redirect.
17079         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
17080         already be marked with region crossing note.
17081         (insert_section_boundary_note): Make non-static, gate on flag
17082         has_bb_partition, rewrite to also check for multiple partitions.
17083         (rest_of_handle_reorder_blocks): Remove call to
17084         insert_section_boundary_note, now done later during free_cfg.
17085         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
17086         * bb-reorder.h (insert_section_boundary_note): Declare.
17087         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
17088         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
17089         invoke insert_section_boundary_note.
17090         (try_redirect_by_replacing_jump): Remove unnecessary
17091         check for region crossing note.
17092         (fixup_partition_crossing): New function.
17093         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
17094         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
17095         in non-cfglayout mode.
17096         (force_nonfallthru_and_redirect): Fixup partition boundaries,
17097         remove old code that tried to do this. Emit barrier correctly
17098         when we are in cfglayout mode.
17099         (last_bb_in_partition): New function.
17100         (rtl_split_edge): Correctly fixup partition boundaries.
17101         (commit_one_edge_insertion): Remove old code that tried to
17102         fixup region crossing edge since this is now handled in
17103         split_block, and set up insertion point correctly since
17104         block may now end in a jump.
17105         (verify_hot_cold_block_grouping): Guard against checking when not in
17106         linearized RTL mode.
17107         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
17108         notes.
17109         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
17110         rtl_verify_flow_info, so not called in cfglayout mode.
17111         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
17112         (fixup_reorder_chain): Remove old code that attempted to fixup region
17113         crossing note as this is now handled in force_nonfallthru_and_redirect.
17114         (duplicate_insn_chain): Don't duplicate switch section notes.
17115         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
17116         note.
17117         * basic-block.h (emit_barrier_after_bb): Declare.
17119 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17121         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
17122         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
17123         arm_usatsihi): Adjust alternatives for arm_restrict_it.
17125 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17127         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
17128         where appropriate.
17129         * config/arm/ldmstm.md: Regenerate.
17131 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17133         * config/arm/sync.md (atomic_loaddi_1):
17134         Disable predication for arm_restrict_it.
17135         (arm_load_exclusive<mode>): Likewise.
17136         (arm_load_exclusivesi): Likewise.
17137         (arm_load_exclusivedi): Likewise.
17138         (arm_load_acquire_exclusive<mode>): Likewise.
17139         (arm_load_acquire_exclusivesi): Likewise.
17140         (arm_load_acquire_exclusivedi): Likewise.
17141         (arm_store_exclusive<mode>): Likewise.
17142         (arm_store_exclusive<mode>): Likewise.
17143         (arm_store_release_exclusivedi): Likewise.
17144         (arm_store_release_exclusive<mode>): Likewise.
17146 2013-06-06  Richard Biener  <rguenther@suse.de>
17148         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
17149         after LTO_null.
17150         (lto_tag_is_tree_code_p): Adjust.
17151         (lto_tag_is_gimple_code_p): Likewise.
17152         (lto_gimple_code_to_tag): Likewise.
17153         (lto_tag_to_gimple_code): Likewise.
17154         (lto_tree_code_to_tag): Likewise.
17155         (lto_tag_to_tree_code): Likewise.
17156         * data-streamer.h (streamer_write_hwi_in_range): Use
17157         uhwi streaming to stream the normalized range.
17158         (streamer_read_hwi_in_range): Likewise.
17160 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17162         * config/arm/arm.md (enabled_for_depr_it): New attribute.
17163         (predicable_short_it): Likewise.
17164         (predicated): Likewise.
17165         (enabled): Handle above.
17166         (define_cond_exec): Set predicated attribute to yes.
17168 2013-06-05  Mike Stump  <mikestump@comcast.net>
17170         * gdbinit.in (__FUNCTION__): Add.
17172 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
17174         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
17175         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
17177 2013-06-05  Jan Hubicka  <jh@suse.cz>
17179         * varasm.c (mark_decl_referenced): Revert the removal until targets
17180         are fixed.
17182 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
17184         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
17185         instead of mark_decl_referenced.
17187 2013-06-05  Jan Hubicka  <jh@suse.cz>
17189         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
17190         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
17191         and symtab_used_from_object_file_p.
17192         (cgraph_make_node_local_1): Clear forced_by_abi.
17193         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
17194         * cgraph.h (symtab_node_base): Add forced_by_abi.
17195         (decide_is_variable_needed): Remove.
17196         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
17197         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
17198         (decide_is_symbol_needed): ... this one; handle symbols in general;
17199         always analyze virtuals; honnor forced_by_abi.
17200         (cgraph_finalize_function): Update.
17201         (varpool_finalize_decl): Update.
17202         (symbol_defined_and_needed): Remove.
17203         (analyze_functions): Update.
17204         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17205         output_refs, input_overwrite_node): Handle forced_by_abi.
17206         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
17207         (address_taken_from_non_vtable_p): ... this one.
17208         (comdat_can_be_unshared_p_1): New function.
17209         (cgraph_comdat_can_be_unshared_p): Rename to ...
17210         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
17211         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
17212         (function_and_variable_visibility): Clear forced_by_abi as needed.
17213         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
17214         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
17215         * symtab.c (dump_symtab_base): Dump forced_by_abi.
17216         * varpool.c (decide_is_variable_needed): Remove.
17218 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17220         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
17221         (arm_option_override): Override arm_restrict_it where appropriate.
17222         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
17223         * config/arm/arm.opt (mrestrict-it): New command-line option.
17224         * doc/invoke.texi: Document -mrestrict-it.
17226 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
17228         * tsan.c (tsan_atomic_table): Make const.
17230 2013-06-05  Richard Biener  <rguenther@suse.de>
17232         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
17233         index associated with the tree we are supposed to replace.
17234         * tree-streamer-out.c (pack_ts_base_value_fields): Output
17235         TREE_ASM_WRITTEN as zero for everything but SSA names.
17237 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
17239         * tree-ssa-structalias.c (call_stmt_vars): Make static.
17241 2013-06-04  Jan Hubicka  <jh@suse.cz>
17243         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
17244         (input_node, input_varpool_node): Handle correctly external same
17245         body aliases.
17246         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
17247         nodes at ltrans stage.
17249 2013-06-04  Jan Hubicka  <jh@suse.cz>
17251         * ipa-inline.c (update_caller_keys): Fix availability test.
17252         (update_callee_keys): Likewise.
17253         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
17254         to follow ELF standard.
17256 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
17258         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
17259         (mips64r5900el-*-elf*): New configurations.
17260         * config/mips/mips-cpus.def (r5900): New processor.
17261         * config/mips/mips-tables.opt: Regenerate.
17262         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
17263         (mips_issue_rate): Handle PROCESSOR_R5900.
17264         (mips_reorg_process_insns): Force reorder mode for the R5900.
17265         * config/mips/mips.h (TARGET_MIPS5900): Define.
17266         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
17267         TARGET_MIPS5900.
17268         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
17269         TARGET_MIPS5900.
17270         * config/mips/mips.md (processor): Add r5900.
17271         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
17273 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17275         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
17276         into function to generate MOVI instruction.
17277         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
17278         (aarch64_preferred_simd_mode): Turn into wrapper.
17279         (aarch64_output_scalar_simd_mov_immediate): New function.
17280         * config/aarch64/aarch64-protos.h: Add prototype for above.
17282 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17284         * config/aarch64/aarch64.c (simd_immediate_info): Remove
17285         element_char member.
17286         (sizetochar): Return signed char.
17287         (aarch64_simd_valid_immediate): Remove elchar and other
17288         unnecessary variables.
17289         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
17290         Calculate element_char as required.
17291         * config/aarch64/aarch64-protos.h: Update and move prototype
17292         for aarch64_output_simd_mov_immediate.
17293         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
17294         Update arguments.
17296 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17298         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
17299         information completed by aarch64_simd_valid_immediate.
17300         (aarch64_legitimate_constant_p): Update arguments.
17301         (aarch64_simd_valid_immediate): Work with struct rather than many
17302         pointers.
17303         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
17304         (aarch64_simd_make_constant): Update arguments.
17305         (aarch64_output_simd_mov_immediate): Work with struct rather than
17306         many pointers.  Output immediate directly rather than as operand.
17307         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
17308         Update prototype.
17309         * config/aarch64/constraints.md (Dn): Update arguments.
17311 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17313         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
17314         longer static.
17315         (aarch64_simd_immediate_valid_for_move): Remove.
17316         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
17317         (aarch64_simd_make_constant): Update call.
17318         (aarch64_output_simd_mov_immediate): Update call.
17319         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
17320         Add prototype.
17321         * config/aarch64/constraints.md (Dn): Update call.
17323 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17325         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
17326         return type to bool for prototype.
17327         (aarch64_legitimate_constant_p): Check for true instead of not -1.
17328         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
17329         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
17331 2013-06-04  Catherine Moore  <clm@codesourcery.com>
17333         * config/mips/mips.opt (meva): New.
17334         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
17335         (ASM_SPEC): Handle -meva.
17336         * doc/invoke.texi (meva):  Document.
17338 2013-06-04  Alan Modra  <amodra@gmail.com>
17340         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
17341         constant output.
17343 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17345         * rtl.def: Add extra fourth optional field to define_cond_exec.
17346         * gensupport.c (process_one_cond_exec): Process attributes from
17347         define_cond_exec.
17348         * doc/md.texi: Document fourth field in define_cond_exec.
17350 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
17352         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
17353         out the processing order as in store_bit_field_1.
17355 2013-06-04  Jan Hubicka  <jh@suse.cz>
17357         PR middle-end/57500
17358         * cgraphunit.c (cgraph_process_same_body_aliases): Create
17359         non-VAR_DECL node if it does not exist yet.
17361 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
17363         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
17364         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
17365         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
17366         target_cpu_default setting.
17368 2013-06-03  Teresa Johnson  <tejohnson@google.com>
17370         * dumpfile.c (opt_info_switch_p): Change -fopt-info
17371         default to -fopt-info=optimized instead of all.
17372         * doc/invoke.texi: Ditto.
17373         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
17374         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
17375         (execute_vect_slp): Emit BB vectorization success under
17376         MSG_OPTIMIZED_LOCATIONS.
17377         * tree-vect-slp.c (vect_slp_transform_bb): Change
17378         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
17379         * tree-vect-loop.c (vect_transform_loop): Ditto.
17381 2013-06-03  Jason Merrill  <jason@redhat.com>
17383         PR c++/57415
17384         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17385         Use TARGET_EXPR for C++.
17387 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
17389         PR rtl-optimization/57268
17390         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
17391         if DEBUG_INSN_P (insn).
17393         Reapply
17394         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
17396         PR rtl-optimization/57268
17397         * sched-deps.c (sched_analyze_2): Flush dependence lists if
17398         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
17400 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
17402         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
17403         (ix86_avoid_lea_for_addr): Likewise.
17404         (exact_dependency_1): Likewise.
17405         (ix86_adjust_cost): Likewise.
17406         (swap_top_of_ready_list): Fix formatting and !reload_completed check
17407         removed.
17408         (do_reorder_for_imul): Fix typo, formatting and
17409         !reload_completed check removed.
17410         (ix86_sched_reorder): Fix typo and formatting.
17411         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
17412         list.
17414 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
17416         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
17418 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
17420         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
17421         <STRING_CST>: Likewise.
17422         <VECTOR_CST>: Likewise.
17424 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
17425             Mikael Morin  <mikael@gcc.gnu.org>
17427         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
17428         * config.in: Regenerated.
17429         * configure: Regenerated.
17431 2013-06-01  Jan Hubicka  <jh@suse.cz>
17433         PR middle-end/57366
17434         * cgraphunit.c (compile): When weakref is not supported,
17435         set up transparent aliases before final output pass.
17436         * varasm.c (assemble_alias): Do not try to do it here.
17438 2013-06-01  Jan Hubicka  <jh@suse.cz>
17440         PR middle-end/57467
17441         * passes.c (for_per_function): Skip unanalyzed functions.
17443 2013-06-01  Jan Hubicka  <jh@suse.cz>
17445         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
17446         (lto_symtab_merge_symbols_1): ... this one.
17447         (lto_symtab_merge_cgraph_nodes): Rename to ...
17448         (lto_symtab_merge_symbols): ... this one; simplify.
17449         * cgraph.c (same_body_aliases_done): Rename to ...
17450         (cpp_implicit_aliases_done): ... this one.
17451         (cgraph_create_function_alias): Update.
17452         (cgraph_same_body_alias): Update.
17453         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
17454         (verify_edge_corresponds_to_fndecl): Simplify.
17455         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
17456         (cgraph_node): Remove same_body_alias.
17457         (varpool_node): Remove alias_of and extra_name_alias.
17458         (same_body_aliases_done): Rename to ..
17459         (cpp_implicit_aliases_done): ... this one.
17460         (symtab_alias_ultimate_target): Add default parameter.
17461         (symtab_resolve_alias): New function.
17462         (fixup_same_cpp_alias_visibility): Declare.
17463         (cgraph_function_node): Add default parameter.
17464         (cgraph_node_asm_name): Likewise.
17465         (cgraph_function_or_thunk_node): Add default parameter; do
17466         not ICE when it is NULL.
17467         (varpool_variable_node): Likewise.
17468         * tree-emutls.c (create_emultls_var): Update.
17469         (ipa_lower_emutls): Update.
17470         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
17471         (cgraph_reset_node): Reset alias info.
17472         (cgraph_finalize_function): Update.
17473         (fixup_same_cpp_alias_visibility): Move to symtab.c.
17474         (analyze_function): Simplify.
17475         (cgraph_process_same_body_aliases): Simplify.
17476         (analyze_functions): Fixup same body aliases.
17477         (handle_alias_pairs): Simplify.
17478         (assemble_thunk): Update.
17479         (assemble_thunks_and_aliases): Update.
17480         (output_weakrefs): Rewrite.
17481         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
17482         (lto_output_varpool_node): Likewise.
17483         (compute_ltrans_boundary): Remve assert.
17484         (get_alias_symbol): New functoin.
17485         (input_node): Rewrite alias handling.
17486         (input_varpool_node): Likewise.
17487         * ipa-pure-const.c (propagate_pure_const): Fix formating.
17488         * ipa.c (process_references): Handle weakrefs correctly.
17489         (symtab_remove_unreachable_nodes): Likewise.
17490         * trans-mem.c (get_cg_data): Update.
17491         (ipa_tm_create_version_alias): Update.
17492         (ipa_tm_execute): Update.
17493         * symtab.c (dump_symtab_base): Dump aliases.
17494         (verify_symtab_base): Verify aliases.
17495         (symtab_node_availability): New function.
17496         (symtab_alias_ultimate_target): Simplify.
17497         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
17498         handle all the fixup cases.
17499         (symtab_resolve_alias): New function.
17500         * passes.c (ipa_write_summaries): Handle weakrefs.
17501         * varpool.c (varpool_analyze_node): Simplify.
17502         (assemble_aliases): Update.
17503         (varpool_create_variable_alias): Simplify.
17504         (varpool_extra_name_alias): Simplify.
17505         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
17506         (lto_symtab_merge_symbols): ... this one.
17508 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
17510         Revert
17511         PR rtl-optimization/57268
17512         * sched-deps.c (sched_analyze_2): Flush dependence lists if
17513         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
17515 2013-06-01  Tobias Burnus  <burnus@net-b.de>
17517         Partially reverted:
17518         2013-05-31  Tobias Burnus  <burnus@net-b.de>
17520         PR middle-end/57073
17521         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
17522         further up.
17524 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
17526         PR rtl-optimization/57268
17527         * sched-deps.c (sched_analyze_2): Flush dependence lists if
17528         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
17530 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
17532         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
17533         unordered comparison operators when -fno-trapping-math is in effect
17534         on the e500.
17535         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
17536         and implement unordered comparison operators properly on the e500.
17538 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
17540         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
17541         for constant scalar integers.
17542         (simplify_relational_operation_1): Likewise.
17544 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
17546         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
17547         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
17548         Fix comment.
17550 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
17551             Igor Zamyatin  <igor.zamyatin@intel.com>
17553         Silvermont (SLM) architecture performance tuning.
17554         * config/i386/i386.h (enum ix86_tune_indices): Add
17555         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
17556         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
17558         * config/i386/i386.c (initial_ix86_tune_features)
17559         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
17560         (ix86_lea_outperforms): Handle Silvermont tuning.
17561         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
17562         call.
17563         (ix86_use_lea_for_mov): Likewise.
17564         (ix86_avoid_lea_for_addr): Likewise.
17565         (ix86_lea_for_add_ok): Likewise.
17566         (exact_dependency_1): New function.
17567         (exact_store_load_dependency): Likewise.
17568         (ix86_adjust_cost): Handle Silvermont tuning.
17569         (do_reoder_for_imul): Likewise.
17570         (swap_top_of_ready_list): New function.
17571         (ix86_sched_reorder): Changed to handle Silvermont tuning.
17573         * config/i386/i386.md (peepholes that split memory operand in fp
17574         converts): New.
17576 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
17578         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
17579         Remove un-necessary braces.
17581 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
17583         * config/aarch64/aarch64.c (aarch64_classify_symbol):
17584         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
17586 2013-05-31  Tobias Burnus  <burnus@net-b.de>
17588         PR middle-end/57073
17589         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
17591 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17593         PR target/56315
17594         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
17595         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
17596         * config/arm/neon.md (iordi3_neon): Remove.
17597         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
17598         * config/arm/predicates.md (imm_for_neon_logic_operand):
17599         Move to earlier in the file.
17600         (neon_logic_op2): Likewise.
17601         (arm_iordi_operand_neon): New predicate.
17603 2013-05-31  Richard Biener  <rguenther@suse.de>
17605         PR tree-optimization/57478
17606         PR tree-optimization/57453
17607         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
17608         are life as well.
17610 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17612         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
17613         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
17615 2013-05-30  Tobias Burnus  <burnus@net-b.de>
17616             Thomas Koenig  <tkoenig@gcc.gnu.org>
17618         PR middle-end/57073
17619         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
17620         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
17622 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
17624         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
17626 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
17628         * target.def (register_usage_leveling_p): New hook.
17629         * targhooks.c (default_register_usage_leveling_p): New.
17630         * targhooks.h (default_register_usage_leveling_p): New prototype.
17631         * lra-assigns.c (register_usage_leveling_p): Use the hook.
17632         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
17633         * doc/tm.texi: Update.
17634         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
17636 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
17638         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
17639         (*insv_reg<mode>): New define_insn.
17641 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
17643         PR rtl-optimization/57439
17644         * postreload.c (move2add_valid_value_p): Check that we have
17645         a zero subreg_regno_offset when accessing the register in
17646         the requested mode.
17648 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
17649             Igor Zamyatin  <igor.zamyatin@intel.com>
17651         Silvermont (SLM) architecture pipeline model, tuning and
17652         insn selection.
17653         * config.gcc: Add slm config options and target.
17655         * config/i386/slm.md: New.
17657         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
17659         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
17660         PROCESSOR_SLM.
17661         (ix86_target_macros_internal): Likewise.
17663         * config/i386/i386.c (slm_cost): New cost.
17664         (m_SLM): New macro flag.
17665         (initial_ix86_tune_features): Set m_SLM.
17666         (x86_accumulate_outgoing_args): Likewise.
17667         (x86_arch_always_fancy_math_387): Likewise.
17668         (processor_target_table): Add slm cost.
17669         (cpu_names): Add slm cpu name.
17670         (x86_option_override_internal): Set SLM ISA.
17671         (ix86_issue_rate): New case PROCESSOR_SLM.
17672         (ia32_multipass_dfa_lookahead): Likewise.
17673         (fold_builtin_cpu): Add slm.
17675         * config/i386/i386.h (TARGET_SLM): New target macro.
17676         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
17677         (processor_type): Add PROCESSOR_SLM.
17679         * config/i386/i386.md (cpu): Add new value "slm".
17680         (slm.md): Include slm.md.
17682 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
17683             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
17685         * config/arm/arm-protos.h: Add and update function protos.
17686         * config/arm/arm.c (use_simple_return_p): New added.
17687         (thumb2_expand_return): Check simple_return flag.
17688         * config/arm/arm.md: Add simple_return and conditional simple_return.
17689         * config/arm/iterators.md: Add iterator for return and simple_return.
17691 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
17693         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
17694         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
17695         (arm_emit_vfp_multi_reg_pop): Likewise.
17696         (thumb2_emit_ldrd_pop): Likewise.
17697         (arm_expand_epilogue): Add misc REG_CFA notes.
17698         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
17700 2013-05-29  Lawrence Crowl  <crowl@google.com>
17702         * config/arm/t-arm: Update for below.
17704         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
17705         Change type to hash_table.  Update dependent calls and types.
17707         * config/i386/t-cygming: Update for below.
17709         * config/i386/t-interix: Update for below.
17711         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
17712         Change type to hash_table.  Update dependent calls and types.
17713         (i386_find_on_wrapper_list::wrappers): Likewise.
17715         * config/ia64/t-ia64: Update for below.
17717         * config/ia64/ia64.c (bundle_state_table):
17718         Change type to hash_table.  Update dependent calls and types.
17720         * config/mips/mips.c (mips_reorg_process_insns::htab):
17721         Change type to hash_table.  Update dependent calls and types.
17723         * config/sol2.c (solaris_comdat_htab):
17724         Change type to hash_table.  Update dependent calls and types.
17726         * config/t-sol2: Update for above.
17728 2013-05-29  Teresa Johnson  <tejohnson@google.com>
17730         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
17731         functions are not yet marked as defined.
17733 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
17734             Pat Haugen  <pthaugen@us.ibm.com>
17735             Peter Bergner  <bergner@vnet.ibm.com>
17737         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
17738         instructions.
17739         (VEC_A): Likewise.
17740         (VEC_C): Likewise.
17741         (vrotl<mode>3): Likewise.
17742         (vashl<mode>3): Likewise.
17743         (vlshr<mode>3): Likewise.
17744         (vashr<mode>3): Likewise.
17746         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17747         support for power8 V2DI builtins.
17749         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
17750         power8 V2DI builtins.
17751         (vupkhsw): Likewise.
17752         (vupklsw): Likewise.
17753         (vaddudm): Likewise.
17754         (vminsd): Likewise.
17755         (vmaxsd): Likewise.
17756         (vminud): Likewise.
17757         (vmaxud): Likewise.
17758         (vpkudum): Likewise.
17759         (vpksdss): Likewise.
17760         (vpkudus): Likewise.
17761         (vpksdus): Likewise.
17762         (vrld): Likewise.
17763         (vsld): Likewise.
17764         (vsrd): Likewise.
17765         (vsrad): Likewise.
17766         (vsubudm): Likewise.
17767         (vcmpequd): Likewise.
17768         (vcmpgtsd): Likewise.
17769         (vcmpgtud): Likewise.
17770         (vcmpequd_p): Likewise.
17771         (vcmpgtsd_p): Likewise.
17772         (vcmpgtud_p): Likewise.
17773         (vupkhsw): Likewise.
17774         (vupklsw): Likewise.
17775         (vaddudm): Likewise.
17776         (vmaxsd): Likewise.
17777         (vmaxud): Likewise.
17778         (vminsd): Likewise.
17779         (vminud): Likewise.
17780         (vpksdss): Likewise.
17781         (vpksdus): Likewise.
17782         (vpkudum): Likewise.
17783         (vpkudus): Likewise.
17784         (vrld): Likewise.
17785         (vsld): Likewise.
17786         (vsrad): Likewise.
17787         (vsrd): Likewise.
17788         (vsubudm): Likewise.
17790         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
17791         support for power8 V2DI instructions.
17793         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
17794         power8 V2DI instructions.  Combine pack and unpack insns to use an
17795         iterator for each mode.  Check whether a particular mode supports
17796         Altivec instructions instead of just checking TARGET_ALTIVEC.
17797         (UNSPEC_VPKUWUM): Likewise.
17798         (UNSPEC_VPKSHSS): Likewise.
17799         (UNSPEC_VPKSWSS): Likewise.
17800         (UNSPEC_VPKUHUS): Likewise.
17801         (UNSPEC_VPKSHUS): Likewise.
17802         (UNSPEC_VPKUWUS): Likewise.
17803         (UNSPEC_VPKSWUS): Likewise.
17804         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
17805         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
17806         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
17807         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
17808         (UNSPEC_VUPKHSB): Likewise.
17809         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
17810         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
17811         (UNSPEC_VUPKHSH): Likewise.
17812         (UNSPEC_VUPKLSB): Likewise.
17813         (UNSPEC_VUPKLSH): Likewise.
17814         (VI2): Likewise.
17815         (VI_char): Likewise.
17816         (VI_scalar): Likewise.
17817         (VI_unit): Likewise.
17818         (VP): Likewise.
17819         (VP_small): Likewise.
17820         (VP_small_lc): Likewise.
17821         (VU_char): Likewise.
17822         (add<mode>3): Likewise.
17823         (altivec_vaddcuw): Likewise.
17824         (altivec_vaddu<VI_char>s): Likewise.
17825         (altivec_vadds<VI_char>s): Likewise.
17826         (sub<mode>3): Likewise.
17827         (altivec_vsubcuw): Likewise.
17828         (altivec_vsubu<VI_char>s): Likewise.
17829         (altivec_vsubs<VI_char>s): Likewise.
17830         (altivec_vavgs<VI_char>): Likewise.
17831         (altivec_vcmpbfp): Likewise.
17832         (altivec_eq<mode>): Likewise.
17833         (altivec_gt<mode>): Likewise.
17834         (altivec_gtu<mode>): Likewise.
17835         (umax<mode>3): Likewise.
17836         (smax<mode>3): Likewise.
17837         (umin<mode>3): Likewise.
17838         (smin<mode>3): Likewise.
17839         (altivec_vpkuhum): Likewise.
17840         (altivec_vpkuwum): Likewise.
17841         (altivec_vpkshss): Likewise.
17842         (altivec_vpkswss): Likewise.
17843         (altivec_vpkuhus): Likewise.
17844         (altivec_vpkshus): Likewise.
17845         (altivec_vpkuwus): Likewise.
17846         (altivec_vpkswus): Likewise.
17847         (altivec_vpks<VI_char>ss): Likewise.
17848         (altivec_vpks<VI_char>us): Likewise.
17849         (altivec_vpku<VI_char>us): Likewise.
17850         (altivec_vpku<VI_char>um): Likewise.
17851         (altivec_vrl<VI_char>): Likewise.
17852         (altivec_vsl<VI_char>): Likewise.
17853         (altivec_vsr<VI_char>): Likewise.
17854         (altivec_vsra<VI_char>): Likewise.
17855         (altivec_vsldoi_<mode>): Likewise.
17856         (altivec_vupkhsb): Likewise.
17857         (altivec_vupkhs<VU_char>): Likewise.
17858         (altivec_vupkls<VU_char>): Likewise.
17859         (altivec_vupkhsh): Likewise.
17860         (altivec_vupklsb): Likewise.
17861         (altivec_vupklsh): Likewise.
17862         (altivec_vcmpequ<VI_char>_p): Likewise.
17863         (altivec_vcmpgts<VI_char>_p): Likewise.
17864         (altivec_vcmpgtu<VI_char>_p): Likewise.
17865         (abs<mode>2): Likewise.
17866         (vec_unpacks_hi_v16qi): Likewise.
17867         (vec_unpacks_hi_v8hi): Likewise.
17868         (vec_unpacks_lo_v16qi): Likewise.
17869         (vec_unpacks_hi_<VP_small_lc>): Likewise.
17870         (vec_unpacks_lo_v8hi): Likewise.
17871         (vec_unpacks_lo_<VP_small_lc>): Likewise.
17872         (vec_pack_trunc_v8h): Likewise.
17873         (vec_pack_trunc_v4si): Likewise.
17874         (vec_pack_trunc_<mode>): Likewise.
17876         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
17877         V2DI builtins.
17878         (vec_vmaxsd): Likewise.
17879         (vec_vmaxud): Likewise.
17880         (vec_vminsd): Likewise.
17881         (vec_vminud): Likewise.
17882         (vec_vpksdss): Likewise.
17883         (vec_vpksdus): Likewise.
17884         (vec_vpkudum): Likewise.
17885         (vec_vpkudus): Likewise.
17886         (vec_vrld): Likewise.
17887         (vec_vsld): Likewise.
17888         (vec_vsrad): Likewise.
17889         (vec_vsrd): Likewise.
17890         (vec_vsubudm): Likewise.
17891         (vec_vupkhsw): Likewise.
17892         (vec_vupklsw): Likewise.
17894 2013-05-29  Jan Hubicka  <jh@suse.cz>
17896         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
17897         flags; reorder rest of fields in more consistent way.
17898         (varpool_node): Remove analyzed, finalized and alias.
17899         (cgraph_ndoe): Likewise.
17900         (symtab_alias_ultimate_target): New function.
17901         (cgraph_function_node): Move offline.
17902         (cgraph_reset_node): Declare.
17903         (cgraph_comdat_can_be_unshared_p): Remove.
17904         (varpool_remove_initializer): Declare.
17905         (varpool_first_defined_variable, varpool_next_defined_variable
17906         cgraph_first_defined_function, cgraph_next_defined_function): Update.
17907         (cgraph_function_with_gimple_body_p): Update.
17908         (varpool_all_refs_explicit_p): Update.
17909         (symtab_alias_target): New function.
17910         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
17911         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
17912         (cgraph_function_or_thunk_node): Simplify using
17913         symtab_alias_ultimate_target.
17914         (varpool_variable_node): Likewise.
17915         * cgraph.c (cgraph_create_function_alias): Update.
17916         (cgraph_add_thunk): Update.
17917         (cgraph_remove_node): Update.
17918         (dump_cgraph_node): Do not dump removed flags.
17919         (cgraph_function_body_availability): Update.
17920         (cgraph_propagate_frequency): Update.
17921         (verify_cgraph_node): Check sanity of local flag.
17922         (cgraph_function_node): Move here from cgraph.h; revamp for
17923         cgraph_function_or_thunk_node.
17924         * lto-symtab.c (lto_varpool_replace_node): Update.
17925         (lto_symtab_resolve_can_prevail_p): Update.
17926         (lto_symtab_merge_cgraph_nodes): Update.
17927         * ipa-cp.c (determine_versionability, initialize_node_lattices,
17928         propagate_constants_accross_call, devirtualization_time_bonus,
17929         ipcp_propagate_stage): Update.
17930         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
17931         * ipa-inline-transform.c (clone_inlined_nodes,
17932         preserve_function_body_p): Update.
17933         * ipa-reference.c (propagate): Update.
17934         (write_node_summary_p): Update.
17935         * toplev.c (wrapup_global_declaration_2): Update.
17936         * cgraphunit.c (cgraph_analyze_function): Rename to ...
17937         (analyze_function) ... this one.
17938         (cgraph_process_new_functions): Update.
17939         (cgraph_reset_node): Export.
17940         (cgraph_finalize_function): Update.
17941         (cgraph_add_new_function): Update.
17942         (process_function_and_variable_attributes): Update.
17943         (varpool_finalize_decl): Update.
17944         (symbol_finalized): Remove.
17945         (symbol_finalized_and_needed): Rename to ...
17946         (symbol_defined_and_needed): ... update.
17947         (cgraph_analyze_functions): Update.
17948         (handle_alias_pairs): Update.
17949         (mark_functions_to_output): Update.
17950         (assemble_thunk): Update.
17951         (output_in_order): Update.
17952         (output_weakrefs): Update.
17953         (finalize_compilation_unit): Update.
17954         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
17955         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
17956         input_node, input_varpool_node): Update.
17957         * dbxout.c (dbxout_expand_expr): Update.
17958         * cgraphclones.c (cgraph_clone_node): Update.
17959         (cgraph_copy_node_for_versioning): Update.
17960         (cgraph_materialize_clone): Update.
17961         (cgraph_materialize_all_clones): Update.
17962         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
17963         propagate_pure_const, propagate_nothrow): Update.
17964         * lto-streamer-out.c (lto_output, write_symbol): Update.
17965         * ipa-utils.c (ipa_reverse_postorder): Update.
17966         * ipa-inline.c (can_inline_edge_p): Update.
17967         (update_caller_keys, ipa_inline): Update.
17968         * dwarf2out.c (reference_to_unused,
17969         premark_types_used_by_global_vars_helper): Update.
17970         * tree-eh.c (tree_could_trap_p): Update.
17971         * ipa-split.c (consider_split, execute_split_functions): Update.
17972         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
17973         has_addr_references_p): Update; move ahead in file for better
17974         readability.
17975         (process_references): Simplify.
17976         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
17977         bodies are removed.
17978         (cgraph_comdat_can_be_unshared_p): Make static.
17979         (cgraph_externally_visible_p): Update.
17980         (varpool_externally_visible_p): Update.
17981         (function_and_variable_visibility): Update.
17982         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
17983         ipa_tm_mark_force_output_node): Update.
17984         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
17985         estimate_edge_devirt_benefit, inline_generate_summary,
17986         inline_write_summary): Update.
17987         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
17988         * ipa-prop.c (ipa_compute_jump_functions): Update.
17989         (ipa_print_node_params, ipa_prop_read_section,
17990         ipa_update_after_lto_read, read_replacements_section): Update.
17991         * varasm.c (mark_decl_referenced): Update.
17992         (assemble_alias, dump_tm_clone_pairs): Update.
17993         * tree-inline.c (copy_bb): Update.
17994         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
17995         Update.
17996         * symtab.c (dump_symtab_base): Print new flags.
17997         (verify_symtab_base): Verify new flags.
17998         (symtab_alias_ultimate_target): New function.
17999         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
18000         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
18001         Update.
18002         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
18003         Update.
18004         * i386.c (ix86_get_function_versions_dispatcher,
18005         ix86_generate_version_dispatcher_body): Update.
18006         (fold_builtin_cpu): Use varpool_add_new_variable.
18007         * varpool.c (varpool_remove_initializer): Break out from ...
18008         (varpool_remove_node): ... this one.
18009         (dump_varpool_node, varpool_node_for_asm,
18010         cgraph_variable_initializer_availability, varpool_analyze_node,
18011         varpool_assemble_decl, varpool_remove_unreferenced_decls,
18012         varpool_finalize_named_section_flags, varpool_create_variable_alias):
18013         Update.
18015 2013-05-29  Jan Hubicka  <jh@suse.cz>
18017         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
18019 2013-05-29  Easwaran Raman  <eraman@google.com>
18021         PR tree-optimization/57442
18022         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
18023         when control exits the main loop.
18025 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
18027         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
18028         and RX600.
18029         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
18030         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
18031         * rx/t-rx: Add rx100 under multi library matches option for nofpu
18032         option.
18034 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18036         PR tree-optimization/57441
18037         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
18038         Don't limit size of incr_vec to number of candidates.
18040 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
18042         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
18043         and mips16 directories.
18044         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
18045         (MULTILIB_DIRNAMES): Ditto.
18046         (MULTILIB_EXCEPTIONS): Add new exceptions.
18047         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
18048         (MULTILIB_DIRNAMES): Ditto.
18049         (MULTILIB_EXCEPTIONS): Add new exceptions.
18051 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18052             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18054         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
18055         SYMBOL_TINY_ABSOLUTE.
18056         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
18057         SYMBOL_TINY_ABSOLUTE.
18058         (aarch64_expand_mov_immediate): Likewise.
18059         (aarch64_classify_symbol): Likewise.
18060         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
18061         Permit SYMBOL_TINY_ABSOLUTE.
18062         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
18064 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18065             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18067         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
18068         Refactor if/switch.  Replace gcc_assert with if.
18070 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
18072         * config/i386/i386.c (initial_ix86_tune_features): Enable
18073         FP Reassociation for AMD bdver1 and bdver2.
18075 2013-05-29  Martin Jambor  <mjambor@suse.cz>
18077         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
18078         and IMAGPART_EXPR do not occur within other handled_components.
18080 2013-05-29  Richard Biener  <rguenther@suse.de>
18082         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
18083         access on whether the use is in the BB we currently try to
18084         vectorize.
18085         (vect_bb_vectorization_profitable_p): Pass the BB we currently
18086         vectorize to vect_bb_slp_scalar_cost.
18088 2013-05-29  Richard Biener  <rguenther@suse.de>
18090         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
18091         computing scalar cost offsetted by stmts that are kept live
18092         by scalar uses.
18093         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
18094         for computation of scalar cost.
18096 2013-05-28  Steve Ellcey  <sellcey@mips.com>
18098         * config/mips/mips-cpus.def (mips32r2): Change processor type.
18100 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
18102         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
18103         array notation built-in reduction functions.
18104         * doc/passes.texi (Passes): Added documentation about changes done
18105         for Cilk Plus.
18106         * doc/invoke.texi (C Dialect Options): Added documentation about
18107         the -fcilkplus flag.
18108         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
18109         (BUILTINS_DEF): Depend on cilkplus.def.
18110         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
18111         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
18112         * cilkplus.def: New file.
18114 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
18116         PR rtl-optimization/57439
18117         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
18119 2013-05-28  Easwaran Raman  <eraman@google.com>
18121         PR tree-optimization/57337
18122         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
18123         (find_insert_point): Correctly identify the insertion point
18124         when two statements with the same UID is compared.
18126 2013-05-28  Richard Biener  <rguenther@suse.de>
18128         PR tree-optimization/56787
18129         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
18130         from the list of data references.
18131         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
18132         clobbers.
18133         (vect_analyze_loop_operations): Likewise.
18134         (vect_transform_loop): Remove clobbers.
18136 2013-05-28  Martin Jambor  <mjambor@suse.cz>
18138         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
18139         and REALPART_EXPRs have scalar type.
18141 2013-05-28  Richard Biener  <rguenther@suse.de>
18143         PR tree-optimization/57411
18144         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
18145         virtual operands.
18146         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
18147         virtual operand propagation.
18149 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18151         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
18152         destination register for bmasksi_vis.
18153         (vector_init_bshuffle): Likewise.
18154         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
18156 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18158         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
18159         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
18160         mode if the instruction isn't available in the original mode.
18161         * config/sparc/sparc.opt (mfix-ut699): New option.
18162         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
18163         (divdf3): Turn into expander.
18164         (divdf3_nofix): New insn.
18165         (divdf3_fix): Likewise.
18166         (divsf3): Disable if -mfix-ut699.
18167         (sqrtdf2): Turn into expander.
18168         (sqrtdf2_nofix): New insn.
18169         (sqrtdf2_fix): Likewise.
18170         (sqrtsf2): Disable if -mfix-ut699.
18172 2013-05-27  Richard Biener  <rguenther@suse.de>
18174         PR middle-end/57412
18175         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
18176         block for the new loop.
18178 2013-05-27  Richard Biener  <rguenther@suse.de>
18180         PR tree-optimization/57343
18181         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
18182         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
18183         (number_of_iterations_cond): Do not build the folded tree.
18185 2013-05-27  Richard Biener  <rguenther@suse.de>
18187         Revert
18188         PR middle-end/57381
18189         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
18190         OEP_CONSTANT_ADDRESS_OF retained.
18192         PR tree-optimization/57417
18193         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
18194         for unchanged base.
18195         (set_ssa_val_to): Compare addresses using
18196         get_addr_base_and_unit_offset.
18198 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
18200         PR rtl-optimization/56833
18201         * postreload.c (move2add_record_mode): New function.
18202         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
18203         (move2add_use_add2_insn): Use move2add_record_sym_value.
18204         (move2add_use_add3_insn): Likewise.
18205         (reload_cse_move2add): Use move2add_valid_value_p and
18206         move2add_record_mode.  Invalidate call-clobbered and REG_INC
18207         affected regs by setting reg_mode to VOIDmode.
18208         (move2add_note_store): Don't pretend the inside of a SUBREG is
18209         the actual destination.  Invalidate single/leading registers by
18210         setting reg_mode to VOIDmode.
18211         Use move2add_record_sym_value, move2add_valid_value_p and
18212         move2add_record_mode.
18214 2013-05-27  Richard Biener  <rguenther@suse.de>
18216         PR tree-optimization/57396
18217         * tree-affine.c (double_int_constant_multiple_p): Properly
18218         return false for val == 0 and div != 0.
18220 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18222         * config/mips/mips.h: Use #elif in preprocessor conditions.
18224 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18226         PR target/53916
18227         * config/mips/constraints.md (kl): New constraint.
18228         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
18229         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
18230         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
18231         from LO for MIPS16.
18232         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
18234 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18236         PR target/55777
18237         * config/mips/mips.c (mips_can_inline_p): New function.
18238         (TARGET_CAN_INLINE_P): Define.
18240 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
18242         * sched-int.h (ds_t, dw_t): Make unsigned int.
18243         Fix documentation that describes how all the ds_t bits are used.
18244         Reserve the last bit for delayed-branch scheduling.
18245         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
18246         (BITS_PER_DEP_WEAK): Fix definition and documentation.
18247         (gen_dep_weak_1): Remove prototype.
18248         * sched-deps.c (get_dep_weak_1): Make static.
18249         * target.def (speculate_insn, needs_block_p, gen_spec_check,
18250         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
18251         * doc/tm.texi: Regenerate.
18252         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
18254 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
18256         PR debug/56950
18257         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
18259 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
18260             Sandra Loosemore  <sandra@codesourcery.com>
18262         * config.gcc (powerpc-*): Allow native for with-cpu.
18264 2013-05-24  Jeff Law  <law@redhat.com>
18266         PR tree-optimization/57124
18267         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
18268         conversion feeding a condition if the range has an overflow
18269         if -fstrict-overflow.  Add warnings for when we do make the
18270         transformation.
18272 2013-05-24  Dehao Chen  <dehao@google.com>
18274         * tree-cfg.c (locus_discrim_map): Fix the typo.
18275         (locus_discrim_hasher): Likewise.
18276         (locus_discrim_hasher::hash): Likewise.
18277         (locus_discrim_hasher::equal): Likewise.
18279 2013-05-24  Martin Jambor  <mjambor@suse.cz>
18281         PR tree-optimization/57294
18282         * cgraph.h (ipa_record_stmt_references): Declare.
18283         * cgraphbuild.c (ipa_record_stmt_references): New function.
18284         (build_cgraph_edges): Use ipa_record_stmt_references.
18285         (rebuild_cgraph_edges): Likewise.
18286         (cgraph_rebuild_references): Likewise.
18287         * ipa-prop.c (ipa_modify_call_arguments): Discard references
18288         associated with the old statement and build references from the
18289         newly built statements.
18290         * ipa-ref.c (ipa_remove_stmt_references): New function.
18291         * ipa-ref.h (ipa_remove_stmt_references): Declare.
18293 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
18295         * lra-constraints.c (emit_spill_move): Use smaller mode for
18296         mem-mem moves.
18297         (check_and_process_move): Consider mem-reg moves for secondary
18298         too.
18299         (curr_insn_transform): Don't lose insns emitted before for
18300         secondary memory moves.
18301         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
18302         reg set up in the current insn.
18304 2013-05-24  Dehao Chen  <dehao@google.com>
18306         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
18307         hash function.
18308         (locus_descrim_hasher::equal): Likewise.
18309         (build_gimple_cfg): New discriminator assignment algorithm.
18310         (make_edges): Likewise.
18311         (next_discriminator_for_locus): Likewise.
18312         (same_line_p): Likewise.
18313         (assign_discriminators): Likewise.
18314         (make_cond_expr_edges): Likewise.
18315         (make_gimple_switch_edges): Likewise.
18316         (make_goto_expr_edges): Likewise.
18317         (make_gimple_asm_edges): Likewise.
18319 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
18321         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
18322         X format specifier to only display bottom 16 bits.
18323         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
18324         immediate to match for operand 2, since it will be masked.
18326 2013-05-24  Richard Biener  <rguenther@suse.de>
18328         PR tree-optimization/57287
18329         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
18330         all SSA names that occur in abnormal PHIs.
18332 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
18334         PR tree-ssa/57385
18335         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
18336         that index is not negative.
18338 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
18340         PR rtl-optimization/55177
18341         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
18342         (simplify_byte_swapping_operation): New.
18343         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
18344         (simplify_relational_operation_1): Deal with BSWAP.
18346 2013-05-23  Richard Henderson  <rth@redhat.com>
18348         PR target/56742
18349         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
18350         (ix86_reorg): Call it.
18352 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
18354         PR target/57379
18355         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
18356         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
18357         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
18359 2013-05-23  Christian Bruel  <christian.bruel@st.com>
18361         PR debug/57351
18362         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
18364 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18365             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18367         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
18368         * config/aarch64/constraints.md (Usa): Remove.
18369         * doc/md.texi (AArch64 Usa): Remove.
18371 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18372             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18374         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
18375         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
18376         * config/aarch64/predicates.md (aarch64_const_address): Remove.
18377         (aarch64_mov_operand): Use aarch64_mov_operand_p.
18379 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
18381         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
18382         instruction (AdvSIMD).
18383         * config/aarch64/aarch64-builtins.c
18384         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
18385         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
18387 2013-05-23  Martin Jambor  <mjambor@suse.cz>
18389         PR middle-end/57347
18390         * tree.h (contains_bitfld_component_ref_p): Declare.
18391         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
18392         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
18393         caller.
18394         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
18395         not access a bit-field.  Assert all final offsets are byte-aligned.
18397 2013-05-23  Richard Biener  <rguenther@suse.de>
18399         PR tree-optimization/57380
18400         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
18401         least one invariant or re-used load.
18402         * passes.c (init_optimization_passes): Move pass_phiprop before
18403         pass_forwprop.
18405 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
18407         * config/aarch64/aarch64-simd.md
18408         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
18410 2013-05-23  Richard Biener  <rguenther@suse.de>
18412         PR middle-end/57381
18413         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
18414         OEP_CONSTANT_ADDRESS_OF retained.
18416 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
18418         PR middle-end/57344
18419         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
18420         don't lower unit.  Handle unit not being always BITS_PER_WORD.
18422 2013-05-23  Richard Biener  <rguenther@suse.de>
18424         PR rtl-optimization/57341
18425         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
18426         instead of true_dependence.
18428 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
18430         * bb-reorder.c (branch_threshold): Make const.
18431         (exec_threshold): Ditto.
18433 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
18434             Pat Haugen  <pthaugen@us.ibm.com>
18435             Peter Bergner  <bergner@vnet.ibm.com>
18437         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
18438         documentation for the power8 crypto builtins.
18440         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
18442         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
18443         macros for defining power8 builtin functions.
18444         (BU_P8V_AV_2): Likewise.
18445         (BU_P8V_AV_P): Likewise.
18446         (BU_P8V_VSX_1): Likewise.
18447         (BU_P8V_OVERLOAD_1): Likewise.
18448         (BU_P8V_OVERLOAD_2): Likewise.
18449         (BU_CRYPTO_1): Likewise.
18450         (BU_CRYPTO_2): Likewise.
18451         (BU_CRYPTO_3): Likewise.
18452         (BU_CRYPTO_OVERLOAD_1): Likewise.
18453         (BU_CRYPTO_OVERLOAD_2): Likewise.
18454         (XSCVSPDP): Fix typo, point to the correct instruction.
18455         (VCIPHER): Add power8 crypto builtins.
18456         (VCIPHERLAST): Likewise.
18457         (VNCIPHER): Likewise.
18458         (VNCIPHERLAST): Likewise.
18459         (VPMSUMB): Likewise.
18460         (VPMSUMH): Likewise.
18461         (VPMSUMW): Likewise.
18462         (VPERMXOR_V2DI): Likewise.
18463         (VPERMXOR_V4SI: Likewise.
18464         (VPERMXOR_V8HI: Likewise.
18465         (VPERMXOR_V16QI: Likewise.
18466         (VSHASIGMAW): Likewise.
18467         (VSHASIGMAD): Likewise.
18468         (VPMSUM): Likewise.
18469         (VPERMXOR): Likewise.
18470         (VSHASIGMA): Likewise.
18472         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
18473         __CRYPTO__ if the crypto instructions are available.
18474         (altivec_overloaded_builtins): Add support for overloaded power8
18475         builtins.
18477         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
18478         support for power8 crypto builtins.
18479         (builtin_function_type): Likewise.
18480         (altivec_init_builtins): Add support for builtins that take vector
18481         long long (V2DI) arguments.
18483         * config/rs6000/crypto.md: New file, define power8 crypto
18484         instructions.
18486 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
18487             Pat Haugen  <pthaugen@us.ibm.com>
18488             Peter Bergner  <bergner@vnet.ibm.com>
18490         * doc/invoke.texi (Option Summary): Add power8 options.
18491         (RS/6000 and PowerPC Options): Likewise.
18493         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
18494         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
18495         wm, wn, wr documentation.
18497         * config/rs6000/constraints.md (wm): New constraint for VSX
18498         registers if direct move instructions are enabled.
18499         (wn): New constraint for no registers.
18500         (wq): New constraint for quad word even GPR registers.
18501         (wr): New constraint if 64-bit instructions are enabled.
18502         (wv): New constraint if power8 vector instructions are enabled.
18503         (wQ): New constraint for quad word memory locations.
18505         * config/rs6000/predicates.md (const_0_to_15_operand): New
18506         constraint for 0..15 for crypto instructions.
18507         (gpc_reg_operand): If VSX allow registers in VSX registers as well
18508         as GPR and floating point registers.
18509         (int_reg_operand): New predicate to match only GPR registers.
18510         (base_reg_operand): New predicate to match base registers.
18511         (quad_int_reg_operand): New predicate to match even GPR registers
18512         for quad memory operations.
18513         (vsx_reg_or_cint_operand): New predicate to allow vector logical
18514         operations in both GPR and VSX registers.
18515         (quad_memory_operand): New predicate for quad memory operations.
18516         (reg_or_indexed_operand): New predicate for direct move support.
18518         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
18519         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
18520         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
18521         (POWERPC_MASKS): Add power8 options.
18522         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
18523         various options.
18525         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
18526         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
18528         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
18529         (-mpower8-fusion): New power8 options.
18530         (-mpower8-fusion-sign): Likewise.
18531         (-mpower8-vector): Likewise.
18532         (-mcrypto): Likewise.
18533         (-mdirect-move): Likewise.
18534         (-mquad-memory): Likewise.
18536         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
18537         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
18538         registers.
18539         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
18540         (rs6000_debug_vector_unit): Add p8_vector.
18541         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
18542         definitions.  Also print fusion state.
18543         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
18544         (rs6000_builtin_mask_calculate): Add power8 builtin support.
18545         (rs6000_option_override_internal): Add support for power8.
18546         (rs6000_common_init_builtins): Add debugging for skipped builtins
18547         if -mdebug=builtin.
18548         (rs6000_adjust_cost): Add power8 support.
18549         (rs6000_issue_rate): Likewise.
18550         (insn_must_be_first_in_group): Likewise.
18551         (insn_must_be_last_in_group): Likewise.
18552         (force_new_group): Likewise.
18553         (rs6000_register_move_cost): Likewise.
18554         (rs6000_opt_masks): Likewise.
18556         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
18557         power8 capable assembler, default to power7 options.
18558         (TARGET_DIRECT_MOVE): Likewise.
18559         (TARGET_CRYPTO): Likewise.
18560         (TARGET_P8_VECTOR): Likewise.
18561         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
18562         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
18563         (VECTOR_MEM_P8_VECTOR_P): Likewise.
18564         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
18565         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
18566         (TARGET_XSCVDPSPN): Likewise.
18567         (TARGET_XSCVSPDPN): Likewsie.
18568         (TARGET_SYNC_HI_QI): Likewise.
18569         (TARGET_SYNC_TI): Likewise.
18570         (MASK_CRYPTO): Likewise.
18571         (MASK_DIRECT_MOVE): Likewise.
18572         (MASK_P8_FUSION): Likewise.
18573         (MASK_P8_VECTOR): Likewise.
18574         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
18575         temporary used by some of the direct move instructions to get two FP
18576         temporary registers does not force creation of a stack frame.
18577         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
18578         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
18579         that any VSX registers are tieable, even if they are also an
18580         Altivec vector mode.
18581         (r6000_reg_class_enum): Add wm, wr, wv constraints.
18582         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
18583         (RS6000_BTM_CRYPTO): Likewise.
18584         (RS6000_BTM_COMMON): Likewise.
18586         * config/rs6000/rs6000.md (cpu attribute): Add power8.
18587         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
18588         (enum rs6000_vector): Add power8 vector support.
18590 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18592         PR target/19599
18593         PR target/57340
18594         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
18595         (any_sibcall_could_use_r3): this and handle indirect calls.
18596         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
18598 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18600         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
18602 2013-05-22  Richard Biener  <rguenther@suse.de>
18604         PR middle-end/57349
18605         * profile.c (branch_prob): Do not split blocks that are
18606         abnormally receiving from ECF_RETURNS_TWICE functions.
18608 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18610         * recog.c (offsettable_address_addr_space_p): Fix calculation of
18611         address mode.  Move pointer mode initialization to the same place.
18613 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
18615         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
18616         while it has any effect.
18618 2013-05-21  Easwaran Raman  <eraman@google.com>
18620         PR tree-optimization/57322
18621         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
18622         UID of the statement added to the BB to be 1.
18624 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
18626         PR tree-optimization/57331
18627         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
18628         of conversion from pointer type to integral type with integer.
18630 2013-05-21  Martin Jambor  <mjambor@suse.cz>
18632         PR lto/57289
18633         * ipa-prop.c (ipa_read_node_info): Process param_used and
18634         controlled_uses in the same order as when writing.
18636 2013-05-21  Magnus Granberg  <baldrick@free.fr>
18638         PR plugins/56754
18639         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
18641 2013-05-21  Richard Biener  <rguenther@suse.de>
18643         PR tree-optimization/57318
18644         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
18645         estimate stmts with side-effects as likely eliminated.
18647 2013-05-21  Richard Biener  <rguenther@suse.de>
18649         PR tree-optimization/57330
18650         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
18651         preserve the call stmts fntype.
18653 2013-05-21  Richard Biener  <rguenther@suse.de>
18655         PR tree-optimization/57303
18656         * tree-ssa-sink.c (statement_sink_location): Improve killing
18657         stmt detection and properly handle self-assignments.
18659 2013-05-21  Christian Bruel  <christian.bruel@st.com>
18661         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
18662         spanning registers. LEAF_REG_REMAP is supported only for contiguous
18663         registers. Set register size out of the PARALLEL loop.
18665 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
18667         PR target/56547
18668         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
18669         (*fmasf4, *fmasf4_media): New insns.
18671 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
18673         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
18674         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
18675         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
18676         (mips_idiv_insns): Update the comments to say that the returned
18677         instruction counts are in units of BASE_INSN_LENGTH.
18678         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
18679         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
18680         using 2 rather than 4 as the length of indirect MIPS16 and
18681         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
18682         length of a NOP.  Don't divide MIPS16 lengths by 2.
18683         (mips16_split_long_branches): Assume a branch is long if the
18684         length is greater than 4 rather than 8.
18685         * config/mips/mips.md (length): Give MIPS16 lengths directly,
18686         rather than multiplying them by 2.  Multiply instruction counts
18687         by BASE_INSN_LENGTH rather than 4.
18688         (*jump_mips16, tls_get_tp_mips16_<mode>)
18689         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
18691 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
18693         * config/mips/mips.md (extended_mips16): Remove branch case.
18694         (length): Remove duplicated extended_mips16 test.
18696 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
18698         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
18700 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
18702         * recog.h (Recog_data): Rename to...
18703         (recog_data_d): ...this.
18704         (recog_data): Update accordingly.
18705         * recog.c (recog_data): Likewise.
18706         * reload.c (save_recog_data): Likewise.
18707         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
18708         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
18710 2013-05-17  Julian Brown  <julian@codesourcery.com>
18712         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
18713         found in a REG_EQUAL note, invalidate it.
18715 2013-05-17  Easwaran Raman  <eraman@google.com>
18717         * tree-ssa-reassoc.c (find_insert_point): New function.
18718         (insert_stmt_after): Likewise.
18719         (get_def_stmt): Likewise.
18720         (ensure_ops_are_available): Likewise.
18721         (not_dominated_by): Likewise.
18722         (rewrite_expr_tree): Do not move statements beyond what is
18723         necessary. Remove call to swap_ops_for_binary_stmt...
18724         (reassociate_bb): ... and move it here.
18725         (build_and_add_sum): Assign UIDs for new statements.
18726         (linearize_expr): Likewise.
18727         (do_reassoc): Renumber gimple statement UIDs.
18729 2013-05-17  Jan Hubicka  <jh@suse.cz>
18731         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
18732         weakrefs.
18733         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
18734         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
18735         weakrefs.
18736         (output_weakrefs): Update.
18738 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
18739             Martin Jambor  <mjambor@suse.cz>
18741         PR middle-end/57276
18742         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
18743         value that corresponds to the given aggval is found in values vector.
18745 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
18747         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
18748         sse, sse2, sse3, ssse3 and sse4a flags to options.
18750 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
18752         * gengtype-state.c: (s_expr_writer): New class, to handle
18753         prettifying of output layout of s-expressions.
18754         (state_writer): New class, to write out gtype.state.
18755         (state_written_type_count): Move this variable into member data of
18756         state_writer.
18757         (s_expr_writer::s_expr_writer): New code: constructor for new class
18758         (state_writer::state_writer(): ditto
18759         (s_expr_writer::write_new_line): New function
18760         (s_expr_writer::write_any_indent): ditto
18761         (s_expr_writer::begin_s_expr): ditto
18762         (s_expr_writer::end_s_expr): ditto
18763         (write_state_fileloc): convert to method of state_writer...
18764         (state_writer:: write_state_fileloc): ...and use methods of
18765         s_expr_writer to write indentation into the gtype.state output file
18766         to visually represent the hierarchical structure of the list
18767         structures
18768         (write_state_fields): ditto, renaming to...
18769         (state_writer::write_state_fields)
18770         (write_state_a_string): ditto, renaming to...
18771         (state_writer::write_state_a_string)
18772         (write_state_string_option): ditto, renaming to...
18773         (state_writer::write_state_string_option)
18774         (write_state_type_option): ditto, renaming to...
18775         (state_writer::write_state_type_option)
18776         (write_state_nested_option): ditto, renaming to...
18777         (state_writer::write_state_nested_option)
18778         (write_state_option): ditto, renaming to...
18779         (state_writer::write_state_option)
18780         (write_state_options): ditto, renaming to...
18781         (state_writer::write_state_options)
18782         (write_state_lang_bitmap): ditto, renaming to...
18783         (state_writer::write_state_lang_bitmap)
18784         (write_state_version): ditto, renaming to...
18785         (state_writer::write_state_version)
18786         (write_state_scalar_type): ditto, renaming to...
18787         (state_writer::write_state_scalar_type)
18788         (write_state_string_type): ditto, renaming to...
18789         (state_writer::write_state_string_type)
18790         (write_state_undefined_type): ditto, renaming to...
18791         (state_writer::write_state_undefined_type)
18792         (write_state_struct_union_type): ditto, renaming to...
18793         (state_writer::write_state_struct_union_type)
18794         (write_state_struct_type): ditto, renaming to...
18795         (state_writer::write_state_struct_type)
18796         (write_state_user_struct_type): ditto, renaming to...
18797         (state_writer::write_state_user_struct_type)
18798         (write_state_lang_struct_type): ditto, renaming to...
18799         (state_writer::write_state_lang_struct_type)
18800         (write_state_param_struct_type): ditto, renaming to...
18801         (state_writer::write_state_param_struct_type)
18802         (write_state_pointer_type): ditto, renaming to...
18803         (state_writer::write_state_pointer_type)
18804         (write_state_array_type): ditto, renaming to...
18805         (state_writer::write_state_array_type)
18806         (write_state_gc_used): ditto, renaming to...
18807         (state_writer::write_state_gc_used)
18808         (write_state_common_type_content): ditto, renaming to...
18809         (state_writer::write_state_common_type_content)
18810         (write_state_type): ditto, renaming to...
18811         (state_writer::write_state_type)
18812         (write_state_pair_list): ditto, renaming to...
18813         (state_writer::write_state_pair_list)
18814         (write_state_pair): ditto, renaming to...
18815         (state_writer::write_state_pair)
18816         (write_state_typedefs): ditto, renaming to...
18817         (state_writer::write_state_typedefs)
18818         (write_state_structures): ditto, renaming to...
18819         (state_writer::write_state_structures)
18820         (write_state_param_structs): ditto, renaming to...
18821         (state_writer::write_state_param_structs)
18822         (write_state_variables): ditto, renaming to...
18823         (state_writer::write_state_variables)
18824         (write_state_srcdir): ditto, renaming to...
18825         (state_writer::write_state_srcdir)
18826         (write_state_files_list): ditto, renaming to...
18827         (state_writer::write_state_files_list)
18828         (write_state_languages): ditto, renaming to...
18829         (state_writer::write_state_languages)
18830         (write_state): create a state_writer instance and use it when
18831         writing out the state file
18833 2013-05-17  Mike Stump  <mikestump@comcast.net>
18835         PR rtl-optimization/57304
18836         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
18837         accessing DF_REF_REAL_LOC.
18839 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
18841         PR rtl-optimization/57281
18842         PR rtl-optimization/57300
18843         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
18844         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
18845         what the other splitter did if the registers are dead.
18847 2013-05-17  Richard Biener  <rguenther@suse.de>
18849         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
18850         MEM_REF offsets.
18852 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
18854         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
18855         linking.
18857 2013-05-17  Marek Polacek  <polacek@redhat.com>
18859         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
18860         length when doing non-zero store of storing '\0' to '\0'.
18862 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
18864         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
18865         vect_external_def oprnd1 with loop_vinfo, try to emit
18866         optional cast, negation and and stmts on the loop preheader
18867         edge instead of into the pattern def seq.
18869         PR tree-optimization/57051
18870         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
18871         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
18873 2013-05-16  Nick Clifton  <nickc@redhat.com>
18875         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
18876         (rl78_is_naked_func): New function.
18877         (rl78_expand_prologue): Skip prologue generation for naked functions.
18878         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
18879         * doc/extend.texi (naked): Add RL78 to the list of processors
18880         that supports this attribute.
18882 2013-05-16  Jeff Law  <law@redhat.com>
18884         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
18886 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
18888         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
18889         cache parameters using detect_caches_amd also for CYRIX,
18890         NSC and TM2 signatures.
18892 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
18893             Dzianis Kahanovich  <mahatma@eu.by>
18895         PR target/45359
18896         PR target/46396
18897         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
18898         VIA/Centaur processors and determine their cache parameters
18899         using detect_caches_amd.
18901 2013-05-16  Teresa Johnson  <tejohnson@google.com>
18903         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
18904         (rtl_verify_edges): New function.
18905         (rtl_verify_bb_insns): Ditto.
18906         (rtl_verify_bb_pointers): Ditto.
18907         (rtl_verify_bb_insn_chain): Ditto.
18908         (rtl_verify_fallthru): Ditto.
18909         (rtl_verify_bb_layout): Ditto.
18910         (rtl_verify_flow_info_1): Outline checks into new functions.
18911         (rtl_verify_flow_info): Ditto.
18913 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
18915         * cfghooks.c (copy_bbs): Add update_dominance argument.
18916         * cfghooks.h (copy_bbs): Update prototype.
18917         * tree-cfg.c (gimple_duplicate_sese_region):
18918         Add update_dominance argument.
18919         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
18920         * tree-ssa-loop-ch.c (copy_loop_headers): Update
18921         gimple_duplicate_sese_region call.
18922         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
18923         Update copy_bbs call.
18924         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
18925         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
18927 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
18929         * tree-vectorizer.h (NUM_PATTERNS): Increment.
18930         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
18931         vect_recog_rotate_pattern.
18932         (vect_recog_rotate_pattern): New function.
18934 2013-05-16  Jason Merrill  <jason@redhat.com>
18936         * Makefile.in (LLINKER): New variable.
18937         (mostlyclean): Remove link mutex.
18938         * configure.ac: Handle --enable-link-mutex.
18939         * lock-and-run.sh: New script.
18941 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18943         PR target/19599
18944         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
18945         for NULL decl.
18947 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18949         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
18951 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
18953         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
18954         * config/arm/arm.c (next_consecutive_mem): New function.
18955         (gen_movmem_ldrd_strd): Likewise.
18956         * config/arm/arm.md (movmemqi): Update condition and code.
18957         (unaligned_loaddi, unaligned_storedi): New patterns.
18959 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18961         * config.gcc: Obsolete *-*-solaris2.9*.
18962         * doc/install.texi (Specific, *-*-solaris2*): Document it.
18964 2013-05-16  Richard Biener  <rguenther@suse.de>
18966         * passes.c (init_optimization_passes): Move pass_parallelize_loops
18967         earlier, after GRAPHITE transforms and IV canonicalization.
18969 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
18971         * omp-low.c (extract_omp_for_data): For collapsed loops,
18972         if at least one of the loops is known at compile time to
18973         iterate zero times, set count to 0.
18974         (expand_omp_regimplify_p): New function.
18975         (expand_omp_for_generic): For collapsed loops, if at least
18976         one of the loops isn't known to iterate at least once,
18977         add runtime check with setting count to 0.
18978         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
18979         For unsigned types if it isn't known at compile time that
18980         the loop will iterate at least once, add runtime check to bypass
18981         the whole loop if initial condition isn't true.
18983 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
18985         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
18987 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
18989         PR middle-end/57286
18990         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
18991         transformations to avoid an infinite loop.
18993 2013-05-16  Marek Polacek  <polacek@redhat.com>
18995         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
18997 2013-05-15  Leif Ekblad  <leif@rdos.net>
18999         * config/i386/i386.c (ix86_decompose_address): Use
19000         DEFAULT_TLS_SEG_REG to access TLS segment register.
19001         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
19002         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
19003         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
19005 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
19007         PR target/57260
19008         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
19009         sibling calls to functions that would normally be lazily bound,
19010         unless $gp is call-clobbered.
19012 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
19014         * config/i386/i386.c (ix86_option_override_internal): Update
19015         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
19016         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
19017         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
19018         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
19019         of TARGET_3DNOW.
19020         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
19022 2013-05-15  Andreas Schwab  <schwab@suse.de>
19024         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
19025         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
19026         third operand.
19028 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19030         * loop-unroll.c (report_unroll_peel): Check decision before
19031         emitting unroll/peel message.
19033 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19035         * function.h (has_bb_partition): New rtl_data flag.
19036         (bb_reorder_complete): Ditto.
19037         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
19038         instead of flag_reorder_blocks_and_partition.
19039         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
19040         with some enhancements.
19041         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
19042         * bb-reorder.c (connect_traces): Check for has_bb_partition
19043         instead of flag_reorder_blocks_and_partition.
19044         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
19045         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
19046         verify_hot_cold_block_grouping.
19047         (partition_hot_cold_basic_blocks): Set has_bb_partition.
19049 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19051         PR target/19599
19052         * config/arm/predicates.md (call_insn_operand): New predicate.
19053         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
19054         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
19055         if insn is not a tail call.
19056         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
19057         registers.
19058         * config/arm/arm.h (enum reg_class): New caller save register class.
19059         (REG_CLASS_NAMES): Likewise.
19060         (REG_CLASS_CONTENTS): Likewise.
19061         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
19062         without decls.
19064 2013-05-15  Richard Biener  <rguenther@suse.de>
19066         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
19067         of MSG_OPTIMIZED_LOCATIONS.
19068         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
19069         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
19070         message.
19071         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
19072         of MSG_OPTIMIZED_LOCATIONS.
19073         (execute_vect_slp): Likewise.
19074         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
19075         (vect_create_cond_for_alias_checks): Likewise.
19076         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
19077         (vect_recog_widen_mult_pattern): Likewise.
19078         (vect_recog_widen_sum_pattern): Likewise.
19079         (vect_recog_over_widening_pattern): Likewise.
19080         (vect_recog_widen_shift_pattern): Likewise.
19081         (vect_recog_vector_vector_shift_pattern): Likewise.
19082         (vect_recog_divmod_pattern): Likewise.
19083         (vect_recog_mixed_size_cond_pattern): Likewise.
19084         (vect_recog_bool_pattern): Likewise.
19085         (vect_pattern_recog_1): Likewise.
19087 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19089         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
19090         non-functions to builtin_unreachable.
19091         * ipa-inline-transform.c (inline_call): Do not assert estimates were
19092         correct when new direct edges were discovered.
19094 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19096         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
19097         header, print symbol order instead of node uid, print more information
19098         about indirect edge targets.
19099         (ipa_make_edge_direct_to_target): Print symbol order instead of node
19100         uids.
19101         (ipa_make_edge_direct_to_target): Likewise.
19102         (remove_described_reference): Likewise.
19103         (propagate_controlled_uses): Likewise.
19104         (ipa_print_node_params): Also print symbol order.
19105         (ipcp_transform_function): Print symbol order instead of node uids.
19106         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
19107         (cgraph_get_create_real_symbol_node): Likewise.
19108         * ipa-cp.c (print_lattice): Likewise.
19109         (print_all_lattices): Likewise.
19110         (determine_versionability): Likewise.
19111         (initialize_node_lattices): Likewise.
19112         (estimate_local_effects): Likewise.
19113         (update_profiling_info): Likewise.
19114         (create_specialized_node): Likewise.
19115         (perhaps_add_new_callers): Likewise.
19116         (decide_about_value): Likewise.
19117         (decide_whether_version_node): Likewise.
19118         (identify_dead_nodes): Likewise.
19119         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
19120         (dump_inline_summary): Likewise.
19121         (estimate_node_size_and_time): Likewise.
19122         (inline_analyze_function): Likewise.
19123         * ipa-inline.c (report_inline_failed_reason): Likewise.
19124         (want_early_inline_function_p): Likewise.
19125         (edge_badness): Likewise.
19126         (update_edge_key): Likewise.
19127         (inline_small_functions): Likewise.  Add dumping of order to two other
19128         dumps.
19129         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
19130         instead of node uids.
19131         (propagate_pure_const): Likewise.
19132         (propagate_pure_const): Likewise.
19133         * ipa-utils.c (dump_cgraph_node_set): Likewise.
19134         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
19135         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
19136         of node uids.
19137         * tree-pretty-print.c (dump_function_header): Likewise.
19138         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
19139         Print symbol order instead of node uids.
19141 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19143         * config/s390/s390.c (s390_register_move_cost): Don't impose the
19144         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
19146 2013-05-15  Richard Biener  <rguenther@suse.de>
19148         PR tree-optimization/57275
19149         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
19150         return value for fail to do runtime alias checks for gather loads.
19152 2013-05-15  Jan Hubicka  <jh@suse.cz>
19154         PR lto/57038
19155         PR lto/47375
19156         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
19157         weakrefs are not external.
19158         (lto_symtab_merge_decls): Fix thinko when dealing with
19159         non-lto_symtab decls.
19160         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
19161         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
19162         * varpool.c (dump_varpool_node): Dump more flags.
19164 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19166         * config/i386/i386.c (processor_alias_table): Add instruction
19167         FSGSBASE for AMD bdver3 architecture.
19169 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19171         * tree.c (warn_deprecated_use): Print file:line using locus color.
19172         * diagnostic.c (diagnostic_report_current_module): Print file:line
19173         and file:line:column using locus color.
19175 2013-05-14  Mike Stump  <mikestump@comcast.net>
19177         * gdbinit.in: Add __null.
19179 2013-05-14  Mike Stump  <mikestump@comcast.net>
19181         * recog.h: Rename struct recog_data to Recog_data.
19182         * recog.c: Likewise.
19183         * reload.c (can_reload_into): Likewise.
19184         * config/picochip/picochip.c: Likewise.
19186 2013-05-14  Mike Stump  <mikestump@comcast.net>
19188         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
19190 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
19192         * resource.h (struct resources): Remove unch_memory member.
19193         (CLEAR_RESOURCE): Don't clear unch_memory.
19194         * resource.c (mark_referenced_resources): Don't set it.
19195         (mark_set_resources): Likewise.
19196         (mark_target_live_regs): Don't clear it.
19197         (init_resource_info): Likewise.
19198         * reorg.c (resource_conflicts_p): Don't compare it.
19199         (redundant_insn): Don't set it.
19201         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
19202         Remove prototypes.
19203         * emit-rtl.c (next_label): Remove unused function.
19204         (skip_consecutive_labels, link_cc0_insns): Move to ...
19205         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
19206         only place where these functions are used, and make them static.
19208 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
19210         * fold-const.c (fold_negate_expr): Handle vectors.
19211         (fold_truth_not_expr): Make it static.
19212         (fold_invert_truthvalue): New static function.
19213         (invert_truthvalue_loc): Handle vectors. Do not call
19214         fold_truth_not_expr directly.
19215         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
19216         <TRUTH_NOT_EXPR>: Do not cast to boolean.
19217         (fold_comparison): Handle vector constants.
19218         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
19219         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
19220         * tree.h (fold_truth_not_expr): Remove declaration.
19222 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
19224         * config/aarch64/aarch64-simd.md
19225         (aarch64_vcond_internal<mode>): Rename to...
19226         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
19227         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
19228         float modes. Clarify all iterator modes.
19229         (vcond<mode><mode>): Use new name for vcond expanders.
19230         (vcond<v_cmp_result><mode>): Likewise.
19231         (vcondu<mode><mode>: Likewise.
19232         * config/aarch64/iterators.md (VDQF_COND): New.
19234 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
19236         PR bootstrap/57266
19237         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
19238         variable for the shift amount. Check that we shift by non-negative
19239         amounts.
19241 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
19243         PR target/42017
19244         * config/arm/arm.h (EPILOGUE_USES): Only return true
19245         for LR_REGNUM after epilogue_completed.
19247 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
19249         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
19250         is error_mark_node.
19252 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19254         PR target/57261
19255         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
19256         and Solaris 11+/x86 with gld.
19257         * configure: Regenerate.
19259 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19261         * expmed.c (expand_shift_1): Canonicalize rotates by
19262         constant bitsize / 2 to bitsize - 1.
19263         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
19264         case ROTATERT>: Likewise.
19266         Revert:
19267         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
19269         * config/i386/i386.md (rotateinv): New code attr.
19270         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
19271         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
19272         roll $31, %eax, etc.
19274 2013-05-14  Richard Biener  <rguenther@suse.de>
19276         PR middle-end/57235
19277         * tree-eh.c (sink_clobbers): Give up for successors with
19278         multiple predecessors and no virtual uses.
19280 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
19282         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
19283         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
19285 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19287         PR middle-end/57251
19288         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
19289         the case when both op0 and op1 have VOIDmode.
19291 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
19293         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
19294         in multiply-accumulate mode.
19296 2013-05-13  Guozhi Wei  <carrot@google.com>
19298         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
19300 2013-05-13  Kai Tietz  <ktietz@redhat.com>
19302         PR target/56975
19303         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
19304         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
19305         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
19306         * config/i386/i386.c (ix86_option_override_internal): Likewise.
19307         (ix86_expand_prologue): Likewise.
19308         (ix86_expand_split_stack_prologue): Likewise.
19309         (legitimate_pic_address_disp_p): Likewise.
19310         (legitimize_pic_address): Likewise.
19311         (legitimize_tls_address): Likewise.
19312         (legitimize_pe_coff_symbol): Likewise.
19313         (output_pic_addr_const): Likewise.
19314         (construct_plt_address): Likewise.
19315         (ix86_expand_call): Likewise.
19316         (x86_output_mi_thunk): Likewise.
19317         (x86_function_profiler): Likewise.
19319 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
19321         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
19322         similar switch cases.
19323         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
19324         (aarch64_simd_mov_to_<mode>low): Delete.
19325         (aarch64_simd_mov_to_<mode>high): Delete.
19326         (move_lo_quad_<mode>): Add w<-r alternative.
19327         (aarch64_simd_move_hi_quad_<mode>): Likewise.
19328         (aarch64_simd_mov_from_*): Update type attribute.
19329         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
19330         statement.
19332 2013-05-13  Jan Hubicka  <jh@suse.cz>
19334         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
19335         * config/i386/i386.c (ix86_compute_frame_layout,
19336         ix86_expand_epilogue, emit_i387_cw_initialization,
19337         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
19338         ix86_local_alignment): Fix use of size/speed predicates.
19340 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
19342         PR tree-optimization/45216
19343         PR tree-optimization/57157
19344         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
19345         the (-Y) & (B - 1) variant if OP is |.
19346         * expmed.c (expand_shift_1): For rotations by const0_rtx just
19347         return shifted.  Use (-op1) & (prec - 1) as other_amount
19348         instead of prec - op1.
19350 2013-05-13  Martin Jambor  <mjambor@suse.cz>
19352         PR middle-end/42371
19353         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
19354         (ipa_constant_data): New type.
19355         (ipa_jump_func): Use ipa_constant_data to hold information about
19356         constant jump functions.
19357         (ipa_get_jf_constant): Adjust to jump function type changes.
19358         (ipa_get_jf_constant_rdesc): New function.
19359         (ipa_param_descriptor): New field controlled_uses.
19360         (ipa_get_controlled_uses): New function.
19361         (ipa_set_controlled_uses): Likewise.
19362         * ipa-ref.h (ipa_find_reference): Declare.
19363         * ipa-prop.c (ipa_cst_ref_desc): New type.
19364         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
19365         changes.
19366         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
19367         New parameter cs.  Adjust all callers.
19368         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
19369         (remove_described_reference): New function.
19370         (jfunc_rdesc_usable): Likewise.
19371         (try_make_edge_direct_simple_call): Decrement controlled use count,
19372         attempt to remove reference if it hits zero.
19373         (combine_controlled_uses_counters): New function.
19374         (propagate_controlled_uses): Likewise.
19375         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
19376         (ipa_edge_duplication_hook): Duplicate reference descriptions.
19377         (ipa_print_node_params): Print described use counter.
19378         (ipa_write_jump_function): Adjust to jump function type changes.
19379         (ipa_read_jump_function): New parameter CS, pass it to
19380         ipa_set_jf_constant.  Adjust caller.
19381         (ipa_write_node_info): Stream controlled use count
19382         (ipa_read_node_info): Likewise.
19383         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
19384         asserting.
19385         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
19386         count.  Remove cloning-added reference if it reaches zero.
19387         * ipa-ref.c (ipa_find_reference): New function.
19389 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19391         * config/i386/i386.c (processor_target_table): Modified default
19392         alignment values for AMD BD and BT architectures.
19394 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
19396         * tree-vect-generic.c (uniform_vector_p): Move ...
19397         * tree.c (uniform_vector_p): ... here.
19398         * tree.h (uniform_vector_p): Declare it.
19399         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
19400         into a scalar.
19402 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
19404         PR tree-optimization/57230
19405         * tree-ssa-strlen.c (handle_char_store): Record length for
19406         array store from STRING_CST.
19408         PR tree-optimization/57230
19409         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
19410         check.
19412 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
19414         * config/epiphany/epiphany.c (epiphany_init): Check size of
19415         NUM_MODES_FOR_MODE_SWITCHING.
19416         (epiphany_expand_prologue):
19417         Remove CONFIG_REGNUM initial value handling code.
19418         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
19419         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
19420         (emit_set_fp_mode, epiphany_mode_after): Likewise.
19421         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
19422         Don't return 1 for FP_MODE_NONE.
19423         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
19424         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
19425         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
19426         * config/epiphany/epiphany.md (save_config): New pattern.
19428 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
19430         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
19432 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
19434         * config/i386/i386.md (memory): Handle sseishft1.
19435         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
19436         (*vec_extractv2di_1): Ditto.
19438 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
19440         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
19441         saved registers.
19443 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
19445         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
19446         Add mthumb/march=armv7-a multilib.
19447         Add mthumb/march=armv7-r multilib.
19448         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
19450 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
19452         * config/v850/t-rtems: Add more multilibs.
19454 2013-05-10  Richard Biener  <rguenther@suse.de>
19456         PR tree-optimization/57214
19457         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
19458         not propagate from SSA names that occur in abnormal PHI nodes.
19460 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
19462         * stor-layout.c (element_precision): New function.
19463         * machmode.h (element_precision): Declare it.
19464         * tree.c (build_minus_one_cst): New function.
19465         (element_precision): Likewise.
19466         * tree.h (build_minus_one_cst): Declare new function.
19467         (element_precision): Likewise.
19468         * fold-const.c (operand_equal_p): Use element_precision.
19469         (fold_binary_loc): Handle vector types.
19470         * convert.c (convert_to_integer): Use element_precision.
19471         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
19472         separately.
19474 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
19476         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
19477         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
19478         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
19479         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
19480         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
19481         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
19482         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
19483         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
19484         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
19485         (Uuw8): New constraints.
19486         (Usb4): Move into alphabetical order.
19487         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
19488         (sd8_operand, ub8_operand, uw8_operand): New predicates.
19489         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
19490         previously unnamed patterns.
19491         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
19492         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
19493         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
19494         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
19495         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
19496         of specific lengths.
19498 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
19500         * config/i386/i386.md (rotateinv): New code attr.
19501         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
19502         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
19503         roll $31, %eax, etc.
19505         PR tree-optimization/45216
19506         PR tree-optimization/57157
19507         * tree-ssa-forwprop.c (simplify_rotate): New function.
19508         (ssa_forward_propagate_and_combine): Call it.
19510 2013-05-10  Richard Biener  <rguenther@suse.de>
19512         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
19513         disable peeling when we version for aliasing.
19514         (vector_alignment_reachable_p): Honor explicit user alignment.
19515         (vect_supportable_dr_alignment): Likewise.
19516         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
19517         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
19518         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
19519         then peeling to arrange for the cost-model check to come first.
19521 2013-05-10  Alan Modra  <amodra@gmail.com>
19523         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
19524         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
19525         * configure: Regenerate.
19527 2013-05-10  Alan Modra  <amodra@gmail.com>
19529         PR target/55033
19530         * varasm.c (default_elf_select_section): Move !DECL_P check..
19531         (get_named_section): ..to here before calling get_section_name.
19532         Adjust assertion.
19533         (default_section_type_flags): Add DECL_P check.
19534         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
19535         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
19537 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
19539         * config/epiphany/epiphany.c (epiphany_expand_prologue):
19540         When using gen_stack_adjust_str with a register offset, add a
19541         REG_FRAME_RELATED_EXPR note.
19543 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
19545         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
19546         (*vec_extractv4si_zext_mem): Ditto.
19547         (*vec_extractv2di): Add 0->x and x->x alternatives.
19548         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
19549         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
19551 2013-05-09  Jason Merrill  <jason@redhat.com>
19553         N3639 C++1y VLA support
19554         * gimplify.c (gimplify_vla_decl): Don't touch an existing
19555         DECL_VALUE_EXPR.
19557         * tree.c (build_constructor_va): New.
19558         * tree.h: Declare it.
19560 2013-05-09  Martin Jambor  <mjambor@suse.cz>
19562         PR lto/57084
19563         * gimple-fold.c (canonicalize_constructor_val): Call
19564         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
19566 2013-05-09  Jan Hubicka  <jh@suse.cz>
19567             Richard Biener  <rguenther@suse.de>
19569         PR lto/54095
19570         * symtab.c (symtab_make_decl_local): Do not add private names.
19572 2013-05-09  Jan Hubicka  <jh@suse.cz>
19574         PR lto/54095
19575         * symtab.c (insert_to_assembler_name_hash): Handle clones.
19576         (unlink_from_assembler_name_hash): Likewise.
19577         (symtab_prevail_in_asm_name_hash, symtab_register_node,
19578         symtab_unregister_node, symtab_initialize_asm_name_hash,
19579         change_decl_assembler_name): Update.
19581 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
19583         * config/aarch64/aarch64.md: New movtf split.
19584         (*movtf_aarch64): Update.
19585         (aarch64_movdi_tilow): Handle TF modes and rename to
19586         aarch64_movdi_<mode>low.
19587         (aarch64_movdi_tihigh): Handle TF modes and rename to
19588         aarch64_movdi_<mode>high
19589         (aarch64_movtihigh_di): Handle TF modes and rename to
19590         aarch64_mov<mode>high_di
19591         (aarch64_movtilow_di): Handle TF modes and rename to
19592         aarch64_mov<mode>low_di
19593         (aarch64_movtilow_tilow): Remove spurious whitespace.
19594         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
19595         splits.
19596         (aarch64_print_operand): Update.
19598 2013-05-09  Alan Modra  <amodra@gmail.com>
19600         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
19601         powerpc64le.
19602         * configure: Regenerate.
19604 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
19606         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
19607         splitter preparation statements.
19608         * config/i386/sse.md (*vec_extract* splitters): Ditto.
19609         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
19610         adjust_address_nv.
19612 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19614         * gimple-ssa-strength-reduction.c (count_candidates): Change
19615         return value to int.
19616         (analyze_candidates_and_replace): Change type of length to int.
19618 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
19620         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
19621         (*vec_extract<mode>): Use VI12_128 mode iterator.
19622         (*vec_extract<mode>_mem): Ditto.
19623         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
19624         attribute.
19626 2013-05-08  Diego Novillo  <dnovillo@google.com>
19628         PR bootstrap/54659
19630         Revert:
19631         2012-08-17  Diego Novillo  <dnovillo@google.com>
19633         PR bootstrap/54281
19634         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
19635         * config.in: Regenerate.
19636         * configure: Regenerate.
19637         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
19639 2013-05-08  Jan Hubicka  <jh@suse.cz>
19641         PR lto/54095
19642         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
19643         * cgraph.h (symtab_node_base): Add unique_name.
19644         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
19645         input_overwrite_node, input_varpool_node): Stream unique_name.
19646         * cgraphclones.c (cgraph_create_virtual_clone,
19647         cgraph_function_versioning): Set unique_name.
19648         * ipa.c (function_and_variable_visibility): Set unique_name.
19650 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19652         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
19653         (alloc_cand_and_find_basis): Restrict conditional candidate
19654         processing to CAND_MULTs.
19656 2013-05-08  Jan Hubicka  <jh@suse.cz>
19658         PR lto/54095
19659         lto-symtab.c (lto_symtab_symbol_p): New function.
19660         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
19661         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
19662         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
19663         Skip static symbols.
19665 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
19667         PR tree-optimization/57200
19668         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
19669         Only call inform if the preceding warning_at returns true.
19671 2013-05-07  Han Shen  <shenhan@google.com>
19673         * cfgexpand.c (record_or_union_type_has_array_p): New function.
19674         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
19675         * common.opt (fstack-protector-strong): New option.
19676         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
19677         * doc/invoke.texi (Optimization Options): Document
19678         "-fstack-protector-strong".
19679         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
19681 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
19683         * config/mips/mips.c (mips_machine_reorg2): Return 0.
19685 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
19687         * ira.c (update_equiv_regs): Add insn having equiv memory even if
19688         it is not lhs of the insn.
19689         (setup_reg_equiv): Remove insn having equiv memory which it is not
19690         lhs of the insn.
19691         * lra-constraints.c (process_address): Try to improve generation
19692         code for address base + disp.
19693         (lra_constraints): Make correct the code for checking insn setting
19694         up backward equivalence.  Remove insn only if it is in the init
19695         insn list.
19696         * lra-eliminations.c (update_reg_eliminate): Change return value.
19697         (lra_eliminate): Use the result.
19699 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
19701         * config/i386/sse.md (ssescalarnummask): New mode attribute.
19702         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
19703         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
19704         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
19705         register target operands.
19706         (*vec_extractv8hi_sse2): New pattern.
19707         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
19708         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
19709         (*vec_extract<mode>_mem): New insn and split pattern.
19711 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
19713         * config/arm/arm.c (arm_asan_shadow_offset): New function.
19714         (TARGET_ASAN_SHADOW_OFFSET): Define.
19715         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
19716         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
19718 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19720         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
19721         (incr_vec_index): Return -1 if increment not found.
19722         (create_add_on_incoming_edge): Assert if increment not found.
19723         (record_increment): Limit number of increments recorded.
19724         (all_phi_incrs_profitable): Return false if an increment not found.
19725         (replace_profitable_candidates): Don't process increments that were
19726         not recorded.
19727         (analyze_candidates_and_replace): Limit size of incr_vec.
19729 2013-05-07  Richard Biener  <rguenther@suse.de>
19731         * calls.c (special_function_p): setjmp-like functions are leaf.
19732         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
19733         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
19735 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
19737         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
19738         (aarch64_simd_mov<mode>): New expander.
19739         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
19740         (aarch64_simd_mov_to_<mode>high): Likewise.
19741         (aarch64_simd_mov_from_<mode>low): Likewise.
19742         (aarch64_simd_mov_from_<mode>high): Likewise.
19743         (aarch64_dup_lane<mode>): Update.
19744         (aarch64_dup_lanedi): New instruction pattern.
19745         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
19746         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
19748 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19750         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
19751         (replace_mult_candidate): Remove unnecessary argument; remove
19752         unnecessary parameter from call to introduce_cast_before_cand.
19753         (replace_unconditional_candidate): Remove unnecessary parameter
19754         from call to replace_mult_candidate.
19755         (replace_conditional_candidate): Likewise.
19756         (insert_initializers): Use make_temp_ssa_name.
19757         (introduce_cast_before_cand): Remove unnecessary argument; use
19758         make_temp_ssa_name.
19759         (replace_one_candidate): Remove unnecessary argument; remove
19760         unnecessary parameter from calls to introduce_cast_before_cand.
19761         (replace_profitable_candidates): Remove unnecessary parameters
19762         from calls to replace_one_candidate.
19764 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19766         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
19767         phi def as possibly hiding a basis for a CAND_ADD whose operands
19768         have been commuted in the analysis.
19769         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
19771 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19773         * config/aarch64/aarch64.md
19774         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
19775         shift value between 0-4.
19777 2013-05-07  Richard Biener  <rguenther@suse.de>
19779         * double-int.h (rshift): New overload.
19780         * double-int.c (rshift): New function.
19781         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
19782         (create_reference_ops_from_ref): Remove.
19783         (vn_reference_insert): Use shared ops for constructing the
19784         reference and copy it.
19786 2013-05-07  Richard Biener  <rguenther@suse.de>
19788         PR middle-end/57190
19789         * tree-eh.c (sink_clobbers): Properly propagate
19790         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
19792 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
19794         PR tree-optimization/57149
19795         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
19796         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
19797         collect_phi_def_edges, execute_late_warn_uninitialized): Use
19798         uninit_undefined_value_p instead of ssa_undefined_value_p.
19800         PR debug/57184
19801         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
19802         for modifier == EXPAND_INITIALIZER.
19804 2013-05-07  Anton Blanchard  <anton@samba.org>
19806         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
19807         for powerpc64 little endian.
19808         * configure: Regenerate.
19810 2013-05-06  Graham Stott  <grahams@btinternet.com>
19812         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
19813         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
19814         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
19815         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
19817 2013-05-06  Graham Stott  <grahams@btinternet.com>
19819         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
19820         codes which allow non-lvalues.
19822 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
19824         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
19825         components are all 1s.
19826         (integer_minus_onep): New function.
19827         * tree.h (integer_minus_onep): Declare it.
19828         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
19829         integer_minus_onep instead of integer_all_onesp.
19831 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
19833         PR target/52933
19834         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
19835         variations of these patterns.
19837 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
19839         * config/i386/i386.md (isa): Add x64_sse4 member.
19840         (enabled): Handle x64_sse4.
19841         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
19842         instruction for 64bit SSE4_1 targets.  Update insn attributes.
19843         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
19844         instruction for SSE4_1 targets.  Update insn attributes.
19845         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
19846         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
19847         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
19848         const_1 selector.
19849         (*vec_extractv4si): Rename from *sse4_1_pextrd.
19850         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
19851         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
19853 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
19855         PR target/57108
19856         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
19858 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
19860         * final.c (do_assembler_dialects): Don't handle curly braces and
19861         vertical bar escaped by % as dialect delimiters.
19862         (output_asm_insn): Print curly braces and vertical bar if escaped
19863         by % and ASSEMBLER_DIALECT defined.
19864         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
19865         * doc/tm.texi: Regenerated.
19867 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
19869         * config/mips/mips.c: Include tree-pass.h.
19870         (mips_reorg): Split in pre- and post-dbr_schedule parts.
19871         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
19872         (pass_mips_machine_reorg2): New machine specific pass.
19873         (insert_pass_mips_machine_reorg2): New pass plugin definition.
19874         (mips_option_override): Register the new pass.
19875         * rtl.h (cleanup_barriers): Remove prototype.
19876         (dbr_schedule): Likewise.
19877         * jump.c (cleanup_barriers): Make static.
19878         * reorg.c (dbr_schedule): Likewise.
19880 2013-05-06  Richard Biener  <rguenther@suse.de>
19882         PR tree-optimization/57185
19883         * tree-parloops.c (add_field_for_reduction): Handle anonymous
19884         SSA names properly.
19886 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
19888         PR target/57106
19889         * config/i386/i386.c (add_parameter_dependencies): Add dependence
19890         between "first_arg" and "insn", not "last" and "insn".
19892 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19894         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
19895         (find_candidates_in_block): Re-enable slsr_process_phi.
19896         (create_phi_basis): Fix double counting of candidate adjustment.
19898 2013-05-06  Richard Biener  <rguenther@suse.de>
19900         PR middle-end/57147
19901         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
19902         the edge is also fallthru, preserve it and just clear the
19903         abnormal flag.
19904         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
19905         also complex, preserve that and just clear the fallthru flag.
19906         * tree-inline.c (update_ssa_across_abnormal_edges): Also
19907         update virtual operands.
19909 2013-05-06  Alan Modra  <amodra@gmail.com>
19911         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
19912         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
19913         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
19914         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
19915         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
19916         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
19918 2013-05-06  Alan Modra  <amodra@gmail.com>
19920         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
19921         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
19922         (DEFAULT_ASM_ENDIAN): Define.
19923         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
19924         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
19925         Update -K PIC clause from sysv4.h.
19926         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
19927         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
19929 2013-05-06  Alan Modra  <amodra@gmail.com>
19931         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
19932         twice for little-endian.
19933         (ashrdi3_no_power, ashrdi3): Support little-endian.
19935 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
19937         PR target/55303
19938         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
19939         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
19940         related expanders.
19941         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
19942         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
19943         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
19944         New predicates.
19946 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
19947             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19949         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
19950         * config/pa/pa.opt: Make mbig-switch a no-op.
19951         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
19952         (CASE_VECTOR_MODE): Always return SImode.
19953         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
19954         for the !TARGET_BIG_SWITCH case.
19955         * config/pa/pa-linux.h: Likewise.
19956         * config/pa/pa-openbsd.h: Likewise.
19957         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
19958         * config/pa/pa.md (short_jump): Remove define_insn.
19959         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
19960         (casesi0): Remove define_insn.
19961         (type): Remove btable_branch.
19962         (pa_combine_type): Likewise.
19963         (in_nullified_branch_delay): Likewise.
19964         (in_call_delay): Likewise.
19965         (define_delay): Likewise.
19966         (define_insn_reservation "Z3"): Likewise.
19967         (define_insn_reservation "Z4"): Likewise.
19968         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
19969         (pa_adjust_insn_length): Remove adjustment for btable branches.
19970         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
19971         and mno-big-switch
19973 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
19975         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
19976         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
19977         Add m->r,x alternatives.
19978         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
19979         splitters using SWI48x mode iterator.
19980         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
19981         TARGET_64BIT.  Add m->x alternative.
19982         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
19983         Add o->x alternative.  Enable for TARGET_SSE.
19984         (sse_storeq): Remove expander.
19985         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
19986         with memory input operand.
19987         (*vec_extractv2di_1 splitter): New.
19988         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
19989         * config/i386/i386.md (ssevecmodelower): New mode attribute.
19991 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
19993         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
19994         (INT_LOWPART): Delete.
19995         (extract_MB): Adjust.
19996         (extract_ME): Adjust.
19997         (print_operand): Adjust.
19999 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20001         * config/rs6000/predicates.md (reg_or_add_cint_operand,
20002         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20003         (reg_or_logical_cint_operand, easy_fp_constant,
20004         logical_const_operand): Delete "CONST_DOUBLE" case.
20005         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
20006         "HOST_BITS_PER_WIDE_INT == 64" test.
20007         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
20008         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
20009         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
20010         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
20011         test.
20012         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
20013         CONST_DOUBLE DImode/VOIDmode case.
20014         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
20015         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20016         CONST_DOUBLE VOIDmode case.
20017         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20018         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
20019         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
20020         Delete CONST_DOUBLE case.
20021         (splitters for mov FMOVE64 const_double): Delete
20022         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20023         "HOST_BITS_PER_WIDE_INT >= 64" test.
20024         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
20025         case.
20026         (mov DI const_double): Delete.
20028 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
20030         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
20031         on op shows all bits zero in mode of a lowpart subreg, return zero.
20033 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
20035         PR target/57150
20036         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
20037         to save TFmode registers and DImode to save TImode registers for
20038         caller save operations.
20039         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
20040         mark being partially clobbered since they only use the first
20041         double word.
20043         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
20044         and TDmode only use the upper 64-bits of each VSX register.
20046 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20048         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
20049         (find_candidates_in_block): Disable slsr_process_phi.
20051 2013-05-03  Guozhi Wei  <carrot@google.com>
20053         * coverage.c (coverage_obj_init): Move the construction of gcov
20054         constructor to ...
20055         (build_init_ctor): ... here.
20057 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20059         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
20060         (slsr_cand_d): Redefine def_phi.
20061         (stride_status, phi_adjust_status, count_phis_status): New enums.
20062         (find_phi_def): New.
20063         (find_basis_for_base_expr): New.
20064         (find_basis_for_candidate): Handle hidden bases.
20065         (alloc_cand_and_find_basis): Handle phi candidates.
20066         (slsr_process_phi): New.
20067         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
20068         (create_mul_imm_cand): Likewise.
20069         (create_add_ssa_cand): Exclude phi base candidates.
20070         (create_add_imm_cand): Likewise.
20071         (slsr_process_cast): Likewise.
20072         (slsr_process_copy): Likewise.
20073         (find_candidates_in_block): Handle phi candidates.
20074         (dump_candidate): Likewise.
20075         (unconditional_cands): Delete.
20076         (unconditional_cands_with_known_stride_p): Delete.
20077         (phi_dependent_cand_p): New.
20078         (cand_increment): Handle phi-dependent candidates.
20079         (replace_dependent): Delete.
20080         (replace_mult_candidate): New.
20081         (replace_unconditional_candidate): New.
20082         (incr_vec_index): Move to avoid forward reference.
20083         (create_add_on_incoming_edge): New.
20084         (create_phi_basis): New.
20085         (replace_dependents): Delete.
20086         (replace_conditional_candidate): New.
20087         (phi_add_costs): New.
20088         (replace_uncond_cands_and_profitable_phis): New.
20089         (record_increment): Handle phi adjustments.
20090         (record_phi_increments): New.
20091         (record_increments): Handle phi adjustments.
20092         (phi_incr_cost): New.
20093         (lowest_cost_path): Handle phis.
20094         (total_savings): Likewise.
20095         (analyze_increments): Likewise.
20096         (ncd_with_phi): New.
20097         (ncd_of_cand_and_phis): New.
20098         (nearest_common_dominator_for_cands): Handle phi increments.
20099         (all_phi_incrs_profitable): New.
20100         (replace_profitable_candidates): Handle phi-dependent candidates.
20101         (analyze_candidates_and_replace): Likewise.
20103 2013-05-03  Teresa Johnson  <tejohnson@google.com>
20105         PR bootstrap/57154
20106         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
20107         do not exceed REG_BR_PROB_BASE.
20109 2013-05-03  Jeff Law  <law@redhat.com>
20111         PR tree-optimization/57144
20112         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
20113         operand of the condition will bit into the new type when eliminating
20114         a cast feeding a condition.
20116 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
20118         PR rtl-optimization/57130
20119         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
20120         of COMPARE as in_code to the recursive call if needed.
20122 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
20124         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
20125         (enabled): Handle new members.
20126         * config/i386/sse.md (*vec_concatv2si): Merge from
20127         *vec_concatv2si_sse2 and vec_concatv2si_sse.
20128         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
20130 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
20132         PR tree-optimization/57027
20133         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
20134         for fnms opportunity, check we got the prerequisite kind
20135         of tree / gimple before using accessor functions.
20137 2013-05-03  Richard Biener  <rguenther@suse.de>
20139         * double-int.h (lshift): New overload without precision
20140         and arith argument.
20141         (operator *=, operator +=, operator -=): Move ...
20142         * double-int.c (operator *=, operator +=, operator -=): ... here
20143         and implement more efficiently.
20144         (mul_double_with_sign): Remove.
20145         (lshift_double): Adjust to take unsinged shift argument, push
20146         dispatching code to callers.
20147         (mul_double_wide_with_sign): Add early out for callers that
20148         are not interested in high parts or overflow.
20149         (lshift): New function.
20150         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
20151         dispatch code here.
20152         (lrotate, rrotate): Use logical shifts.
20153         * expr.c (get_inner_reference): Use lshift.
20154         * fixed-value.c (do_fixed_divide): Likewise.
20155         * tree-dfa.c (get_ref_base_and_extent): Likewise.
20156         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
20157         (indirect_refs_may_alias_p): Likewise.
20158         (stmt_kills_ref_p_1): Likewise.
20160 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20162         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
20164 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20166         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
20167         scalar form of FABD instruction.
20169 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
20171         * lra-constraints.c (process_alt_operands): Add checking alt
20172         number to choose the best alternative.
20174 2013-05-02  Richard Biener  <rguenther@suse.de>
20176         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
20177         bitmap and its handling.
20178         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
20180 2013-05-02  Richard Biener  <rguenther@suse.de>
20182         PR middle-end/57140
20183         * tree-inline.c (copy_loops): Properly handle removed loops.
20184         (copy_cfg_body): Mark destination loops for fixup if source
20185         loops needed fixup.
20187 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
20189         PR target/56732
20190         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
20191         generating simple_return for naked functions.
20193 2013-05-02  Martin Jambor  <mjambor@suse.cz>
20195         PR middle-end/56988
20196         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
20197         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
20198         flags match.
20199         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
20200         ipa_agg_replacement_value structures.
20201         (known_aggs_to_agg_replacement_list): Likewise.
20202         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
20203         (read_agg_replacement_chain): Likewise.
20204         (ipcp_transform_function): Also check that by_ref flags match.
20206 2013-05-02  Richard Biener  <rguenther@suse.de>
20208         * graphds.h (struct graph): Add obstack member.
20209         * graphds.c (new_graph): Initialize obstack and allocate
20210         vertices from it.
20211         (add_edge): Allocate edge from the obstack.
20212         (free_graph): Free the obstack instead of all edges and vertices.
20214 2013-05-02  Teresa Johnson  <tejohnson@google.com>
20216         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
20217         divides.
20218         * cfg.c (update_bb_profile_for_threading): Ditto.
20219         * tree-inline.c (copy_bb): Ditto.
20220         (copy_edges_for_bb): Ditto.
20221         (initialize_cfun): Ditto.
20222         (copy_cfg_body): Ditto.
20223         (expand_call_inline): Ditto.
20224         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
20225         (estimate_node_size_and_time): Ditto.
20226         (inline_merge_summary): Ditto.
20227         * cgraphclones.c (cgraph_clone_edge): Ditto.
20228         (cgraph_clone_node): Ditto.
20229         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20230         (compute_trg_info): Ditto.
20232 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
20234         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
20235         S reg when fp attribute set.
20236         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
20238 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
20240         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
20241         New pattern.
20242         (*and_one_cmplsi3_compare0_uxtw): Likewise.
20243         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
20244         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
20246 2013-05-02  Richard Biener  <rguenther@suse.de>
20248         * tree-scalar-evolution.c (scev_info_hasher): Remove.
20249         (struct instantiate_cache_entry): New type.
20250         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
20251         (struct instantiate_cache_type): New type.
20252         (set_instantiated_value, get_instantiated_value): Remove.
20253         (get_instantiated_value_entry): New function.
20254         (instantiate_scev_name): Use the new cache and adjust.
20255         (instantiate_scev_poly): Adjust.
20256         (instantiate_scev_binary): Likewise.
20257         (instantiate_array_ref): Likewise.
20258         (instantiate_scev_convert): Likewise.
20259         (instantiate_scev_not): Likewise.
20260         (instantiate_scev_3): Likewise.
20261         (instantiate_scev_2): Likewise.
20262         (instantiate_scev_r): Likewise.
20263         (instantiate_scev): Likewise.
20264         (resolve_mixers): Likewise.
20266 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
20268         PR target/57091
20269         * lra-constraints.c (best_small_class_operands_num): Remove.
20270         (process_alt_operands): Remove small_class_operands_num.  Take
20271         small classes operands into losers and only if the operand is not
20272         matched.  Modify debugging output.
20273         (curr_insn_transform): Remove best_small_class_operands_num.
20274         Print insn name.
20276 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20278         * config/aarch64/aarch64-builtins.c
20279         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
20280         * config/aarch64/aarch64-simd-builtins.def
20281         (reduc_splus_): Add new modes.
20282         (reduc_uplus_): New.
20283         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
20284         (reduc_uplus_v4sf): Likewise.
20285         (reduc_splus_v4sf): Likewise.
20286         (aarch64_addv<mode>): Likewise.
20287         (reduc_uplus_<mode>): Likewise.
20288         (reduc_splus_<mode>): Likewise.
20289         (aarch64_addvv2di): Likewise.
20290         (reduc_uplus_v2di): Likewise.
20291         (reduc_splus_v2di): Likewise.
20292         (aarch64_addvv2si): Likewise.
20293         (reduc_uplus_v2si): Likewise.
20294         (reduc_splus_v2si): Likewise.
20295         (reduc_<sur>plus_<mode>): New.
20296         (reduc_<sur>plus_v2di): Likewise.
20297         (reduc_<sur>plus_v2si): Likewise.
20298         (reduc_<sur>plus_v4sf): Likewise.
20299         (aarch64_addpv4sf): Likewise.
20300         * config/aarch64/arm_neon.h
20301         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
20302         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
20303         add UNSPEC_SADDV, UNSPEC_UADDV.
20304         (SUADDV): New.
20305         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
20307 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20309         * config/aarch64/arm_neon.h
20310         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
20312 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20314         * config/aarch64/aarch64-builtins
20315         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
20317 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20319         * config/aarch64/aarch64-simd-builtins.def
20320         (reduc_smax_): New.
20321         (reduc_smin_): Likewise.
20322         (reduc_umax_): Likewise.
20323         (reduc_umin_): Likewise.
20324         (reduc_smax_nan_): Likewise.
20325         (reduc_smin_nan_): Likewise.
20326         (fmax): Remove.
20327         (fmin): Likewise.
20328         (smax): Update for V2SF, V4SF and V2DF modes.
20329         (smin): Likewise.
20330         (smax_nan): New.
20331         (smin_nan): Likewise.
20332         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
20333         (<su><maxmin><mode>3): ...This, refactor.
20334         (s<maxmin><mode>3): New.
20335         (<maxmin_uns><mode>3): Likewise.
20336         (reduc_<maxmin_uns>_<mode>): Refactor.
20337         (reduc_<maxmin_uns>_v4sf): Likewise.
20338         (reduc_<maxmin_uns>_v2si): Likewise.
20339         (aarch64_<fmaxmin><mode>: Remove.
20340         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
20341         new builtin names.
20342         (vmin<q>_f<32,64>): Likewise.
20343         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
20344         (FMAXMIN): New.
20345         (su): Add mappings for smax, smin, umax, umin.
20346         (maxmin): New.
20347         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
20348         (FMAXMIN): Rename as...
20349         (FMAXMIN_UNS): ...This.
20350         (maxminv): Remove.
20351         (fmaxminv): Likewise.
20352         (fmaxmin): Likewise.
20353         (maxmin_uns): New.
20354         (maxmin_uns_op): Likewise.
20356 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20358         * config/aarch64/arm_neon.h
20359         (vac<ge, gt><sd>_f<32, 64>): Rename to...
20360         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
20361         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
20363 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20365         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
20366         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
20368 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20370         * config/aarch64/aarch64-simd.md
20371         (vcond<mode>_internal): Handle special cases for constant masks.
20372         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
20373         (vcondu<mode><mode>): Likewise.
20374         (vcond<v_cmp_result><mode>): New.
20376 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20378         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
20379         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
20380         * config/aarch64/aarch64-simd-builtins.def
20381         (cmeq): Update to BUILTIN_VALLDI.
20382         (cmgt): Likewise.
20383         (cmge): Likewise.
20384         (cmle): Likewise.
20385         (cmlt): Likewise.
20386         * config/aarch64/arm_neon.h
20387         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
20388         to builtins or C as appropriate.
20390 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20392         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
20393         (cmgeu): ...This.
20394         (cmhi): Rename to...
20395         (cmgtu): ...This.
20396         * config/aarch64/aarch64-simd.md
20397         (simd_mode): Add SF.
20398         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
20399         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
20400         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
20401         (cstore<mode>_neg): ...This.
20402         * config/aarch64/iterators.md
20403         (VALLF): new.
20404         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
20405         (COMPARISONS): New.
20406         (UCOMPARISONS): Likewise.
20407         (optab): Add missing comparisons.
20408         (n_optab): New.
20409         (cmp_1): Likewise.
20410         (cmp_2): Likewise.
20411         (CMP): Likewise.
20412         (cmp): Remove.
20413         (VCMP_S): Likewise.
20414         (VCMP_U): Likewise.
20415         (V_cmp_result): Add DF, SF modes.
20416         (v_cmp_result): Likewise.
20417         (v): Likewise.
20418         (vmtype): Likewise.
20419         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
20421 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
20423         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
20424         define_insn to define_insn_and_split.
20425         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
20426         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
20427         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
20428         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
20429         (thumb2_negscc): Likewise.
20431 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
20433         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
20435 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
20437         * config/arm/thumb2.md: Remove trailing whitespaces.
20439 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
20441         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
20442         Use gen_int_mode rather than GEN_INT.
20444 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
20446         * value-prof.c (stream_in_histogram_value): Remove the strayed
20447         debug_gimple_stmt.
20449 2013-04-30  Richard Biener  <rguenther@suse.de>
20451         PR middle-end/57122
20452         * cfghooks.c (split_edge): Properly check for the loop latch edge.
20454 2013-04-30  Richard Biener  <rguenther@suse.de>
20456         PR middle-end/57107
20457         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
20459 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
20461         PR rtl-optimization/56957
20462         PR rtl-optimization/57105
20463         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
20464         variable.  Use just INSN_UID for determining whether an insn
20465         should be only disconnected from the insn stream.
20466         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
20468 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
20470         PR tree-optimization/57104
20471         * tsan.c (instrument_expr): Don't instrument accesses to
20472         DECL_HARD_REGISTER VAR_DECLs.
20474 2013-04-30  Richard Biener  <rguenther@suse.de>
20476         * function.h (loops_for_fn): New inline function.
20477         (set_loops_for_fn): Likewise.
20478         * cfgloop.h (place_new_loop): Add struct function parameter.
20479         (get_loop): Likewise.
20480         (get_loops): Likewise.
20481         (number_of_loops): Likewise.
20482         (fel_next): Adjust.
20483         (fel_init): Likewise.
20484         * cfg.c (get_loop_copy): Adjust.
20485         * cfgloop.c (flow_loops_dump): Likewise.
20486         (record_loop_exits): Likewise.
20487         (verify_loop_structure): Likewise.
20488         * cfgloopanal.c (mark_irreducible_loops): Likewise.
20489         (estimate_reg_pressure_cost): Likewise.
20490         (mark_loop_exit_edges): Likewise.
20491         * cfgloopmanip.c (place_new_loop): Likewise.
20492         (add_loop): Likewise.
20493         (duplicate_loop): Likewise.
20494         * graph.c (draw_cfg_nodes): Likewise.
20495         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
20496         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
20497         (extract_affine_chrec): Likewise.
20498         (build_scop_iteration_domain): Likewise.
20499         * graphite.c (graphite_initialize): Likewise.
20500         * ira-build.c (create_loop_tree_nodes): Likewise.
20501         (more_one_region_p): Likewise.
20502         (rebuild_regno_allocno_maps): Likewise.
20503         (mark_loops_for_removal): Likewise.
20504         (mark_all_loops_for_removal): Likewise.
20505         (remove_unnecessary_regions): Likewise.
20506         (ira_build): Likewise.
20507         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
20508         * loop-init.c (fix_loop_structure): Likewise.
20509         (gate_rtl_move_loop_invariants): Likewise.
20510         (gate_rtl_unswitch): Likewise.
20511         (gate_rtl_unroll_and_peel_loops): Likewise.
20512         (rtl_doloop): Likewise.
20513         * lto-streamer-in.c (input_cfg): Likewise.
20514         * lto-streamer-out.c (output_cfg): Likewise.
20515         * modulo-sched.c (sms_schedule): Likewise.
20516         * predict.c (tree_estimate_probability): Likewise.
20517         (tree_estimate_probability_driver): Likewise.
20518         (estimate_loops): Likewise.
20519         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
20520         (move_sese_region_to_fn): Likewise.
20521         (debug_loop_num): Likewise.
20522         * tree-chrec.c (chrec_evaluate): Likewise.
20523         (hide_evolution_in_other_loops_than_loop): Likewise.
20524         (chrec_component_in_loop_num): Likewise.
20525         (reset_evolution_in_loop): Likewise.
20526         (evolution_function_is_invariant_rec_p): Likewise.
20527         * tree-if-conv.c (main_tree_if_conversion): Likewise.
20528         * tree-inline.c (copy_loops): Likewise.
20529         (copy_cfg_body): Likewise.
20530         (tree_function_versioning): Likewise.
20531         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
20532         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
20533         Likewise.
20534         (add_to_evolution_1): Likewise.
20535         (scev_const_prop): Likewise.
20536         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
20537         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
20538         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
20539         (tree_ssa_lim_initialize): Likewise.
20540         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
20541         (verify_loop_closed_ssa): Likewise.
20542         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
20543         (tree_ssa_loop_im): Likewise.
20544         (tree_ssa_loop_unswitch): Likewise.
20545         (tree_vectorize): Likewise.
20546         (check_data_deps): Likewise.
20547         (tree_ssa_loop_ivcanon): Likewise.
20548         (tree_ssa_loop_bounds): Likewise.
20549         (tree_complete_unroll): Likewise.
20550         (tree_complete_unroll_inner): Likewise.
20551         (tree_parallelize_loops): Likewise.
20552         (tree_ssa_loop_prefetch): Likewise.
20553         (tree_ssa_loop_ivopts): Likewise.
20554         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
20555         * tree-vectorizer.c (vectorize_loops): Likewise.
20557 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
20559         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
20560         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
20561         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
20562         with EABI_LINK_SPEC.
20564 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
20566         PR target/44578
20567         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
20568         alternative.
20570 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
20572         PR target/57097
20573         * lra-constraints.c (process_alt_operands): Discourage a bit more
20574         using memory for pseudos.  Print cost dump for alternatives.
20575         Modify cost values for conflicts with early clobbers.
20576         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
20578 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
20580         PR target/57098
20581         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
20583 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
20585         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
20586         from/to S register.
20587         (movdi_aarch64): Support LDR/STR from/to D register.
20589 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
20591         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
20592         or higher by default.
20594 2013-04-29  Richard Biener  <rguenther@suse.de>
20596         PR middle-end/57075
20597         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
20598         even if not adding abnormal edges for calls that can make
20599         abnormal gotos.
20601 2013-04-29  Richard Biener  <rguenther@suse.de>
20603         PR middle-end/57103
20604         * tree-cfg.c (move_stmt_op): Fix condition under which to update
20605         TREE_BLOCK.
20606         (move_stmt_r): Remove redundant checking.
20608 2013-04-29  Teresa Johnson  <tejohnson@google.com>
20610         PR bootstrap/57077
20611         * basic-block.h (apply_scale): New function.
20612         (apply_probability): Use apply_scale.
20613         * gimple-streamer-in.c (input_bb): Ditto.
20614         * lto-streamer-in.c (input_cfg): Ditto.
20615         * lto-cgraph.c (merge_profile_summaries): Ditto.
20616         * tree-optimize.c (execute_fixup_cfg): Ditto.
20617         * tree-inline.c (copy_bb): Update comment to use apply_scale.
20618         (copy_edges_for_bb): Ditto.
20619         (copy_cfg_body): Ditto.
20621 2013-04-29  Tom de Vries  <tom@codesourcery.com>
20623         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
20624         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
20625         (tail_merge_optimize): Handle current_loops == NULL.
20627 2013-04-26  Jeff Law  <law@redhat.com>
20629         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
20630         (simplify_cond_using_ranges): Generalize code to simplify
20631         COND_EXPRs where one argument is a constant and the other
20632         is an SSA_NAME created by an integral type conversion.
20634 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20636         * config/arm/arm.md (store_minmaxsi): Use only when
20637         optimize_insn_for_size_p.
20639 2013-04-29  Christian Bruel  <christian.bruel@st.com>
20641         PR target/57108
20642         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
20644 2013-04-29  Richard Biener  <rguenther@suse.de>
20646         PR middle-end/57089
20647         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
20648         loop tree make sure to schedule a fixup for the child as well.
20649         (expand_omp_for_generic): Properly add loops.
20650         (expand_omp_for_static_nochunk): Likewise.
20651         (expand_omp_for_static_chunk): Likewise.
20652         (expand_omp_for): For the degenerate case fixup loops.
20653         (expand_omp_sections): Fix default bb placement in loops.
20654         (expand_omp_atomic_pipeline): Properly add loops.
20656 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20658         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
20660 2013-04-29  Tom de Vries  <tom@codesourcery.com>
20662         * tree-ssa-tail-merge.c: Update header comment.
20664 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20666         * config/aarch64/arm_neon.h
20667         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
20668         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
20669         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
20670         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
20671         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
20672         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
20673         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
20674         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
20676 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20678         * config/aarch64/aarch64-simd.md
20679         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
20680         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
20681         fix_trunc, fixuns_trunc.
20682         (ftrunc<VDQF:mode>2): New.
20683         * config/aarch64/iterators.md (optab): Add fix, fixuns.
20684         (fix_trunc_optab): New.
20686 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20688         * config/aarch64/aarch64-builtins.c
20689         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
20690         iceilf, lround, iroundf.
20692 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
20694         PR target/54349
20695         * config/i386/i386.h (enum ix86_tune_indices)
20696         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
20697         New, split from X86_TUNE_INTER_UNIT_MOVES.
20698         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
20699         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
20700         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
20701         (TARGET_INTER_UNIT_MOVES): Remove.
20702         * config/i386/i386.c (initial_ix86_tune_features): Update.
20703         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
20704         (ix86_expand_convert_uns_didf_sse): Use
20705         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
20706         (ix86_expand_vector_init_one_nonzero): Ditto.
20707         (ix86_expand_vector_init_interleave): Ditto.
20708         (inline_secondary_memory_needed): Return true for moves from SSE class
20709         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
20710         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20711         * config/i386/constraints.md (Yi, Ym): Depend on
20712         TARGET_INTER_UNIT_MOVES_TO_VEC.
20713         (Yj, Yn): New constraints.
20714         * config/i386/i386.md (*movdi_internal): Change constraints of
20715         operand 1 from Yi to Yj and from Ym to Yn.
20716         (*movsi_internal): Ditto.
20717         (*movdf_internal): Ditto.
20718         (*movsf_internal): Ditto.
20719         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
20720         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
20721         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
20722         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
20723         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
20724         * config/i386/sse.md (movdi_to_sse): Ditto.
20725         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
20726         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
20727         TARGET_INTER_UNIT_MOVES.
20728         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
20729         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
20730         instead of TARGET_INTER_UNIT_MOVES.
20731         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
20732         operand 1 from Yi to Yj and from Ym to Yn.
20734 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20736         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
20737         (float_truncate_hi_): Likewise.
20738         (float_extend_lo_): Likewise.
20739         (float_truncate_lo_): Likewise.
20740         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
20741         (aarch64_float_extend_lo_v2df): Likewise.
20742         (vec_unpacks_hi_v4sf): Likewise.
20743         (aarch64_float_truncate_lo_v2sf): Likewise.
20744         (aarch64_float_truncate_hi_v4sf): Likewise.
20745         (vec_pack_trunc_v2df): Likewise.
20746         (vec_pack_trunc_df): Likewise.
20748 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20750         * config/aarch64/aarch64-builtins.c
20751         (aarch64_fold_builtin): Fold float conversions.
20752         * config/aarch64/aarch64-simd-builtins.def
20753         (floatv2si, floatv4si, floatv2di): New.
20754         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
20755         * config/aarch64/aarch64-simd.md
20756         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
20757         * config/aarch64/iterators.md (FLOATUORS): New.
20758         (optab): Add float, floatuns.
20759         (su_optab): Likewise.
20761 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20763         * config/aarch64/aarch64-builtins.c
20764         (aarch64_builtin_vectorized_function): Use new names for
20765         fcvt builtins.
20766         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
20767         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
20768         (fcvtzu): Split as...
20769         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
20770         (fcvtas): Split as...
20771         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
20772         (fcvtau): Split as...
20773         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
20774         (fcvtps): Split as...
20775         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
20776         (fcvtpu): Split as...
20777         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
20778         (fcvtms): Split as...
20779         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
20780         (fcvtmu): Split as...
20781         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
20782         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
20783         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
20784         (lfrintnusf, lfrintnudf): Likewise.
20785         * config/aarch64/aarch64-simd.md
20786         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
20787         define_insn.
20788         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
20789         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
20790         (fcvt_pattern): Likewise.
20792 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20794         * config/aarch64/aarch64-simd.md
20795         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
20796         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
20798 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20800         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
20801         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
20802         (vrnd<a,m,n,p>_f32): Implement using builtins.
20803         (vrnd<i,x><q>_f<32, 64>): New.
20805 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
20807         * config/aarch64/aarch64-builtins.c
20808         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
20809         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
20810         (frintz): Rename to...
20811         (btrunc): ...this.
20812         (frintp): Rename to...
20813         (ceil): ...this.
20814         (frintm): Rename to...
20815         (floor): ...this.
20816         (frinti): Rename to...
20817         (nearbyint): ...this.
20818         (frintx): Rename to...
20819         (rint): ...this.
20820         (frinta): Rename to...
20821         (round): ...this.
20822         * config/aarch64/aarch64-simd.md
20823         (aarch64_frint<frint_suffix><mode>): Delete.
20824         (<frint_pattern><mode>2): Convert to insn.
20825         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
20826         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
20827         (frint_pattern): Likewise.
20828         (frint_suffix): Likewise.
20830 2013-04-29  Richard Biener  <rguenther@suse.de>
20832         PR tree-optimization/57081
20833         * loop-init.c: Include tree-flow.h.
20834         (loop_optimizer_finalize): Free number of iteration estimates.
20835         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
20837 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
20839         PR tree-optimization/57083
20840         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
20841         non-singleton shift count range, zero extend low_bound for uns case.
20843         * config/i386/predicates.md (general_vector_operand): New predicate.
20844         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
20845         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
20846         if they aren't nonimmediate operands.  If their original values
20847         satisfy const_vector_equal_evenodd_p, don't shift them.
20848         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
20849         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
20850         if not nonimmediate_operand.
20851         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
20852         instead of register_operand.
20853         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
20855 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
20857         * stor-layout.c (finalize_size_functions): Allocate a structure and
20858         reset cfun before dumping the functions.
20860 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
20862         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
20864         PR target/56866
20865         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
20866         use xop_pmacsdqh if uns_p.
20867         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
20868         the immediate rotate count.
20870 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
20872         * rtl.h (struct rtx_def): Add comment for field jump.
20873         (LRA_SUBREG_P): New macro.
20874         * recog.c (register_operand): Check LRA_SUBREG_P.
20875         * lra.c (lra): Add note at the end of RTL code. Align non-empty
20876         stack frame.
20877         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
20878         (lra_final_code_change): Skip subreg change for operators.
20879         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
20880         if there are no operand changes.
20881         * lra-constraints.c (curr_insn_set): New.
20882         (match_reload): Set LRA_SUBREG_P.
20883         (emit_spill_move): Ditto.
20884         (check_and_process_move): Use curr_insn_set. Process only single
20885         set insns.  Don't initialize sec_mem_p and change_p.
20886         (simplify_operand_subreg): Use LRA_SUBREG_P.
20887         (reg_in_class_p): New function.
20888         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
20889         of #ifdef.  Add code to remove cycling.
20890         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
20891         non-null disp.  Reload inner instead of disp when base and index
20892         are null.  Try to put lo_sum into register.
20893         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
20894         (check_and_process_move): Move code for move cost check to
20895         simple_move_p.  Remove equiv_substitution.
20896         (simple_move_p): New function.
20897         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
20898         curr_insn_set.  Call check_and_process_move only for single set
20899         insns.  Use the new function.  Move call of check_and_process_move
20900         after operand equiv substitution and address process.
20902 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
20904         PR go/57045
20905         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
20906         with nonlocal goto receivers or returns twice calls, ignore
20907         unininitialized values from abnormal edges to nl goto receiver
20908         or returns twice call.
20910 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
20912         PR tree-optimization/57051
20913         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
20914         and VEC_RSHIFT_EXPR if shift count is a multiple of element
20915         bitsize.
20917 2013-04-26  Richard Biener  <rguenther@suse.de>
20919         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
20920         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
20921         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
20922         (fixup_loop_arrays_after_move): New function.
20923         (move_sese_region_to_fn): Properly outline the loop tree parts
20924         of the SESE region.
20926 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
20928         * config/i386/i386.md (type, unit): Fix long lines.
20930 2013-04-26  Richard Biener  <rguenther@suse.de>
20932         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
20933         (lto-streamer-out.o): Likewise.
20934         * cfgloop.c (init_loops_structure): Export, add struct function
20935         argument and adjust.
20936         (flow_loops_find): Adjust.
20937         * cfgloop.h (enum loop_estimation): Add EST_LAST.
20938         (init_loops_structure): Declare.
20939         * lto-streamer-in.c: Include cfgloop.h.
20940         (input_cfg): Input the loop tree.
20941         * lto-streamer-out.c: Include cfgloop.h.
20942         (output_cfg): Output the loop tree.
20943         (output_struct_function_base): Do not drop PROP_loops.
20945 2013-03-26  Richard Biener  <rguenther@suse.de>
20947         * tree-cfg.c (execute_build_cfg): Build the loop tree.
20948         (pass_build_cfg): Provide PROP_loops.
20949         (move_sese_region_to_fn): Remove loops that are outlined into fn
20950         for now.
20951         * tree-inline.c: Include cfgloop.h.
20952         (initialize_cfun): Do not drop PROP_loops.
20953         (copy_loops): New function.
20954         (copy_cfg_body): Copy loop structure.
20955         (tree_function_versioning): Initialize destination loop tree.
20956         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
20957         (pass_parallelize_loops): Do IL verification.
20958         * loop-init.c (loop_optimizer_init): Fixup loops if required.
20959         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
20960         the CFG make sure we fixup loops as well.
20961         * tree-ssa-tail-merge.c: Include cfgloop.h.
20962         (replace_block_by): When merging loop latches mark loops for fixup.
20963         * lto-streamer-out.c (output_struct_function_base): Drop
20964         PROP_loops for now.
20965         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
20966         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
20967         * ipa-split.c: Include cfgloop.h.
20968         (split_function): Add the new return block to the loop tree root.
20969         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
20970         whether we have removed the forwarder block.
20971         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
20972         * cfgloop.h (place_new_loop): Declare.
20973         * cfgloopmanip.c (place_new_loop): Export.
20974         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
20975         (tree-switch-conversion.o): Likewise.
20976         (tree-complex.o): Likewise.
20977         (tree-inline.o): Likewise.
20978         (tree-ssa-tailmerge.o): Likewise.
20979         (ipa-split.o): Likewise.
20980         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
20981         (tree-ssa-copy.o): Likewise.
20982         * tree-switch-conversion.c: Include cfgloop.h
20983         (process_switch): If we emit a bit-test cascade, schedule loops
20984         for fixup.
20985         * tree-complex.c: Include cfgloop.h.
20986         (expand_complex_div_wide): Properly add new basic-blocks to loops.
20987         * asan.c: Include cfgloop.h.
20988         (create_cond_insert_point): Properly add new basic-blocks to
20989         loops, schedule loop fixup.
20990         * cfgloop.c (verify_loop_structure): Check that looks are not
20991         marked for fixup.
20992         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
20993         to loops.
20994         (expand_omp_for_generic): Likewise.
20995         (expand_omp_sections): Likewise.
20996         (expand_omp_atomic_pipeline): Schedule loops for fixup.
20997         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
20998         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
20999         is initialized, not when loops are present.
21000         * tree-parloops.c (parallelize_loops): Remove checking here.
21001         * passes.c (init_optimization_passes): Schedule a copy-propagation
21002         pass before complete unrolling of inner loops.
21004 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21006         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
21007         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
21008         (colorize_init): Add argument to _WIN32 version.
21009         * toplev.c: Include diagnostic-color.h.
21010         (process_options): Default to -fdiagnostics-color=auto if
21011         GCC_COLORS env var is in the environment.
21012         * common.opt (fdiagnostics-color=): Add Var and Init.
21013         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
21014         env var is in the environment, the default is auto rather than never.
21016         * diagnostic.h (file_name_as_prefix): Add context argument.
21017         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
21018         the string as locus.
21019         * langhooks.c (lhd_print_error_function): Adjust caller.
21021 2013-04-25  Lawrence Crowl  <crowl@google.com>
21023         * var-tracking.c (shared_hash_def::htab):
21024         Change type to hash_table.  Update dependent calls and types.
21026 2013-04-25  Lawrence Crowl  <crowl@google.com>
21028         * Makefile.in: Update as needed below.
21030         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
21031         Move declaration to after the type's method definitons.
21033         * attribs.c (htab_t scoped_attributes::attribute_hash):
21034         Change type to hash_table.  Update dependent calls and types.
21036         * bitmap.c (htab_t bitmap_desc_hash):
21037         Change type to hash_table.  Update dependent calls and types.
21039         * cselib.c (htab_t cselib_hash_table):
21040         Change type to hash_table.  Update dependent calls and types.
21042         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
21043         (hash_string_slot_node): Move implementation into lto-streamer.h
21044         struct string_slot_hasher.
21045         (eq_string_slot_node): Likewise.
21047         * data-streamer-out.c: Update output_block::string_hash_table
21048         dependent calls and types.
21050         * dwarf2cfi.c (htab_t trace_index):
21051         Change type to hash_table.  Update dependent calls and types.
21053         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
21054         Change type to hash_table.  Update dependent calls and types.
21055         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
21056         (htab_t optimize_external_refs::map): Likewise.
21057         (htab_t output_comp_unit::extern_map): Likewise.
21058         (htab_t output_comdat_type_unit::extern_map): Likewise.
21059         (htab_t output_macinfo::macinfo_htab): Likewise.
21060         (htab_t optimize_location_lists::htab): Likewise.
21061         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
21063         * except.c (htab_t ehspec_hash_type):
21064         Change type to hash_table.  Update dependent calls and types.
21065         (assign_filter_values::ttypes): Likewise.
21066         (assign_filter_values::ehspec): Likewise.
21067         (sjlj_assign_call_site_values::ar_hash): Likewise.
21068         (convert_to_eh_region_ranges::ar_hash): Likewise.
21070         * gcse.c (htab_t pre_ldst_table):
21071         Change type to hash_table.  Update dependent calls and types.
21073         * ggc-common.c (htab_t saving_htab):
21074         Change type to hash_table.  Update dependent calls and types.
21075         (htab_t loc_hash): Likewise.
21076         (htab_t ptr_hash): Likewise.
21077         (call_count): Rename ggc_call_count.
21078         (call_alloc): Rename ggc_call_alloc.
21079         (loc_descriptor): Rename make_loc_descriptor.
21080         (add_statistics): Rename ggc_add_statistics.
21082         * ggc-common.c (saving_htab):
21083         Change type to hash_table.  Update dependent calls and types.
21085         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
21086         (push_gimplify_context): Likewise.
21087         (pop_gimplify_context): Likewise.
21088         (struct gimple_temp_hash_elt): Added.
21089         (struct gimplify_hasher): Likewise.
21090         (struct gimplify_ctx.temp_htab):
21091         Change type to hash_table.  Update dependent calls and types.
21093         * gimple-fold.c: Include gimplify-ctx.h.
21095         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
21096         Change type to hash_table.  Update dependent calls and types.
21097         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
21098         avoid potential global name collision.
21100         * gimplify.c: Include gimplify-ctx.h.
21101         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
21102         (htab_t gimplify_ctx::temp_htab):
21103         Update dependent calls and types for new type hash_table.
21104         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
21105         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
21107         * gimplify-ctx.h: New.
21108         (struct gimple_temp_hash_elt): Move from gimplify.c.
21109         (class gimplify_hasher): New.
21110         (struct gimplify_ctx): Move from gimple.h.
21111         (htab_t gimplify_ctx::temp_htab):
21112         Change type to hash_table.  Update dependent calls and types.
21114         * graphite-clast-to-gimple.c: Include graphite-htab.h.
21115         (htab_t ivs_params::newivs_index):
21116         Change type to hash_table.  Update dependent calls and types.
21117         (htab_t ivs_params::params_index): Likewise.
21118         (htab_t print_generated_program::params_index): Likewise.
21119         (htab_t gloog::newivs_index): Likewise.
21120         (htab_t gloog::params_index): Likewise.
21122         * graphite.c: Include graphite-htab.h.
21123         4htab_t graphite_transform_loops::bb_pbb_mapping):
21124         Change type to hash_table.  Update dependent calls and types.
21126         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
21127         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
21128         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
21130         * graphite-dependences.c: Include graphite-htab.h.
21131         (loop_is_parallel_p): Change hash table type of parameter.
21133         * graphite-htab.h: New.
21134         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
21135         (extern find_pbb_via_hash): Move from graphite-poly.h.
21136         (extern loop_is_parallel_p): Move from graphite-poly.h.
21137         (extern get_loop_body_pbbs): Move from graphite-poly.h.
21139         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
21140         (extern loop_is_parallel_p): Move to graphite-htab.h.
21141         (extern get_loop_body_pbbs): Move to graphite-htab.h.
21143         * haifa-sched.c (htab_t delay_htab):
21144         Change type to hash_table.  Update dependent calls and types.
21145         (htab_t delay_htab_i2): Likewise.
21147         * ira-color.c (htab_t allocno_hard_regs_htab):
21148         Change type to hash_table.  Update dependent calls and types.
21150         * ira-costs.c (htab_t cost_classes_htab):
21151         Change type to hash_table.  Update dependent calls and types.
21153         * loop-invariant.c (htab_t merge_identical_invariants::eq):
21154         Change type to hash_table.  Update dependent calls and types.
21156         * loop-iv.c (htab_t bivs):
21157         Change type to hash_table.  Update dependent calls and types.
21159         * loop-unroll.c (htab_t opt_info::insns_to_split):
21160         Change type to hash_table.  Update dependent calls and types.
21161         (htab_t opt_info::insns_with_var_to_expand): Likewise.
21163         * lto-streamer.h (struct string_slot): Move from data-streamer.h
21164         (struct string_slot_hasher): New.
21165         (htab_t output_block::string_hash_table):
21166         Change type to hash_table.  Update dependent calls and types.
21168         * lto-streamer-in.c (freeing_string_slot_hasher): New.
21169         (htab_t file_name_hash_table):
21170         Change type to hash_table.  Update dependent calls and types.
21172         * lto-streamer-out.c: Update output_block::string_hash_table dependent
21173         calls and types.
21175         * lto-streamer.c (htab_t tree_htab):
21176         Change type to hash_table.  Update dependent calls and types.
21178         * omp-low.c: Include gimplify-ctx.h.
21180         * passes.c (htab_t name_to_pass_map):
21181         Change type to hash_table.  Update dependent calls and types.
21182         (pass_traverse): Rename to passes_pass_traverse.
21184         * plugin.c (htab_t event_tab):
21185         Change type to hash_table.  Update dependent calls and types.
21187         * postreload-gcse.c (htab_t expr_table):
21188         Change type to hash_table.  Update dependent calls and types.
21189         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
21191         * sese.c (debug_rename_map_1): Make extern.
21192         (htab_t copy_bb_and_scalar_dependences::rename_map):
21193         Change type to hash_table.  Update dependent calls and types.
21195         * sese.h (extern debug_rename_map): Move to .c file.
21197         * store-motion.c (htab_t store_motion_mems_table):
21198         Change type to hash_table.  Update dependent calls and types.
21200         * trans-mem.c (htab_t tm_new_mem_hash):
21201         Change type to hash_table.  Update dependent calls and types.
21203         * tree-browser.c (htab_t TB_up_ht):
21204         Change type to hash_table.  Update dependent calls and types.
21206         * tree-cfg.c (htab_t discriminator_per_locus):
21207         Change type to hash_table.  Update dependent calls and types.
21209         * tree-complex.c: Include tree-hasher.h
21210         (htab_t complex_variable_components):
21211         Change type to hash_table.  Update dependent calls and types.
21213         * tree-eh.c (htab_t finally_tree):
21214         Change type to hash_table.  Update dependent calls and types.
21216         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
21217         struct int_tree_hasher.
21218         (extern int_tree_map_eq): Likewise.
21219         (uid_decl_map_hash): Removed.
21220         (extern decl_tree_map_eq): Likewise.
21222         * tree-hasher.h: New.
21223         (struct int_tree_hasher): New.
21224         (typedef int_tree_htab_type): New.
21226         * tree-inline.c: Include gimplify-ctx.h.
21228         * tree-mudflap.c: Include gimplify-ctx.h.
21230         * tree-parloops.c: Include tree-hasher.h.
21231         (htab_t eliminate_local_variables_stmt::decl_address):
21232         Change type to hash_table.  Update dependent calls and types.
21233         (htab_t separate_decls_in_region::decl_copies): Likewise.
21235         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
21236         Change type to hash_table.  Update dependent calls and types.
21238         * tree-sra.c (candidates):
21239         Change type to hash_table.  Update dependent calls and types.
21241         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
21242         in tree-flow.h.
21243         (int_tree_map_hash): Likewise.
21245         * tree-ssa-dom.c (htab_t avail_exprs):
21246         Change type to hash_table.  Update dependent calls and types.
21248         * tree-ssa-live.c (var_map_base_init::tree_to_index):
21249         Change type to hash_table.  Update dependent calls and types.
21251         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
21252         Change type to hash_table.  Update dependent calls and types.
21254         * tree-ssa-phiopt.c (seen_ssa_names):
21255         Change type to hash_table.  Update dependent calls and types.
21257         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
21258         Change type to hash_table.  Update dependent calls and types.
21260         * tree-ssa-uncprop.c (equiv):
21261         Change type to hash_table.  Update dependent calls and types.
21263 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
21265         PR rtl-optimization/57003
21266         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
21267         call note_stores with kill_clobbered_value callback again after
21268         killing regs_invalidated_by_call.
21270 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21272         * config/aarch64/aarch64-simd.md
21273         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
21274         (aarch64_simd_bsl<mode>): Likewise.
21275         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
21277 2013-04-25  Marek Polacek  <polacek@redhat.com>
21279         PR tree-optimization/57066
21280         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
21282 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21284         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
21286 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21288         * config/aarch64/aarch64-builtins.c
21289         (aarch64_fold_builtin): New.
21290         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
21291         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
21292         * config/aarch64/aarch64-simd-builtins.def (abs): New.
21293         * config/aarch64/arm_neon.h
21294         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
21296 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21297             Tejas Belagod  <tejas.belagod@arm.com>
21299         * config/aarch64/aarch64-builtins.c
21300         (aarch64_gimple_fold_builtin): New.
21301         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
21302         * config/aarch64/aarch64-simd-builtins.def (addv): New.
21303         * config/aarch64/aarch64-simd.md (addpv4sf): New.
21304         (addvv4sf): Update.
21305         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
21307 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21309         * config/aarch64/aarch64.md
21310         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
21312 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21314         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
21315         (*ngcsi_uxtw): New pattern.
21317 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21318             Julian Brown  <julian@codesourcery.com>
21320         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
21321         (TB_DREG): Add T_V4HF.
21322         (v4hf_UP): New macro.
21323         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21324         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21325         Handle initialisation of V4HF. Adjust initialisation of reinterpret
21326         built-ins.
21327         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21328         (arm_vector_mode_supported_p): Handle V4HF.
21329         (arm_mangle_map): Handle V4HFmode.
21330         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
21331         * config/arm/arm_neon_builtins.def: Add entries for
21332         vcvtv4hfv4sf, vcvtv4sfv4hf.
21333         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
21334         (neon_vcvtv4hfv4sf): Likewise.
21335         * config/arm/neon-gen.ml: Handle half-precision floating point
21336         features.
21337         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
21338         * config/arm/arm_neon.h: Regenerate.
21339         * config/arm/neon.ml (type elts): Add F16.
21340         (type vectype): Add T_float16x4, T_floatHF.
21341         (type vecmode): Add V4HF.
21342         (type features): Add Requires_FP_bit feature.
21343         (elt_width): Handle F16.
21344         (elt_class): Likewise.
21345         (elt_of_class_width): Likewise.
21346         (mode_of_elt): Refactor.
21347         (type_for_elt): Handle F16, fix error messages.
21348         (vectype_size): Handle T_float16x4.
21349         (vcvt_sh): New function.
21350         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
21351         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
21352         (string_of_mode): Handle V4HF.
21353         * doc/arm-neon-intrinsics.texi: Regenerate.
21355 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21357         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
21358         format specifier in 'X' case.
21360 2013-04-25  Alan Modra  <amodra@gmail.com>
21362         PR target/57052
21363         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
21364         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
21365         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
21366         Repeat for many other rotate/shift and mask patterns using subregs.
21367         Name lshiftrt insns.
21368         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
21369         on WORDS_BIG_ENDIAN.
21371 2013-04-25  Alan Modra  <amodra@gmail.com>
21373         * config.gcc: Support little-endian powerpc-linux targets.
21374         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
21375         (LINK_OS_LINUX_SPEC): Define.
21376         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
21377         Preserve MASK_LITTLE_ENDIAN.
21378         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
21379         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
21380         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
21381         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
21382         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
21383         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
21384         Correct fp word order for little-endian.  Don't shift toc entries
21385         smaller than a word for little-endian.
21386         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
21387         (bswapdi2 splits): Correct low-part subreg for little-endian.
21388         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
21389         low/high where such is correct only for be.
21390         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
21391         little-endian for -mcall-aixdesc.
21393 2013-04-25  Alan Modra  <amodra@gmail.com>
21395         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
21396         replace_equiv_address_nv.
21398 2013-04-25  Alan Modra  <amodra@gmail.com>
21400         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
21402 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21404         Revert:
21405         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21406         * rtl.h (struct rtx_def): ...
21408 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21410         PR rtl-optimizations/57046
21411         * lra-constraints (split_reg): Set up lra_risky_transformations_p
21412         for multi-reg splits.
21414 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
21416         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
21418 2013-04-24  Sterling Augustine  <saugustine@google.com>
21420         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
21421         (comp_dir_string, debug_str_dwo_section): New.
21422         (DEBUG_STR_DWO_SECTION): Rename to ...
21423         (DEBUG_DWO_STR_SECTION): ... this.
21424         (DEBUG_NORM_STR_SECTION): Delete.
21425         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
21426         (DEBUG_STR_DWO_SECTION_FLAGS): New.
21427         (find_AT_string): Move most logic to ...
21428         (find_AT_string_in_table): ... here.  New.
21429         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
21430         add_skeleton_AT_string.  Delete logic.
21431         (output_skeleton_debug_sections): Remove call to
21432         add_top_level_skeleton_die_attrs.
21433         (add_comp_dir_attribute): Move logic to comp_dir_string.
21434         (dwarf2out_init): Initialize debug_str_dwo_section.
21435         (output_indirect_string): Call find_string_form.
21436         (output_indirect_strings): Rewrite.
21437         (prune_unused_types): Empty skeleton_debug_str_hash.
21438         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
21439         (dwarf2out_finish):  Call output_indirect_strings.
21441 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
21443         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
21445 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21447         * rtl.h (struct rtx_def): Add comment for field jump.
21448         (LRA_SUBREG_P): New macro.
21449         * recog.c (register_operand): Check LRA_SUBREG_P.
21450         * lra.c (lra): Add note at the end of RTL code. Align non-empty
21451         stack frame.
21452         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
21453         (lra_final_code_change): Skip subreg change for operators.
21454         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
21455         if there are no operand changes.
21456         * lra-constraints.c (curr_insn_set): New.
21457         (match_reload): Set LRA_SUBREG_P.
21458         (emit_spill_move): Ditto.
21459         (check_and_process_move): Use curr_insn_set. Process only single
21460         set insns.  Don't initialize sec_mem_p and change_p.
21461         (simplify_operand_subreg): Use LRA_SUBREG_P.
21462         (reg_in_class_p): New function.
21463         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
21464         of #ifdef.  Add code to remove cycling.
21465         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
21466         non-null disp.  Reload inner instead of disp when base and index
21467         are null.  Try to put lo_sum into register.
21468         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
21469         (check_and_process_move): Move code for move cost check to
21470         simple_move_p.  Remove equiv_substitution.
21471         (simple_move_p): New function.
21472         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
21473         curr_insn_set.  Call check_and_process_move only for single set
21474         insns.  Use the new function.  Move call of check_and_process_move
21475         after operand equiv substitution and address process.
21477 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
21479         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
21480         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
21481         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
21483 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
21485         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
21487 2013-04-24  Marek Polacek  <polacek@redhat.com>
21489         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
21490         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
21491         (select_loops_exit_conditions): Likewise.
21492         (number_of_iterations_for_all_loops): Likewise.
21493         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
21494         (scev_analysis): Likewise.
21496 2013-04-02  Catherine Moore  <clm@codesourcery.com>
21497             Chao-ying Fu  <fu@mips.com>
21499         * config/mips/micromips.md (jraddiusp): New pattern.
21500         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
21501         instruction if possible.
21503 2013-04-24  Alan Modra  <amodra@gmail.com>
21505         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
21507 2013-04-24  Julian Brown  <julian@codesourcery.com>
21508             Chung-Lin Tang  <cltang@codesourcery.com>
21510         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
21511         dependency behavior in enumeration type DIE generation. Add TODO note
21512         to comments about future DW_FORM_sdata/udata re-work of related code.
21514 2013-04-23  Lawrence Crowl  <crowl@google.com>
21516         * Makefile.in: Update as needed below.
21518         * hash-table.h (class hash_table):
21519         Correct many methods with parameter types compare_type to the correct
21520         value_type.  (Correct code was unlikely to notice the change.)
21521         (hash_table::elements_with_deleted) New.
21522         (class hashtable::iterator): New.
21523         (hashtable::begin()): New.
21524         (hashtable::end()): New.
21525         (FOR_EACH_HASH_TABLE_ELEMENT): New.
21527         * statistics.c (statistics_hashes):
21528         Change type to hash_table.  Update dependent calls and types.
21530         * tree-into-ssa.c (var_infos):
21531         Change type to hash_table.  Update dependent calls and types.
21533         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
21534         Change type to hash_table.  Update dependent calls and types.
21536         * tree-ssa-loop-im.c (struct mem_ref.refs):
21537         Change type to hash_table.  Update dependent calls and types.
21539         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
21540         Change type to hash_table.  Update dependent calls and types.
21542         * tree-ssa-sccvn.c (vn_tables_s::nary):
21543         Change type to hash_table.  Update dependent calls and types.
21544         (vn_tables_s::phis): Likewise.
21545         (vn_tables_s::references): Likewise.
21547         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
21548         (vn_reference_eq): Update parameter and return types.
21550         * tree-ssa-structalias.c (pointer_equiv_class_table):
21551         Change type to hash_table.  Update dependent calls and types.
21552         (location_equiv_class_table): Likewise.
21554         * tree-vect-data-refs.c: Consequential changes for making
21555         peeling a hash_table.
21557         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
21558         (destroy_loop_vec_info): Dependent hash_table update.
21560         * tree-vectorizer.h (peeling_htab):
21561         Change type to hash_table.  Update dependent calls and types.
21563 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
21565         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
21566         to check the register content is equal or not.
21567         * lra-constraints.c (match_reload): Use lra_assign_reg_val
21568         to assign register content record.
21569         * lra-eliminations.c (update_reg_eliminate): Use
21570         lra_update_reg_val_offset to update register content offset.
21571         * lra-int.h (struct lra_reg): Add offset member.
21572         (lra_reg_val_equal_p): New static inline function.
21573         (lra_update_reg_val_offset): New static inline function.
21574         (lra_assign_reg_val): New static inline function.
21575         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
21576         to assign register content record.
21577         (initialize_lra_reg_info_element): Initial offset to zero.
21579 2013-04-23  Catherine Moore  <clm@codesourcery.com>
21581         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
21582         operands.  Record compression.
21584 2013-04-23  Xinliang David Li  <davidxl@google.com>
21586         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
21588 2013-04-23  Richard Biener  <rguenther@suse.de>
21590         PR middle-end/57036
21591         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
21592         parameter, only add abnormal goto edges from the copied body
21593         if the call could perform abnormal gotos.
21594         (copy_cfg_body): Adjust.
21596 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
21598         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
21600 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
21602         * coretypes.h (gimple_stmt_iterator): Add struct to make
21603         compatible with C.
21605 2013-04-23  Richard Biener  <rguenther@suse.de>
21607         PR tree-optimization/57026
21608         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
21609         from SSA names occuring in abnormal PHI nodes.
21611 2013-04-22  Andi Kleen  <ak@linux.intel.com>
21613         * lto/lto.c (print_lto_report_1): Fix LTO report names.
21615 2013-04-22  Andi Kleen  <ak@linux.intel.com>
21617         * lto/lto.c (print_lto_report_1): Declare early.
21618         (read_cgraph_and_symbols): Call print_lto_report_1 early.
21620 2013-04-22  Andi Kleen  <ak@linux.intel.com>
21622         * common.opt (-flto-report-wpa): Add.
21623         * doc/invoke.texi (-flto-report-wpa): Add.
21624         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
21625         (lto_main): dito.
21627 2013-04-22  Xinliang David Li  <davidxl@google.com>
21629         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
21630         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
21631         * Makefile.in: New dependency
21633             David Daney <ddaney.cavm@gmail.com>
21635         * configure.ac (gcc_cv_as_micromips_support):  Use the
21636         --fatal-warnings option.
21637         * configure: Regenerate.
21639 2013-04-22  Marek Polacek  <polacek@redhat.com>
21641         PR sanitizer/56990
21642         * tsan.c (instrument_expr): Don't instrument expression
21643         in case its size is zero.
21645 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
21647         PR target/57032
21648         Revert:
21649         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
21651         * config/alpha/alpha.c (TARGET_LRA_P): New define.
21653 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
21655         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
21656         (gimple_stmt_iterator): New typedef.
21657         * gimple.h (gimple_stmt_iterator): Rename to...
21658         (gimple_stmt_iterator_d): ... This.
21659         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
21660         trees be valid for GIMPLE and GENERIC.
21661         (TARGET_GIMPLE_FOLD_BUILTIN): New.
21662         * gimple-fold.c (gimple_fold_call): Call target hook
21663         gimple_fold_builtin.
21664         * hooks.c (hook_bool_gsiptr_false): New.
21665         * hooks.h (hook_bool_gsiptr_false): New.
21666         * target.def (fold_stmt): New.
21667         * doc/tm.texi: Regenerate.
21669 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
21671         PR target/57018
21672         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
21673         a set sp if no stack realignment.
21675 2013-04-22  Nick Clifton  <nickc@redhat.com>
21677         * config.gcc (tilegx-linux): Extend extra_objs rather than
21678         overwriting it.
21679         (tilepro-linux): Likewise.
21681 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
21683         * config/aarch64/aarch64-builtins.c
21684         (CF): Remove.
21685         (CF0, CF1, CF2, CF3, CF4, CF10): New.
21686         (VAR<1-12>): Add MAP parameter.
21687         (BUILTIN_*): Likewise.
21688         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
21689         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
21690         (aarch64_ushl_n<mode>): Likewise.
21691         (aarch64_sshr_n<mode>): Likewise.
21692         (aarch64_ushr_n<mode>): Likewise.
21693         (aarch64_<maxmin><mode>): Likewise.
21694         (aarch64_sqrt<mode>): Likewise.
21695         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
21696         (vshr<q>_n_*): Likewise.
21698 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
21700         * config/aarch64/aarch64-builtins.c
21701         (aarch64_simd_builtin_type_mode): Handle SF types.
21702         (sf_UP): Define.
21703         (BUILTIN_GPF): Define.
21704         (aarch64_init_simd_builtins): Handle SF types.
21705         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
21706         (frecps): Likewise.
21707         (frecpx): Likewise.
21708         * config/aarch64/aarch64-simd.md
21709         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
21710         (aarch64_frecpe<mode>): New.
21711         (aarch64_frecps<mode>): Likewise.
21712         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
21713         (v8type): Add frecp<esx>.
21714         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
21715         (aarch64_frecps<mode>): Likewise.
21716         * config/aarch64/iterators.md (FRECP): New.
21717         (frecp_suffix): Likewise.
21718         * config/aarch64/arm_neon.h
21719         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
21721 2013-04-22  Christian Bruel  <christian.bruel@st.com>
21723         PR target/56995
21724         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
21725         (REG_CLASS_NAMES): Idem.
21726         (REG_CLASS_CONTENTS): Idem.
21727         (REGCLASS_HAS_FP_REG): Idem.
21728         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
21729         (sh_conditional_register_usage): Idem.
21731 2013-04-21  Jeff Law  <law@redhat.com>
21733         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
21734         (ssa_forward_propagate_and_combine): Use it.
21736 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
21738         * lra.c: Update the flow chart diagram.
21740 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
21742         PR rtl-optimization/56847
21743         * lra-constraints.c (process_alt_operands): Discourage alternative
21744         with non-matche doffsettable memory constraint fro memory with
21745         known offset.
21747 2013-04-19  Richard Biener  <rguenther@suse.de>
21749         PR tree-optimization/56982
21750         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
21751         function.
21752         * gimplify.c (gimplify_call_expr): Notice special calls.
21753         (gimplify_modify_expr): Likewise.
21754         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
21755         abnormal control flow receivers.
21756         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
21757         in the same way as cfun->has_nonlocal_labels.
21758         (gimple_purge_dead_abnormal_call_edges): Likewise.
21759         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
21760         receivers start a basic-block.
21762 2013-04-19  Richard Biener  <rguenther@suse.de>
21764         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
21765         member ...
21766         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
21767         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
21768         (SLP_TREE_LOAD_PERMUTATION): Add.
21769         (vect_transform_slp_perm_load): Adjust prototype.
21770         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
21771         (vect_free_slp_instance): Likewise.
21772         (vect_create_new_slp_node): Likewise.
21773         (vect_supported_slp_permutation_p): Remove.
21774         (vect_slp_rearrange_stmts): Adjust.
21775         (vect_supported_load_permutation_p): Likewise.  Inline
21776         vect_supported_slp_permutation_p here.
21777         (vect_analyze_slp_instance): Compute load permutations per
21778         slp node instead of per instance.
21779         (vect_get_slp_defs): Adjust.
21780         (vect_transform_slp_perm_load): Likewise.
21781         (vect_schedule_slp_instance): Remove redundant code.
21782         (vect_schedule_slp): Remove hack for PR56270, add it ...
21783         * tree-vect-stmts.c (vectorizable_load): ... here, do not
21784         CSE loads for SLP.  Adjust.
21786 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
21788         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
21789         spelling in two comments.
21791 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
21793         PR target/56797
21794         * config/arm/arm.c (load_multiple_sequence): Require SP
21795         as base register for loads if SP is in the register list.
21797 2013-04-19  Martin Jambor  <mjambor@suse.cz>
21799         PR tree-optimization/56718
21800         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
21801         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
21802         and made public.  Adjusted all callers.
21803         (ipa_intraprocedural_devirtualization): New function.
21804         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
21805         (ipa_intraprocedural_devirtualization): Likewise.
21806         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
21808 2013-04-19  Richard Biener  <rguenther@suse.de>
21810         PR tree-optimization/57000
21811         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
21813 2013-04-19  Terry Guo  <terry.guo@arm.com>
21815         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
21816         Replace with ...
21817         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
21818         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
21819         (cortex_m4_fmacs): Use new reservations.
21820         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
21822 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
21824         PR rtl-optimization/56999
21825         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
21826         related code.
21827         (lra_coalesce): Remove split_origin_bitmap and related code.
21828         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
21829         ranges if necessary.
21831 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
21833         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
21834         New array.
21835         (ix86_expand_call): Remove clobbered_registers array and use
21836         x86_64_ms_sysv_extra_clobbered_registers instead.
21837         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
21838         Declare here.
21839         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
21840         predicate.
21841         * config/i386/i386.md (*call_rex64_ms_sysv): Use
21842         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
21843         (*call_value_rex64_ms_sysv): Ditto.
21845 2013-04-18  Cary Coutant  <ccoutant@google.com>
21847         * dwarf2out.c (output_pubnames): Check die_perennial_p of
21848         parent instead of die_mark.
21850 2013-04-18  Diego Novillo  <dnovillo@google.com>
21852         * gimple.c (create_gimple_tmp): New.
21853         (get_expr_type): New.
21854         (build_assign): New.
21855         (build_type_cast): New.
21856         * gimple.h (enum ssa_mode): Define.
21857         (gimple_seq_set_location): New.
21858         * asan.c (build_check_stmt): Change some gimple_build_* calls
21859         to use build_assign and build_type_cast.
21861 2013-04-18  Richard Biener  <rguenther@suse.de>
21863         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
21864         handle negative step.  Remove redundant checks.
21865         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
21866         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
21867         for negative step and grouped loads fail to vectorize.
21869 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
21871         * emit-rtl.c (reset_insn_used_flags): New function.
21872         (reset_all_used_flags): Use it.
21873         (verify_insn_sharing): New function.
21874         (verify_rtl_sharing): Fix verification for SEQUENCEs.
21876 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
21878         PR tree-optimization/56984
21879         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
21880         and (x >> M) >= N don't register any assertion if N << M is the
21881         minimum value.
21883 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
21885         * lower-subreg.c (resolve_simple_move): If called self-recursive,
21886         do not delete_insn insns that have not yet been emitted, only
21887         unlink them with remove_insn.
21888         * df-scan.c (df_insn_delete): Revert r197492.
21890 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
21892         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
21893         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
21895 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
21897         * config/arm/arm.md (movsicc_insn): Convert define_insn into
21898         define_insn_and_split.
21899         (and_scc,ior_scc,negscc): Likewise.
21900         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
21902 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
21904         * config/arm/arm.c (use_return_insn): Return 0 for targets that
21905         can benefit from using a sequence of LDRD instructions in epilogue
21906         instead of a single LDM instruction.
21908 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
21910         PR 45688
21911         * doc/extend.texi: Fix typo.
21913 2013-04-17  Richard Biener  <rguenther@suse.de>
21915         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
21916         (vect_build_slp_tree): ... here.
21917         (vect_build_slp_tree_1): Compute which stmts of the SLP group
21918         match.  Remove special-casing of mismatched complex loads.
21919         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
21920         re-try the match with swapped commutative operands.
21921         (vect_supported_load_permutation_p): Remove special-casing of
21922         mismatched complex loads.
21923         (vect_analyze_slp_instance): Adjust.
21925 2013-04-17  Richard Biener  <rguenther@suse.de>
21927         PR rtl-optimization/56921
21928         * cfgloop.h (struct loop): Add simple_loop_desc member.
21929         (struct niter_desc): Mark with GTY(()).
21930         (simple_loop_desc): Do not use aux field but simple_loop_desc.
21931         * loop-iv.c (get_simple_loop_desc): Likewise.
21932         (free_simple_loop_desc): Likewise.
21934         Revert
21935         2013-04-16  Richard Biener  <rguenther@suse.de>
21937         PR rtl-optimization/56921
21938         * loop-init.c (pass_rtl_move_loop_invariants): Add
21939         TODO_do_not_ggc_collect to todo_flags_finish.
21940         (pass_rtl_unswitch): Same.
21941         (pass_rtl_unroll_and_peel_loops): Same.
21942         (pass_rtl_doloop): Same.
21944 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
21946         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
21947         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
21948         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
21949         references.
21950         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
21951         * tree-streamer.c (record_common_node): Adjust reference in comment.
21953 2013-04-17  Terry Guo  <terry.guo@arm.com>
21955         * config/arm/cortex-m4.md: Add a new bypass.
21957 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21959         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
21960         New pattern.
21961         (*subs_<optab><mode>_multp2): New pattern.
21962         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
21963         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
21965 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21967         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
21968         (*subs_mul_imm_<mode>): New pattern.
21970 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
21972         PR target/56948
21973         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
21974         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
21975         (vsx_movti_32bit): Change j->wa to O->wa.
21977 2013-04-16  Richard Biener  <rguenther@suse.de>
21979         PR rtl-optimization/56921
21980         * loop-init.c (pass_rtl_move_loop_invariants): Add
21981         TODO_do_not_ggc_collect to todo_flags_finish.
21982         (pass_rtl_unswitch): Same.
21983         (pass_rtl_unroll_and_peel_loops): Same.
21984         (pass_rtl_doloop): Same.
21986 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
21988         * config/arm/arm.c (emit_multi_reg_push): New declaration
21989         for an existing function.
21990         (arm_emit_strd_push): New function.
21991         (arm_expand_prologue): Used here.
21992         (arm_emit_ldrd_pop): New function.
21993         (arm_expand_epilogue): Used here.
21994         (arm_get_frame_offsets): Update condition.
21995         (arm_emit_multi_reg_pop): Add a special case for load of a single
21996         register with writeback.
21998 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
22000         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
22001         description.
22003 2013-04-16  Richard Biener  <rguenther@suse.de>
22005         PR tree-optimization/56756
22006         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
22007         (first_mem_ref_loc): New.
22008         (execute_sm): Place the load temporarily before a previous
22009         access instead of in the latch edge to ensure its SSA dependencies
22010         are defined at points dominating the load.
22012 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
22014         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
22015         correct fix by moving header and footer insn to the footer of
22016         the merged basic block.  Clear BB_END of the merged-away block.
22018         PR middle-end/43631
22019         * emit-rtl.c (make_note_raw): New function.
22020         (link_insn_into_chain): New static inline function.
22021         (add_insn): Use it.
22022         (add_insn_before, add_insn_after): Factor insn chain linking code...
22023         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
22024         using link_insn_into_chain.
22025         (note_outside_basic_block_p): New helper function for emit_note_after
22026         and emit_note_before.
22027         (emit_note_after): Use nobb variant of add_insn_after if the note
22028         should not be contained in a basic block.
22029         (emit_note_before): Use nobb variant of add_insn_before if the note
22030         should not be contained in a basic block.
22031         (emit_note_copy): Use make_note_raw.
22032         (emit_note): Likewise.
22033         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
22034         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
22035         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
22036         the moved barrier the tail of the basic block it follows.
22037         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
22039 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
22041         PR tree-optimization/56962
22042         * gimple-ssa-strength-reduction.c (record_increment): Only set
22043         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
22044         either rhs1 or rhs2 is equal to c->base_expr.
22046 2013-04-15  Richard Biener  <rguenther@suse.de>
22048         PR tree-optimization/56933
22049         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
22050         member.
22051         (GROUP_READ_WRITE_DEPENDENCE): Remove.
22052         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
22053         * tree-vect-data-refs.c (vect_analyze_group_access): Move
22054         dependence check ...
22055         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
22056         ... here.
22057         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
22058         GROUP_READ_WRITE_DEPENDENCE.
22060 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22062         * emit-rtl.c (reset_all_used_flags): New function.
22063         (verify_rtl_sharing): Call reset_all_used_flags before and after
22064         performing the checks.
22066 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22068         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
22069         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
22070         * config/arm/constraints.md (De): New constraint.
22071         * config/arm/neon.md (anddi3_neon): Delete.
22072         (neon_vand<mode>): Expand to standard anddi3 pattern.
22073         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
22074         Move earlier in the file.
22075         (neon_inv_logic_op2): Likewise.
22076         (arm_anddi_operand_neon): New predicate.
22078 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22080         * configure.ac (gcc_cv_ld_as_needed): Set
22081         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
22082         Use -z ignore, -z record on *-*-solaris2*.
22083         (HAVE_LD_AS_NEEDED): Update comment.
22084         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
22085         * configure: Regenerate.
22086         * config.in: Regenerate.
22087         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
22088         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
22089         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
22090         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
22091         equivalents.  Fix markup.
22092         * doc/tm.texi: Regenerate.
22094 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
22096         * config/i386/i386.opt: New option mstack-protector-guard=.
22097         * config/i386/i386-opts.h: Add enum stack_protector_guard.
22098         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
22099         TARGET_SSP_TLS_GUARD.
22100         * config/i386/i386.c (ix86_option_override_internal): Set
22101         ix86_stack_protector_guard.
22102         * config/i386/i386.md (stack_protect_set): Enable for
22103         TARGET_SSP_TLS_GUARD only.
22104         (stack_protect_set_<mode>): Ditto.
22105         (stack_protect_test): Ditto.
22106         (stack_protect_test_<mode>): Ditto.
22107         * doc/invoke.texi (i386 Option): Document.
22109 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
22111         PR target/56890
22112         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
22113         (S_MODES): Set H_MODE bit.
22114         (SF_MODES): Set only S_MODE and SF_MODE bits.
22115         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
22116         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
22117         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
22118         <MODE_FLOAT>: Likewise.
22120 2013-04-15  Joey Ye  <joey.ye@arm.com>
22122         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
22124 2013-04-15  Joey Ye  <joey.ye@arm.com>
22126         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
22127         for real far jump.
22128         (thumb_far_jump_used_p): Count instruction size and set
22129         far_jump_used.
22131 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
22133         * reorg.c (fill_simple_delay_slots): Reindent block of code.
22134         * resource.c (mark_target_live_regs): Reformat conditional block.
22136 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22138         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
22139         notes, they are emitted only just before final.
22140         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
22142 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22144         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
22145         * cfgrtl.c (delete_insn): Call it here instead.
22146         * lra-spills.c (lra_final_code_change): Use delete_insn.
22147         * haifa-sched.c (sched_remove_insn): Likewise.
22148         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
22149         returning to the nop pool.
22150         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
22151         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
22153 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22155         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
22156         * doc/tm.texi: Regenerated.
22158 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
22160         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
22161         QImode checks.
22163 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22165         * df-core.c (df_find_def): Compare register numbers.
22166         (df_find_use): Likewise.
22168 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
22170         PR target/56903
22171         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
22172         lra_in_progress for return.
22174 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
22176         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
22177         define_insn into define_insn_and_split and emit movsicc patterns.
22179 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
22181         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
22183 2013-04-12  Richard Biener  <rguenther@suse.de>
22185         * tree-pass.h (TODO_do_not_ggc_collect): New.
22186         * passes.c (execute_one_ipa_transform_pass): Honor
22187         TODO_do_not_ggc_collect.
22188         (execute_one_pass): Likewise.
22190         Revert
22191         2013-04-10  Richard Biener  <rguenther@suse.de>
22193         * passes.c (init_optimization_passes): Remove reload pass.
22194         * ira.c (do_reload): Merge into ...
22195         (ira): ... this.
22196         (rest_of_handle_reload): Remove.
22197         (pass_reload): Likewise.
22198         * config/i386/i386.c (ix86_option_override): Refer to ira instead
22199         of reload for vzeroupper pass placement.
22201 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
22203         PR tree-optimization/56918
22204         PR tree-optimization/56920
22205         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
22206         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
22207         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
22208         use wide_mul_with_sign method.
22210 2013-04-12  Richard Biener  <rguenther@suse.de>
22212         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
22213         not be considered a gimple constant.
22215 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
22217         * fold-const.c (const_binop): Handle vector shifts by a scalar.
22218         (fold_binary_loc): Call const_binop also for mixed vector-scalar
22219         operations.
22221 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
22222             Jakub Jelinek  <jakub@redhat.com>
22224         * opts.c: Include diagnostic-color.h.
22225         (common_handle_option): Handle OPT_fdiagnostics_color_.
22226         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
22227         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
22228         (diagnostic-color.o): New.
22229         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
22230         (diagnostic_color_rule): New enum.
22231         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
22232         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
22233         the location string.
22234         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
22235         either NULL, or color kind.
22236         * diagnostic-color.c: New file.
22237         * diagnostic-color.h: New file.
22238         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
22239         arguments.
22240         * doc/invoke.texi (-fdiagnostics-color): Document.
22241         * pretty-print.h (pp_show_color): Define.
22242         (struct pretty_print_info): Add show_color field.
22243         * diagnostic.c: Include diagnostic-color.h.
22244         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
22245         macros.  Colorize error:, warning: etc. strings and also the location
22246         string.
22247         (diagnostic_show_locus): Colorize the caret line.
22248         * pretty-print.c: Include diagnostic-color.h.
22249         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
22250         inside of %< %> quotes or quoted through q format modifier.
22252 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22254         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
22256 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22258         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
22259         code in CC_NZ mode.
22260         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
22261         pattern.
22263 2013-04-11  Marek Polacek  <polacek@redhat.com>
22265         PR tree-optimization/48184
22266         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
22268 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
22270         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
22271         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
22272         (skip_simple_arithmetic): Tidy up.
22273         * tree.h (skip_simple_constant_arithmetic): Declare.
22275 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22277         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
22279 2013-04-11  Richard Biener  <rguenther@suse.de>
22281         * tree-vect-loop.c (get_initial_def_for_induction): Properly
22282         generate vector constants.
22284 2013-04-11  Richard Biener  <rguenther@suse.de>
22286         PR tree-optimization/56878
22287         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
22288         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
22289         New function.
22290         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22291         Prefer to align the DR with the most invariant base address.
22293 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
22295         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
22296         comment.
22298 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
22300         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
22301         floating-point vector comparisons against 0.
22303 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
22305         PR tree-optimization/56899
22306         * fold-const.c (extract_muldiv_1): Apply distributive law
22307         only if TYPE_OVERFLOW_WRAPS (ctype).
22309 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
22311         PR target/56124
22312         * ira-costs.c (scan_one_insn): Check whether the source rtx of
22313         loading has side effect.
22315 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
22317         * config/sparc/sparc.c: Include tree-pass.h.
22318         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
22319         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
22320         head of file.  Change return type.  Split off gate function.
22321         (sparc_gate_work_around_errata): New function.
22322         (pass_work_around_errata): New pass definition.
22323         (insert_pass_work_around_errata) New pass insert definition to
22324         insert pass_work_around_errata just after delayed-branch scheduling.
22325         (sparc_option_override): Insert the pass.
22326         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
22328 2013-04-10  David S. Miller  <davem@davemloft.net>
22330         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
22331         or -mcpu=hypersparc.
22333         * target.def (cstore_mode): New hook.
22334         * target.h: Include insn-codes.h
22335         * targhooks.c: Likewise.
22336         (default_cstore_mode): New function.
22337         * targhooks.h: Declare it.
22338         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
22339         * doc/tm.texi: Rebuild.
22340         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
22341         target hook, rather than inspecting the insn_data.
22342         * config/sparc/sparc.c (sparc_cstore_mode): New function.
22343         (TARGET_CSTORE_MODE): Redefine.
22344         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
22345         result patterns.
22346         * config/sparc/predicates.md (cstore_result_operand): New special
22347         predicate.
22348         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
22349         Use it for operand 0.
22350         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
22351         (*snesi_special): Likewise.
22352         (*snesi_zero): Likewise.
22353         (*seqsi_zero): Likewise.
22354         (*sltu_insn): Likewise.
22355         (*sgeu_insn): Likewise.
22356         (*seqdi_special): Make operand 0 and comparison operation be of
22357         DImode.
22358         (*snedi_special): Likewise.
22359         (*snedi_special_vis3): Likewise.
22360         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
22361         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
22362         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
22363         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
22364         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
22365         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
22366         (*sltu_extend_sp64): Likewise.
22367         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
22368         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
22369         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
22370         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
22371         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
22373 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
22375         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
22376         (aarch64_start_file): Use the new function.
22378 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
22379             Jason Merrill  <jason@redhat.com>
22381         * common.opt: Add -gdwarf.
22382         * opts.c (common_handle_option): Handle it.
22383         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
22385 2013-04-10  Richard Biener  <rguenther@suse.de>
22387         * passes.c (execute_todo): Do not call ggc_collect conditional here.
22388         (execute_one_ipa_transform_pass): But unconditionally here.
22389         (execute_one_pass): And here.
22390         (init_optimization_passes): Remove reload pass.
22391         * tree-pass.h (TODO_ggc_collect): Remove.
22392         (pass_reload): Likewise.
22393         * ira.c (do_reload): Merge into ...
22394         (ira): ... this.
22395         (rest_of_handle_reload): Remove.
22396         (pass_reload): Likewise.
22397         * config/i386/i386.c (ix86_option_override): Refer to ira instead
22398         of reload for vzeroupper pass placement.
22399         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
22400         and todo_flags_finish of all passes.
22402 2013-04-10  Richard Biener  <rguenther@suse.de>
22404         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
22405         first_const_oprnd field, rename first_def_type to first_op_type.
22406         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
22407         (vect_get_and_check_slp_defs): Always use the type of the
22408         operand.  Allow mixed vect_external_def, vect_constant_def types.
22409         (vect_get_constant_vectors): Handle mixed vect_external_def,
22410         vect_constant_def types.
22412 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
22414         PR tree-optimization/55524
22415         * tree-ssa-math-opts.c
22416         (convert_mult_to_fma): Don't use an fms construct
22417         when we don't have an fms operation, but fnma, and it looks
22418         likely that we'll be able to use the latter.
22420 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
22422         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
22423         function.
22424         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
22425         inline fail caused by overwritable functions.
22427 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
22429         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
22430         unnecessary bits in the constant power of two case.
22432 2013-04-10  Richard Biener  <rguenther@suse.de>
22434         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
22435         broken code swapping operands.
22436         (vect_build_slp_tree): Do not compute load permutations here.
22437         (vect_analyze_slp_instance): Compute load permutations here,
22438         after building the SLP tree.
22440 2013-04-09  Christian Bruel  <christian.bruel@st.com>
22442         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
22443         of next/prev_real_insn.
22445 2013-04-09  Jan Hubicka  <jh@suse.cz>
22447         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
22448         Drop aliased parameter.
22449         (function_and_variable_visibility): Do not handle alias pairs.
22450         * cgraph.c (varpool_externally_visible_p): Update prototype.
22451         * varpool.c (varpool_add_new_variable): Update.
22453 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22455         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
22457 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
22459         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
22461         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
22463 2013-04-09  Marek Polacek  <polacek@redhat.com>
22465         PR tree-optimization/48762
22466         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
22468 2013-04-09  Richard Biener  <rguenther@suse.de>
22470         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
22471         dealing with cost.
22472         (vect_build_slp_tree): Likewise.
22473         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
22474         calculating the cost of a SLP instance.
22475         (vect_analyze_slp_instance): Use it from here, after building
22476         the SLP tree.
22478 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
22480         PR middle-end/56883
22481         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
22482         expand_omp_for_static_chunk): Use simple_p = true in
22483         force_gimple_operand_gsi calls when assigning to addressable decls.
22485 2013-04-09  Jeff Law  <law@redhat.com>
22487         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
22488         when the boolean was created by converting a wider object which
22489         had a boolean range.
22491 2013-04-09  Richard Biener  <rguenther@suse.de>
22493         * tree-vectorizer.h (slp_void_p): Remove.
22494         (slp_tree): Typedef before _slp_tree declaration.
22495         (struct _slp_tree): Use a vector of slp_tree as children.
22496         (vect_get_place_in_interleaving_chain): Remove.
22497         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
22498         Move ...
22499         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
22500         and make static.
22501         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
22502         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
22503         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
22504         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
22505         Use slp_node instead of slp_void_p and adjust.
22507 2013-04-09  Richard Biener  <rguenther@suse.de>
22509         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
22510         work that is not necessary.
22512 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
22514         PR tree-optimization/56854
22515         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
22516         forward into clobber stmts if it would change MEM_REF lhs into
22517         non-MEM_REF.
22519 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
22521         * tree.c (type_hash_lookup, type_hash_add): Make static.
22522         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
22524 2013-04-09  Richard Biener  <rguenther@suse.de>
22526         * tree.h (unsave_expr_now): Remove.
22527         * tree-inline.c (mark_local_for_remap_r): Remove.
22528         (unsave_expr_1): Likewise.
22529         (unsave_r): Likewise.
22530         (unsave_expr_now): Likewise.
22531         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
22532         (propagate_tree_value): Likewise.
22534 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
22536         * doc/rtl.texi (sequence): Rewrite documentation to match the
22537         current use of SEQUENCE rtl objects.
22538         * rtl.def (SEQUENCE): Likewise.
22540         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
22541         Update documentation.
22542         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
22543         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
22545         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
22547 2013-04-08  Teresa Johnson  <tejohnson@google.com>
22549         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
22550         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
22551         methods.
22552         (estimate_edge_size_and_time): Add comment to suggest using rounding
22553         methods.
22554         (estimate_node_size_and_time): Ditto.
22555         (remap_edge_change_prob): Use helper rounding divide methods.
22556         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
22557         (gimple_mod_pow2_value_transform): Ditto.
22558         (gimple_mod_subtract_transform): Ditto.
22559         (gimple_ic_transform): Ditto.
22560         (gimple_stringops_transform): Ditto.
22561         * stmt.c (conditional_probability): Ditto.
22562         (emit_case_dispatch_table): Ditto.
22563         * lto-cgraph.c (merge_profile_summaries): Ditto.
22564         * tree-optimize.c (execute_fixup_cfg): Ditto.
22565         * cfgcleanup.c (try_forward_edges): Ditto.
22566         * cfgloopmanip.c (scale_loop_profile): Ditto.
22567         (loopify): Ditto.
22568         (duplicate_loop_to_header_edge): Ditto.
22569         (lv_adjust_loop_entry_edge): Ditto.
22570         * tree-vect-loop.c (vect_transform_loop): Ditto.
22571         * profile.c (compute_branch_probabilities): Ditto.
22572         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
22573         * lto-streamer-in.c (input_cfg): Ditto.
22574         * gimple-streamer-in.c (input_bb): Ditto.
22575         * ipa-cp.c (update_profiling_info): Ditto.
22576         (update_specialized_profile): Ditto.
22577         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
22578         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
22579         rounding methods.
22580         * sched-rgn.c (compute_dom_prob_ps): Ditto.
22581         (compute_trg_info): Ditto.
22582         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
22583         (purge_dead_edges): Ditto.
22584         * loop-unswitch.c (unswitch_loop): Ditto.
22585         * cgraphclones.c (cgraph_clone_edge): Ditto.
22586         (cgraph_clone_node): Ditto.
22587         * tree-inline.c (copy_bb): Ditto.
22588         (copy_edges_for_bb): Ditto.
22589         (initialize_cfun): Ditto.
22590         (copy_cfg_body): Ditto.
22591         (expand_call_inline): Ditto.
22593 2013-04-08  Kai Tietz  <ktietz@redhat.com>
22595         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
22596         TARGET_CYGWIN64 by TARGET_64BIT.
22598 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
22600         * config/epiphany/epiphany.md (GPR_1): New constant.
22601         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
22602         * config/epiphany/epiphany.c (gen_compare_reg):
22603         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
22604         is already in place.
22605         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
22606         Don't require being called during rtl expansion; If y operlaps r0,
22607         return 0.
22608         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
22609         (epiphany_expand_epilogue): Likewise.
22611         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
22612         Don't use CC_FPmode for ORDERED / UNORDERED.
22613         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
22615         * config/epiphany/constraints.md (CnL): New constraint.
22616         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
22617         * config/epiphany/predicates.md (add_operand): Allow 1024.
22619         * config/epiphany/epiphany.md (logical_op): New code iterator.
22620         (op_mnc): New code attribute.
22621         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
22622         (mov_f+1, mov_f+2): New peephole2 patterns.
22624         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
22625         (cstoresi4): Also allow re-use of zero result when doing a NE
22626         comparison to a non-zero operand.
22627         Use (clobber (scratch)) for first insn if the gpr output is not needed.
22629         * config/epiphany/epiphany.md (<insn_opname>v2si3):
22630         Use gen_addsi3_i / gen_subsi3_i.
22632 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
22634         PR c++/34949
22635         PR c++/50243
22636         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
22637         contain anything but clobbers, at most one __builtin_stack_restore,
22638         optionally debug stmts and final resx, and if it has at least one
22639         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
22640         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
22641         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
22642         which isn't defaut definition, remove them.
22643         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
22644         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
22645         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
22646         with MEM_REF LHS with SSA_NAME address.
22648 2013-04-08  Jeff Law  <law@redhat.com>
22650         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
22652 2013-04-08  Richard Biener  <rguenther@suse.de>
22654         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
22655         extra newline.
22656         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
22657         determined vector type.
22658         (vect_analyze_data_refs): Likewise.
22659         (vect_get_new_vect_var): Adjust.
22660         (vect_create_destination_var): Preserve SSA name versions.
22661         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
22662         not dump anything here.
22664 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
22666         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
22667         Add member lr_slot_known.
22668         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
22669         if necessary.
22670         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
22671         Remove code that sets lr_slot_offset according to what a previous
22672         version of epiphany_emit_save_restore used to do.
22673         (epiphany_emit_save_restore): When doing an lr save or restore,
22674         set/verify lr_slot_known and lr_slot_offset.
22676 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
22678         PR target/54338
22679         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
22680         in ALL_REGS.
22682 2013-04-08  Richard Biener  <rguenther@suse.de>
22684         * alias.c (find_base_term): Fix thinko in previous change.
22686 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
22688         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
22689         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
22690         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
22691         if possible to compute val.
22692         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
22693         For QImode integers don't require anything about precision.  Use
22694         const_with_all_bytes_same to find out if the constant doesn't have
22695         repeated bytes in it.
22697 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22699         * config/s390/s390.c (s390_expand_insv): Only accept insertions
22700         within mode size.
22702 2013-04-08  Marek Polacek  <polacek@redhat.com>
22704         PR rtl-optimization/48182
22705         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
22706         value to 1.
22708 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22710         PR target/55487
22711         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
22712         nuses, make sure we have a label.
22714 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22716         PR target/56843
22717         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
22718         (rs6000_emit_swdiv_low_precision): Remove.
22719         (rs6000_emit_swdiv): Rewrite to handle between one and four
22720         iterations of Newton-Raphson generally; modify required number of
22721         iterations for some cases.
22722         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
22724 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
22726         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
22727         set-but-unused variable.
22729         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
22730         basic blocks of released function bodies garbage-collectable.
22732         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
22733         (struct rtl_opt_pass): Add TODO_df_finish.
22735         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
22737 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22739         * config/arm/constraints.md (q): New constraint.
22740         * config/arm/ldrdstrd.md: New file.
22741         * config/arm/arm.md (ldrdstrd.md) New include.
22742         (arm_movdi): Use "q" instead of "r" constraint
22743         for double-word memory access.
22744         (movdf_soft_insn): Likewise.
22745         * config/arm/vfp.md (movdi_vfp): Likewise.
22746         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
22747         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
22748         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
22749         (mem_ok_for_ldrd_strd): Likewise.
22750         (output_move_double): Update assertion.
22752 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22754         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
22756 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22758         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
22759         define_insn_and_split.
22760         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
22762 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22764         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
22765         define_insn_and_split.
22766         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
22767         (shiftsi3_compare): New pattern.
22768         (rrx): New pattern.
22769         * config/arm/unspecs.md (UNSPEC_RRX): New.
22771 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22773         * config/arm/arm.md (negdi_extendsidi): New pattern.
22774         (negdi_zero_extendsidi): Likewise.
22776 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22778         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
22779         define_insn_and_split.
22780         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
22781         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
22783 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22785         * config/arm/arm.md (arm_subdi3): Convert define_insn into
22786         define_insn_and_split.
22787         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
22788         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
22790 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22792         * config/arm/arm.md (subsi3_carryin): New pattern.
22793         (subsi3_carryin_const): Likewise.
22794         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
22795         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
22797 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22799         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
22801 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
22803         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
22804         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
22806 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22808         * config/arm/arm.c (arm_expand_builtin): Change fcode
22809         type to unsigned int.
22811 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
22813         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
22815 2013-04-04  Ian Lance Taylor  <iant@google.com>
22817         * doc/standards.texi (Standards): The Go frontend supports the Go 1
22818         language standard.
22820 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
22822         PR middle-end/56729
22823         * df-scan.c (df_insn_delete): Disable failing assert.
22825 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22827         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
22828         New function prototype.
22829         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
22830         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
22831         (arm_builtin_vectorized_function): New function.
22833 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22835         * config/arm/arm_neon_builtins.def: New file.
22836         * config/arm/arm.c (neon_builtin_data): Move contents to
22837         arm_neon_builtins.def.
22838         (enum arm_builtins): Include neon builtin definitions.
22839         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
22840         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
22842 2013-04-04  Marek Polacek  <polacek@redhat.com>
22844         PR tree-optimization/48186
22845         * predict.c (maybe_hot_frequency_p): Return false if
22846         HOT_BB_FREQUENCY_FRACTION is 0.
22847         (cgraph_maybe_hot_edge_p): Likewise.
22849 2013-04-04  Richard Biener  <rguenther@suse.de>
22851         PR tree-optimization/56826
22852         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
22853         more accurately.
22855 2013-04-04  Richard Biener  <rguenther@suse.de>
22857         PR tree-optimization/56213
22858         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
22859         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
22861 2013-04-04  Richard Biener  <rguenther@suse.de>
22863         PR tree-optimization/56837
22864         * tree-loop-distribution.c (classify_partition): For non-zero
22865         values require that the value has the same precision as its
22866         mode to be useful as memset value.
22868 2013-04-03  Nick Clifton  <nickc@redhat.com>
22870         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
22871         (fmssf4): Use fmsf.s on E3V5 architectures.
22872         (fnmasf4): Use fnmaf.s on E3V5 architectures.
22873         (fnmssf4): Use fnmsf.s on E3V5 architectures.
22875 2013-04-03  Jeff Law  <law@redhat.com>
22877         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
22878         (lra-eliminations.o): Likewise.
22880 2013-04-03  Teresa Johnson  <tejohnson@google.com>
22882         * gcov-io.c (compute_working_sets): Moved most of body of old
22883         compute_working_sets here from profile.c.
22884         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
22885         (gcov_working_set_t): Moved typedef here from basic-block.h
22886         (compute_working_set): Declare.
22887         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
22888         (get_working_sets): Renamed from compute_working_set,
22889         replace most of body with call to new compute_working_sets.
22890         (get_exec_counts): Replace call to compute_working_sets
22891         to get_working_sets.
22892         * profile.h (get_working_sets): Renamed from compute_working_set.
22893         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
22894         to get_working_sets.
22895         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
22896         * gcov-dump.c (dump_working_sets): New function.
22898 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
22900         * hwint.c (sext_hwi, zext_hwi): New functions.
22901         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
22902         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
22903         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
22904         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
22905         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
22906         (sext_hwi, zext_hwi): New functions.
22908 2013-04-03  Jeff Law  <law@redhat.com>
22910         PR tree-optimization/56799
22911         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
22912         back test for widening conversion erroneously dropped in prior change.
22914 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22916         PR target/56809
22917         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
22918         instead of next_real_insn.
22920 2013-04-03  Marek Polacek  <polacek@redhat.com>
22922         PR sanitizer/55702
22923         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
22925 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22927         PR target/56809
22928         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
22929         next_real_insn.
22930         (thumb1_output_casesi): Likewise.
22931         (thumb2_output_casesi): Likewise.
22933 2013-04-03  Richard Biener  <rguenther@suse.de>
22935         PR tree-optimization/56817
22936         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
22937         Split out ...
22938         (tree_unroll_loops_completely_1): ... new function to manually
22939         walk the loop tree, properly defering outer loops of unrolled
22940         loops to later iterations.
22942 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
22944         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
22945         (vectorizable_load): Likewise.
22946         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
22947         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
22949 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
22951         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
22952         BIT_FIELD_REF.
22954 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
22956         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
22958 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
22960         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
22962 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
22964         PR tree-optimization/56790
22965         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
22966         folding.
22968 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
22970         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
22971         Handle VEC_MERGE.
22972         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
22973         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
22974         equal arguments.
22976 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
22978         PR c/19449
22979         * tree.h (force_folding_builtin_constant_p): New decl.
22980         * builtins.c (force_folding_builtin_constant_p): New variable.
22981         (fold_builtin_constant_p): Fold immediately also if
22982         force_folding_builtin_constant_p.
22984 2013-04-03  Richard Biener  <rguenther@suse.de>
22986         PR tree-optimization/56812
22987         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
22988         DRs of the same interleaving chain are independent.
22990 2013-04-02  Jason Merrill  <jason@redhat.com>
22992         * gdbinit.in (pbb): Use debug fn.
22994 2013-04-02  Lawrence Crowl  <crowl@google.com>
22996         * sese.h (struct ivtype_map_elt_s): Remove unused.
22997         (extern debug_ivtype_map): Remove unused.
22998         (extern eq_ivtype_map_elts): Remove unused.
22999         * sese.c (debug_ivtype_map): Removed unused.
23000         (debug_ivtype_map_1): Removed unused.
23001         (debug_ivtype_elt): Remove unused.
23002         (eq_ivtype_map_elts): Remove unused.
23005 2013-04-02  Kai Tietz  <ktietz@redhat.com>
23007         PR target/52790
23008         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
23009         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
23010         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
23011         function.
23012         (legitimize_pe_coff_symbol): Likewise.
23013         (is_imported_p): New helper-function.
23014         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
23015         for Windows x64 targets.
23016         (ix86_expand_prologue): Optimize for pe-coff targets.
23017         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
23018         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
23019         medium/large code-model.
23020         (legitimize_pic_address): Likewise.
23021         (legitimize_tls_address): Likewise.
23022         (ix86_expand_call): Likewise.
23023         (x86_output_mi_thunk): Likewise.
23024         (get_dllimport_decl): Add new beimport argument.
23025         (construct_plt_address): Don't assert for x64 pe-coff targets.
23026         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
23027         targets.
23028         (SYMBOL_FLAG_STUBVAR): New macro.
23029         (SYMBOL_REF_STUBVAR_P): Likewise.
23030         * config/i386/winnt.c (stub_list): New structure.
23031         (stub_head): New local variable.
23032         (i386_pe_record_stub): New function.
23033         (i386_pe_file_end): Emit refptr-stubs.
23035 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
23037         PR rtl-optimization/56745
23038         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
23039         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
23041         PR c++/34949
23042         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
23043         and both of them are MEM_REFs, just compare first argument for
23044         equality and attempt to deal even with differing offsets.
23046         PR c++/34949
23047         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
23048         of gimple_clobber_p to be MEM_REF.
23049         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
23050         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
23051         after gimplification.
23052         * asan.c (get_mem_ref_of_assignment): Don't instrument
23053         gimple_clobber_p stmts.
23054         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
23055         gimple_clobber_p stmt if they have MEM_REF lhs and
23056         are dead because of another gimple_clobber_p stmt.
23057         * tree-ssa-live.c (clear_unused_block_pointer): Treat
23058         gimple_clobber_p stmts like debug stmts.
23059         (remove_unused_locals): Remove clobbers with MEM_REF lhs
23060         that refer to unused VAR_DECLs or uninitialized values.
23061         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
23062         gimple_clobber_p stmts if they refer to removed parameters.
23063         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
23064         formatting.
23066 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
23068         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
23069         using SWI48 mode attribute.
23071 2013-04-02  Wei Mi  <wmi@google.com>
23073         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
23074         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
23075         *<rotate_insn><mode>3_mask in i386.md.
23077 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
23079         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
23081 2013-04-02  Richard Biener  <rguenther@suse.de>
23083         PR tree-optimization/56778
23084         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
23085         Runtime alias tests are not supported for gather loads.
23086         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
23087         stmts referenced from SSA operands before updating SSA form.
23089 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
23090             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23092         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
23093         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
23094         * config/arm/cortex-a53.md: New file.
23095         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
23096         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
23097         * config/arm/arm.c (arm_issue_rate): Likewise.
23098         * config/arm/arm-tune.md: Regenerate
23099         * config/arm/arm-tables.opt: Regenerate.
23100         * config/arm/arm-cores.def: Add cortex-a53.
23102 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
23104         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
23105         non-static link.
23107 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23109         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
23110         scalar load/store operations using B/H registers.
23111         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
23113 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23115         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
23116         scalar move.
23117         * config/aarch64/aarch64.c
23118         (aarch64_simd_scalar_immediate_valid_for_move): New.
23119         * config/aarch64/aarch64-protos.h
23120         (aarch64_simd_scalar_immediate_valid_for_move): New.
23121         * config/aarch64/constraints.md (Dh, Dq): New.
23122         * config/aarch64/iterators.md (hq): New.
23124 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
23126         * reorg.c (get_branch_condition): Deal with conditional returns.
23127         (fill_simple_delay_slots): Remove dead code dealing with jumps.
23129 2013-04-01  Wei Mi  <wmi@google.com>
23131         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
23132         Truncate operand 2 using %b asm operand modifier.
23133         (*<shift_insn><mode>3_mask): Ditto.
23134         (*<rotate_insn><mode>3_mask): Ditto.
23136 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
23138         PR middle-end/56798
23139         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
23141 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
23143         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
23144         of next_real_insn.
23145         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
23147 2013-03-30  Lawrence Crowl  <crowl@google.com>
23149         * dse.c (clear_alias_sets): Remove never set.
23150         (disqualified_clear_alias_sets): Remove never set.
23151         (clear_alias_mode_pool): Remove never set.
23152         (dse_step0): Remove condition that is never true.
23153         (canon_address): Remove condition that is never true.
23154         (dse_step7): Remove condition that is never true.
23155         (rest_of_handle_dse): Remove condition that is never true.
23156         (rest_of_handle_dse::did_global): Remove never read from above.
23157         (dse_step2_spill): Remove never called from above.
23158         (dse_step5_spill): Remove never called from above.
23160 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
23162         * doc/md.texi (Standard Names) <casesi>: Update documentation for
23163         JUMP_TABLE_DATA changes.
23164         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
23165         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
23166         (Insns) <jump_table_data>: New entry.
23167         * doc/tm.texi: Regenerate.
23169         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
23171         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
23172         for table jump at the end of a basic block using tablejump_p.
23173         * targhooks.c (default_invalid_within_doloop): Likewise.
23174         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
23175         target hook implementation that is identical to the default hook.
23176         (rs6000_invalid_within_doloop): Remove.
23178         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
23179         unused variable from tablejump_p call.
23181         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
23182         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
23183         (INSN_DELETED_P): Likewise.
23184         (emit_jump_table_data): New prototype.
23185         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
23186         after 4th as unused.
23187         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
23188         * sched-vis.c (print_insn): Likewise.
23189         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
23190         insn for compatibility with back ends that use next_active_insn to
23191         identify jump table data.
23192         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
23193         (remove_insn): Likewise.
23194         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
23195         to be emitted.
23196         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
23197         (emit_jump_table_data): New function.
23199         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
23200         basic block, a JUMP_TABLE_DATA never is.
23201         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
23202         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
23203         off from code handling real insns.
23204         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
23205         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
23206         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
23207         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
23208         is not a NONDEBUG_INSN_P.
23209         * ira-costs.c (scan_one_insn): Likewise.
23210         * jump.c (mark_all_labels): Likewise.
23211         (mark_jump_label_1): Likewise.
23212         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
23213         * lra.c (get_insn_freq): Expect all insns reaching here to be in
23214         a basic block.
23215         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
23216         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
23217         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
23218         JUMP_TABLE_DATA_P insns.
23219         (calculate_elim_costs_all_insns): Likewise.
23220         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
23221         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
23222         (delete_output_reload): Code style fixups.
23223         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
23224         insn flags on this non-insn.
23225         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
23226         as scheduling barriers, for pre-change compatibility.
23227         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
23228         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
23230         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
23231         redundant JUMP_TABLE_DATA_P test.
23232         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
23233         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
23234         (frv_for_each_packet): Likewise.
23235         * config/i386/i386.c (min_insn_size): Likewise.
23236         (ix86_avoid_jump_mispredicts): Likewise.
23237         * config/m32r/m32r.c (m32r_is_insn): Likewise.
23238         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
23239         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
23240         (mips16_insn_length): Robustify.
23241         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
23242         (mips16_split_long_branches): Likewise.
23243         * config/pa/pa.c (pa_combine_instructions): Likewise.
23244         * config/rs6000/rs6000.c (get_next_active_insn): Treat
23245         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
23246         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
23247         as contributing to pool range lengths.
23248         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
23249         Remove redundant JUMP_TABLE_DATA_P test.
23250         (sh_loop_align): Likewise.
23251         (split_branches): Likewise.
23252         (sh_insn_length_adjustment): Likewise.
23253         * config/spu/spu.c (get_branch_target): Likewise.
23255 2013-03-29  Jan Hubicka  <jh@suse.cz>
23257         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
23258         gcov streaming; stream hot bb threshold to ltrans.
23259         * predict.c (get_hot_bb_threshold): Break out from ....
23260         (maybe_hot_count_p): ... here.
23261         (set_hot_bb_threshold): New function.
23262         * lto-section-in.c (lto_section_name): Add profile.
23263         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
23264         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
23265         and data-streamer.h
23266         (histogram_entry): New structure.
23267         (histogram, histogram_pool): New global vars.
23268         (histogram_hash): New structure.
23269         (histogram_hash::hash): New method.
23270         (histogram_hash::equal): Likewise.
23271         (account_time_size): New function.
23272         (cmp_counts): New function.
23273         (dump_histogram): New function.
23274         (ipa_profile_generate_summary): New function.
23275         (ipa_profile_write_summary): New function.
23276         (ipa_profile_read_summary): New function.
23277         (ipa_profile): Decide on threshold.
23278         (pass_ipa_profile): Add ipa_profile_write_summary and
23279         ipa_profile_read_summary.
23280         * Makefile.in (ipa.o): Update dependencies.
23281         * lto-streamer.h (LTO_section_ipa_profile): New section.
23283 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
23285         * tree.h (VAR_P): New.
23287 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
23289         PR lto/56777
23290         * doc/invoke.texi ([-fwhole-program]): Fix typo.
23292 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
23294         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
23295         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
23296         (control_flow_insn_p): Likewise.
23297         * cfgrtl.c (duplicate_insn_chain): Likewise.
23298         * final.c (get_attr_length_1): Likewise.
23299         (shorten_branches): Likewise.
23300         (final_scan_insn): Likewise.
23301         * function.c (instantiate_virtual_regs): Likewise.
23302         * gcse.c (insert_insn_end_basic_block): Likewise.
23303         * ira-costs.c (scan_one_insn): Likewise.
23304         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
23305         * lra.c (check_rtl): Likewise.
23306         * reload1.c (elimination_costs_in_insn): Likewise.
23307         * reorg.c (follow_jumps): Likewise.
23309         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
23310         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
23311         (thumb_far_jump_used_p): Likewise.
23312         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
23313         (workaround_speculation): Likewise.
23314         (add_sched_insns_for_speculation): Likewise.
23315         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
23316         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
23317         (frv_for_each_packet): Likewise.
23318         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
23319         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
23320         (final_emit_insn_group_barriers): Likewise.
23321         * config/m32r/m32r.c (m32r_is_insn): Likewise.
23322         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
23323         (mips16_insn_length): Likewise.
23324         * config/pa/pa.c (pa_reorg): Likewise.
23325         (pa_combine_instructions): Likewise.
23326         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
23327         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
23328         (sh_reorg): Likewise.
23329         (split_branches): Likewise.
23330         * config/spu/spu.c (get_branch_target): Likewise.
23332         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
23333         JUMP_TABLE_DATA_P.
23335 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
23337         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
23338         Fix declaration name.
23340 2013-03-28  Lawrence Crowl  <crowl@google.com>
23342         * graphds.h (struct graph.indicies): Remove unused.
23343         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
23344         (SCOP_ORIGINAL_PDDRS): Remove unused.
23345         * sese.h (extern insert_loop_close_phis): Removed unused.
23346         (extern insert_guard_phis): Removed unused.
23347         (extern ivtype_map_elt_info): Removed unused.
23348         (new_ivtype_map_elt): Removed unused.
23349         * sese.c (ivtype_map_elt_info): Removed unused.
23351 2013-03-28  Lawrence Crowl  <crowl@google.com>
23353         * Makefile.in: Add several missing include dependences.
23354         (DUMPFILE_H): New.
23355         (test-dump.o): New.  This object is not added to any executable,
23356         but is present for ad-hoc testing.
23357         * bitmap.c
23358         (debug (const bitmap_head_def &)): New.
23359         (debug (const bitmap_head_def *)): New.
23360         * bitmap.h
23361         (extern debug (const bitmap_head_def &)): New.
23362         (extern debug (const bitmap_head_def *)): New.
23363         * cfg.c
23364         (debug (edge_def &)): New.
23365         (debug (edge_def *)): New.
23366         * cfghooks.c
23367         (debug (basic_block_def &)): New.
23368         (debug (basic_block_def *)): New.
23369         * dumpfile.h
23370         (dump_node (const_tree, int, FILE *)): Correct source file.
23371         * dwarf2out.c
23372         (debug (die_struct &)): New.
23373         (debug (die_struct *)): New.
23374         * dwarf2out.h
23375         (extern debug (die_struct &)): New.
23376         (extern debug (die_struct *)): New.
23377         * gimple-pretty-print.c
23378         (debug (gimple_statement_d &)): New.
23379         (debug (gimple_statement_d *)): New.
23380         * gimple-pretty-print.h
23381         (extern debug (gimple_statement_d &)): New.
23382         (extern debug (gimple_statement_d *)): New.
23383         * ira-build.c
23384         (debug (ira_allocno_copy &)): New.
23385         (debug (ira_allocno_copy *)): New.
23386         (debug (ira_allocno &)): New.
23387         (debug (ira_allocno *)): New.
23388         * ira-int.h
23389         (extern debug (ira_allocno_copy &)): New.
23390         (extern debug (ira_allocno_copy *)): New.
23391         (extern debug (ira_allocno &)): New.
23392         (extern debug (ira_allocno *)): New.
23393         * ira-lives.c
23394         (debug (live_range &)): New.
23395         (debug (live_range *)): New.
23396         * lra-int.h
23397         (debug (lra_live_range &)): New.
23398         (debug (lra_live_range *)): New.
23399         * lra-lives.c
23400         (debug (lra_live_range &)): New.
23401         (debug (lra_live_range *)): New.
23402         * omega.c
23403         (debug (omega_pb_d &)): New.
23404         (debug (omega_pb_d *)): New.
23405         * omega.h
23406         (extern debug (omega_pb_d &)): New.
23407         (extern debug (omega_pb_d *)): New.
23408         * print-rtl.c
23409         (debug (const rtx_def &)): New.
23410         (debug (const rtx_def *)): New.
23411         * print-tree.c
23412         (debug_tree (tree): Move within file.
23413         (debug_raw (const tree_node &)): New.
23414         (debug_raw (const tree_node *)): New.
23415         (dump_tree_via_hooks (const tree_node *, int)): New.
23416         (debug (const tree_node &)): New.
23417         (debug (const tree_node *)): New.
23418         (debug_verbose (const tree_node &)): New.
23419         (debug_verbose (const tree_node *)): New.
23420         (debug_head (const tree_node &)): New.
23421         (debug_head (const tree_node *)): New.
23422         (debug_body (const tree_node &)): New.
23423         (debug_body (const tree_node *)): New.
23424         (debug_vec_tree (tree): Move and reimplement in terms of dump.
23425         (debug (vec<tree, va_gc> &)): New.
23426         (debug (vec<tree, va_gc> *)): New.
23427         * rtl.h
23428         (extern debug (const rtx_def &)): New.
23429         (extern debug (const rtx_def *)): New.
23430         * sbitmap.c
23431         (debug_raw (simple_bitmap_def &)): New.
23432         (debug_raw (simple_bitmap_def *)): New.
23433         (debug (simple_bitmap_def &)): New.
23434         (debug (simple_bitmap_def *)): New.
23435         * sbitmap.h
23436         (extern debug (simple_bitmap_def &)): New.
23437         (extern debug (simple_bitmap_def *)): New.
23438         (extern debug_raw (simple_bitmap_def &)): New.
23439         (extern debug_raw (simple_bitmap_def *)): New.
23440         * sel-sched-dump.c
23441         (debug (vinsn_def &)): New.
23442         (debug (vinsn_def *)): New.
23443         (debug_verbose (vinsn_def &)): New.
23444         (debug_verbose (vinsn_def *)): New.
23445         (debug (expr_def &)): New.
23446         (debug (expr_def *)): New.
23447         (debug_verbose (expr_def &)): New.
23448         (debug_verbose (expr_def *)): New.
23449         (debug (vec<rtx> &)): New.
23450         (debug (vec<rtx> *)): New.
23451         * sel-sched-dump.h
23452         (extern debug (vinsn_def &)): New.
23453         (extern debug (vinsn_def *)): New.
23454         (extern debug_verbose (vinsn_def &)): New.
23455         (extern debug_verbose (vinsn_def *)): New.
23456         (extern debug (expr_def &)): New.
23457         (extern debug (expr_def *)): New.
23458         (extern debug_verbose (expr_def &)): New.
23459         (extern debug_verbose (expr_def *)): New.
23460         (extern debug (vec<rtx> &)): New.
23461         (extern debug (vec<rtx> *)): New.
23462         * sel-sched-ir.h
23463         (_list_iter_cond_expr): Make inline instead of static.
23464         * sreal.c
23465         (debug (sreal &)): New.
23466         (debug (sreal *)): New.
23467         * sreal.h
23468         (extern debug (sreal &)): New.
23469         (extern debug (sreal *)): New.
23470         * tree.h
23471         (extern debug_raw (const tree_node &)): New.
23472         (extern debug_raw (const tree_node *)): New.
23473         (extern debug (const tree_node &)): New.
23474         (extern debug (const tree_node *)): New.
23475         (extern debug_verbose (const tree_node &)): New.
23476         (extern debug_verbose (const tree_node *)): New.
23477         (extern debug_head (const tree_node &)): New.
23478         (extern debug_head (const tree_node *)): New.
23479         (extern debug_body (const tree_node &)): New.
23480         (extern debug_body (const tree_node *)): New.
23481         (extern debug (vec<tree, va_gc> &)): New.
23482         (extern debug (vec<tree, va_gc> *)): New.
23483         * tree-cfg.c
23484         (debug (struct loop &)): New.
23485         (debug (struct loop *)): New.
23486         (debug_verbose (struct loop &)): New.
23487         (debug_verbose (struct loop *)): New.
23488         * tree-dump.c: Add header dependence.
23489         * tree-flow.h
23490         (extern debug (struct loop &)): New.
23491         (extern debug (struct loop *)): New.
23492         (extern debug_verbose (struct loop &)): New.
23493         (extern debug_verbose (struct loop *)): New.
23494         * tree-data-ref.c
23495         (debug (data_reference &)): New.
23496         (debug (data_reference *)): New.
23497         (debug (vec<data_reference_p> &)): New.
23498         (debug (vec<data_reference_p> *)): New.
23499         (debug (vec<ddr_p> &)): New.
23500         (debug (vec<ddr_p> *)): New.
23501         * tree-data-ref.h
23502         (extern debug (data_reference &)): New.
23503         (extern debug (data_reference *)): New.
23504         (extern debug (vec<data_reference_p> &)): New.
23505         (extern debug (vec<data_reference_p> *)): New.
23506         (extern debug (vec<ddr_p> &)): New.
23507         (extern debug (vec<ddr_p> *)): New.
23508         * tree-ssa-alias.c
23509         (debug (pt_solution &)): New.
23510         (debug (pt_solution *)): New.
23511         * tree-ssa-alias.h
23512         (extern debug (pt_solution &)): New.
23513         (extern debug (pt_solution *)): New.
23514         * tree-ssa-alias.c
23515         (debug (_var_map &)): New.
23516         (debug (_var_map *)): New.
23517         (debug (tree_live_info_d &)): New.
23518         (debug (tree_live_info_d *)): New.
23519         * tree-ssa-alias.h
23520         (extern debug (_var_map &)): New.
23521         (extern debug (_var_map *)): New.
23522         (extern debug (tree_live_info_d &)): New.
23523         (extern debug (tree_live_info_d *)): New.
23525 2013-03-28  Jan Hubicka  <jh@suse.cz>
23527         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
23529 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
23531         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
23532         record only when desired or required.
23534 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
23536         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
23537         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
23539 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23541         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
23542         (*andsi3_compare0_uxtw): New pattern.
23543         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
23544         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
23546 2013-03-28  Jan Hubicka  <jh@suse.cz>
23548         * data-streamer-in.c (streamer_read_gcov_count): New function.
23549         * gimple-streamer-out.c: Include value-prof.h.
23550         (output_gimple_stmt): Output histogram.
23551         (output_bb): Use streamer_write_gcov_count.
23552         * value-prof.c: Include data-streamer.h
23553         (dump_histogram_value): Add HIST_TYPE_MAX.
23554         (stream_out_histogram_value): New function.
23555         (stream_in_histogram_value): New function.
23556         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
23557         (stream_out_histogram_value, stream_in_histogram_value): Declare.
23558         * data-streamer-out.c (streamer_write_gcov_count): New function.
23559         (streamer_write_gcov_count_stream): New function.
23560         * lto-cgraph.c (lto_output_edge): Update counter streaming.
23561         (lto_output_node): Likewise.
23562         (input_node, input_edge): Likewise.
23563         * lto-streamer-out.c (output_cfg): Update streaming.
23564         * lto-streamer-in.c (input_cfg): Likewise.
23565         * data-streamer.h (streamer_write_gcov_count,
23566         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
23567         * gimple-streamer-in.c: Include value-prof.h
23568         (input_gimple_stmt): Input histograms.
23569         (input_bb): Update profile streaming.
23571 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
23573         * genmodes.c (emit_max_int): New function.
23574         (emit_insn_modes_h): Added call to emit_max_function.
23575         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
23576         Added doc.
23577         * machmode.def: Fixed comment.
23579 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
23581         * combine.c (try_combine): Removed useless assert.
23582         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
23584 2013-03-28  Marek Polacek  <polacek@redhat.com>
23585             Richard Biener  <rguenther@suse.de>
23587         PR tree-optimization/56695
23588         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
23589         build signed result of a vector comparison.
23590         * tree-cfg.c (verify_gimple_comparison): Check that a result
23591         of a vector comparison has signed type.
23593 2013-03-28  Richard Biener  <rguenther@suse.de>
23595         PR tree-optimization/37021
23596         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
23597         do not restrict gaps between groups.
23598         * tree-vect-stmts.c (vectorizable_load): Properly account for
23599         a gap between groups.
23601 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
23603         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
23604         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
23605         is not enabled.
23607 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
23609         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
23610         * doc/extend.texi (Named Address Spaces): Ditto.
23611         (Variable Attributes): Ditto.
23613 2013-03-27  Kai Tietz  <ktietz@redhat.com>
23615         * config.build: Add support for cygwin x64 target.
23616         * config.gcc: Likewise.
23617         * config.host: Likewise.
23618         * configure.ac: Likewise
23619         * configure: Regenerated.
23621 2013-03-27  Kai Tietz  <ktietz@redhat.com>
23623         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
23624         * config/i386/t-cygwin-w64: New file.
23625         * config/i386/cygwin-w64.h: New file.
23626         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
23627         and add support for x64-cygwin target.
23628         (CPP_SPEC): Likewise.
23629         (CXX_WRAP_SPEC_LIST): Undefine before define.
23630         (LIBGCJ_SONAME): Use 15 as version.
23632 2013-03-27  Richard Biener  <rguenther@suse.de>
23634         PR tree-optimization/56716
23635         * tree-ssa-structalias.c (perform_var_substitution): Adjust
23636         dumping for ref nodes.
23638 2013-03-27  Martin Jambor  <mjambor@suse.cz>
23640         PR tree-optimization/55334
23641         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
23642         restricted pointers to arrays.
23644 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
23646         * Makefile.in (.SUFFIXES): Add .cc.
23647         (.c.o): Apply same recipe for implicit rule .cc.o.
23649 2013-03-27  Richard Biener  <rguenther@suse.de>
23651         PR tree-optimization/37021
23652         * tree-vect-data-refs.c (vect_check_strided_load): Allow
23653         REALPART/IMAGPART_EXPRs around the supported refs.
23654         * tree-ssa-structalias.c (find_func_aliases): Assume that
23655         floating-point values are not used to transfer pointers.
23657 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
23659         * target.def (TARGET_HAS_IFUNC_P): New target hook.
23660         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
23661         * doc/tm.texi: Regenerate.
23662         * targhooks.h (default_has_ifunc_p): New.
23663         * targhooks.c (default_has_ifunc_p): Ditto.
23664         * config/linux-protos.h: New file.
23665         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
23666         hook for linux which disables support of indirect functions in android.
23667         * config/linux-android.c: New file.
23668         * config/t-linux-android.c: Ditto.
23669         * config.gcc: Added new object file linux-android.o.
23670         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
23671         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
23672         * varasm.c (do_assemble_alias): Likewise.
23673         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
23674         doesn't support indirect functions.
23675         * configure: Regenerate.
23677 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
23679         PR target/56102
23680         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
23681         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
23682         mult-word mode.
23684 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23686         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
23688 2013-03-27  Terry Guo  <terry.guo@arm.com>
23690         * config/arm/arm-cores.def: Added core cortex-r7.
23691         * config/arm/arm-tune.md: Regenerated.
23692         * config/arm/arm-tables.opt: Regenerated.
23693         * doc/invoke.texi: Added entry for core cortex-r7.
23695 2013-03-27  Walter Lee  <walt@tilera.com>
23697         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
23698         double-decrement of next_scratch_regno.
23700 2013-03-27  Walter Lee  <walt@tilera.com>
23702         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
23703         input operands.
23704         (insn_v1mulus): Ditto.
23705         (insn_v2muls): Ditto.
23707 2013-03-27  Walter Lee  <walt@tilera.com>
23709         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
23710         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
23712 2013-03-27  Walter Lee  <walt@tilera.com>
23714         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
23715         (*sibcall_value): Ditto.
23717 2013-03-27  Walter Lee  <walt@tilera.com>
23719         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
23720         (insn_mnz_v8qi): ... this ...
23721         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
23722         vector equivalent.
23723         (insn_v<n>mnz): Replaced by ...
23724         (insn_v1mnz): ... this ...
23725         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
23726         equivalent.
23727         (insn_mz_<mode>): Replaced by ...
23728         (insn_mz_v8qi): ... this ...
23729         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
23730         vector equivalent.
23731         (insn_v<n>mz): Replaced by ...
23732         (insn_v1mz): ... this ...
23733         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
23734         equivalent.
23736 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
23738         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
23740 2013-03-26  Roland McGrath  <mcgrathr@google.com>
23742         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
23743         than fprintf with a non-constant, non-format string.
23745 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
23747         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
23748         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
23749         operand 0 predicate.
23750         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
23751         attribute.  Use general_x64nomem_operand as operand 1 predicate.
23752         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
23753         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
23754         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
23755         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
23756         (mov<mode>_insv_1): Remove expander.  Merge insn with
23757         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
23758         Use general_x64nomem_operand as operand 1 predicate.
23759         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
23760         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
23761         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
23762         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
23763         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
23764         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
23765         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
23766         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
23767         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
23768         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
23769         (general_x64nomem_operand): Ditto.
23771 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
23773         * config/rtems.opt: Add -pthread option.
23775 2013-03-26  Richard Biener  <rguenther@suse.de>
23777         * alias.c (find_base_term): Avoid redundant and not used recursion.
23778         (base_alias_check): Get the initial base term from the caller.
23779         (true_dependence_1): Compute and pass base terms to base_alias_check.
23780         (write_dependence_p): Likewise.
23781         (may_alias_p): Likewise.
23783 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
23785         * config/aarch64/aarch64.c (aarch64_classify_address): Support
23786         PC-relative load in SI modes and above only.
23788 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
23790         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
23791         * config/arm/iwmmxt.md (WCGR0): Update.
23792         (WCGR1, WCGR2, WCGR3): Likewise.
23794 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
23796         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
23797         Use x64 and nox64 isa attributes.
23799 2013-03-26  Richard Biener  <rguenther@suse.de>
23801         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
23802         alignment computations and rely on get_object_alignment_1
23803         for the !TYPE_P case.
23804         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
23806 2013-03-26  Walter Lee  <walt@tilera.com>
23808         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
23809         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
23811 2013-03-25  Jeff Law  <law@redhat.com>
23813         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
23814         check for INTEGRAL_TYPE_P that was missing due to checking in
23815         wrong version of prior patch.
23817 2013-03-25  Walter Lee  <walt@tilera.com>
23819         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
23820         TILEGX_INSN_SHUFFLEBYTES1.
23821         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
23822         shufflebytes1.
23823         (tilegx_builtins): Ditto.
23824         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
23826 2013-03-25  Walter Lee  <walt@tilera.com>
23828         * config/tilegx/tilegx.md (floatsisf2): New pattern.
23829         (floatunssisf2): New pattern.
23830         (floatsidf2): New pattern.
23831         (floatunssidf2): New pattern.
23833 2013-03-25  Walter Lee  <walt@tilera.com>
23835         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
23836         tests for constraint J, K, N, P.
23838 2013-03-25  Walter Lee  <walt@tilera.com>
23840         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
23841         Use indirect/pcrel encoding.
23842         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
23843         Ditto.
23845 2013-03-25  Steve Ellcey  <sellcey@mips.com>
23847         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
23848         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
23849         * config/mips/mips.c (mips_option_override): Set IMADD default.
23850         * config/mips/mips.h (PTF_AVOID_IMADD): New.
23851         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
23852         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
23853         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
23854         * doc/invoke.texi (-mimadd/-mno-imadd): New.
23856 2013-03-25  Jeff Law  <law@redhat.com>
23858         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
23859         slightly to avoid creating and folding useless trees.  Simplify
23860         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
23862 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
23864         * config/i386/i386.md (*zero_extendsidi2): Merge with
23865         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
23866         * config/i386/predicates.md (x86_64_zext_operand): Rename from
23867         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
23868         targets.  Clarify comment.
23870 2013-03-25  Martin Jambor  <mjambor@suse.cz>
23872         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
23873         pass-through jump functions differently.
23874         (ipa_read_jump_function): Likewise.  Also use setter functions to set
23875         up jump functions.
23877 2013-03-25  Martin Jambor  <mjambor@suse.cz>
23879         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
23880         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
23881         process it.
23882         (ipa_get_indirect_edge_target): New function.
23883         (devirtualization_time_bonus): New parameter known_aggs, pass it to
23884         ipa_get_indirect_edge_target.  Update all callers.
23885         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
23886         ipa_get_indirect_edge_target_1 instead of calling
23887         ipa_get_indirect_edge_target.
23888         (create_specialized_node): Pass aggvlas to
23889         ipcp_discover_new_direct_edges.
23891 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23893         * config/arm/arm.md (f_sels, f_seld): New types.
23894         (*cmov<mode>): New pattern.
23895         * config/arm/predicates.md (arm_vsel_comparison_operator): New
23896         predicate.
23898 2013-03-25  Kai Tietz  <ktietz@redhat.com>
23900         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
23901         POSIX-printf for mingw-hosted  builds.
23903 2013-03-25  Richard Biener  <rguenther@suse.de>
23905         PR middle-end/56694
23906         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
23907         must-not-throw stmt location.
23909 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23911         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
23912         Emit load-acquire versions when acq is true.
23913         (arm_emit_store_exclusive): Add rel parameter.
23914         Emit store-release versions when rel is true.
23915         (arm_split_compare_and_swap): Use acquire-release instructions
23916         instead.
23917         of barriers when appropriate.
23918         (arm_split_atomic_op): Likewise.
23919         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
23920         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
23921         (VUNSPEC_SLX): Likewise.
23922         (VUNSPEC_LDA): Likewise.
23923         (VUNSPEC_STL): Likewise.
23924         * config/arm/sync.md (atomic_load<mode>): New pattern.
23925         (atomic_store<mode>): Likewise.
23926         (arm_load_acquire_exclusive<mode>): Likewise.
23927         (arm_load_acquire_exclusivesi): Likewise.
23928         (arm_load_acquire_exclusivedi): Likewise.
23929         (arm_store_release_exclusive<mode>): Likewise.
23931 2013-03-25  Catherine Moore  <clm@codesourcery.com>
23933         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
23934         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
23935         * config/mip/predicates.md (lwsp_swsp_operand,
23936         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
23937         sb16_operand, db4_operand, db7_operand, ib3_operand,
23938         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
23939         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
23940         andi16_operand): New predicates.
23941         * config/mips/mips.md (compression): New attribute.
23942         (enabled): New attribute.
23943         (length): Consider compression in computing length.
23944         (shift_compression): New code attribute.
23945         (*add<mode>3): New operands. Record compression.
23946         (sub<mode>3): Likewise.
23947         (one_cmpl<mode>2): Likewise.
23948         (*and<mode>3): Likewise.
23949         (*ior<mode>3): Likewise.
23950         (unnamed pattern for xor): Likewise.
23951         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
23952         (*<optab><mode>3): Likewise.
23953         (*mov<mode>_internal: Likewise.
23954         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
23955         (mips_unsigned_immediate_p): New.
23956         (umips_lwsp_swsp_address_p): New.
23957         (m16_based_address_p): New.
23958         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
23959         (mips_unsigned_immediate_p): New prototype.
23960         (lwsp_swsp_address_p): New prototype.
23961         (m16_based_address_p): New prototype.
23962         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
23963         (mips_signed_immediate_p): New function.
23964         (m16_based_address_p): New function.
23965         (lwsp_swsp_address_p): New function.
23966         (mips_print_operand_punctuation): Recognize short delay slot insns
23967         for microMIPS.add<mode>3"
23969 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23971         PR target/56720
23972         * config/arm/iterators.md (v_cmp_result): New mode attribute.
23973         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
23975 2013-03-25  Richard Biener  <rguenther@suse.de>
23977         PR tree-optimization/56689
23978         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
23979         any edge.
23981 2013-03-25  Richard Biener  <rguenther@suse.de>
23983         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
23984         of bitmap.
23985         (memory_references): Likewise.
23986         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
23987         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
23988         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
23989         (gather_mem_refs_in_loops): Fold into ...
23990         (analyze_memory_references): ... this.  Move initialization
23991         to tree_ssa_lim_initialize.
23992         (fill_always_executed_in): Rename to ...
23993         (fill_always_executed_in_1): ... this.
23994         (fill_always_executed_in): Move contains_call computation to
23995         this new function from ...
23996         (tree_ssa_lim_initialize): ... here.
23997         (tree_ssa_lim): Call fill_always_executed_in.
23999 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24001         * postreload.c (reload_combine): Fix code detecting returns.
24003 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24005         * function.c (emit_use_return_register_into_block): On cc0 targets,
24006         do not emit the sequence between cc0 setter and user.
24008 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24010         * config/i386/predicates.md (local_symbolic_operand): Interpret
24011         dll-imported symbols as none-local.
24013 2013-03-25  Richard Biener  <rguenther@suse.de>
24015         * tree-ssa-loop-im.c (struct depend): Remove.
24016         (struct lim_aux_data): Make depends a vec of gimples.
24017         (free_lim_aux_data): Adjust.
24018         (add_dependency): Likewise.
24019         (set_level): Likewise.
24021 2013-03-25  Richard Biener  <rguenther@suse.de>
24023         PR middle-end/56434
24024         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
24025         the pointer returned by calls with ECF_MALLOC set.
24027 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24029         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
24031 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24033         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
24034         using MMXMODE mode iterator.
24035         (*move<mode>_internal): Merge with *movv2sf_internal and
24036         *movv2sf_internal_rex64 using MMXMODE mode iterator.
24038 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24040         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
24041         (record_last_mem_set_info): Likewise.
24043         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
24044         of XNEWVEC followed by memset.
24045         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
24047 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24049         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
24050         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
24051         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
24052         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
24053         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
24054         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
24055         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
24056         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
24057         BARRIER_P instead of GET_CODE.
24059 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24061         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
24062         inaccuracy in the probing code.
24064         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
24065         (ctrapdi4): Likewise.
24067 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24069         * calls.c (expand_call): Add missing guard to code handling return
24070         of non-BLKmode structures in MSB.
24071         * function.c (expand_function_end): Likewise.
24073 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24075         * combine.c (try_combine): Adjust comment.  Do not add the set of
24076         insn #0 if the destination indirectly is set or dies in insn #2.
24077         Tidy up code to distribute a new note.
24079 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24081         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
24082         also for alternatives 16 and 17.
24084 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24086         * config/i386/sse.md (*mov<mode>_internal): Merge with
24087         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
24088         Emit insn template depending on type attribute.  Use
24089         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24090         movd instead of movq mnemonic for interunit moves.  Rewrite mode
24091         attribute calculation.  Remove unit attribute calculation.
24092         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
24093         Set prefix_data16 attribute for DImode ssemov types.
24094         Use Ym instead of y for SSE-MMX conversion alternatives.
24095         Reorder operand constraints.
24097 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
24099         * df.h (df_insn_delete): Adjust prototype.
24100         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
24101         and let it decide whether mark the basic block dirty.
24102         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
24103         * df-scan.c (df_insn_info_delete): New helper function, split
24104         off from df_insn_delete.
24105         (df_scan_free_bb_info): Use it.
24106         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
24107         Likewise.
24108         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
24109         that the insn is actually an insn and it has a non-NULL basic block.
24110         Do not mark basic block dirty if only deleting a DEBUG_INSN.
24112 2013-03-22  Richard Biener  <rguenther@suse.de>
24114         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
24115         dep_ref members.
24116         (mem_ref_alloc): Do not allocate them.
24117         (refs_independent_p): Do not query or maintain a cache.
24119 2013-03-22  Richard Biener  <rguenther@suse.de>
24121         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
24122         (gather_mem_refs_in_loops): Do not compute it.
24123         (analyze_memory_references): Do not allocate it.
24124         (tree_ssa_lim_finalize): Do not free it.
24125         (for_all_locs_in_loop): Do not query all_refs_in_loop.
24127 2013-03-22  Richard Biener  <rguenther@suse.de>
24129         * is-a.h (as_a): Use gcc_checking_assert.
24131 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
24133         * config/aarch64/aarch64.c (aarch64_print_operand): New
24134         format specifier for printing a constant in hex.
24135         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
24136         format specifier for printing second operand.
24138 2013-03-22  Richard Biener  <rguenther@suse.de>
24140         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
24141         bitmaps.
24142         (gather_mem_refs_in_loops): Perform store accumulation here.
24143         (create_vop_ref_mapping_loop): Remove.
24144         (create_vop_ref_mapping): Likewise.
24145         (analyze_memory_references): Initialize refs_stored_in_loop.
24146         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
24147         (record_indep_loop): Remove.
24148         (record_dep_loop): New function.
24149         (ref_indep_loop_p_1): Adjust to only walk over references
24150         in the loop, not its subloops.
24151         (ref_indep_loop_p): Rename to ...
24152         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
24153         maintaining a more fine-grained cache.
24154         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
24155         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
24157 2013-03-22  Richard Biener  <rguenther@suse.de>
24159         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
24160         (struct mem_ref): Make accesses_in_loop a vec of a vec of
24161         aggregate mem_ref_loc.
24162         (free_mem_ref_locs): Inline into ...
24163         (memref_free): ... this and adjust.
24164         (mem_ref_alloc): Adjust.
24165         (mem_ref_locs_alloc): Remove.
24166         (record_mem_ref_loc): Adjust.
24167         (get_all_locs_in_loop): Rewrite into ...
24168         (for_all_locs_in_loop): ... this iterator.
24169         (rewrite_mem_ref_loc): New functor.
24170         (rewrite_mem_refs): Use for_all_locs_in_loop.
24171         (sm_set_flag_if_changed): New functor.
24172         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
24173         (ref_always_accessed): New functor.
24174         (ref_always_accessed_p): Use for_all_locs_in_loop.
24176 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24178         * tree-pass.h (PROP_gimple_lvec): New.
24179         * passes.c (dump_properties): Handle PROP_gimple_lvec.
24180         (init_optimization_passes): Move pass_lower_vector.
24181         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
24182         PROP_gimple_lvec.
24183         (pass_lower_vector): Provide PROP_gimple_lvec.
24184         (pass_lower_vector_ssa): Likewise.
24185         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
24187 2013-03-21  Mark Wielaard  <mjw@redhat.com>
24189         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
24191 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
24193         * config/i386/i386.md (*movdi_internal): Disparage slightly
24194         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
24195         conversion alternatives.
24197 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
24199         PR middle-end/48087
24200         * diagnostic.def (DK_WERROR): New kind.
24201         * diagnostic.h (werrorcount): Define.
24202         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
24203         promoted to DK_ERROR, increment DK_WERROR counter instead of
24204         DK_ERROR counter.
24205         * toplev.c (toplev_main): Call print_ignored_options even if
24206         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
24207         even if just werrorcount is non-zero.
24209         PR debug/55608
24210         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
24211         on failure.
24212         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
24213         (string_cst_pool_decl): New function.
24214         (optimize_one_addr_into_implicit_ptr): New function.
24215         (resolve_addr_in_expr): Optimize DWARF location expression
24216         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
24217         which doesn't live in memory, but has DW_AT_location or
24218         DW_AT_const_value, or refers to a string literal, into
24219         DW_OP_GNU_implicit_pointer.
24220         (optimize_location_into_implicit_ptr): New function.
24221         (resolve_addr): If removing DW_AT_location of a variable because
24222         it was DW_OP_addr of address of the variable, but the variable doesn't
24223         live in memory, try to emit const value attribute for the initializer.
24225 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24227         * tree.h (VECTOR_TYPE_P): New macro.
24228         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
24229         TYPE_MODE): Use it.
24230         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
24231         VEC_COND_EXPR cannot be lvalues.
24232         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
24234 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24236         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
24237         Restrict the transformation to equal modes.
24239 2013-03-21  Richard Biener  <rguenther@suse.de>
24241         PR tree-optimization/39326
24242         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
24243         (MEM_ANALYZABLE): Adjust.
24244         (record_mem_ref_loc): Move bitmap ops ...
24245         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
24246         unanalyzable refs, do not record locations for it.
24247         (analyze_memory_references): Allocate ref zero as shared
24248         unanalyzable ref.
24249         (refs_independent_p): Do not test for unanalyzed mems here.
24250         (ref_indep_loop_p_1): Special-case disambiguation against
24251         the unanalyzed ref.
24252         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
24254 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
24256         * config/arm/arm-protos.h (tune_params): Add
24257         prefer_neon_for_64bits field.
24258         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
24259         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
24260         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
24261         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
24262         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
24263         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
24264         (arm_option_override): Handle -mneon-for-64bits new option.
24265         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
24266         (prefer_neon_for_64bits): Declare new variable.
24267         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
24268         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
24269         (arch_enabled): Handle new arch types. Remove support for onlya8
24270         and nota8.
24271         (one_cmpldi2): Use new arch names.
24272         (zero_extend<mode>di2, extend<mode>di2): Ditto.
24273         * config/arm/arm.opt (mneon-for-64bits): Add option.
24274         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
24275         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
24276         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
24277         of onlya8.
24278         * doc/invoke.texi (-mneon-for-64bits): Document.
24280 2013-03-21  Richard Biener  <rguenther@suse.de>
24282         PR tree-optimization/39326
24283         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
24284         (sort_bbs_in_loop_postorder_cmp): New function.
24285         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
24287 2013-03-21  Richard Biener  <rguenther@suse.de>
24289         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
24290         (vect_insert_into_interleaving_chain): Likewise.
24291         (vect_drs_dependent_in_basic_block): Inline ...
24292         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
24293         split out from ...
24294         (vect_analyze_data_ref_dependence): ... here.  Simplify.
24295         (vect_check_interleaving): Simplify.
24296         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
24297         (vect_slp_analyze_data_ref_dependences): ... this new function.
24298         (dr_group_sort_cmp): New function.
24299         (vect_analyze_data_ref_accesses): Compute data-reference groups
24300         here instead of in vect_analyze_data_ref_dependence.  Use
24301         a more efficient algorithm.
24302         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
24303         vect_slp_analyze_data_ref_dependences.  Call
24304         vect_analyze_data_ref_accesses earlier.
24305         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
24306         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
24307         (vect_slp_analyze_data_ref_dependences): New prototype.
24309 2013-03-21  Richard Biener  <rguenther@suse.de>
24311         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
24312         ref is stored in the loop.
24313         (find_refs_for_sm): Walk only over all stores.
24314         (store_motion_loop): Allocate from lim_bitmap_obstack.
24315         (store_motion): Likewise.
24317 2013-03-21  Richard Biener  <rguenther@suse.de>
24319         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
24320         Update virtual SSA form.
24322 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
24324         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
24325         * configure: Regenerate.
24326         * config.in: Regenerate.
24327         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
24328         if !HAVE_LD_EH_FRAME_CIEV3.
24330 2013-03-21  Richard Biener  <rguenther@suse.de>
24332         * tree-cfg.c (verify_expr_no_block): New function.
24333         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
24334         nor DECL_VALUE_EXPR have locations with associated blocks.
24335         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
24336         (clear_unused_block_pointer): Remove code dealing with
24337         blocks in DECL_DEBUG_EXPR locations.
24339 2013-03-21  Richard Biener  <rguenther@suse.de>
24341         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
24342         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
24343         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
24344         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
24345         instead of DECL_DEBUG_EXPR_IS_FROM.
24346         * gimplify.c (gimplify_modify_expr): Likewise.
24347         * tree-cfg.c (verify_expr_location_1): Likewise.
24348         * tree-complex.c (create_one_component_var): Likewise.
24349         * tree-sra.c (create_access_replacement): Likewise.
24350         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
24351         (clear_unused_block_pointer): Likewise.
24352         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
24353         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
24354         * var-tracking.c (var_debug_decl): Likewise.
24355         (track_expr_p): Likewise.
24356         * tree-inline.c (add_local_variables): Likewise.  Set
24357         DECL_HAS_DEBUG_EXPR_P after copying it.
24358         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
24359         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
24361 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
24363         PR bootstrap/56656
24364         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
24365         * configure: Regenerate.
24366         * config.in: Regenerate.
24367         * config/i386/i386.md (*movdf_internal): Use
24368         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24369         movd instead of movq mnemonic for interunit moves.
24370         (*movdi_internal): Ditto.
24372 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24374         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
24375         (abd<mode>_3): New pattern.
24376         (aba<mode>_3): New pattern.
24377         (fabd<mode>_3): New pattern.
24379 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24381         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
24382         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
24383         occurrence of REGISTER_PREFIX as its empty string.
24385 2013-03-20  Jeff Law  <law@redhat.com>
24387         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
24388         addititional equivalences for equality comparisons between an SSA_NAME
24389         and a constant where the SSA_NAME was set from a widening conversion.
24391 2013-03-20  Walter Lee  <walt@tilera.com>
24393         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
24395 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
24397         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
24398         depending on type attribute.
24399         (*movti_internal): Ditto.
24400         (*movtf_internal): Ditto.
24401         (*movxf_internal): Ditto.
24402         (*movdf_internal): Ditto.
24403         (*movsf_internal): Ditto.
24405 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
24407         * config/i386/i386.md (*movti_internal): Set prefix attribute to
24408         maybe_vex for sselog1 and ssemov types.
24409         (*movdi_internal): Reorder operand constraints.
24410         (*movsi_internal): Ditto.  Set prefix attribute to
24411         maybe_vex for sselog1 and ssemov types.
24412         (*movtf_internal): Set prefix attribute to maybe_vex
24413         for sselog1 and ssemov types.
24414         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
24415         DImode ssemov types.  Reorder operand constraints.
24416         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
24417         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
24418         attribute for SImode ssemov types.  Reorder operand constraints.
24420 2013-03-20  Martin Jambor  <mjambor@suse.cz>
24422         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
24423         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
24425 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
24427         * config/rs6000/predicates.md (indexed_address, update_address_mem
24428         update_indexed_address_mem): New predicates.
24429         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
24430         attribute for load/store instructions.
24431         * config/rs6000/dfp.md (movsd_store): Likewise.
24432         (movsd_load): Likewise.
24433         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
24434         (unnamed HI->DI extend define_insn): Likewise.
24435         (unnamed SI->DI extend define_insn): Likewise.
24436         (unnamed QI->SI extend define_insn): Likewise.
24437         (unnamed QI->HI extend define_insn): Likewise.
24438         (unnamed HI->SI extend define_insn): Likewise.
24439         (unnamed HI->SI extend define_insn): Likewise.
24440         (extendsfdf2_fpr): Likewise.
24441         (movsi_internal1): Likewise.
24442         (movsi_internal1_single): Likewise.
24443         (movhi_internal): Likewise.
24444         (movqi_internal): Likewise.
24445         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
24446         attribute for load/store instructions.
24447         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
24448         instructions.
24449         (mov<mode>_softfloat): Likewise.
24450         (mov<mode>_hardfloat32): Likewise.
24451         (mov<mode>_hardfloat64): Likewise.
24452         (mov<mode>_softfloat64): Likewise.
24453         (movdi_internal32): Likewise.
24454         (movdi_internal64): Likewise.
24455         (probe_stack_<mode>): Likewise.
24457 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
24459         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
24460         floating point, and decimal floating point to reload iterator.
24462         * config/rs6000/constraints.md (wl constraint): New constraints to
24463         return FLOAT_REGS if certain options are used to reduce the number
24464         of separate patterns that exist in the file.
24465         (wx constraint): Likewise.
24466         (wz constraint): Likewise.
24468         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
24469         -mdebug=reg, print wg, wl, wx, and wz constraints.
24470         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
24471         Initialize the reload functions for 64-bit binary/decimal floating
24472         point types.
24473         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
24474         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
24475         create the buffer on the stack to overcome not having a 32-bit
24476         load and store.
24477         (rs6000_emit_move): Likewise.
24478         (rs6000_secondary_memory_needed_rtx): Likewise.
24479         (rs6000_alloc_sdmode_stack_slot): Likewise.
24480         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
24481         via xxlxor, just like DFmode 0.0.
24483         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
24484         define as 1 if we are running on a power7 or newer.
24485         (enum r6000_reg_class_enum): Add new constraints.
24487         * config/rs6000/dfp.md (movsd): Delete, combine with binary
24488         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
24489         with other moves by using conditional constraits (wg).  Use LFIWZX
24490         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
24491         (movsd splitter): Likewise.
24492         (movsd_hardfloat): Likewise.
24493         (movsd_softfloat): Likewise.
24495         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
24496         binary and decimal floating point moves.
24497         (fmove_ok): New attributes to combine binary and decimal floating
24498         point moves, and to combine power6x (mfpgpr) moves along normal
24499         floating moves.
24500         (real_value_to_target): Likewise.
24501         (f32_lr): Likewise.
24502         (f32_lm): Likewise.
24503         (f32_li): Likewise.
24504         (f32_sr): Likewise.
24505         (f32_sm): Likewise.
24506         (f32_si): Likewise.
24507         (movsf): Combine binary and decimal floating point moves.  Combine
24508         power6x (mfpgpr) moves with other moves by using conditional
24509         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
24510         (mov<mode> for SFmode/SDmode); Likewise.
24511         (SFmode/SDmode splitters): Likewise.
24512         (movsf_hardfloat): Likewise.
24513         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
24514         (movsf_softfloat): Likewise.
24515         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
24517         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
24518         wx and wz constraints.
24520         * config/rs6000/constraints.md (wg constraint): New constraint to
24521         return FLOAT_REGS if -mmfpgpr (power6x) was used.
24523         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
24524         constraint.
24526         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
24527         -mdebug=reg, print wg, wl, wx, and wz constraints.
24528         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
24529         Initialize the reload functions for 64-bit binary/decimal floating
24530         point types.
24531         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
24532         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
24533         create the buffer on the stack to overcome not having a 32-bit
24534         load and store.
24535         (rs6000_emit_move): Likewise.
24536         (rs6000_secondary_memory_needed_rtx): Likewise.
24537         (rs6000_alloc_sdmode_stack_slot): Likewise.
24538         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
24539         via xxlxor, just like DFmode 0.0.
24541         * config/rs6000/dfp.md (movdd): Delete, combine with binary
24542         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
24543         with other moves by using conditional constraits (wg).  Use LFIWZX
24544         and STFIWX for loading SDmode on power7.
24545         (movdd splitters): Likewise.
24546         (movdd_hardfloat32): Likewise.
24547         (movdd_softfloat32): Likewise.
24548         (movdd_hardfloat64_mfpgpr): Likewise.
24549         (movdd_hardfloat64): Likewise.
24550         (movdd_softfloat64): Likewise.
24552         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
24553         64-bit binary and decimal floating point moves.
24554         (FMOVE64X): Likewise.
24555         (movdf): Combine 64-bit binary and decimal floating point moves.
24556         Combine power6x (mfpgpr) moves with other moves by using
24557         conditional constraits (wg).
24558         (mov<mode> for DFmode/DDmode): Likewise.
24559         (DFmode/DDmode splitters): Likewise.
24560         (movdf_hardfloat32): Likewise.
24561         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
24562         (movdf_softfloat32): Likewise.
24563         (movdf_hardfloat64_mfpgpr): Likewise.
24564         (movdf_hardfloat64): Likewise.
24565         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
24566         (movdf_softfloat64): Likewise.
24567         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
24568         (reload_<mode>_load): Move to later in the file so they aren't in
24569         the middle of the floating point move insns.
24570         (reload_<mode>_store): Likewise.
24572         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
24573         constraint.
24575         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
24576         constraint if -mdebug=reg.
24577         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
24578         Enable using dd reload support if needed.
24580         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
24581         binary and decimal floating point moves in rs6000.md.
24582         (movtd_internal): Likewise.
24584         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
24585         decimal floating point moves.
24586         (movtf): Likewise.
24587         (movtf_internal): Likewise.
24588         (mov<mode>_internal, TDmode/TFmode): Likewise.
24589         (movtf_softfloat): Likewise.
24590         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
24592         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
24593         movdi_internal64, using wg constraint for move direct operations.
24594         (movdi_internal64): Likewise.
24596         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
24597         MODES_TIEABLE_P for selected modes.  Print the numerical value of
24598         the various virtual registers. Use GPR/FPR first/last values,
24599         instead of hard coding the register numbers.  Print which modes
24600         have reload functions registered.
24601         (rs6000_option_override_internal): If -mdebug=reg, trace the options
24602         settings before/after setting cpu, target and subtarget settings.
24603         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
24604         and for secondary reload failures in rs6000_secondary_reload_inner.
24605         (rs6000_secondary_reload_fail): Likewise.
24606         (rs6000_secondary_reload_inner): Likewise.
24608         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
24609         macros for first/last GPR and FPR registers.
24610         (LAST_GPR_REGNO): Likewise.
24611         (FIRST_FPR_REGNO): Likewise.
24612         (LAST_FPR_REGNO): Likewise.
24614         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
24615         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
24616         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
24617         (vcond<mode><mode>): Likewise.
24618         (vcondu<mode><mode>): Likewise.
24619         (vector_gtu<mode>): Likewise.
24620         (vector_gte<mode>): Likewise.
24621         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
24622         to prevent the compiler from converting DImode operations to TImode.
24623         (ior<mode>3): Likewise.
24624         (and<mode>3): Likewise.
24625         (one_cmpl<mode>2): Likewise.
24626         (nor<mode>3): Likewise.
24627         (andc<mode>3): Likewise.
24629         * config/rs6000/constraints.md (wt constraint): New constraint
24630         that returns VSX_REGS if TImode is allowed in VSX registers.
24632         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
24633         constant under VSX.
24635         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
24636         similar to TImode, but it is restricted to being in the GPRs.
24638         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
24639         TImode to occupy a single VSX register.
24641         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
24642         -mvsx-timode for power7/power8.
24643         (power7 cpu): Likewise.
24644         (power8 cpu): Likewise.
24646         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
24647         sure that TFmode/TDmode take up two registers if they are ever
24648         allowed in the upper VSX registers.
24649         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
24650         registers.
24651         (rs6000_init_hard_regno_mode_ok): Likewise.
24652         (rs6000_debug_reg_global): Add debugging for PTImode and wt
24653         constraint.  Print if LRA is turned on.
24654         (rs6000_option_override_internal): Give an error if -mvsx-timode
24655         and VSX is not enabled.
24656         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
24657         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
24658         to reg+offset addressing.  Use PTImode when checking offset
24659         addresses for validity.
24660         (reg_offset_addressing_ok_p): Likewise.
24661         (rs6000_legitimate_offset_address_p): Likewise.
24662         (rs6000_legitimize_address): Likewise.
24663         (rs6000_legitimize_reload_address): Likewise.
24664         (rs6000_legitimate_address_p): Likewise.
24665         (rs6000_eliminate_indexed_memrefs): Likewise.
24666         (rs6000_emit_move): Likewise.
24667         (rs6000_secondary_reload): Likewise.
24668         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
24669         reloads to fpr registers to continue to use reg+offset addressing,
24670         but 64-bit reloads to altivec registers need reg+reg addressing.
24671         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
24672         it.  Treat LO_SUM like a PLUS operation.
24673         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
24674         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
24675         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
24676         registers to share a register with a smaller sized type, since VSX
24677         puts scalars in the upper 64-bits.
24678         (print_operand): Add support for PTImode.
24679         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
24680         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
24681         registers, but don't have arithmetic support.
24682         (rs6000_memory_move_cost): Add test for VSX.
24683         (rs6000_opt_masks): Add -mvsx-timode.
24685         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
24686         for TImode.
24687         (VSs): Likewise.
24688         (VSr): Use wt constraint for TImode.
24689         (VSv): Drop TImode support.
24690         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
24691         (vsx_movti_64bit): Likewise.
24692         (vsx_movti_32bit): Likewise.
24693         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
24694         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
24695         one '?' on the appropriate output constraint.  Do not allow TImode
24696         logical operations on 32-bit systems.
24697         (vsx_ior<mode>3): Likewise.
24698         (vsx_xor<mode>3): Likewise.
24699         (vsx_one_cmpl<mode>2): Likewise.
24700         (vsx_nor<mode>3): Likewise.
24701         (vsx_andc<mode>3): Likewise.
24702         (vsx_concat_<mode>): Likewise.
24703         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
24705         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
24706         OPTION_MASK_VSX_TIMODE.
24707         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
24708         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
24710         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
24711         (TI2 iterator): New iterator for TImode, PTImode.
24712         (wd mode attribute): Add values for vector types.
24713         (movti_string): Replace TI move operations with operations for TImode
24714         and PTImode.  Add support for TImode being allowed in VSX registers.
24715         (mov<mode>_string, TImode/PTImode): Likewise.
24716         (movti_ppc64): Likewise.
24717         (mov<mode>_ppc64, TImode/PTImode): Likewise.
24718         (TI mode splitters): Likewise.
24720         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
24721         constraint.
24723 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
24725         PR tree-optimization/56355
24726         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
24727         Also handle integers with undefined overflow.
24729 2013-03-20  Catherine Moore  <clm@codesourcery.com>
24730             Maciej W. Rozycki  <macro@codesourcery.com>
24731             Tom de Vries  <tom@codesourcery.com>
24732             Nathan Sidwell  <nathan@codesourcery.com>
24733             Iain Sandoe  <iain@codesourcery.com>
24734             Nathan Froyd  <froydnj@codesourcery.com>
24735             Chao-ying Fu  <fu@mips.com>
24737         * doc/extend.texi: (micromips, nomicromips, nocompression):
24738         Document new function attributes.
24739         * doc/invoke.texi (minterlink-compressed, mmicromips,
24740         m14k, m14ke, m14kec): Document new options.
24741         (minterlink-mips16): Update documentation.
24742         * doc/md.texi (ZC, ZD): Document new constraints.
24743         * configure.ac (gcc_cv_as_micromips): Check if linker
24744         supports the .set micromips directive.
24745         * configure: Regenerate.
24746         * config.in: Regenerate.
24747         * config/mips/mips-tables.opt: Regenerate.
24748         * config/mips/micromips.md: New file.
24749         * constraints.md (ZC, ZD): New constraints.
24750         * config/mips/predicates.md (movep_src_register): New predicate.
24751         (movep_src_operand): New predicate.
24752         (non_volatile_mem_operand): New predicate.
24753         * config/mips/mips.md (multimem): New type.
24754         (length): Differentiate between 17-bit and 18-bit branch offsets.
24755         (MOVEP1, MOVEP2): New mode iterator.
24756         (mov_<load>l): Use ZC constraint.
24757         (mov_<load>r): Likewise.
24758         (mov_<store>l): Likewise.
24759         (mov_<store>r): Likewise.
24760         (*branch_equality<mode>_inverted): Add microMIPS support.
24761         (*branch_equality<mode>): Likewise.
24762         (*jump_absolute): Likewise.
24763         (indirect_jump_<mode>): Likewise.
24764         (tablejump_<mode>): Likewise.
24765         (<optab>_internal): Likewise.
24766         (sibcall_internal): Likewise.
24767         (sibcall_value_internal): Likewise.
24768         (prefetch): Use constraint ZD.
24769         * config/mips/mips.opt (minterlink-compressed): New option.
24770         (minterlink-mips16): Now an alias for minterlink-compressed.
24771         (mmicromips): New option.
24772         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24773         (compare_and_swap_12): Likewise.
24774         (sync_add<mode>): Likewise.
24775         (sync_<optab>_12): Likewise.
24776         (sync_old_<optab>_12): Likewise.
24777         (sync_new_<optab>_12): Likewise.
24778         (sync_nand_12): Likewise.
24779         (sync_old_nand_12): Likewise.
24780         (sync_new_nand_12): Likewise.
24781         (sync_sub<mode>): Likewise.
24782         (sync_old_add<mode>): Likewise.
24783         (sync_old_sub<mode>): Likewise.
24784         (sync_new_add<mode>): Likewise.
24785         (sync_new_sub<mode>): Likewise.
24786         (sync_<optab><mode>): Likewise.
24787         (sync_old_<optab><mode>): Likewise.
24788         (sync_new_<optab><mode>): Likewise.
24789         (sync_nand<mode>): Likewise.
24790         (sync_old_nand<mode>): Likewise.
24791         (sync_new_nand<mode>): Likewise.
24792         (sync_lock_test_and_set<mode>): Likewise.
24793         (test_and_set_12): Likewise.
24794         (atomic_compare_and_swap<mode>): Likewise.
24795         (atomic_exchange<mode>_llsc): Likewise.
24796         (atomic_fetch_add<mode>_llsc): Likewise.
24797         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24798         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24799         (umips_save_restore_pattern_p): Likewise.
24800         (umips_load_store_pair_p): Likewise.
24801         (umips_output_load_store_pair): Likewise.
24802         (umips_movep_target_p): Likewise.
24803         (umips_12bit_offset_address_p): Likewise.
24804         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24805         (mips_base_mips16): Rename this...
24806         (mips_base_compression_flags): ...to this. Update all uses.
24807         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24808         (mips_mips16_decl_p): Delete.
24809         (mips_nomips16_decl_p): Delete.
24810         (mips_get_compress_on_flags): New function.
24811         (mips_get_compress_off_flags): New function.
24812         (mips_get_compress_mode): New function.
24813         (mips_get_compress_on_name): New function.
24814         (mips_get_compress_off_name): New function.
24815         (mips_insert_attributes): Support multiple compression types.
24816         (mips_merge_decl_attributes): Likewise.
24817         (umips_12bit_offset_address_p): New function.
24818         (mips_start_function_definition): Emit .set micromips directive.
24819         (mips_call_may_need_jalx_p): New function.
24820         (mips_function_ok_for_sibcall): Add microMIPS support.
24821         (mips_print_operand_punctuation): Support short delay slots and
24822         compact jumps.
24823         (umips_swm_mask, umips_swm_encoding): New.
24824         (umips_build_save_restore): New function.
24825         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24826         (was_mips16_p): Remove.
24827         (old_compression_mode): New.
24828         (mips_set_compression_mode): New function.
24829         (mips_set_current_function): Add microMIPS support.
24830         (mips_option_override): Likewise.
24831         (umips_save_restore_pattern_p): New function.
24832         (umips_output_save_restore): New function.
24833         (umips_load_store_pair_p_1): New function.
24834         (umips_load_store_pair_p): New function.
24835         (umips_output_load_store_pair_1): New function.
24836         (umips_output_load_store_pair): New function.
24837         (umips_movep_target_p) New function.
24838         (mips_prepare_pch_save): Add microMIPS support.
24839         * config/mips/mips.h (TARGET_COMPRESSION): New.
24840         (TARGET_CPU_CPP_BUILTINS): Update macro
24841         to use new compression flags and to support microMIPS.
24842         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24843         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24844         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24845         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24846         (ASM_SPEC): Support mmicromips and mno-micromips.
24847         (M16STORE_REG_P): New macro.
24848         (MIPS_CALL): Support TARGET_MICROMIPS.
24849         (MICROMIPS_J): New macro.
24850         (mips_base_mips16): Rename this...
24851         (mips_base_compression_flags): ...to this.
24852         (UMIPS_12BIT_OFFSET_P): New macro.
24853         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24854         (MULTILIB_DIRNAMES): Likewise.
24855 2013-03-20  Richard Biener  <rguenther@suse.de>
24857         PR tree-optimization/56661
24858         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
24859         the result does not have to be distinct.
24861 2013-03-20  Richard Biener  <rguenther@suse.de>
24863         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
24864         remap_gimple_op_r.
24866 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
24867             Steven Bosscher  <steven@gcc.gnu.org>
24869         PR rtl-optimization/56605
24870         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
24872 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
24874         PR bootstrap/56656
24875         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
24876         that require movd instead of movq.
24878 2013-03-20  Richard Biener  <rguenther@suse.de>
24880         * tree-ssa-structalias.c (struct variable_info): Add pointer
24881         to the first field of an aggregate with sub-vars.  Make
24882         this and the pointer to the next subfield its ID.
24883         (vi_next): New function.
24884         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
24885         storedanything_id, integer_id): Increment by one.
24886         (new_var_info, get_call_vi, lookup_call_clobber_vi,
24887         get_call_clobber_vi): Adjust.
24888         (solution_set_expand): Simplify and speedup.
24889         (solution_set_add): Inline into ...
24890         (set_union_with_increment): ... this.  Adjust accordingly.
24891         (do_sd_constraint): Likewise.
24892         (do_ds_constraint): Likewise.
24893         (do_complex_constraint): Simplify.
24894         (build_pred_graph): Adjust.
24895         (solve_graph): Likewise.  Simplify and speedup.
24896         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
24897         get_constraint_for_component_ref, get_constraint_for_1,
24898         first_vi_for_offset, first_or_preceding_vi_for_offset,
24899         create_function_info_for, create_variable_info_for_1,
24900         create_variable_info_for, intra_create_variable_infos): Adjust.
24901         (init_base_vars): Push NULL for ID zero.
24902         (compute_points_to_sets): Adjust.
24904 2013-03-20  Richard Biener  <rguenther@suse.de>
24906         * cfgloop.c (verify_loop_structure): Streamline and avoid
24907         ICEing on corrupt loop tree.
24908         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
24909         loop tree.
24911 2013-03-20  Richard Biener  <rguenther@suse.de>
24913         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
24914         check whether an SSA update is needed.
24916 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
24918         * config/mips/constraints.md (T): Rename to...
24919         (Yf): ...this.
24920         (U): Rename to...
24921         (Yd): ...this.
24922         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
24923         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
24925 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
24927         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
24928         (*subsi3_carryin_uxtw): Likewise.
24930 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
24932         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
24933         (*rorsi3_insn_uxtw): Likewise.
24935 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
24937         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
24938         (*extrsi5_insn_uxtw): Likewise.
24940 2013-03-19  Richard Biener  <rguenther@suse.de>
24942         PR tree-optimization/56273
24943         * passes.c (init_optimization_passes): Move second VRP after DOM.
24945 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
24947         * config/i386/i386.md (*movti_internal): Merge from
24948         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
24949         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
24950         nox64 isa attributes.
24952 2013-03-18  Richard Biener  <rguenther@suse.de>
24954         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
24955         (unite): Likewise.
24956         (merge_node_constraints): Likewise.
24957         (build_succ_graph): Likewise.
24958         (valid_graph_edge): Inline into single caller.
24959         (unify_nodes): Likewise.  Use bitmap_set_bit return value
24960         and cache varinfo.
24961         (scc_visit): Fix formatting and variable use.
24962         (do_sd_constraint): Use gcc_checking_assert.
24963         (do_ds_constraint): Likewise.
24964         (do_complex_constraint): Likewise.
24965         (condense_visit): Likewise.  Cleanup.
24966         (dump_pred_graph): New function.
24967         (perform_var_substitution): Dump the pred-graph before
24968         variable substitution.
24969         (find_equivalent_node): Use gcc_checking_assert.
24970         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
24972 2013-03-18  Richard Biener  <rguenther@suse.de>
24974         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
24975         Remove cond_expr_stmt_list argument and do not gimplify the
24976         built expression.
24977         (vect_loop_versioning): Adjust.
24978         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
24979         Cleanup to use less temporaries.
24980         (vect_create_data_ref_ptr): Cleanup.
24982 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
24984         PR tree-optimization/56635
24985         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
24986         require types_compatible_p types.
24988 2013-03-18  Nick Clifton  <nickc@redhat.com>
24990         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
24991         spurious backslash.
24993         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
24994         Add missing line to comment describing function.
24996 2013-03-18  Richard Biener  <rguenther@suse.de>
24998         PR tree-optimization/56210
24999         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25000         Handle string / character search functions.
25001         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
25003 2013-03-18  Richard Biener  <rguenther@suse.de>
25005         PR middle-end/56483
25006         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
25007         and implement properly.
25008         * gimple.h (gimple_cond_single_var_p): Remove.
25010 2013-03-18  Richard Biener  <rguenther@suse.de>
25012         * tree-data-ref.h (find_data_references_in_loop): Declare.
25013         * tree-data-ref.c (get_references_in_stmt): Use a stack
25014         vector pre-allocated in the callers.
25015         (find_data_references_in_stmt): Adjust.
25016         (graphite_find_data_references_in_stmt): Likewise.
25017         (create_rdg_vertices): Likewise.
25018         (find_data_references_in_loop): Export.
25019         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
25020         Compute dependences here...
25021         (vect_analyze_data_refs): ...not here.  When we encounter
25022         a non-vectorizable data reference in basic-block vectorization
25023         truncate the data reference vector.  Do not bother to
25024         fixup data-dependence information for gather loads.
25025         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
25026         of data references, as reported.
25028 2013-03-18  Richard Biener  <rguenther@suse.de>
25030         PR tree-optimization/3713
25031         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
25032         has_constants and expr.
25033         (stmt_has_constants): Properly valueize SSA names when deciding
25034         whether the stmt has constants.
25036 2013-03-18  Richard Biener  <rguenther@suse.de>
25038         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
25039         whole function when there is nothing to do.
25040         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
25041         * tree-vectorizer.c (vectorize_loops): Update virtual and
25042         loop-closed SSA once.
25043         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
25045 2013-03-18  Richard Biener  <rguenther@suse.de>
25047         PR middle-end/56113
25048         * domwalk.c (bb_postorder): New global static.
25049         (cmp_bb_postorder): New function.
25050         (walk_dominator_tree): Replace scheme imposing an order for
25051         visiting dominator sons by one sorting them at the time they
25052         are pushed on the stack.
25054 2013-03-18  Richard Biener  <rguenther@suse.de>
25056         PR tree-optimization/39326
25057         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
25058         (struct mem_ref): Replace mem member with ao_ref typed member.
25059         (MEM_ANALYZABLE): Adjust.
25060         (memref_eq): Likewise.
25061         (mem_ref_alloc): Likewise.
25062         (gather_mem_refs_stmt): Likewise.
25063         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
25064         (execute_sm_if_changed_flag_set): Adjust.
25065         (execute_sm): Likewise.
25066         (ref_always_accessed_p): Likewise.
25067         (refs_independent_p): Likewise.
25068         (can_sm_ref_p): Likewise.
25070 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25072         PR c/56566
25073         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
25074         return 1 even for !unsignedp.
25076 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25078         * config/i386/i386.md (isa): Add x64 and nox64.
25079         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
25080         (*pushtf): Enable *roF alternative for x64 isa only.
25081         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
25082         mode attribute of integer alternatives to DImode for TARGET_64BIT.
25083         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
25084         (*movtf_internal): Merge from *movtf_internal_rex64 and
25085         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
25086         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
25087         nox64 isa attributes.
25088         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
25089         nox64 isa attributes.
25090         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
25092 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25094         * config/alpha/alpha.c (TARGET_LRA_P): New define.
25096 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
25098         PR target/56640
25099         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
25100         class names.  Remove trailing comma after "ALL_REGS".
25102 2013-03-16  Jan Hubicka  <jh@suse.cz>
25104         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
25105         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
25106         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
25107         of cgraph_get_create_node.
25108         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
25110 2013-03-16  Jason Merrill  <jason@redhat.com>
25112         PR debug/49090
25113         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
25114         with DW_AT_default_value.
25116 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
25118         * BASE-VER: Set to 4.9.0.
25120 2013-03-14  Andi Kleen  <ak@linux.intel.com>
25122         PR target/56619
25123         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
25124         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
25125         Document _x* TSX intrinsics.
25127 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
25128             David Holsgrove  <david.holsgrove@xilinx.com>
25130         * configure.ac: Add MicroBlaze TLS support detection.
25131         * configure: Regenerate.
25132         * config/microblaze/microblaze-protos.h
25133         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
25134         symbol_mentioned_p, label_mentioned_p): Add prototypes.
25135         * config/microblaze/microblaze.c (microblaze_address_type): Add
25136         ADDRESS_TLS and tls_reloc address types.
25137         (microblaze_address_info): Add tls_reloc.
25138         (TARGET_HAVE_TLS): Define.
25139         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
25140          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
25141          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
25142         load_tls_operand,  microblaze_call_tls_get_addr,
25143         microblaze_legitimize_tls_address): New functions.
25144         (microblaze_classify_unspec): Handle UNSPEC_TLS.
25145         (get_base_reg): Use microblaze_tls_symbol_p.
25146         (microblaze_classify_address): Handle TLS.
25147         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
25148         label_mentioned_p and microblaze_tls_referenced_p.
25149         (microblaze_legitimize_address): Handle TLS.
25150         (microblaze_address_insns): Handle ADDRESS_TLS.
25151         (pic_address_needs_scratch): Handle TLS.
25152         (print_operand_address): Handle TLS.
25153         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
25154         (microblaze_expand_move): Handle TLS.
25155         (microblaze_legitimate_constant_p): Check
25156         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
25157         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
25158         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
25159         (PIC_OFFSET_TABLE_REGNUM): Set.
25160         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
25161         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
25162         (addsi3, movsi_internal2, movdf_internal): Update constraints
25163         * config/microblaze/predicates.md (arith_plus_operand): Define
25164         (move_operand): Redefine as move_src_operand,
25165         check microblaze_tls_referenced_p.
25167 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25169         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
25170         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
25172 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25174         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
25175         CC mode for AND.
25177 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
25179         PR tree-optimization/53265
25180         * common.opt (Waggressive-loop-optimizations): New option.
25181         * tree-ssa-loop-niter.c: Include tree-pass.h.
25182         (do_warn_aggressive_loop_optimizations): New function.
25183         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
25184         if number_of_latch_executions returned constant.
25185         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
25186         early.  If number_of_latch_executions returned constant, set
25187         nb_iterations_upper_bound back to it.
25188         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
25189         field.
25190         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
25191         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
25193         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
25194         (MULTILIB_OSDIRNAMES): Set.
25195         * genmultilib: If defaultosdirname doesn't start with :: , set
25196         defaultosdirname2 instead, clear it and emit two . multilib_raw
25197         entries instead of just one.
25199 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
25201         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
25202         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
25203         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
25204         (SUBTARGET_OVERRIDE_OPTIONS): New.
25206 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
25208         PR target/49880
25209         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
25210         (musermode): Convert to Var(TARGET_USERMODE).
25211         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
25212         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
25213         * config/sh/sh.c (sh_option_override): Use
25214         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
25215         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
25216         condition.
25217         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
25218         TARGET_SH4.
25219         (udivsi3_i4_single, divsi3_i4_single): Use
25220         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
25222 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
25224         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
25225         default setting.
25227 2013-03-13  Richard Biener  <rguenther@suse.de>
25229         PR tree-optimization/56608
25230         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
25231         calls when vectorizing basic-blocks.
25233 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
25235         PR plugins/45078
25236         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
25237         tm_file.
25239 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
25241         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
25243 2013-03-11  Jan Hubicka  <jh@suse.cz>
25245         PR lto/56557
25246         * lto-streamer-out.c (output_symbol_p): Skip references from
25247         constructors of external variables.
25249 2013-03-11  Jan Hubicka  <jh@suse.cz>
25251         PR middle-end/56571
25252         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
25253         from pseudos.
25254         * emit-rtl.c (verify_rtx_sharing): Likewise.
25255         (copy_insn_1): Likewise.
25256         * rtl.c (copy_rtx): Likewise.
25258 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
25260         PR target/56591
25261         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
25262         output_operand_lossage message.
25264 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
25266         PR target/56470
25267         * arm.c (shift_op): Validate RTL pattern on the fly.
25268         (arm_print_operand, case 'S'): Don't use shift_operator to validate
25269         the RTL.
25271 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25273         PR target/56347
25274         * config/pa/pa.md (call_value): Check for calls to powf and direct to
25275         new call patterns that clobber %fr12.
25276         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
25277         split and postreload patterns.
25278         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
25279         registers %fr12 and %fr12R as call used.
25281 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
25283         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
25284         (canon_address, record_store, replace_read, check_mem_read_rtx,
25285         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
25286         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
25287         rest_of_handle_dse): Likewise.
25289 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
25291         PR middle-end/56524
25292         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
25293         Add base_optabs.
25294         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
25295         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
25296         (save_optabs_if_changed): Replace with...
25297         (init_tree_optimization_optabs): ...this.
25298         * optabs.c (save_optabs_if_changed): Rename to...
25299         (init_tree_optimization_optabs): ...this.  Take the optimization node
25300         as argument.  Do nothing if the base optabs are already correct.
25301         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
25302         to recompute optabs.
25303         * function.h (function): Remove optabs field.
25304         * function.c (invoke_set_current_function_hook): Call
25305         init_tree_optimization_optabs.  Use the result to initialize
25306         this_fn_optabs.
25308 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
25310         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
25311         if GTMA_HAS_NO_INSTRUMENTATION.
25312         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
25313         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
25314         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
25315         * gimple-pretty-print.c (dump_gimple_transaction): Handle
25316         GTMA_HAS_NO_INSTRUMENTATION.
25318 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
25320         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
25321         libasan_preinit.o.
25323 2013-03-08  Marek Polacek  <polacek@redhat.com>
25324             Jakub Jelinek  <jakub@redhat.com>
25326         PR tree-optimization/56478
25327         * predict.c (is_comparison_with_loop_invariant_p): Change the
25328         type of loop_step to tree.
25329         (predict_loops): Adjust.
25330         (predict_iv_comparison): Perform the computations on double_ints.
25332 2013-03-08  Richard Biener  <rguenther@suse.de>
25334         PR tree-optimization/56570
25335         * tree-cfg.c (verify_expr_location_1): Verify locations for
25336         DECL_DEBUG_EXPR.
25337         * tree-sra.c (create_access_replacement): Strip locations
25338         from DECL_DEBUG_EXPRs.
25340 2013-03-08  Richard Biener  <rguenther@suse.de>
25342         * tree-inline.c (expand_call_inline): Do not associate
25343         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
25344         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
25346 2013-03-08  Richard Biener  <rguenther@suse.de>
25348         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
25349         or block changes with -Og.  Fix for location / block encoding
25350         changes and PHI arguments with locations.
25352 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
25354         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
25355         for all counters.
25356         (struct output_info): Likewise.
25357         (register_overhead): Remove bad gcc_assert.
25358         (bitmap_find_bit): If there is only a single bitmap element, do not
25359         count a miss as a search.
25360         (print_statistics): Update for counter type changes.
25361         (dump_bitmap_statistics): Likewise.  Print headers such that they
25362         are properly lined up with the printed counters.
25364 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
25366         PR tree-optimization/56559
25367         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
25368         check that it has only a single use.
25370 2013-03-07  Richard Biener  <rguenther@suse.de>
25372         * doc/invoke.texi (fwhole-program): Discourage use in combination
25373         with -flto.
25375 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
25377         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
25379         PR tree-optimization/56539
25380         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
25381         instead of GSI_CONTINUE_LINKING as last argument to
25382         force_gimple_operand_gsi.  Adjust function comment.
25384         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
25385         aarch64-cores.def.
25387         PR middle-end/56548
25388         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
25389         promoted mode, convert the result back to the original mode.
25391 2013-03-06  Richard Biener  <rguenther@suse.de>
25393         PR middle-end/56294
25394         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
25395         (insert_updated_phi_nodes_compare_uids): New function.
25396         (update_ssa): Sort symbols_to_rename after UID before
25397         traversing it to insert PHI nodes.
25399 2013-03-06  Richard Biener  <rguenther@suse.de>
25401         PR middle-end/50494
25402         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
25403         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
25405         Revert
25406         2013-02-13  Richard Biener  <rguenther@suse.de>
25408         PR lto/50494
25409         * varasm.c (output_constant_def_1): Get the decl representing
25410         the constant as argument.
25411         (output_constant_def): Wrap output_constant_def_1.
25412         (make_decl_rtl): Use output_constant_def_1 with the decl
25413         representing the constant.
25414         (build_constant_desc): Optionally re-use a decl already
25415         representing the constant.
25416         (tree_output_constant_def): Adjust.
25418 2013-03-06  Joey Ye  <joey.ye@arm.com>
25420         PR lto/50293
25421         * gcc.c (convert_white_space): New function.
25422         (main): Handles white space in function name.
25424 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
25426         PR target/56529
25427         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
25428         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
25429         to SH_DIV_CALL_TABLE for TARGET_SH2.
25430         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
25431         list.
25432         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
25433         call-table options.
25435 2013-03-05  Sterling Augustine  <saugustine@google.com>
25436             Cary Coutant  <ccoutant@google.com>
25438         PR debug/55364
25439         * dwarf2out.c (resolve_addr): Don't call
25440         remove_loc_list_addr_table_entries a second time for the same
25441         expression.
25443 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
25445         PR debug/56510
25446         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
25447         (avoid_complex_debug_insns): New function.
25448         (expand_debug_locations): Call it.
25450         PR rtl-optimization/56484
25451         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
25452         lifetimes of hard registers on small register class machines.
25454 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
25456         * config/microblaze/microblaze-protos.h: Rename
25457         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
25458         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
25459         fast_interrupt.
25460         (microblaze_fast_interrupt_function_p): New function.
25461         (microblaze_is_interrupt_handler): Rename to
25462         microblaze_is_interrupt_variant and add fast_interrupt check.
25463         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
25464         (save_restore_insns): Likewise.
25465         (compute_frame_size): Likewise.
25466         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
25467         (microblaze_globalize_label): Likewise.
25468         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
25469         * config/microblaze/microblaze.md: Use wrapper
25470         microblaze_is_interrupt_variant.
25472 2013-03-05  Kai Tietz  <ktietz@redhat.com>
25474         * sdbout.c (sdbout_one_type): Switch to current function's section
25475         supporting cold/hot.
25477 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
25479         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
25480         -mxl-reorder.
25482 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
25484         PR middle-end/56461
25485         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
25486         if VALGRIND_GET_VBITS is defined, temporarily make object
25487         memory all defined, and restore previous valgrind addressability
25488         and definability afterwards.  Free this_object at the end.
25490         PR middle-end/56461
25491         * lra.c (lra): Call lra_clear_live_ranges if live_p,
25492         right before calling lra_create_live_ranges, also call it
25493         when clearing live_p.  Only call lra_clear_live_ranges
25494         at the end if live_p.
25496         PR middle-end/56461
25497         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
25499 2013-03-05  Richard Biener  <rguenther@suse.de>
25501         PR tree-optimization/56521
25502         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
25503         value-id.
25505 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
25507         PR c++/55135
25508         * except.h (remove_unreachable_eh_regions): New prototype.
25509         * except.c (remove_eh_handler_splicer): New function, split out
25510         of remove_eh_handler.
25511         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
25512         warning about running it on many EH regions one at a time.
25513         (remove_unreachable_eh_regions_worker): New function, walk the
25514         EH tree in depth-first order and remove non-marked regions.
25515         (remove_unreachable_eh_regions): New function.
25516         * tree-eh.c (mark_reachable_handlers): New function, split out
25517         from remove_unreachable_handlers.
25518         (remove_unreachable_handlers): Use mark_reachable_handlers and
25519         remove_unreachable_eh_regions.
25520         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
25521         and remove_unreachable_eh_regions.
25523 2013-03-05  Richard Biener  <rguenther@suse.de>
25525         PR middle-end/56525
25526         * loop-init.c (fix_loop_structure): Remove loops in two stages,
25527         not freeing them until the end.
25529 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
25531         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
25533 2013-03-05  Richard Biener  <rguenther@suse.de>
25535         PR tree-optimization/56270
25536         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
25537         of loads after scheduling an SLP instance.
25539 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
25541         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
25542         tic6x.exp.
25543         (check_gcc_parallelize): Run guality.exp as a separate job from
25544         vect.exp with unsorted.exp and $(dg_target_exps) separately from
25545         struct-layout-1.exp with stackalign.exp.
25547         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
25549         PR middle-end/56461
25550         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
25551         load_index sbitmap even if some bit in it isn't set.
25553         PR middle-end/56461
25554         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
25555         (discover_iteration_bound_by_body_walk): Change queues to
25556         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
25557         spelling in comment.  Call safe_push on queues[bound_index] directly.
25558         Release queues[queue_index] in every iteration unconditionally.
25559         Release bounds vector.
25561         PR middle-end/56461
25562         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
25563         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
25564         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
25565         inner_phis vector.
25567 2013-03-05  Richard Biener  <rguenther@suse.de>
25569         PR lto/56515
25570         * tree-inline.c (remap_blocks_to_null): New function.
25571         (expand_call_inline): When expanding a call stmt without
25572         an associated block inline remap all callee blocks to NULL.
25574 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
25576         PR rtl-optimization/56494
25577         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
25578         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
25579         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
25581         PR middle-end/56461
25582         * sel-sched-ir.c (free_sched_pools): Release
25583         succs_info_pool.stack[succs_info_pool.max_top] vectors too
25584         if succs_info_pool.max_top isn't -1.
25586         PR bootstrap/56509
25587         * opts.c (opts_obstack, opts_concat): Moved to...
25588         * opts-common.c (opts_obstack, opts_concat): ... here.
25590 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
25592         PR middle-end/56461
25593         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
25595 2013-03-04  Martin Jambor  <mjambor@suse.cz>
25597         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
25598         all appropriate places.
25600 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
25602         PR tree-optimization/56424
25603         * ipa-split.c (split_function): Do not set the RSO flag if result is
25604         not by reference and its type is a register type.
25606 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
25608         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
25609         (microblaze_legitimate_pic_operand): Likewise
25610         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
25611         new function microblaze_legitimate_pic_operand
25612         * config/microblaze/microblaze-protos.h
25613         (microblaze_legitimate_pic_operand): Declare.
25615 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25617         * config/microblaze/predicates.md (call_insn_simple_operand):
25618         New predicate for supported rtx code types.
25619         * config/microblaze/microblaze.md (call_internal1): Use
25620         call_insn_simple_operand predicate.
25622 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
25624         PR middle-end/56461
25625         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
25626         partitions.ordered_remove.
25628         PR middle-end/56461
25629         * tree-vect-stmts.c (vectorizable_conversion): Don't call
25630         vec_oprnds0.create (1) for modifier == NONE.
25632         PR middle-end/56461
25633         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
25634         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
25635         vec_oprnds1 right before pushing anything to it for
25636         scalar_shift_arg.
25638         PR middle-end/56461
25639         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
25640         set nbbs to 0 instead of having separate code path.
25641         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
25642         instead of false as last argument if returning NULL.
25644 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
25646         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
25647         the attribute is now called "target" instead of "option".
25648         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
25649         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
25650         attribute/pragma name for TARGET_OPTION_VALID_P and
25651         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
25652         * doc/tm.texi: Regenerated.
25654 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
25656         * config/microblaze/microblaze.c:
25657         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
25658         * config/microblaze/microblaze.h: Add -mxl-reorder to
25659         DRIVER_SELF_SPECS.
25660         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
25661         instructions emitted if TARGET_REORDER.
25662         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
25663         or 0 for -m/-mno case, but initialises as 2 to detect default use case
25664         separately.
25666 2013-03-01  Xinliang David Li  <davidxl@google.com>
25668         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
25669         walk length.
25671 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
25673         PR middle-end/56461
25674         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
25675         vector even when returning true.  Fix up function comment formatting.
25677         PR middle-end/56461
25678         * ira-build.c (ira_loop_nodes_count): New variable.
25679         (create_loop_tree_nodes): Initialize it.
25680         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
25682         PR middle-end/56461
25683         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
25684         method on dr_chain and result_chain.
25685         * tree-vect-stmts.c (vectorizable_store): Only call
25686         result_chain.create if j == 0.
25688         PR middle-end/56461
25689         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
25690         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
25691         before overwriting it.
25693 2013-03-01  Tobias Burnus  <burnus@net-b.de>
25695         * doc/extended.texi (C Extensions): Change order in @menu
25696         to match @node.
25697         (Other MIPS Built-in Functions): Move last MIPS entry before
25698         "picoChip Built-in Functions".
25699         (SH Built-in Functions): Move after RX Built-in Functions.
25700         * doc/gcc.texi (Introduction): Change order in @menu
25701         to match @node.
25702         * doc/md.texi (Constraints): Ditto.
25703         * gty.texi (Type Information): Ditto.
25704         (User-provided marking routines for template types): Make
25705         subsection.
25706         * doc/invoke.texi (AArch64 Options): Move before
25707         "Adapteva Epiphany Options".
25709 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
25710             Jakub Jelinek  <jakub@redhat.com>
25712         PR sanitizer/56454
25713         * asan.c (gate_asan): Lookup no_sanitize_address instead of
25714         no_address_safety_analysis attribute.
25715         * doc/extend.texi (no_address_safety_attribute): Rename to
25716         no_sanitize_address attribute, mention no_address_safety_analysis
25717         attribute as deprecated alias.
25719 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
25721         PR middle-end/56461
25722         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
25723         type to vec<vec<tree> > *.
25724         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
25725         to be vec<tree> instead of vec<tree> *, set vec_defs
25726         to vNULL and call vec_defs.create (number_of_vects), adjust other
25727         uses of vec_defs.
25728         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
25729         vectorizable_condition): Adjust vect_get_slp_defs callers.
25731 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
25733         * config/aarch64/aarch64.c
25734         (aarch64_float_const_representable): Remove unused variable.
25736 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
25738         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
25740 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
25742         * config/aarch64/aarch64-builtins.c
25743         (aarch64_init_simd_builtins): Make static.
25745 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
25747         * config/aarch64/aarch64.c
25748         (aarch64_simd_make_constant): Make static.
25750 2013-02-28  Martin Jambor  <mjambor@suse.cz>
25752         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
25753         with no initialization to the RHS of debug statements.
25755 2013-02-28  Martin Jambor  <mjambor@suse.cz>
25757         PR tree-optimization/56294
25758         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
25759         Adjust dumping.
25760         (get_access_replacement): Do not call create_access_replacement.
25761         Assert a replacement exists.
25762         (get_repl_default_def_ssa_name): Create the replacement declaration
25763         itself.
25765 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25767         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
25768         final_end_function.
25770 2013-02-28  Marek Polacek  <polacek@redhat.com>
25772         PR rtl-optimization/56466
25773         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
25774         if we're changing a loop.
25775         (peel_loops_completely): Likewise.
25777 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
25779         PR c++/55813
25780         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
25782 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
25784         PR target/56445
25785         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
25786         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
25787         INTX_FTYPE_FX, FX_FTYPE_INTX.
25788         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
25790 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
25792         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
25793         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
25794         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
25795         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
25796         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
25797         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
25798         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
25799         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
25800         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
25801         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
25802         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
25803         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
25804         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
25805         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
25806         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
25807         (avrxmega6): Increase max flash segments from 5 to 6.
25808         * config/avr/t-multilib: Regenerate.
25809         * config/avr/avr-tables.opt: Regenerate.
25810         * doc/avr-mmcu.texi: Regenerate.
25812 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
25814         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
25815         (avr_device_to_arch): Rename to avr_device_to_ld.
25816         (avr_device_to_as): New prototype.
25817         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
25818         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
25819         * config/avr/driver-avr.c (avr_device_to_as): New.
25820         (avr_device_to_arch): Rename to avr_device_to_ld.
25822 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
25824         PR middle-end/56461
25825         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
25826         method on dr_chain and result_chain.
25828         PR middle-end/56461
25829         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
25830         pointer_set_destroy on not_executed_last_iteration.
25832         PR middle-end/56461
25833         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
25835         PR middle-end/56461
25836         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
25837         FOR_EACH_DEFINED_FUNCTION when freeing state.
25839         PR middle-end/56461
25840         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
25841         pool_free.
25842         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
25843         overwriting it.
25845         PR middle-end/56461
25846         * ipa-cp.c (decide_whether_version_node): Call vec_free on
25847         known_aggs[i].items and release known_aggs vector.
25849         PR middle-end/56461
25850         * ipa-reference.c (propagate): Free node_info even for alias nodes.
25852 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25854         * config/microblaze/microblaze.c (microblaze_emit_compare):
25855         Use xor for EQ/NE comparisions.
25856         * config/microblaze/microblaze.md (cstoresf4): Add constraints
25857         (cbranchsf4): Adjust operator to comparison_operator.
25859 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
25861         PR middle-end/56461
25862         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
25863         vector.
25864         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
25865         vec_safe_push, always update *slot.
25866         (redirect_edge_var_map_clear): Use vec_free.
25867         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
25868         (free_var_map_entry): Use vec_free.
25869         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
25870         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
25872 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
25874         PR middle-end/45472
25875         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
25876         when the may_trap_p bit of the exprs being merged differs.
25877         Reorder tests for speculativeness in the logical and operator.
25879 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
25881         * incpath.c (add_standard_paths): Use reconcat instead of concat
25882         where appropriate and avoid leaking memory.
25884         * opts.h: Include obstack.h.
25885         (opts_concat): New prototype.
25886         (opts_obstack): New declaration.
25887         * opts.c (opts_concat): New function.
25888         (opts_obstack): New variable.
25889         (init_options_struct): Call gcc_init_obstack on opts_obstack.
25890         (finish_options): Use opts_concat instead of concat
25891         and XOBNEWVEC instead of XNEWVEC.
25892         * opts-common.c (generate_canonical_option, decode_cmdline_option,
25893         generate_option): Likewise.
25894         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
25895         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
25897         PR target/56455
25898         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
25899         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
25901 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
25903         PR middle-end/56461
25904         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
25906 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
25908         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
25909         (arm_block_move_unaligned_straight): Likewise.
25910         (arm_adjust_block_mem): Likewise.
25912 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
25914         PR target/48901
25915         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
25916         temp, cond and label.
25917         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
25919         PR target/52500
25920         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
25921         * config/c6x/c6x.h (dbx_register_map): Update declaration.
25923         PR target/52501
25924         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
25925         of prologue/epilogue functions.
25927         PR target/52550
25928         * config/tilegx/tilegx.c (tilegx_expand_prologue):
25929         Remove unused variable cfa_offset.
25930         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
25932         PR target/54639
25933         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
25934         type promotion to unsigned.
25936         PR target/54640
25937         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
25938         for HOST_WIDE_INT of 32 bit / same size as int.
25939         (arm_block_move_unaligned_straight): Likewise.
25940         (arm_adjust_block_mem): Likewise.
25942         PR target/54662
25943         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
25944         ALL_CFLAGS.
25946 2013-02-26  Marek Polacek  <polacek@redhat.com>
25948         PR tree-optimization/56426
25949         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
25951 2013-02-26  Richard Biener  <rguenther@suse.de>
25953         PR target/56444
25954         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
25955         unused variable loops.
25957 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
25959         PR tree-optimization/56448
25960         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
25961         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
25962         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
25963         later operands of the references, or even first operand for
25964         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
25966         PR tree-optimization/56443
25967         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
25968         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
25969         to type_for_mode langhook.
25971 2013-02-25  Matt Turner  <mattst88@gmail.com>
25973         * doc/invoke.texi: Document r4700.
25975 2013-02-25  Richard Biener  <rguenther@suse.de>
25977         PR tree-optimization/56175
25978         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
25979         split out from ...
25980         (simplify_bitwise_binary): ... here.  Also guard the conversion
25981         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
25983 2013-02-25  Catherine Moore  <clm@codesourcery.com>
25985         Revert:
25986         2013-02-24  Catherine Moore  <clm@codesourcery.com>
25987             Maciej W. Rozycki  <macro@codesourcery.com>
25988             Tom de Vries  <tom@codesourcery.com>
25989             Nathan Sidwell  <nathan@codesourcery.com>
25990             Iain Sandoe  <iain@codesourcery.com>
25991             Nathan Froyd  <froydnj@codesourcery.com>
25992             Chao-ying Fu  <fu@mips.com>
25994         * doc/extend.texi: (micromips, nomicromips, nocompression):
25995         Document new function attributes.
25996         * doc/invoke.texi (minterlink-compressed, mmicromips,
25997         m14k, m14ke, m14kec): Document new options.
25998         (minterlink-mips16): Update documentation.
25999         * doc/md.texi (ZC, ZD): Document new constraints.
26000         * configure.ac (gcc_cv_as_micromips): Check if linker
26001         supports the .set micromips directive.
26002         * configure: Regenerate.
26003         * config.in: Regenerate.
26004         * config/mips/mips-tables.opt: Regenerate.
26005         * config/mips/micromips.md: New file.
26006         * constraints.md (ZC, AD): New constraints.
26007         * config/mips/predicates.md (movep_src_register): New predicate.
26008         (movep_src_operand): New predicate.
26009         (non_volatile_mem_operand): New predicate.
26010         * config/mips/mips.md (multimem): New type.
26011         (length): Differentiate between 17-bit and 18-bit branch offsets.
26012         (MOVEP1, MOVEP2): New mode iterator.
26013         (mov_<load>l): Use ZC constraint.
26014         (mov_<load>r): Likewise.
26015         (mov_<store>l): Likewise.
26016         (mov_<store>r): Likewise.
26017         (*branch_equality<mode>_inverted): Add microMIPS support.
26018         (*branch_equality<mode>): Likewise.
26019         (*jump_absolute): Likewise.
26020         (indirect_jump_<mode>): Likewise.
26021         (tablejump_<mode>): Likewise.
26022         (<optab>_internal): Likewise.
26023         (sibcall_internal): Likewise.
26024         (sibcall_value_internal): Likewise.
26025         (prefetch): Use constraint ZD.
26026         * config/mips/mips.opt (minterlink-compressed): New option.
26027         (minterlink-mips16): Now an alias for minterlink-compressed.
26028         (mmicromips): New option.
26029         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26030         (compare_and_swap_12): Likewise.
26031         (sync_add<mode>): Likewise.
26032         (sync_<optab>_12): Likewise.
26033         (sync_old_<optab>_12): Likewise.
26034         (sync_new_<optab>_12): Likewise.
26035         (sync_nand_12): Likewise.
26036         (sync_old_nand_12): Likewise.
26037         (sync_new_nand_12): Likewise.
26038         (sync_sub<mode>): Likewise.
26039         (sync_old_add<mode>): Likewise.
26040         (sync_old_sub<mode>): Likewise.
26041         (sync_new_add<mode>): Likewise.
26042         (sync_new_sub<mode>): Likewise.
26043         (sync_<optab><mode>): Likewise.
26044         (sync_old_<optab><mode>): Likewise.
26045         (sync_new_<optab><mode>): Likewise.
26046         (sync_nand<mode>): Likewise.
26047         (sync_old_nand<mode>): Likewise.
26048         (sync_new_nand<mode>): Likewise.
26049         (sync_lock_test_and_set<mode>): Likewise.
26050         (test_and_set_12): Likewise.
26051         (atomic_compare_and_swap<mode>): Likewise.
26052         (atomic_exchange<mode>_llsc): Likewise.
26053         (atomic_fetch_add<mode>_llsc): Likewise.
26054         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26055         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26056         (umips_save_restore_pattern_p): Likewise.
26057         (umips_load_store_pair_p): Likewise.
26058         (umips_output_load_store_pair): Likewise.
26059         (umips_movep_target_p): Likewise.
26060         (umips_12bit_offset_address_p): Likewise.
26061         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26062         (mips_base_mips16): Rename this...
26063         (mips_base_compression_flags): ...to this. Update all uses.
26064         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26065         (mips_mips16_decl_p): Delete.
26066         (mips_nomips16_decl_p): Delete.
26067         (mips_get_compress_on_flags): New function.
26068         (mips_get_compress_off_flags): New function.
26069         (mips_get_compress_mode): New function.
26070         (mips_get_compress_on_name): New function.
26071         (mips_get_compress_off_name): New function.
26072         (mips_insert_attributes): Support multiple compression types.
26073         (mips_merge_decl_attributes): Likewise.
26074         (umips_12bit_offset_address_p): New function.
26075         (mips_start_function_definition): Emit .set micromips directive.
26076         (mips_call_may_need_jalx_p): New function.
26077         (mips_function_ok_for_sibcall): Add microMIPS support.
26078         (mips_print_operand_punctuation): Support short delay slots and
26079         compact jumps.
26080         (umips_swm_mask, umips_swm_encoding): New.
26081         (umips_build_save_restore): New function.
26082         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26083         (was_mips16_p): Remove.
26084         (old_compression_mode): New.
26085         (mips_set_compression_mode): New function.
26086         (mips_set_current_function): Add microMIPS support.
26087         (mips_option_override): Likewise.
26088         (umips_save_restore_pattern_p): New function.
26089         (umips_output_save_restore): New function.
26090         (umips_load_store_pair_p_1): New function.
26091         (umips_load_store_pair_p): New function.
26092         (umips_output_load_store_pair_1): New function.
26093         (umips_output_load_store_pair): New function.
26094         (umips_movep_target_p) New function.
26095         (mips_prepare_pch_save): Add microMIPS support.
26096         * config/mips/mips.h (TARGET_COMPRESSION): New.
26097         (TARGET_CPU_CPP_BUILTINS): Update macro
26098         to use new compression flags and to support microMIPS.
26099         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
26100         (MIPS_ARCH_FLOAT_SPEC): Likewise.
26101         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
26102         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
26103         (ASM_SPEC): Support mmicromips and mno-micromips.
26104         (M16STORE_REG_P): New macro.
26105         (MIPS_CALL): Support TARGET_MICROMIPS.
26106         (MICROMIPS_J): New macro.
26107         (mips_base_mips16): Rename this...
26108         (mips_base_compression_flags): ...to this.
26109         (UMIPS_12BIT_OFFSET_P): New macro.
26110         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
26111         (MULTILIB_DIRNAMES): Likewise.
26113 2013-02-25  Tom de Vries  <tom@codesourcery.com>
26115         PR rtl-optimization/56131
26116         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
26117         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
26118         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
26120 2013-02-25  Tobias Burnus  <burnus@net-b.de>
26122         * doc/invoke.texi (-fsanitize=): Move from optimization
26123         to debugging options.
26125 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26127         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
26129 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26130             Alexander Monakov  <amonakov@ispras.ru>
26132         PR middle-end/56077
26133         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
26134         flush pending lists also on non-jumps.  Adjust comment.
26136 2013-02-24  Catherine Moore  <clm@codesourcery.com>
26137             Maciej W. Rozycki  <macro@codesourcery.com>
26138             Tom de Vries  <tom@codesourcery.com>
26139             Nathan Sidwell  <nathan@codesourcery.com>
26140             Iain Sandoe  <iain@codesourcery.com>
26141             Nathan Froyd  <froydnj@codesourcery.com>
26142             Chao-ying Fu  <fu@mips.com>
26144         * doc/extend.texi: (micromips, nomicromips, nocompression):
26145         Document new function attributes.
26146         * doc/invoke.texi (minterlink-compressed, mmicromips,
26147         m14k, m14ke, m14kec): Document new options.
26148         (minterlink-mips16): Update documentation.
26149         * doc/md.texi (ZC, ZD): Document new constraints.
26150         * configure.ac (gcc_cv_as_micromips): Check if linker
26151         supports the .set micromips directive.
26152         * configure: Regenerate.
26153         * config.in: Regenerate.
26154         * config/mips/mips-tables.opt: Regenerate.
26155         * config/mips/micromips.md: New file.
26156         * constraints.md (ZC, AD): New constraints.
26157         * config/mips/predicates.md (movep_src_register): New predicate.
26158         (movep_src_operand): New predicate.
26159         (non_volatile_mem_operand): New predicate.
26160         * config/mips/mips.md (multimem): New type.
26161         (length): Differentiate between 17-bit and 18-bit branch offsets.
26162         (MOVEP1, MOVEP2): New mode iterator.
26163         (mov_<load>l): Use ZC constraint.
26164         (mov_<load>r): Likewise.
26165         (mov_<store>l): Likewise.
26166         (mov_<store>r): Likewise.
26167         (*branch_equality<mode>_inverted): Add microMIPS support.
26168         (*branch_equality<mode>): Likewise.
26169         (*jump_absolute): Likewise.
26170         (indirect_jump_<mode>): Likewise.
26171         (tablejump_<mode>): Likewise.
26172         (<optab>_internal): Likewise.
26173         (sibcall_internal): Likewise.
26174         (sibcall_value_internal): Likewise.
26175         (prefetch): Use constraint ZD.
26176         * config/mips/mips.opt (minterlink-compressed): New option.
26177         (minterlink-mips16): Now an alias for minterlink-compressed.
26178         (mmicromips): New option.
26179         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26180         (compare_and_swap_12): Likewise.
26181         (sync_add<mode>): Likewise.
26182         (sync_<optab>_12): Likewise.
26183         (sync_old_<optab>_12): Likewise.
26184         (sync_new_<optab>_12): Likewise.
26185         (sync_nand_12): Likewise.
26186         (sync_old_nand_12): Likewise.
26187         (sync_new_nand_12): Likewise.
26188         (sync_sub<mode>): Likewise.
26189         (sync_old_add<mode>): Likewise.
26190         (sync_old_sub<mode>): Likewise.
26191         (sync_new_add<mode>): Likewise.
26192         (sync_new_sub<mode>): Likewise.
26193         (sync_<optab><mode>): Likewise.
26194         (sync_old_<optab><mode>): Likewise.
26195         (sync_new_<optab><mode>): Likewise.
26196         (sync_nand<mode>): Likewise.
26197         (sync_old_nand<mode>): Likewise.
26198         (sync_new_nand<mode>): Likewise.
26199         (sync_lock_test_and_set<mode>): Likewise.
26200         (test_and_set_12): Likewise.
26201         (atomic_compare_and_swap<mode>): Likewise.
26202         (atomic_exchange<mode>_llsc): Likewise.
26203         (atomic_fetch_add<mode>_llsc): Likewise.
26204         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26205         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26206         (umips_save_restore_pattern_p): Likewise.
26207         (umips_load_store_pair_p): Likewise.
26208         (umips_output_load_store_pair): Likewise.
26209         (umips_movep_target_p): Likewise.
26210         (umips_12bit_offset_address_p): Likewise.
26211         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26212         (mips_base_mips16): Rename this...
26213         (mips_base_compression_flags): ...to this. Update all uses.
26214         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26215         (mips_mips16_decl_p): Delete.
26216         (mips_nomips16_decl_p): Delete.
26217         (mips_get_compress_on_flags): New function.
26218         (mips_get_compress_off_flags): New function.
26219         (mips_get_compress_mode): New function.
26220         (mips_get_compress_on_name): New function.
26221         (mips_get_compress_off_name): New function.
26222         (mips_insert_attributes): Support multiple compression types.
26223         (mips_merge_decl_attributes): Likewise.
26224         (umips_12bit_offset_address_p): New function.
26225         (mips_start_function_definition): Emit .set micromips directive.
26226         (mips_call_may_need_jalx_p): New function.
26227         (mips_function_ok_for_sibcall): Add microMIPS support.
26228         (mips_print_operand_punctuation): Support short delay slots and
26229         compact jumps.
26230         (umips_swm_mask, umips_swm_encoding): New.
26231         (umips_build_save_restore): New function.
26232         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26233         (was_mips16_p): Remove.
26234         (old_compression_mode): New.
26235         (mips_set_compression_mode): New function.
26236         (mips_set_current_function): Add microMIPS support.
26237         (mips_option_override): Likewise.
26238         (umips_save_restore_pattern_p): New function.
26239         (umips_output_save_restore): New function.
26240         (umips_load_store_pair_p_1): New function.
26241         (umips_load_store_pair_p): New function.
26242         (umips_output_load_store_pair_1): New function.
26243         (umips_output_load_store_pair): New function.
26244         (umips_movep_target_p) New function.
26245         (mips_prepare_pch_save): Add microMIPS support.
26246         * config/mips/mips.h (TARGET_COMPRESSION): New.
26247         (TARGET_CPU_CPP_BUILTINS): Update macro
26248         to use new compression flags and to support microMIPS.
26249         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
26250         (MIPS_ARCH_FLOAT_SPEC): Likewise.
26251         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
26252         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
26253         (ASM_SPEC): Support mmicromips and mno-micromips.
26254         (M16STORE_REG_P): New macro.
26255         (MIPS_CALL): Support TARGET_MICROMIPS.
26256         (MICROMIPS_J): New macro.
26257         (mips_base_mips16): Rename this...
26258         (mips_base_compression_flags): ...to this.
26259         (UMIPS_12BIT_OFFSET_P): New macro.
26260         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
26261         (MULTILIB_DIRNAMES): Likewise.
26263 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
26265         PR target/52555
26266         * target-globals.c (save_target_globals): For init_reg_sets and
26267         target_reinit remporarily set this_fn_optabs to this_target_optabs.
26269 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
26271         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
26272         * config/aarch64/t-aarch64
26273         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
26275 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
26277         PR inline-asm/56148
26278         * lra-constraints.c (process_alt_operands): Reload operand
26279         conflicting with earlier clobber only if no more other conflicting
26280         operands.
26282 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
26284         PR sanitizer/56393
26285         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
26286         if not linking a shared library.
26288 2013-02-22  Seth LaForge  <sethml@google.com>
26290         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
26292 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
26294         * config/arm/arm.md (split for extendsidi): Update condition.
26295         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
26296         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
26297         (qhs_zextenddi_cstr): Likewise.
26299 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
26301         PR middle-end/56420
26302         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
26303         avoid signed wrapping.
26304         (expand_mult): Handle properly multiplication by
26305         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
26306         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
26307         in the compiler if coeff is HOST_WIDE_INT_MIN.
26308         (expand_divmod): Don't make ext_op1 static, change it's type to
26309         uhwi.  Avoid undefined behavior in -INTVAL (op1).
26311         PR rtl-optimization/50339
26312         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
26313         field.
26314         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
26315         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
26316         into splitting_ashiftrt field.
26317         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
26318         ASHIFTRT.
26319         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
26320         choices.
26322 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
26324         PR middle-end/56108
26325         * trans-mem.c (execute_tm_mark): Do not expand transactions that
26326         are sure to go irrevocable.
26328 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
26330         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
26331         scalars are valid operands.
26333 2013-02-21  Martin Jambor  <mjambor@suse.cz>
26335         PR tree-optimization/56310
26336         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
26337         only matching indices and non-negative final offsets.
26338         (intersect_aggregates_with_edge): Pass src_idx to
26339         agg_replacements_to_vector.  Pass src_idx insstead of index to
26340         intersect_with_agg_replacements.
26342 2013-02-21  Martin Jambor  <mjambor@suse.cz>
26344         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
26345         instead of hard-wired defaults.
26347 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
26349         * doc/invoke.texi (MIPS Options): Update documentation of the
26350         floating-point multiply-accumulate instruction restrictions.
26352 2013-02-21  Kostya Serebryany  <kcc@google.com>
26354         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
26355         asan_shadow_offset on x86_64 linux.
26357 2013-02-21  Richard Biener  <rguenther@suse.de>
26359         PR tree-optimization/56415
26360         Revert
26361         2013-02-11  Richard Biener  <rguenther@suse.de>
26363         PR tree-optimization/56273
26364         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
26365         first VRP run.
26367 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
26369         PR bootstrap/56258
26370         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
26371         instead of @itemx.
26373         PR inline-asm/56405
26374         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
26375         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
26377 2013-02-20  Jan Hubicka  <jh@suse.cz>
26379         PR tree-optimization/56265
26380         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
26381         when target is referenced for first time.
26383 2013-02-20  Richard Biener  <rguenther@suse.de>
26385         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
26386         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
26387         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
26388         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
26389         not return anything.
26390         (rename_ssa_copies): Do not remove unused locals.
26391         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
26392         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
26393         * passes.c (execute_function_todo): Do not schedule unused locals
26394         removal if cleanup_tree_cfg did something.
26395         * tree-ssa-live.c (remove_unused_locals): Dump statistics
26396         about the number of removed locals.
26398 2013-02-20  Richard Biener  <rguenther@suse.de>
26400         PR tree-optimization/56398
26401         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
26403 2013-02-20  Martin Jambor  <mjambor@suse.cz>
26405         PR tree-optimization/55334
26406         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
26407         restricted pointers to arrays.
26409 2013-02-20  Richard Biener  <rguenther@suse.de>
26410             Jakub Jelinek  <jakub@redhat.com>
26412         PR tree-optimization/56396
26413         * tree-ssa-ccp.c (n_const_val): New static variable.
26414         (get_value): Return NULL for SSA names we don't have a lattice
26415         entry for.
26416         (ccp_initialize): Initialize n_const_val.
26417         * tree-ssa-copy.c (n_copy_of): New static variable.
26418         (init_copy_prop): Initialize n_copy_of.
26419         (get_value): Return NULL_TREE for SSA names we don't have a
26420         lattice entry for.
26422 2013-02-20  Martin Jambor  <mjambor@suse.cz>
26424         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
26426 2013-02-20  Richard Biener  <rguenther@suse.de>
26428         * genpreds.c (write_lookup_constraint): Do not compare first
26429         letter of the constraint again.
26431 2013-02-20  Richard Biener  <rguenther@suse.de>
26433         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
26434         and ceil_log2.
26435         (get_use_iv_cost): Terminate hashtable walk when coming across
26436         an empty entry.
26438 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
26440         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
26441         reassociation for avx2 targets.
26443 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26445         * config/microblaze/microblaze.c: microblaze_has_clz = 0
26446         Add version check for v8.10.a to enable microblaze_has_clz
26447         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
26448         version and TARGET_PATTERN_COMPARE check
26449         * config/microblaze/microblaze.md: New clzsi2 instruction
26451 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26453         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
26454         function before branching.
26456 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
26458         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
26459         DUMP_INSN_RTX_UID.
26460         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
26462 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
26464         PR middle-end/55889
26465         * sel-sched.c: Include ira.h.
26466         (implicit_clobber_conflict_p): New function.
26467         (moveup_expr): Use it.
26468         * Makefile.in (sel-sched.o): Depend on ira.h.
26470 2013-02-19  Richard Biener  <rguenther@suse.de>
26472         PR tree-optimization/56384
26473         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
26474         (vn_hash_type): Split out from ...
26475         (vn_hash_constant_with_type): ... here.
26476         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
26477         (vn_phi_eq): Compare types from vn_phi_s structure.
26478         (vn_phi_lookup): Populate vn_phi_s type.
26479         (vn_phi_insert): Likewise.
26481 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
26483         PR tree-optimization/56350
26484         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
26485         if haven't found reduction or nested cycle operand, rather than
26486         asserting we must find it.
26488         PR tree-optimization/56381
26489         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
26490         to fold_build3.
26492 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
26493             Jakub Jelinek  <jakub@redhat.com>
26495         PR target/52555
26496         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
26497         (swap_optab_enable): Same.
26498         (init_all_optabs): Use argument instead of global.
26499         * tree.h (struct tree_optimization_option): New field target_optabs.
26500         * expr.h (init_all_optabs): Add argument to prototype.
26501         (TREE_OPTIMIZATION_OPTABS): New.
26502         (save_optabs_if_changed): Protoize.
26503         * optabs.h: Declare this_fn_optabs.
26504         * optabs.c (save_optabs_if_changed): New.
26505         Declare this_fn_optabs.
26506         (init_optabs): Add argument to init_all_optabs() call.
26507         * function.c (invoke_set_current_function_hook): Handle per
26508         function optabs.
26509         * function.h (struct function): New field optabs.
26510         * config/mips/mips.c (mips_set_mips16_mode): Handle when
26511         optimization_current_node has changed.
26512         * target-globals.h (save_target_globals_default_opts): Protoize.
26513         * target-globals.c (save_target_globals_default_opts): New.
26515 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26517         PR target/56347
26518         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
26519         registers %fr12 and %fr12R as call used.
26521         PR target/56214
26522         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
26523         and HImode, require all displacements to be an integer multiple of
26524         their mode size.
26525         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
26526         only allow QImode and HImode when reload is in progress and strict is
26527         true.  Likewise for symbolic addresses.  Use base14_operand to check
26528         displacements in REG+BASE addresses.
26530 2013-02-18  Richard Biener  <rguenther@suse.de>
26532         PR tree-optimization/56366
26533         * tree-vect-loop.c (get_initial_def_for_induction): Properly
26534         handle sign-conversion of outer-loop initial induction value.
26536 2013-02-18  Richard Biener  <rguenther@suse.de>
26538         PR middle-end/56349
26539         * cfghooks.c (merge_blocks): If we merge a latch into another
26540         block adjust references to it.
26541         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
26542         (verify_loop_structure): Verify that a recorded latch is in fact
26543         a latch.
26545 2013-02-18  Richard Biener  <rguenther@suse.de>
26547         PR tree-optimization/56321
26548         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
26549         order SSA name release and virtual operand unlinking.
26551 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26553         * config/microblaze/microblaze.md (save_stack_block): Define.
26554         (restore_stack_block): Likewise.
26556 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26558         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
26559         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
26560         * config/microblaze/microblaze.c (microblaze_option_override):
26561         Bail out early for PIC modes when target does not support PIC.
26563 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26565         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
26566         Replace with a microblaze version.
26567         (microblaze_trampoline_init): Adapt for microblaze.
26568         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
26569         microblaze.
26571 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
26572             Dodji Seketeli  <dodji@redhat.com>
26574         PR asan/56330
26575         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
26576         (instrument_mem_region_access): Do not forget to always put
26577         instrumentation of the of 'base' and 'base + len' in a "if (len !=
26578         0) statement, even for cases where either 'base' or 'base + len'
26579         are not instrumented -- because they have been previously
26580         instrumented.  Simplify the logic by putting all the statements
26581         instrument 'base + len' inside a sequence, and then insert that
26582         sequence right before the current insertion point.  Then, to
26583         instrument 'base + len', just get an iterator on that statement.
26584         And do not forget to update the pointer to iterator the function
26585         received as argument.
26587 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
26589         PR rtl-optimization/56348
26590         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
26592 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
26594         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
26595         (clean_graph_dump_file): Pass base to start_graph_dump.
26597 2013-02-14  Richard Henderson  <rth@redhat.com>
26599         PR target/55941
26600         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
26602 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
26604         * collect2-aix.h: Define F_LOADONLY.
26606 2013-02-14  Richard Biener  <rguenther@suse.de>
26608         PR lto/50494
26609         * varasm.c (output_constant_def_1): Get the decl representing
26610         the constant as argument.
26611         (output_constant_def): Wrap output_constant_def_1.
26612         (make_decl_rtl): Use output_constant_def_1 with the decl
26613         representing the constant.
26614         (build_constant_desc): Optionally re-use a decl already
26615         representing the constant.
26616         (tree_output_constant_def): Adjust.
26618 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
26620         Fix an asan crash
26621         * asan.c (instrument_builtin_call):  Really put the length of the
26622         second source argument into src1_len.
26624 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
26626         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
26627         argument.  If it is false, don't create edge from then_bb to
26628         fallthru_bb.
26629         (insert_if_then_before_iter): Pass true to it.
26630         (build_check_stmt): Pass false to it.
26631         (transform_statements): Flush hash table only on extended basic
26632         block boundaries, rather than at the beginning of every bb.
26633         Don't flush hash table on nonfreeing_call_p calls.
26634         * tree-flow.h (nonfreeing_call_p): New prototype.
26635         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
26637 2013-02-13  David S. Miller  <davem@davemloft.net>
26639         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
26641 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
26643         PR target/56184
26644         * ira.c (max_regno_before_ira): Move from ...
26645         (ira): ... here.
26646         (fix_reg_equiv_init): Use max_regno_before_ira instead of
26647         vec_safe_length.
26649 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
26651         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
26653 2013-02-13  Richard Biener  <rguenther@suse.de>
26655         PR lto/56295
26656         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
26657         globals in MEM_REFs.
26659 2013-02-13  Richard Biener  <rguenther@suse.de>
26661         * loop-init.c (loop_optimizer_init): Clear loop state when
26662         re-initializing preserved loops.
26663         * loop-unswitch.c (unswitch_single_loop): Return whether
26664         we unswitched the loop.  Do not verify loop state here.
26665         (unswitch_loops): When we unswitched a loop discover new loops.
26667 2013-02-13  Kostya Serebryany  <kcc@google.com>
26669         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
26670         on x86_64 linux.
26671         * sanitizer.def: Rename __asan_init to __asan_init_v1.
26673 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
26675         Avoid instrumenting duplicated memory access in the same basic block
26676         * Makefile.in (asan.o): Add new dependency on hash-table.h
26677         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
26678         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
26679         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
26680         (free_mem_ref_resources, has_mem_ref_been_instrumented)
26681         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
26682         (get_mem_ref_of_assignment): New functions.
26683         (get_mem_refs_of_builtin_call): Extract from
26684         instrument_builtin_call and tweak a little bit to make it fit with
26685         the new signature.
26686         (instrument_builtin_call): Use the new
26687         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
26688         of is_gimple_builtin_call.
26689         (instrument_derefs, instrument_mem_region_access): Insert the
26690         instrumented memory reference into the hash table.
26691         (maybe_instrument_assignment): Renamed instrument_assignment into
26692         this, and change it to advance the iterator when instrumentation
26693         actually happened and return true in that case.  This makes it
26694         homogeneous with maybe_instrument_assignment, and thus give a
26695         chance to callers to be more 'regular'.
26696         (transform_statements): Clear the memory reference hash table
26697         whenever we enter a new BB, when we cross a function call, or when
26698         we are done transforming statements.  Use
26699         maybe_instrument_assignment instead of instrumentation.  No more
26700         need to special case maybe_instrument_assignment and advance the
26701         iterator after calling it; it's now handled just like
26702         maybe_instrument_call.  Update comment.
26704 2013-02-13  Richard Biener  <rguenther@suse.de>
26706         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
26707         Fix loop discovery code.
26709 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
26711         PR inline-asm/56148
26712         * lra-constraints.c (process_alt_operands): Match early clobber
26713         operand with itself.  Check conflicts with earlyclobber only if
26714         the operand is not reloaded.  Prefer to reload conflicting operand
26715         if earlyclobber and matching operands are the same.
26717 2013-02-12  Richard Biener  <rguenther@suse.de>
26719         PR lto/56297
26720         * lto-streamer-out.c (write_symbol): Do not output symbols
26721         for hard register variables.
26723 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
26725         PR target/54222
26726         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
26727         (umulsidi3_insn, mulsidi3_insn): New insns.
26729 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
26731         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
26732         (struct tune_params): Add vec_costs field.
26733         * config/arm/arm.c (arm_builtin_vectorization_cost)
26734         (arm_add_stmt_cost): New functions.
26735         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
26736         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
26737         (arm_default_vec_cost): New struct of type cpu_vec_costs.
26738         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
26739         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
26740         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
26741         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
26743 2013-02-12  Richard Biener  <rguenther@suse.de>
26745         PR lto/56295
26746         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
26747         decls again if possible.
26749 2013-02-12  Richard Biener  <rguenther@suse.de>
26751         PR middle-end/56288
26752         * tree-ssa.c (verify_ssa_name): Fix check, move
26753         SSA_NAME_IN_FREE_LIST check up.
26755 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
26756             Steven Bosscher   <steven@gcc.gnu.org>
26758         PR rtl-optimization/56151
26759         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
26760         equal to op0 or op1, and last_insn pattern is CODE operation
26761         with MEM dest and one of the operands matches that MEM.
26763 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
26765         * doc/extend.texi: Document Function Multiversioning and "default"
26766         parameter string to target attribute.
26767         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
26768         target attribute parameter is "default".
26769         (ix86_compare_version_priority): Remove checks for target attribute.
26770         (ix86_mangle_function_version_assembler_name): Change error to sorry.
26771         Remove check for target attribute equal to NULL. Add assert.
26772         (ix86_generate_version_dispatcher_body): Change error to sorry.
26774 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
26775             Jack Howarth  <howarth@bromo.med.uc.edu>
26776             Patrick Marlier  <patrick.marlier@gmail.com>
26778         PR libitm/55693
26779         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
26780         define ENDFILE_SPEC as TM_DESTRUCTOR.
26781         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
26783 2013-02-11  Alexander Potapenko  <glider@google.com>
26784             Jack Howarth  <howarth@bromo.med.uc.edu>
26785             Jakub Jelinek  <jakub@redhat.com>
26787         PR sanitizer/55617
26788         * config/darwin.c (cdtor_record): Rename ctor_record.
26789         (sort_cdtor_records): Rename sort_ctor_records.
26790         (finalize_dtors): New routine to sort destructors by
26791         priority before use in assemble_integer.
26792         (machopic_asm_out_destructor): Use finalize_dtors if needed.
26794 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
26796         PR rtl-optimization/56275
26797         * simplify-rtx.c (avoid_constant_pool_reference): Check that
26798         offset is non-negative and less than cmode size before
26799         calling simplify_subreg.
26801 2013-02-11  Richard Biener  <rguenther@suse.de>
26803         PR tree-optimization/56264
26804         * cfgloop.h (fix_loop_structure): Adjust prototype.
26805         * loop-init.c (fix_loop_structure): Return the number of
26806         newly discovered loops.
26807         * tree-cfgcleanup.c (repair_loop_structures): When new loops
26808         are discovered, do a full loop-closed SSA rewrite.
26810 2013-02-11  Richard Biener  <rguenther@suse.de>
26812         PR tree-optimization/56273
26813         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
26814         first VRP run.
26815         (check_array_ref): Fix missing newline in dumps.
26816         (search_for_addr_array): Likewise.
26818 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
26820         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
26822 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
26824         PR target/56256
26825         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
26827 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
26829         PR rtl-optimization/56246
26830         * lra-constraints.c (simplify_operand_subreg): Try to reuse
26831         reload pseudo.
26832         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
26833         constraints are satisfied.
26835 2013-02-08  Jeff Law  <law@redhat.com>
26837         PR debug/53948
26838         * emit-rtl.c (reg_is_parm_p): New function.
26839         * regs.h (reg_is_parm_p): New prototype.
26840         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
26841         callee-clobbered registers.
26843 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
26845         PR target/56043
26846         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
26847         If there is no implicit builtin declaration, just return NULL.
26849 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
26851         * config/i386/sse.md (FMAMODEM): New mode iterator.
26852         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
26853         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
26855 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
26857         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
26858         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
26859         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
26861 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26863         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
26864         (microblaze*-*-elf): Likewise.
26865         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
26866         LINK_SPEC.
26867         * config/microblaze/microblaze-c.c: Add builtin defines for
26868         _LITTLE_ENDIAN and _BIG_ENDIAN.
26869         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
26870         add to TARGET_DEFAULT flags.
26871         Expand ASM_SPEC and LINK_SPEC.
26872         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
26873         * config/microblaze/microblaze.md: Update extendsidi2 and
26874         movdi_internal instructions to use low-order / high-order reg
26875         print_operands.
26876         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
26877         options and inversemask / mask of LITTLE_ENDIAN.
26878         * config/microblaze/t-microblaze: Expand multilib options to
26879         include mlittle-endian (le) and update exceptions patterns.
26881 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
26883         PR rtl-optimization/56195
26884         * lra-constraints.c (get_reload_reg): Don't reuse regs
26885         if they have smaller mode than requested, if they have
26886         wider mode than requested, try to return a SUBREG.
26888         PR tree-optimization/56250
26889         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
26890         if type is unsigned and code isn't MULT_EXPR.
26892 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
26894         PR tree-optimization/56064
26895         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
26896         bits according to mode.
26897         * fixed-value.h (fixed_from_double_int)
26898         (const_fixed_from_double_int): Adjust comments.
26900 2013-02-08  Richard Biener  <rguenther@suse.de>
26902         PR lto/56231
26903         * lto-streamer.h (struct data_in): Remove current_file, current_line
26904         and current_col members.
26905         * lto-streamer-out.c (lto_output_location): Stream changed bits
26906         en-block for efficiency.
26907         * lto-streamer-in.c (clear_line_info): Remove.
26908         (lto_input_location): Cache current file, line and column
26909         globally via local statics.  Read changed bits en-block.
26910         (input_function): Do not call clear_line_info.
26911         (lto_read_body): Likewise.
26912         (lto_input_toplevel_asms): Likewise.
26914 2013-02-08  Michael Matz  <matz@suse.de>
26916         PR tree-optimization/52448
26917         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
26918         (nt_call_phase): New static.
26919         (add_or_mark_expr): Only mark accesses with newer phase than any
26920         call seen.
26921         (nonfreeing_call_p): New.
26922         (nt_init_block): Update nt_call_phase, mark blocks as visited.
26923         (nt_fini_block): Keep blocks marked as visited.
26924         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
26926 2013-02-08  Richard Biener  <rguenther@suse.de>
26928         * ira.c (ira): Free broken dominator information.
26930 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
26932         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
26934 2013-02-08  Marek Polacek  <polacek@redhat.com>
26936         * cfgloop.c (verify_loop_structure): Add more checking of headers.
26938 2013-02-08  Richard Biener  <rguenther@suse.de>
26940         PR middle-end/56181
26941         * cfgloop.h (flow_loops_find): Adjust.
26942         (bb_loop_header_p): Declare.
26943         * cfgloop.c (bb_loop_header_p): New function split out from ...
26944         (flow_loops_find): ... here.  Adjust function signature,
26945         support incremental loop structure update.
26946         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
26947         * cfgloopmanip.c (fix_loop_structure): Move ...
26948         * loop-init.c (fix_loop_structure): ... here.
26949         (apply_loop_flags): Split out from ...
26950         (loop_optimizer_init): ... here.
26951         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
26952         in incremental mode, only remove dead loops here.
26954 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
26956         PR target/54222
26957         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
26958         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
26959         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
26960         (*round<mode>3.libgcc): New insns for fixed-modes.
26961         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
26962         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
26963         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
26964         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
26965         implementations.  Define to __builtin_avr_absFX,
26966         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
26967         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
26968         __builtin_avr_countlsFX, respectively.
26969         * config/avr/avr-c.c (target.h): Include it.
26970         (enum avr_builtin_id): New enum.
26971         (avr_resolve_overloaded_builtin): New static function.
26972         (avr_register_target_pragmas): Use it to set
26973         targetm.resolve_overloaded_builtin.
26974         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
26975         tree nodes used by DEF_BUILTIN.
26976         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
26977         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
26978         <AVR_BUILTIN_xxBITS>: Same.
26980 2013-02-08  Richard Biener  <rguenther@suse.de>
26982         * cfgloop.c (verify_loop_structure): Properly handle
26983         a loop exiting to another loop header.
26984         * ira-int.h (ira_loops): Remove.
26985         * ira.c (ira_loops): Remove.
26986         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
26987         (do_reload): Use loop_optimizer_finalize.
26988         * ira-build.c (create_loop_tree_nodes): Use get_loops and
26989         number_of_loops to access the loop tree.
26990         (more_one_region_p): Likewise.
26991         (finish_loop_tree_nodes): Likewise.
26992         (rebuild_regno_allocno_maps): Likewise.
26993         (mark_loops_for_removal): Likewise.
26994         (mark_all_loops_for_removal): Likewise.
26995         (remove_unnecessary_regions): Likewise.
26996         (ira_build): Likewise.
26997         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
26999 2013-02-08  Richard Biener  <rguenther@suse.de>
27001         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
27002         * ipa-pure-const.c (analyze_function): Avoid calling
27003         mark_irreducible_loops twice.
27004         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
27006 2013-02-07  David S. Miller  <davem@davemloft.net>
27008         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
27009         on 'reg'.
27010         * var-tracking.c (vt_add_function_parameter): Test the presence of
27011         HAVE_window_save properly and do not remap argument registers when
27012         we have a leaf function.
27014 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
27016         PR bootstrap/56227
27017         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
27018         instead of "ll".
27019         * config/i386/i386.c (ix86_print_operand): Ditto.
27021 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27023         * lra-constraints.c (process_alt_operands): Fix recently added comment.
27025 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27027         PR rtl-optimization/56225
27028         * lra-constraints.c (process_alt_operands): Check that reload hard
27029         reg can hold value for strict_low_part.
27031 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27033         PR debug/56154
27034         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
27035         dwarf2out_end_function.
27036         (in_first_function_p, maybe_at_text_label_p,
27037         first_loclabel_num_not_at_text_label): New variables.
27038         (dwarf2out_var_location): In the first function find out
27039         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
27040         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
27041         functions.
27043 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
27045         PR rtl-optimization/56178
27046         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
27047         SUBREG of a register.  Tidy up related block of code.
27048         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
27049         note if the source is a register or a SUBREG of a register.
27051 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27053         PR target/56228
27054         * config/rs6000/rs6000.md (ptrm): New mode attr.
27055         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
27056         call_value_indirect_aix<pttrsize>,
27057         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
27058         m in constraints.
27060 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
27062         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
27063         if -bnortl. Convert to strcmp and strncmp.
27065 2013-02-07  Alan Modra  <amodra@gmail.com>
27067         PR target/54009
27068         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
27069         addresses won't wrap when offsetting.
27070         (rs6000_secondary_reload): Provide secondary reloads needed for
27071         wrapping LO_SUM addresses.
27073 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
27075         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
27076         MACH, just __MACH__.
27078 2013-02-06  Richard Biener  <rguenther@suse.de>
27080         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
27081         instead of calling fix_loop_structure.
27083 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
27085         PR middle-end/56217
27086         * omp-low.c (use_pointer_for_field): Return false if
27087         lower_send_shared_vars doesn't generate any copy-out code.
27089 2013-02-06  Tom de Vries  <tom@codesourcery.com>
27091         PR rtl-optimization/56131
27092         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
27093         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
27094         of the label is NULL.  Add comment.
27096 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
27098         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
27100         PR sanitizer/55374
27101         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
27102         (STATIC_LIBTSAN_LIBS): Likewise.
27103         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
27104         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
27105         is defined, don't add anything else beyond that.
27106         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
27107         (LINK_COMMAND_SPEC): Use them.
27109         PR tree-optimization/56205
27110         * tree-stdarg.c (check_all_va_list_escapes): Return true if
27111         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
27112         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
27114 2013-02-05  Richard Biener  <rguenther@suse.de>
27116         PR tree-optimization/53342
27117         PR tree-optimization/53185
27118         * tree-vectorizer.h (vect_check_strided_load): Remove.
27119         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
27120         not disallow peeling for vectorized strided loads.
27121         (vect_check_strided_load): Make static and simplify.
27122         (vect_analyze_data_refs): Adjust.
27123         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
27124         correctly when vectorizing strided loads.
27126 2013-02-05  Richard Biener  <rguenther@suse.de>
27128         * doc/install.texi: Refer to ISL, not PPL.
27130 2013-02-05  Jan Hubicka  <jh@suse.cz>
27132         PR tree-optimization/55789
27133         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
27135 2013-02-05  Jan Hubicka  <jh@suse.cz>
27137         PR tree-optimization/55789
27138         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
27139         the dead call anyway.
27141 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
27143         PR sanitizer/55374
27144         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
27146 2013-02-04  Alexander Potapenko  <glider@google.com>
27147             Jack Howarth  <howarth@bromo.med.uc.edu>
27148             Jakub Jelinek  <jakub@redhat.com>
27150         PR sanitizer/55617
27151         * config/darwin.c (sort_ctor_records): Stabilized qsort
27152         on constructor priority by using original position.
27153         (finalize_ctors): New routine to sort constructors by
27154         priority before use in assemble_integer.
27155         (machopic_asm_out_constructor): Use finalize_ctors if needed.
27157 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
27159         PR libstdc++/54314
27160         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
27161         about visibility on artificial decls.
27162         * config/sol2.c (solaris_assemble_visibility): Likewise.
27164 2013-02-04  Kai Tietz  <ktietz@redhat.com>
27166         PR target/56186
27167         * config/i386/i386.c (function_value_ms_64): Add additional valtype
27168         argument and improve checking of return-argument types for 16-byte
27169         modes.
27170         (ix86_function_value_1): Add additional valtype argument on call
27171         of function_value_64.
27172         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
27173         handling infunction_value_64 function.
27175 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
27177         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
27179 2013-02-04  Richard Biener  <rguenther@suse.de>
27181         PR tree-optimization/56188
27182         * tree-ssa-structalias.c (label_visit): Consider case with
27183         initially non-empty points-to set.
27184         (perform_var_substitution): Dump node mapping and clean up.
27186 2013-02-04  Richard Guenther  <rguenther@suse.de>
27188         PR lto/56168
27189         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
27190         node prevail as last resort.
27191         (lto_symtab_merge_decls): Remove guard on LTRANS here.
27192         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
27194 2013-02-04  Richard Biener  <rguenther@suse.de>
27196         PR tree-optimization/56113
27197         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
27198         Merge into ...
27199         (equiv_class_lookup_or_add): ... this.
27200         (label_visit): Adjust and fix error in previous patch.
27201         (perform_var_substitution): Adjust.
27203 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
27205         * config/sh/divtab.c: Fix formatting and comments throughout the file.
27206         * config/sh/sh4-300.md: Likewise.
27207         * config/sh/sh4a.md: Likewise.
27208         * config/sh/constraints.md: Likewise.
27209         * config/sh/sh.md: Likewise.
27210         * config/sh/netbsd-elf.h: Likewise.
27211         * config/sh/predicates.md: Likewise.
27212         * config/sh/sh-protos.h: Likewise.
27213         * config/sh/ushmedia.h: Likewise.
27214         * config/sh/linux.h: Likewise.
27215         * config/sh/sh.c: Likewise.
27216         * config/sh/superh.h: Likewise.
27217         * config/sh/elf.h: Likewise.
27218         * config/sh/sh4.md: Likewise.
27219         * config/sh/sh.h: Likewise.
27221 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27223         * config/pa/constraints.md: Adjust unused letters.  Change "T"
27224         constraint to match_test floating_point_store_memory_operand().
27225         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
27226         (base14_operand): New.
27227         (floating_point_store_memory_operand): New.
27228         (integer_store_memory_operand): Revise to use base14_operand and
27229         reg_plus_base_memory_operand.
27230         (move_dest_operand): Allow symbolic_memory_operands.
27231         (symbolic_memory_operand): Check for LO_SOM.
27232         (symbolic_operand): Change default case to break.
27233         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
27234         CONST_DOUBLE values to be reloaded by putting them into memory when
27235         the destination is a floating point register.
27236         (movdf): Remove code to handle CONST_DOUBLE.
27237         (movsf): Likewise.
27238         (reload_indf_r1): New.
27239         (reload_insf_r1): New.
27240         Consistently use "Q" and "T" constraints with integer and floating
27241         point move instructions, respectively.
27242         (movdi): Remove FAIL.
27243         Change predicate for source operand unamed DImode move from
27244         general_operand to move_src_operand.
27245         (umulsidi3): Change predicate for destination operand to
27246         register_operand.
27247         Likewise for similar unamed patterns.
27248         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
27249         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
27250         (hppa_legitimize_address): Simplify mask calculation.
27251         (pa_emit_move_sequence): Revised handling of secondary reloads from
27252         REG+D addresses for floating point loads and stores.  Directly handle
27253         loading CONST0_RTX (mode) to a floating point register.
27254         (pa_secondary_reload): Handle reloading DF and SFmode constant values
27255         to floating point registers.  Don't restrict secondary reloads to
27256         floating point registers to integer modes.  Revise some comments and
27257         cleanup some code.
27258         (TARGET_LEGITIMATE_ADDRESS_P): Define.
27259         (pa_legitimate_address_p): New.
27260         (pa_legitimize_reload_address): New.
27261         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
27262         (STRICT_REG_OK_FOR_BASE_P): New.
27263         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
27264         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
27266 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
27267             Andrew Dixie  <andrewd@gentrack.com>
27269         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
27270         flag set.
27272 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
27274         * expmed.c (extract_bit_field_1): Pass the full width of the
27275         structure to get_best_reg_extraction_insn.
27277 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
27279         PR target/54601
27280         * configure.ac (use_cxa_atexit): Add AIX.
27281         * configure: Regenerate.
27283         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
27285 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
27287         PR debug/54793
27288         * final.c (need_profile_function): New variable.
27289         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
27290         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
27291         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
27292         notes, targetm.asm_out.function_prologue doesn't emit anything,
27293         HAVE_prologue and profiler should be emitted before prologue,
27294         set need_profile_function instead of emitting it.
27295         (final_scan_insn): If need_profile_function, emit
27296         profile_function on the first NOTE_INSN_BASIC_BLOCK or
27297         NOTE_INSN_FUNCTION_BEG note.
27299 2013-02-01  Richard Henderson  <rth@redhat.com>
27301         * config/rs6000/rs6000.md (smulditi3): New.
27302         (umulditi3): New.
27304         * config/alpha/alpha.md (umulditi3): New.
27306 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
27308         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
27309         (ASM_OUTPUT_ALIGNED_LOCAL): New.
27311 2013-02-01  Richard Biener  <rguenther@suse.de>
27313         PR tree-optimization/56113
27314         * tree-ssa-structalias.c (label_visit): Reduce work for
27315         single-predecessor nodes.
27317 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
27319         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
27320         range isn't testing for zero.
27322 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
27324         PR middle-end/56113
27325         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
27327 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
27328             Nick Clifton  <nickc@redhat.com>
27330         * config/v850/constraints.md (Q): Define as a memory constraint.
27331         * config/v850/predicates.md (label_ref_operand): New predicate.
27332         (e3v5_shift_operand): New predicate.
27333         (ior_operator): New predicate.
27334         * config/v850/t-v850: Add e3v5 multilib.
27335         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
27336         (v850_gen_movdi): Prototype.
27337         * config/v850/v850.c: Add support for e3v5 architecture.
27338         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
27339         TARGET_V850E_UP.
27340         (construct_save_jarl): Add e3v5 long JARL support.
27341         (v850_adjust_insn_length): New function.  Adjust length of call
27342         insns when using e3v5 instructions.
27343         (v850_gen_movdi): New function: Generate instructions to move a
27344         DImode value.
27345         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
27346         (CPP_SPEC): Define __v850e3v5__ as appropriate.
27347         (TARGET_USE_FPU): Enable for e3v5.
27348         (CONST_OK_FOR_W): New macro.
27349         (ADJUST_INSN_LENGTH): Define.
27350         * config/v850/v850.md (UNSPEC_LOOP): Define.
27351         (attr cpu): Add v850e3v5.
27352         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
27353         (movdi): New pattern.
27354         (movdi_internal): New pattern.
27355         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
27356         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
27357         (cstoresf4): Likewise.
27358         (cstoredf4): Likewise.
27359         (insv): New pattern.
27360         (rotlso3_a): New pattern.
27361         (rotlsi3_b): New pattern
27362         (rotlsi3_v850e3v5): New pattern.
27363         (doloop_begin): New pattern.
27364         (fix_loop_counter): New pattern.
27365         (doloop_end): New pattern.
27366         (branch_normal): Add e3v5 long branch support.
27367         (branch_invert): Likewise.
27368         (branch_z_normal): Likewise.
27369         (branch_z_invert): Likewise.
27370         (branch_nz_normal): Likewise.
27371         (branch_nz_invert): Likewise.
27372         (call_internal_short): Add e3v5 register-indirect JARL support.
27373         (call_internal_long): Likewise.
27374         (call_value_internal_short): Likewise.
27375         (call_value_internal_long): Likewise.
27376         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
27377         (mloop): New option.
27378         * config.gcc: Add support for configuring v840e3v5 target.
27379         * doc/invoke.texi: Document new v850 specific command line options.
27381 2013-01-31  Paul Koning  <ni1d@arrl.net>
27383         PR debug/55059
27384         PR debug/54508
27385         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
27386         children if parent is a class.
27387         (prune_unused_types_prune): Don't add DW_AT_declaration.
27389 2013-01-31  Richard Biener  <rguenther@suse.de>
27391         PR tree-optimization/56157
27392         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
27393         match up operand with SLP child.
27395 2013-01-31  Jason Merrill  <jason@redhat.com>
27397         PR debug/54410
27398         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
27399         parameters the first time.
27400         (gen_scheduled_generic_parms_dies): Check completeness here.
27402 2013-01-31  Richard Biener  <rguenther@suse.de>
27404         PR middle-end/53073
27405         * common.opt (faggressive-loop-optimizations): New flag,
27406         enabled by default.
27407         * doc/invoke.texi (faggressive-loop-optimizations): Document.
27408         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
27409         infer_loop_bounds_from_undefined by it.
27411 2013-01-31  Richard Biener  <rguenther@suse.de>
27413         PR tree-optimization/56150
27414         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
27415         visit virtual operands.
27416         (find_uses_to_rename_bb): Likewise.
27418 2013-01-31  Richard Biener  <rguenther@suse.de>
27420         PR tree-optimization/56150
27421         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
27422         mixed store non-store stmts.
27424 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
27426         PR sanitizer/55374
27427         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
27428         LIBASAN_EARLY_SPEC is defined.
27429         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
27430         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
27431         before %o.
27432         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
27434         PR c++/55742
27435         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
27436         invalid args instead of ICEing on it.
27437         (ix86_valid_target_attribute_tree): Return error_mark_node if
27438         ix86_valid_target_attribute_inner_p failed.
27439         (ix86_valid_target_attribute_p): Return false only if
27440         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
27441         target("default") attribute.
27442         (sorted_attr_string): Change argument from const char * to tree,
27443         merge in all target attribute arguments rather than just one.
27444         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
27445         instead of free.  Avoid using strcat.
27446         (ix86_mangle_function_version_assembler_name): Mangle
27447         target("default") as if no target attribute is present.  Adjust
27448         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
27449         instead of xmalloc and XDELETEVEC instead of free.
27450         (ix86_function_versions): Don't return true if one of the decls
27451         doesn't have target attribute.  If they don't and one of the decls
27452         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
27453         sorted_attr_string caller.  Use XDELETEVEC instead of free.
27454         (ix86_supports_function_versions): Remove.
27455         (make_name): Fix up formatting.
27456         (make_dispatcher_decl): Remove resolver_name and its initialization.
27457         Avoid leaking memory.
27458         (is_function_default_version): Return true if there is
27459         target("default") attribute rather than no target attribute at all.
27460         (make_resolver_func): Avoid leaking memory.
27461         (ix86_generate_version_dispatcher_body): Likewise.
27462         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
27463         * target.def (supports_function_versions): Remove.
27464         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
27465         * doc/tm.texi: Regenerated.
27467 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
27469         PR rtl-optimization/56144
27470         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
27471         for values with side effects.
27473 2013-01-30  Richard Biener  <rguenther@suse.de>
27475         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
27476         (sparseset_pop): Likewise.
27477         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
27478         to be able to use quick_push in the worker loop.
27480 2013-01-30  Marek Polacek  <polacek@redhat.com>
27482         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
27484 2013-01-30  Richard Biener  <rguenther@suse.de>
27486         PR lto/56147
27487         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
27489 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
27491         PR tree-optimization/56064
27492         * fixed-value.c (fixed_from_double_int): New function.
27493         * fixed-value.h (fixed_from_double_int): New prototype.
27494         (const_fixed_from_double_int): New static inline function.
27495         * fold-const.c (native_interpret_fixed): New static function.
27496         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
27497         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
27498         (native_encode_fixed): New static function.
27499         (native_encode_expr) <FIXED_CST>: Use it.
27500         (native_interpret_int): Move double_int worker code to...
27501         * double-int.c (double_int::from_buffer): ...this new static method.
27502         * double-int.h (double_int::from_buffer): Prototype it.
27504 2013-01-30  Richard Biener  <rguenther@suse.de>
27506         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
27507         New pointer-map and obstack.
27508         (init_alias_vars): Allocate pointer-map and obstack.
27509         (delete_points_to_sets): Free them.
27510         (find_what_var_points_to): Cache result.
27511         (find_what_p_points_to): Adjust for changed interface of
27512         find_what_var_points_to.
27513         (compute_points_to_sets): Likewise.
27514         (ipa_pta_execute): Likewise.
27516 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
27518         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
27519         * configure: Regenerate.
27520         * config.in: Regenerate.
27521         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
27522         #nobits/#progbits if supported.
27524 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
27526         PR target/56121
27527         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
27528         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
27529         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
27531 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
27533         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
27534         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
27536 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
27538         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
27539         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
27541 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
27543         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
27544         declaration.
27545         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
27546         * config/arm/cortex-a7.md: New bypasses using
27547         arm_mac_accumulator_is_result.
27549 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
27551         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
27552         (cortex_a7_neon_mla): Likewise.
27553         (cortex_a7_fpfmad): New reservation.
27554         (cortex_a7_fpmacs): Use ffmas and update required units.
27555         (cortex_a7_fpmuld): Update required units and latency.
27556         (cortex_a7_fpmacd): Likewise.
27557         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
27558         (cortex_a7_neon). Likewise.
27559         (bypass) Update participating units.
27561 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
27563         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
27564         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
27565         from fmac to ffma.
27566         * config/arm/vfp11.md (vfp_farith): Use ffmas.
27567         (vfp_fmul): Use ffmad.
27568         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
27569         (cortex_r4_fmacd): Use ffmad.
27570         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
27571         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
27572         (cortex_a9_fmacd): Use ffmad.
27573         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
27574         (cortex_a8_vfp_macd): Use ffmad.
27575         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
27576         (cortex_a5_fpmacd): Use ffmad.
27577         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
27578         (cortex_a15_vfp_macd): Use ffmad.
27579         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
27581 2013-01-29  Jason Merrill  <jason@redhat.com>
27583         PR libstdc++/54314
27584         * varasm.c (default_assemble_visibility): Don't warn about
27585         visibility on artificial decls.
27587 2013-01-29  Richard Biener  <rguenther@suse.de>
27589         PR tree-optimization/56113
27590         * tree-ssa-structalias.c (equiv_class_lookup): Also return
27591         the bitmap leader.
27592         (label_visit): Free duplicate bitmaps and record the leader instead.
27593         (perform_var_substitution): Adjust.
27595 2013-01-29  Richard Biener  <rguenther@suse.de>
27597         PR tree-optimization/55270
27598         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
27599         the CFG, schedule loops for fixup.
27601 2013-01-29  Nick Clifton  <nickc@redhat.com>
27603         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
27604         SP_REG.
27606 2013-01-28  Leif Ekblad  <leif@rdos.net>
27608         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
27609         * config/i386/i386.h (TARGET_RDOS): New macro.
27610         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
27611         * config/i386/i386.c (ix86_option_override_internal): For 64bit
27612         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
27613         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
27614         DEFAULT_LARGE_SECTION_THRESHOLD.
27615         * config/i386/i386.md (R14_REG, R15_REG): New constants.
27616         * config/i386/rdos.h: New file.
27617         * config/i386/rdos64.h: New file.
27619 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
27621         PR other/54814
27622         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
27623         TEST_HARD_REG_BIT.
27625 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
27627         PR rtl-optimization/56117
27628         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
27629         call cselib_lookup_from_insn on the MEM before calling
27630         add_insn_mem_dependence.
27632 2013-01-28  Richard Biener  <rguenther@suse.de>
27634         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
27635         to a stmt that didn't have one.
27636         (copy_phis_for_bb): Likewise for PHI arguments.
27637         (copy_debug_stmt): Likewise for debug stmts.
27639 2013-01-28  Richard Biener  <rguenther@suse.de>
27641         PR tree-optimization/56034
27642         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
27643         (partition_builtin_p): Adjust.
27644         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
27645         it is the last partition.
27646         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
27647         up the vertex for the definition.
27648         (classify_partition): Classify whether a partition is a
27649         PKIND_REDUCTION, thus has uses outside of the loop.
27650         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
27651         Merge all PKIND_REDUCTION partitions into the last partition.
27652         (tree_loop_distribution): Seed partitions from reductions as well.
27654 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
27656         PR tree-optimization/56125
27657         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
27658         pow(x,c) into sqrt(x) * powi(x, n/2) or
27659         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
27660         optimizing for size.
27661         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
27662         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
27663         integer.
27665         PR tree-optimization/56094
27666         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
27667         to UNKNOWN_LOCATION while gimplifying expr.
27669 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
27671         PR target/56114
27672         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
27673         operand 0 in movabs insn template for -masm=intel asm alternative.
27674         (*movabs<mode>_2): Ditto for operand 1.
27676 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
27678         PR target/54663
27679         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
27680         of microblaze-c.o
27682 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27684         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
27685         tm_file.
27687 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
27689         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
27690         Undef to avoid warning.
27692 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
27694         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
27695         * configure: Regenerate.
27697 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
27699         PR tree-optimization/56098
27700         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
27701         for stmts with volatile ops.
27702         (cond_store_replacement): Don't optimize if assign has volatile ops.
27703         (cond_if_else_store_replacement_1): Don't optimize if either
27704         then_assign or else_assign have volatile ops.
27705         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
27706         volatile ops.
27708 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
27710         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
27712 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
27714         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
27715         missing ':' in asm example.
27717 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
27719         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
27720         entries into lane and laneq entries.
27721         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
27722         Remove AdvSIMD scalar modes.
27723         (aarch64_sq<r>dmulh_laneq<mode>): New.
27724         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
27725         modes.
27726         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
27727         builtin implementations to relfect changes in RTL in aarch64-simd.md.
27728         * config/aarch64/iterators.md (VCOND): New.
27729         (VCONQ): New.
27731 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
27733         PR target/54222
27734         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
27735         Add NULL LIBNAME argument to existing definitions.
27736         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
27737         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
27738         * config/avr/avr.c (DEF_BUILTIN): Same.
27739         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
27740         (avr_expand_builtin): Expand to a vanilla call if a libgcc
27741         implementation is available (DECL_ASSEMBLER_NAME is set).
27742         (avr_fold_absfx): New static function.
27743         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
27744         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
27745         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
27746         AVR_BUILTIN_ABSLLK.
27747         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
27748         (abshk, absk, abslk, absllk): Provide as static inline functions.
27750 2013-01-25  Marek Polacek  <polacek@redhat.com>
27752         PR tree-optimization/56035
27753         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
27755 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
27757         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
27758         (*movtf_internal_rex64): Add (!o,C) alternative
27759         (*movxf_internal_rex64): Ditto.
27760         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
27762 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
27764         * doc/invoke.texi: fix typo.
27765         * doc/objc.texi: fix typo.
27767 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
27769         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
27770         for the first two alternatives.
27772 2013-01-24  Diego Novillo  <dnovillo@google.com>
27774         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
27775         (ggc-zone.o): Remove.
27776         * configure.ac: Remove option --with-gc.
27777         * configure: Re-generate.
27778         * doc/install.texi: Remove documentation for --with-gc.
27779         * gengtype.c (write_enum_defn): Remove.  Update all users.
27780         (write_Types_process_field): Remove generation of gt_e_* argument.
27781         (output_type_enum): Remove.  Update all users.
27782         (write_enum_defn): Remove.  Update all users.
27783         (enum alloc_zone): Remove.  Update all users.
27784         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
27785         * ggc-common.c (ggc_splay_alloc): Remove first argument.
27786         Update all callers.
27787         (struct ptr_data): Remove field TYPE.  Update all users.
27788         (gt_pch_note_object): Remove argument TYPE.  Update all users.
27789         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
27790         Update all users.
27791         * ggc-none.c (ggc_alloc_typed_stat): Remove.
27792         (struct alloc_zone): Remove.
27793         (ggc_internal_alloc_zone_stat): Remove.
27794         (ggc_internal_cleared_alloc_zone_stat): Remove.
27795         * ggc-page.c (ggc_alloc_typed_stat): Remove.
27796         (ggc_pch_count_object): Remove last argument.  Update all users.
27797         (ggc_pch_alloc_object): Remove last argument.  Update all users.
27798         (struct alloc_zone): Remove.
27799         * ggc-zone.c: Remove.
27800         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
27801         (struct alloc_zone): Remove.
27802         (ggc_alloc_typed_stat): Remove.
27803         (ggc_alloc_typed): Remove.
27804         (ggc_splay_alloc): Remove first argument.
27805         (rtl_zone): Remove.  Update all users.
27806         (tree_zone): Remove.  Update all users.
27807         (tree_id_zone): Remove.  Update all users.
27808         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
27809         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
27810         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
27811         * tree-ssanames.c: Remove references to zone allocator in comments.
27813 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
27815         * config/avr/avr.c (avr_out_fract): Make register numbers that
27816         might be outside of source operand signed.
27818 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
27820         * config/i386/constraints.md (Yf): New constraint.
27821         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
27822         of f constraint to conditionaly disable x87 register preferences.
27823         (*movdf_internal): Ditto.
27824         (*movsf_internal): Ditto.
27826 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
27828         PR inline-asm/55934
27829         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
27830         that have operands with impossible constraints.
27831         Add a FIXME for a speed-up opportunity.
27832         * lra-constraints.c (process_alt_operands): Verify that a class
27833         selected from constraints on asms is valid for the operand mode.
27834         (curr_insn_transform): Remove incorrect comment.
27836 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
27838         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
27839         TOC operand is a valid symbol ref in the constant pool.
27841 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27843         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
27845 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
27847         PR target/54222
27848         * config/avr/stdfix.h: New file.
27849         * t-avr (stdfix-gcc.h): New rule to build it.
27850         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
27852 2013-01-23  Kostya Serebryany  <kcc@google.com>
27854         * config/darwin.h: remove dependency on
27855         CoreFoundation (asan on Mac OS).
27857 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
27859         PR target/49069
27860         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
27861         instead of cmpdi_operand for first comparison operand.
27862         Don't assert that comparison operands aren't both constants.
27864 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
27866         * doc/install.texi (Downloading the Source): Update references to
27867         downloading separate components.
27869 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
27871         * doc/extend.texi (__int128): Improve grammar.
27873 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
27875         PR target/56028
27876         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
27877         alternative to (o,r).
27878         (*movdi_internal_rex64): Remove (!o,n) alternative.
27879         (DImode immediate->memory splitter): Remove.
27880         (DImode immediate->memory peephole2): Remove.
27881         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
27882         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
27883         alternative to (!o,*r).
27884         (*movtf_internal_sse): New pattern.
27885         (*movxf_internal_rex64): New pattern.
27886         (*movxf_internal): Disable for TARGET_64BIT.
27887         (*movdf_internal_rex64): Remove (!o,F) alternative.
27889 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
27891         PR middle-end/56074
27892         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
27893         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
27894         * tree-vect-loop-manip.c (find_loop_location): Also ignore
27895         stmt locations where LOCATION_LOCUS of the stmt location is
27896         UNKNOWN_LOCATION or BUILTINS_LOCATION.
27898         PR target/55686
27899         * config/i386/i386.md (UNSPEC_STOS): New.
27900         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
27901         *strsetqi_1): Add UNSPEC_STOS.
27903 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
27905         PR c++/56067
27906         * doc/invoke.texi: Remove left over -Wsynth example.
27908 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
27910         PR tree-optimization/56051
27911         * fold-const.c (fold_binary_loc): Don't fold
27912         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
27913         a narrowing conversion, or widening conversion from signed
27914         to unsigned.
27916 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
27918         PR rtl-optimization/56023
27919         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
27920         dependent on debug instruction.
27922 2013-01-21  Martin Jambor  <mjambor@suse.cz>
27924         PR middle-end/56022
27925         * function.c (allocate_struct_function): Call
27926         invoke_set_current_function_hook earlier.
27928 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
27930         * reload1.c (init_reload): Only initialize reload_obstack
27931         during the first call.
27933 2013-01-21  Marek Polacek  <polacek@redhat.com>
27935         * cfgloop.c (verify_loop_structure): Fix up grammar.
27937 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
27939         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
27940         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
27942 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
27944         PR target/56058
27945         * config/arm/marvell-pj4.md: Update copyright year.
27946         Fix up use of alu to alu_reg and simple_alu_imm.
27948 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
27950         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
27952 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
27954         PR target/55433
27955         * lra-constraints.c (curr_insn_transform): Don't reuse original
27956         insn for secondary memory move when memory mode should be different.
27958 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27960         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
27961         atomic_storedi_1): New patterns.
27963 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
27965         btver2 pipeline descriptions.
27966         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
27967         descriptions.
27968         * config/i386/i386.md (btver2_decode): New type attributes.
27969         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
27970         type attributes.
27971         * config/i386/btver2.md: New file describing btver2 pipelines.
27973 2013-01-19  Andrew Pinski  <apinski@cavium.com>
27975         PR tree-optimization/52631
27976         * tree-ssa-sccvn (visit_use): Before looking up the original
27977         statement, try looking up the simplified expression.
27979 2013-01-19  Anthony Green  <green@moxielogic.com>
27981         * config/moxie/moxie.c (moxie_expand_prologue): Set
27982         current_function_static_stack_size.
27984 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
27986         PR tree-optimization/56029
27987         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
27988         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
27990 2013-01-18  Sharad Singhai  <singhai@google.com>
27992         PR tree-optimization/55995
27993         * dumpfile.c (dump_loc): Print location only if available.
27994         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
27996 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
27998         PR target/55433
27999         * lra-constraints.c (curr_insn_transform): Reuse original insn for
28000         secondary memory move.
28001         (inherit_reload_reg): Use rclass instead of cl for
28002         check_secondary_memory_needed_p.
28004 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28006         PR middle-end/56015
28007         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
28008         the case where writing real complex part of target modifies op1.
28010 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
28012         * config/aarch64/aarch64-simd.md
28013         (aarch64_vcond_internal<mode>): Handle unordered cases.
28014         * config/aarch64/iterators.md (v_cmp_result): New.
28016 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
28017             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28019         * config/arm/marvell-pj4.md: New file.
28020         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
28021         * config/arm/arm.md (generic_sched): Add marvell_pj4.
28022         (generic_vfp): Likewise.
28023         * config/arm/arm-cores.def: Add marvell-pj4.
28024         * config/arm/arm-tune.md: Regenerate.
28025         * config/arm/arm-tables.opt: Regenerate.
28026         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
28027         * doc/invoke.texi: Document marvell-pj4.
28029 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
28031         * config/aarch64/arm_neon.h: Map scalar types to standard types.
28033 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28035         PR debug/54114
28036         PR debug/54402
28037         PR debug/49888
28038         * var-tracking.c (negative_power_of_two_p): New.
28039         (global_get_addr_cache, local_get_addr_cache): New.
28040         (get_addr_from_global_cache, get_addr_from_local_cache): New.
28041         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
28042         heading comment.
28043         (vt_stack_offset_p): Remove.
28044         (vt_canon_true_dep): Always canonicalize loc's address.
28045         (clobber_overlapping_mems): Make sure we have a MEM.
28046         (local_get_addr_clear_given_value): New.
28047         (val_reset): Clear local cached entries.
28048         (compute_bb_dataflow): Create and release the local cache.
28049         Disable duplicate MEMs clobbering.
28050         (emit_notes_in_bb): Clobber MEMs likewise.
28051         (vt_emit_notes): Create and release the local cache.
28052         (vt_initialize, vt_finalize): Create and release the global
28053         cache, respectively.
28054         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
28056 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28058         PR libmudflap/53359
28059         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
28060         not found in the symtab.
28062 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28064         PR debug/56006
28065         PR rtl-optimization/55547
28066         PR rtl-optimization/53827
28067         PR debug/53671
28068         PR debug/49888
28069         * alias.c (offset_overlap_p): New, factored out of...
28070         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
28071         the conservative special case for symbolic constants.  Don't
28072         adjust zero sizes on alignment.
28074 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
28076         PR rtl-optimization/52573
28077         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
28078         REG_UNUSED for the same register.
28080 2013-01-17  Richard Biener  <rguenther@suse.de>
28081             Marek Polacek  <polacek@redhat.com>
28083         PR rtl-optimization/55833
28084         * loop-unswitch.c (unswitch_loops): Move loop verification...
28085         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
28086         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
28087         Set it to true when we're removing a loop from hierarchy tree in
28088         an irreducible region.
28089         (fix_bb_placements): Adjust caller.
28090         (fix_loop_placements): Likewise.
28092 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
28094         * config/avr/builtins.def (DEF_BUILTIN): Factor out
28095         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
28096         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
28097         Remove ID.  Adjust comments.
28098         * config/avr/avr-c.c (avr_builtin_name): Remove.
28099         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
28100         * config/avr/avr.c (avr_tolower): New static function.
28101         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
28102         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
28103         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
28104         default expansion.
28106 2013-01-17  Jan Hubicka  <jh@suse.cz>
28108         PR tree-optimization/55273
28109         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
28111 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
28113         PR target/55981
28114         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
28115         store through atomic_store<mode>_1.
28116         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
28118 2013-01-17  Martin Jambor  <mjambor@suse.cz>
28120         PR tree-optimizations/55264
28121         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
28122         for virtual methods.
28123         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
28124         virtual methods before inlining is over.
28125         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
28126         virtual functions.
28127         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
28128         non-virtual.
28130 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
28132         PR rtl-optimization/56005
28133         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
28134         pending reads for prefetch.
28136 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
28138         * config/aarch64/aarch64.md
28139         (*cstoresi_neg_uxtw): New pattern.
28140         (*cmovsi_insn_uxtw): New pattern.
28141         (*<optab>si3_uxtw): New pattern.
28142         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
28143         (*<optab>si3_insn_uxtw): New pattern.
28144         (*bswapsi2_uxtw): New pattern.
28146 2013-01-16  Richard Biener  <rguenther@suse.de>
28148         * tree-inline.c (tree_function_versioning): Remove set but
28149         never used variable.
28151 2013-01-16  Richard Biener  <rguenther@suse.de>
28153         PR tree-optimization/55964
28154         * tree-flow.h (rename_variables_in_loop): Remove.
28155         (rename_variables_in_bb): Likewise.
28156         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
28157         (copy_loop_before): Adjust and delete update-ssa status.
28158         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
28159         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
28160         (rename_variables_in_loop): Remove.
28161         (slpeel_update_phis_for_duplicate_loop): Likewise.
28162         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
28163         use available cfg machinery instead of duplicating it.
28164         Update PHI nodes and perform poor-mans SSA update here.
28165         (slpeel_tree_peel_loop_to_edge): Adjust.
28167 2013-01-16  Richard Biener  <rguenther@suse.de>
28169         PR tree-optimization/54767
28170         PR tree-optimization/53465
28171         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
28172         (vrp_visit_phi_node): For PHI arguments coming via backedges
28173         drop all symbolical range information.
28174         (execute_vrp): Compute backedges.
28176 2013-01-16  Richard Biener  <rguenther@suse.de>
28178         * doc/install.texi: Update CLooG and ISL requirements to
28179         0.18.0 and 0.11.1.
28181 2013-01-16  Christian Bruel  <christian.bruel@st.com>
28183         PR target/55301
28184         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
28185         (broken_move): Handle UNSPECV_SP_SWITCH_B.
28186         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
28188 2013-01-16  DJ Delorie  <dj@redhat.com>
28190         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
28191         (UNSPECV_SP_SWITCH_E): New.
28192         (sp_switch_1): Change to an unspec.
28193         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
28194         replace $r15.
28196 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
28198         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
28199         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
28200         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
28201         (expand_mem_thread_fence): Ditto.
28202         (expand_mem_signal_fence): Ditto.
28203         (expand_atomic_load): Ditto.
28204         (expand_atomic_store): Ditto.
28206 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
28208         PR rtl-optimization/55547
28209         PR rtl-optimization/53827
28210         PR debug/53671
28211         PR debug/49888
28212         * alias.c (memrefs_conflict_p): Set sizes to negative after
28213         AND adjustments.
28215 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28217         PR target/55940
28218         * function.c (thread_prologue_and_epilogue_insns): Always
28219         add crtl->drap_reg to set_up_by_prologue.set, even if
28220         stack_realign_drap is false.
28222 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
28224         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
28225         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
28226         *call): Fix indention.
28228 2013-01-15  Tom de Vries  <tom@codesourcery.com>
28230         PR target/55876
28231         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
28232         Update comment.
28234 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
28236         PR rtl-optimization/55153
28237         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
28239 2013-01-15  Martin Jambor  <mjambor@suse.cz>
28241         PR tree-optimization/55920
28242         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
28243         accesses as grp_to_be_debug_replaced.
28245 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28247         PR tree-optimization/55920
28248         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
28249         there is non-useless type conversion needed from debug rhs to lhs,
28250         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
28252 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
28253             Mikael Pettersson  <mikpe@it.uu.se>
28255         PR target/43961
28256         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
28257         Thumb.
28258         (ASM_OUTPUT_CASE_LABEL): Remove.
28259         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
28260         * final.c (shorten_branches): Update alignment of labels before
28261         jump tables if CASE_VECTOR_SHORTEN_MODE.
28263 2013-01-15  Richard Biener  <rguenther@suse.de>
28265         PR bootstrap/55961
28266         * system.h: Do not include gmp.h for building host tools.
28268 2013-01-15  Richard Biener  <rguenther@suse.de>
28270         PR middle-end/55882
28271         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
28272         account for bitpos when computing alignment.
28274 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
28276         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
28277         (ix86_target_macros_internal): Likewise.
28279         * config/i386/i386.c (m_CORE2I7): Removed.
28280         (m_CORE_HASWELL): New macro.
28281         (m_CORE_ALL): Likewise.
28282         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
28283         (initial_ix86_arch_features): Likewise.
28284         (processor_target_table): Initializations for Core avx2.
28285         (cpu_names): New names "core-avx2".
28286         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
28287         PROCESSOR_CORE_HASWELL.
28288         (ix86_issue_rate): New case.
28289         (ia32_multipass_dfa_lookahead): Likewise.
28290         (ix86_sched_init_global): Likewise.
28292         * config/i386/i386.h (TARGET_HASWELL): New macro.
28293         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
28294         (processor_type): New PROCESSOR_HASWELL.
28296 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28298         PR tree-optimization/55955
28299         * tree-vect-loop.c (vectorizable_reduction): Give up early on
28300         *SHIFT_EXPR and *ROTATE_EXPR codes.
28302         PR tree-optimization/48766
28303         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
28304         -ftrapv disable -fwrapv.
28306 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28308         PR target/55974
28309         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
28310         etc. to 1 and not to __flash.
28311         Use LL suffix for __INT24_MAX__ with -mint8.
28312         Use ULL suffix for __UINT24_MAX__ with -mint8.
28314 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28316         * config/avr/avr-arch.h
28317         (struct base_arch_s): Use typedef avr_arch_t instead.
28318         (struct arch_info_s): Use typedef avr_arch_info_t instead.
28319         (struct mcu_type_s): Use typedef avr_mcu_t instead.
28320         * config/avr/avr.c: Same.
28321         * config/avr/avr-devices.c: Same.
28322         * config/avr/driver-avr.c: Same.
28323         * config/avr/gen-avr-mmcu-texi.c: Same.
28324         * config/avr/avr-mcus.def: Adjust comment.
28326 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
28328         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
28329         * config/aarch64/iterators.md (VALLDI): New.
28331 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
28332             Andi Kleen  <ak@linux.intel.com>
28334         PR target/55948
28335         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
28336         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
28337         memmodel flag.
28339 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28341         * config/avr/avr-stdint.h: Remove trailing blanks.
28342         * config/avr/avr-log.h: Same.
28343         * config/avr/avr-arch.h: Same.
28344         * config/avr/avr-devices.c: Same.
28345         * config/avr/avr-dimode.md: Same.
28346         * config/avr/predicates.md: Same.
28347         * config/avr/avr-c.c: Same.  And fix typo.
28349         * config/avr/avr-protos.h: Same.  And:
28350         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
28351         (init_cumulative_args): Rename to avr_init_cumulative_args.
28352         (expand_prologue): Rename to avr_expand_prologue.
28353         (expand_epilogue): Rename to avr_expand_epilogue.
28354         (adjust_insn_length): Rename to avr_adjust_insn_length.
28355         (notice_update_cc): Rename to avr_notice_update_cc.
28356         (final_prescan_insn): Rename to avr_final_prescan_insn.
28357         * config/avr/avr.c: Same.
28358         * config/avr/avr.h: Same.
28359         * config/avr/avr.md: Remove trailing blanks.
28360         (prologue): Use avr_expand_prologue.
28361         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
28363 2013-01-14  Richard Biener  <rguenther@suse.de>
28365         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
28366         verify_location, collect_subblocks): New functions.
28367         (verify_gimple_in_cfg): Verify that locations only reference
28368         BLOCKs in the functions BLOCK tree.
28370 2013-01-14  Richard Biener  <rguenther@suse.de>
28372         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
28373         PHI argument.
28374         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
28375         unshare reference.
28376         (insert_out_of_ssa_copy_on_edge): Likewise.
28377         (rewrite_close_phi_out_of_ssa): Likewise.
28378         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
28379         debug expressions.
28380         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
28381         propagated constants.
28382         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
28383         can not be shared.
28385 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28387         * config/avr/avr-modes.def: Add GPL copyright notice.
28389 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
28391         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
28392         MEMMODEL_MASK to determine memory model.
28393         (atomic_store<mode>): Ditto from operands[2].
28394         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
28396 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
28398         PR fortran/55935
28399         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
28400         (fold_gimple_assign): Don't call unshare_expr here.
28401         (fold_ctor_reference): Call unshare_expr.
28403 2013-01-13  Terry Guo  <terry.guo@arm.com>
28405         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
28406         * doc/fragments.texi: Document MULTILIB_REUSE.
28407         * gcc.c (multilib_reuse): New internal spec.
28408         (set_multilib_dir): Also search multilib from multilib_reuse.
28409         * genmultilib (tmpmultilib3): Refactor code.
28410         (tmpmultilib4): Ditto.
28411         (multilib_reuse): New multilib argument.
28413 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
28415         * Makefile.in: Update copyright.
28417 2013-01-12  Tom de Vries  <tom@codesourcery.com>
28419         PR middle-end/55890
28420         * calls.c (expand_call): Check if arg_nr is valid.
28422 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
28424         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
28425         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
28426         documentation.  Add missing '__' in front of
28427         __builtin_ia32_packssdw256.
28429 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
28431         PR target/55719
28432         * config/s390/s390.c (s390_preferred_reload_class): Do not return
28433         NO_REGS for larl operands.
28434         (s390_reload_larl_operand): Use s390_load_address instead of
28435         emit_move_insn.
28437 2013-01-11  Richard Biener  <rguenther@suse.de>
28439         * tree-cfg.c (verify_node_sharing_1): Split out from ...
28440         (verify_node_sharing): ... here.
28441         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
28443 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
28445         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
28446         Substitute TREECHECKING.
28447         * configure: Regenerate.
28448         * Makefile.in (TREECHECKING): New.
28450 2013-01-11  Richard Guenther  <rguenther@suse.de>
28452         PR tree-optimization/44061
28453         * tree-vrp.c (extract_range_basic): Compute zero as
28454         value-range for __builtin_constant_p of function parameters.
28456 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
28458         Update copyright years.
28460 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
28462         PR rtl-optimization/55672
28463         * lra-eliminations.c (mark_not_eliminable): Permit addition with
28464         const to be eliminable.
28466 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
28468         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
28469         * configure: Regenerate.
28471 2013-01-10  Richard Biener  <rguenther@suse.de>
28473         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
28475 2013-01-10  Richard Biener  <rguenther@suse.de>
28477         PR bootstrap/55792
28478         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
28479         locations for virtual PHI arguments.
28480         (rewrite_update_phi_arguments): Likewise.
28482 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
28484         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
28485         on to assembler.
28487 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
28489         PR tree-optimization/55921
28490         * tree-complex.c (expand_complex_asm): New function.
28491         (expand_complex_operations_1): Call it for GIMPLE_ASM.
28493 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
28495         PR target/55718
28496         * config/s390/s390.c (s390_symref_operand_p)
28497         (s390_loadrelative_operand_p): Merge the two functions.
28498         (s390_check_qrst_address, print_operand_address): Add parameters
28499         to s390_loadrelative_operand_p invokation.
28500         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
28501         (s390_reload_larl_operand, s390_secondary_reload): Use
28502         s390_loadrelative_operand_p instead of s390_symref_operand_p.
28503         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
28505 2013-01-09  Mike Stump  <mikestump@comcast.net>
28507         * dse.c (record_store): Remove unnecessary assert.
28509 2013-01-09  Jan Hubicka  <jh@suse.cz>
28511         PR tree-optimization/55569
28512         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
28513         * cfgloop.h (scale_loop_profile): Likewise.
28515 2013-01-09  Jan Hubicka  <jh@suse.cz>
28517         PR lto/45375
28518         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
28519         functions.
28520         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
28522 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
28524         PR middle-end/55114
28525         * expr.h (maybe_emit_group_store): Declare.
28526         * expr.c (maybe_emit_group_store): New function.
28527         * builtins.c (expand_builtin_int_roundingfn): Call it.
28528         (expand_builtin_int_roundingfn_2): Likewise.
28530 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
28532         PR rtl-optimization/55829
28533         * lra-constraints.c (match_reload): Add code for absent output.
28534         (curr_insn_transform): Add code for reloads of matched inputs
28535         without output.
28537 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
28539         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
28540         attribute of movddup insn to DF.
28541         (*vec_interleave_lowv2df): Ditto.
28542         (vec_dupv2df): Ditto.
28544 2013-01-09  Jan Hubicka  <jh@suse.cz>
28546         PR tree-optimiation/55875
28547         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
28548         EVERY_ITERATION parameter.
28549         (number_of_iterations_exit): Check if exit is executed every iteration.
28550         (idx_infer_loop_bounds): Similarly here.
28551         (n_of_executions_at_most): Simplify
28552         to only test for cases where statement is dominated by the
28553         particular bound; handle correctly the "postdominance" test.
28554         (scev_probably_wraps_p): Use max loop iterations info
28555         as a global bound first.
28557 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
28558             Nick Clifton  <nickc@redhat.com>
28560         * config/v850/v850.md (cbranchsf4): New pattern.
28561         (cstoresf4): New pattern.
28562         (cbranchdf4): New pattern.
28563         (cstoredf4): New pattern.
28564         (movsicc): Disallow floating point comparisons.
28565         (cmpsf_le_insn): Fix order of operators.
28566         (cmpsf_lt_insn): Likewise.
28567         (cmpsf_eq_insn): Likewise.
28568         (cmpdf_le_insn): Likewise.
28569         (cmpdf_lt_insn): Likewise.
28570         (cmpdf_eq_insn): Likewise.
28571         (cmpsf_ge_insn): Use LE comparison.
28572         (cmpdf_ge_insn): Likewise.
28573         (cmpsf_gt_insn): Use LT comparison.
28574         (cmpdf_gt_insn): Likewise.
28575         (cmpsf_ne_insn): Delete pattern.
28576         (cmpdf_ne_insn): Delete pattern.
28577         * config/v850/v850.c (v850_gen_float_compare): Use
28578         gen_cmpdf_eq_insn for NE comparison.
28579         (v850_float_z_comparison_operator)
28580         (v850_float_nz_comparison_operator): Move from here ...
28581         * config/v850/predicates.md: ... to here.  Move GT and GE
28582         comparisons into v850_float_z_comparison_operator.
28583         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
28584         Delete prototype.
28585         (v850_float_nz_comparison_operator): Likewise.
28587 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28589         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
28590         with calls to gen_insvsi/gen_insvdi.
28592 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
28594         * config/i386/i386.c (initial_ix86_tune_features): Set up
28595         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
28597 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
28598             Jakub Jelinek  <jakub@redhat.com>
28600         PR tree-optimization/48189
28601         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
28602         If nitercst is 0, don't predict the exit edge.
28604 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
28606         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
28607         in asm_fprintf with reg_names.
28608         (aarch64_print_operand_address): Likewise.
28609         (aarch64_return_addr): Likewise.
28610         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
28612 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28614         * config/pa/pa.h (VAL_U6_BITS_P): Define.
28615         (INT_U6_BITS): Likewise.
28616         * config/pa/predicates.md (uint6_operand): New predicate.
28617         (shift5_operand, shift6_operand): Likewise.
28618         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
28619         arith32_operand.
28620         (lshrdi3): Use shift6_operand.
28621         (shrpsi4, shrpdi4): New insn patterns.
28622         (extzv): Delete expander.
28623         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
28624         predicates in unamed zero extract patterns.  Tighten common constraint.
28625         (extv): Delete expander.
28626         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
28627         predicates in unamed sign extract patterns.  Tighten common constraint.
28628         (insv): Delete expander.
28629         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
28630         predicates in unamed insert patterns.  Tighten common constraint.
28631         Change uint32_operand predicate to uint6_operand predicate in unamed
28632         DImode pattern to insert constant values of type 1...1xxxx.
28634 2013-01-04  Jan Hubicka  <jh@suse.cz>
28636         PR tree-optimization/55823
28637         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
28638         issue.
28640 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
28641             Uros Bizjak  <ubizjak@gmail.com>
28643         PR rtl-optimization/55845
28644         * df-problems.c (can_move_insns_across): Stop scanning at
28645         volatile_insn_p source instruction or give up if
28646         across_from .. across_to range contains any volatile_insn_p
28647         instructions.
28649 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
28651         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
28652         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
28653         Declare.
28654         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
28655         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
28657 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
28659         PR fortran/55341
28660         * asan.c (asan_clear_shadow): New function.
28661         (asan_emit_stack_protection): Use it.
28663 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
28665         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
28666         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
28667         with tab instead of space.
28669 2013-01-08  Nick Clifton  <nickc@redhat.com>
28671         * config/rl78/rl78.c (rl78_expand_prologue): Always select
28672         register bank 0 at the start of an interrupt handler.
28673         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
28674         MDBH registers.
28676 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
28678         * config/aarch64/aarch64-simd.md
28679         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
28680         (aarch64_simd_bsl): Likewise.
28681         (aarch64_vcond_internal<mode>): Likewise.
28682         (vcond<mode><mode>): Likewise.
28683         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
28684         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
28686 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
28688         * config/aarch64/aarch64-builtins.c
28689         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
28691 2013-01-08  Martin Jambor  <mjambor@suse.cz>
28693         PR debug/55579
28694         * tree-sra.c (analyze_access_subtree): Return true also after
28695         potentially creating a debug-only replacement.
28697 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
28699         PR middle-end/55890
28700         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
28702         PR tree-optimization/54120
28703         * tree-vrp.c (range_fits_type_p): Don't allow
28704         src_precision < precision from signed vr to unsigned_p
28705         if vr->min or vr->max is negative.
28706         (simplify_float_conversion_using_ranges): Test can_float_p
28707         against CODE_FOR_nothing.
28709 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
28710             Richard Biener  <rguenther@suse.de>
28712         PR middle-end/55851
28713         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
28714         types instead of just INTEGER_TYPE types.
28716 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
28718         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
28719         TF_SIZE): Define.
28721 2013-01-07  Steve Ellcey  <sellcey@mips.com>
28723         PR target/42661
28724         * config/mips/mips.opt: Change mad to mmad to match documentation.
28726 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
28728         PR target/55897
28729         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
28730         .progmemx.data now.
28732 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
28734         PR target/55897
28735         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
28736         (avr_addrspace_t): Add .section_name field.
28737         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
28738         array size.
28739         (avr_addrspace): Same.  Initialize .section_name.  Remove last
28740         NULL entry.  Put __memx into .progmemx.data.
28741         (progmem_section_prefix): Remove.
28742         (avr_asm_init_sections): No need to initialize progmem_section.
28743         (avr_asm_named_section): Use avr_addrspace[].section_name to get
28744         section name prefix.
28745         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
28746         retrieve the progmem section.
28747         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
28748         boundary to run over avr_addrspace[].
28749         (avr_register_target_pragmas): Ditto.
28751 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
28753         * varasm.c (output_constant_def_contents): For asan_protect_global
28754         protected strings, adjust DECL_ALIGN if needed, before testing for
28755         anchored symbols.
28756         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
28757         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
28758         normal decls.
28759         (output_object_block): For asan protected decls, emit asan padding
28760         after their contents.
28761         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
28762         (asan_finish_file): Test it here instead.
28764 2013-01-07  Nick Clifton  <nickc@redhat.com>
28765             Matthias Klose  <doko@debian.org>
28766             Doug Kwan  <dougkwan@google.com>
28767             H.J. Lu  <hongjiu.lu@intel.com>
28769         PR driver/55470
28770         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
28772         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
28774         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
28776         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
28778         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
28780 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
28782         PR target/54461
28783         * doc/install.texi (Cross-Compiler-Specific Options): Document
28784         --with-avrlibc.
28786 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
28788         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
28789         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
28790         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
28791         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
28792         vqmovun_high_s64): Fix source operand number and update copyright.
28794 2013-01-07  Richard Biener  <rguenther@suse.de>
28796         PR middle-end/55890
28797         * gimple.h (gimple_call_builtin_p): New overload.
28798         * gimple.c (validate_call): New function.
28799         (gimple_call_builtin_p): Likewise.
28800         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28801         Use gimple_call_builtin_p.
28802         (find_func_clobbers): Likewise.
28803         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
28804         (strlen_optimize_stmt): Likewise.
28806 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
28808         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
28809         (vld1q_dup_*): Likewise.
28810         (vld1_*): Likewise.
28811         (vld1q_*): Likewise.
28812         (vld1_lane_*): Likewise.
28813         (vld1q_lane_*): Likewise.
28815 2013-01-07  Richard Biener  <rguenther@suse.de>
28817         * lto-streamer.h (LTO_minor_version): Bump to 2.
28819 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
28821         * config/aarch64/aarch64-protos.h
28822         (aarch64_const_double_zero_rtx_p): Rename to...
28823         (aarch64_float_const_zero_rtx_p): ...this.
28824         (aarch64_float_const_representable_p): New.
28825         (aarch64_output_simd_mov_immediate): Likewise.
28826         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
28827         move immediate case.
28828         * config/aarch64/aarch64.c
28829         (aarch64_const_double_zero_rtx_p): Rename to...
28830         (aarch64_float_const_zero_rtx_p): ...this.
28831         (aarch64_print_operand): Allow printing of new constants.
28832         (aarch64_valid_floating_const): New.
28833         (aarch64_legitimate_constant_p): Check for valid floating-point
28834         constants.
28835         (aarch64_simd_valid_immediate): Likewise.
28836         (aarch64_vect_float_const_representable_p): New.
28837         (aarch64_float_const_representable_p): Likewise.
28838         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
28839         (aarch64_output_simd_mov_immediate): New.
28840         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
28841         (*movdf_aarch64): Likewise.
28842         * config/aarch64/constraints.md (Ufc): New.
28843         (Y): call aarch64_float_const_zero_rtx.
28844         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
28846 2013-01-07  Richard Biener  <rguenther@suse.de>
28848         PR tree-optimization/55888
28849         PR tree-optimization/55862
28850         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
28851         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
28852         not if it is contained therein.
28854 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
28856         * config/avr/t-avr: Typo.
28858 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
28860         PR55243
28861         * config/avr/t-avr: Don't automatically rebuild
28862         $(srcdir)/config/avr/t-multilib
28863         $(srcdir)/config/avr/avr-tables.opt
28864         $(srcdir)/doc/avr-mmcu.texi
28865         (avr-mcus): New phony target to build them on request.
28866         (s-avr-mlib, s-avr-mmcu-texi): Remove.
28867         * avr/avr-mcus.def: Adjust comments.
28869 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
28871         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
28873 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
28875         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
28877 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
28879         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
28881 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
28883         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
28884         to generate profiling.
28885         * config/rs6000/aix64.h (LIB_SPEC): Same.
28887 2013-01-04  Andrew Pinski  <apinski@cavium.com>
28889         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
28890         New function.
28891         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
28893 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
28895         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
28896         unconditionally.
28897         (ix86_expand_move): Ditto.
28898         (ix86_zero_extend_to_Pmode): Ditto.
28899         (ix86_expand_call): Ditto.
28900         (ix86_expand_special_args_builtin): Ditto.
28901         (ix86_expand_builtin): Ditto.
28903 2013-01-04  Richard Biener  <rguenther@suse.de>
28905         PR tree-optimization/55862
28906         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
28907         translating them through PHI nodes.
28909 2013-01-04  Martin Jambor  <mjambor@suse.cz>
28911         PR tree-optimization/55755
28912         * tree-sra.c (sra_modify_assign): Do not check that an access has no
28913         children when trying to avoid producing a VIEW_CONVERT_EXPR.
28915 2013-01-04  Marek Polacek  <polacek@redhat.com>
28917         PR middle-end/55859
28918         * opts.c (default_options_optimization): Clarify error message.
28920 2013-01-04  Richard Biener  <rguenther@suse.de>
28922         PR middle-end/55863
28923         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
28924         reassociation.
28926 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28928         PR target/53789
28929         * config/pa/pa.md (movsi): Revert previous change.
28930         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
28931         references.
28933 2013-01-03  Richard Henderson  <rth@redhat.com>
28935         * config/i386/i386.c (ix86_expand_move): Always assign to op1
28936         after eliminating TLS symbols.
28938 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
28940         PR bootstrap/50167
28941         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
28942         * graphite-poly.c (debug_gmp_value): Likewise.
28944 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
28946         PR target/55712
28947         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
28948         selected code model, define __code_mode_small__, __code_model_medium__,
28949         __code_model_large__, __code_model_32__ or __code_model_kernel__.
28950         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
28951         xchg temporary register with %k.  Declare temporary register as
28952         early clobbered.
28953         [__x86_64__]: For medium and large code models, preserve %rbx register.
28955 2013-01-03  Richard Biener  <rguenther@suse.de>
28957         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
28958         (dump_subscript): Adjust.
28959         (finalize_ddr_dependent): Do not dump redundant info.
28960         (analyze_siv_subscript): Adjust.
28961         (subscript_dependence_tester): Likewise.
28962         (compute_affine_dependence): Likewise.
28964 2013-01-03  Richard Biener  <rguenther@suse.de>
28966         Revert
28967         2013-01-03  Richard Biener  <rguenther@suse.de>
28969         PR tree-optimization/55857
28970         * tree-vect-stmts.c (vectorizable_load): Do not setup
28971         re-alignment for invariant loads.
28973         2013-01-02  Richard Biener  <rguenther@suse.de>
28975         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
28976         invariant load do not generate a vector load from the scalar location.
28978 2013-01-03  Richard Biener  <rguenther@suse.de>
28980         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
28981         for not vectorizing.
28982         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
28983         not build INDIRECT_REFs, call get_name once only.
28984         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
28985         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
28987 2013-01-03  Richard Biener  <rguenther@suse.de>
28989         PR tree-optimization/55857
28990         * tree-vect-stmts.c (vectorizable_load): Do not setup
28991         re-alignment for invariant loads.
28993 2013-01-03  Richard Biener  <rguenther@suse.de>
28995         PR lto/55848
28996         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
28997         prefer a built-in decl.
28999 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29001         * gcc.c (process_command): Update copyright notice dates.
29002         * gcov.c (print_version): Likewise.
29003         * gcov-dump.c (print_version): Likewise.
29005         PR rtl-optimization/55838
29006         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
29007         iv0.step, iv1.step and step.
29009 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29010             Marc Glisse  <marc.glisse@inria.fr>
29012         PR tree-optimization/55832
29013         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
29014         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
29015         integer_{one,zero}_node.
29017 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29019         PR debug/54402
29020         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
29021         * var-tracking.c (reverse_op): Don't add reverse ops to
29022         VALUEs that have already
29023         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
29025 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
29027         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
29029 2013-01-02  Teresa Johnson  <tejohnson@google.com>
29031         * dumpfile.c (dump_loc): Print filename with location.
29032         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
29033         new location_t parameter to emit complete unroll message with
29034         new dump framework.
29035         (canonicalize_loop_induction_variables): Compute loops location
29036         and pass to try_unroll_loop_completely.
29037         * loop-unroll.c (report_unroll_peel): New function.
29038         (peel_loops_completely): Use new dump format with location
29039         for main dumpfile message, and invoke report_unroll_peel on success.
29040         (decide_unrolling_and_peeling): Ditto.
29041         (decide_peel_once_rolling): Remove old dumpfile message subsumed
29042         by report_unroll_peel.
29043         (decide_peel_completely): Ditto.
29044         (decide_unroll_constant_iterations): Ditto.
29045         (decide_unroll_runtime_iterations): Ditto.
29046         (decide_peel_simple): Ditto.
29047         (decide_unroll_stupid): Ditto.
29048         * cfgloop.c (get_loop_location): New function.
29049         * cfgloop.h (get_loop_location): Declare.
29051 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29053         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
29054         NULL.
29056 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29058         PR middle-end/55198
29059         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
29060         BLKmode objects when EXPAND_MEMORY is specified.
29062 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29064         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
29065         in loop predicate.
29066         (fold_builtin_cpu): Do not share cpu model decls across statements.
29068 2013-01-02  Jason Merrill  <jason@redhat.com>
29070         PR c++/55804
29071         * tree.c (build_array_type_1): Revert earlier change.
29073 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
29075         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
29076         "cortex-a57".
29077         * config/aarch64/aarch64-tune.md: Re-generate.
29079 2013-01-02  Richard Biener  <rguenther@suse.de>
29081         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29082         invariant load do not generate a vector load from the scalar location.
29084 2013-01-02  Richard Biener  <rguenther@suse.de>
29086         PR bootstrap/55784
29087         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
29088         * configure: Regenerate.
29090 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29092         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
29093         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
29094         (expand_builtin_int_roundingfn_2): Keep the original target around
29095         for the fallback case.
29097 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29099         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
29100         to be clear for sign changes.
29102 2013-01-01  Jan Hubicka  <jh@suse.cz>
29104         * ipa-inline-analysis.c: Fix formatting.
29106 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
29108         PR tree-optimization/55831
29109         * tree-vect-loop.c (get_initial_def_for_induction): Use
29110         gsi_after_labels instead of gsi_start_bb.
29112 Copyright (C) 2013 Free Software Foundation, Inc.
29114 Copying and distribution of this file, with or without modification,
29115 are permitted in any medium without royalty provided the copyright
29116 notice and this notice are preserved.