1 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
2 Jakub Jelinek <jakub@redhat.com>
4 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
5 (BUILT_IN_FABSD64): Likewise.
6 (BUILT_IN_FABSD128): Likewise.
7 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8 (fold_builtin_1): Likewise.
9 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
11 (*abstd2_fpr): Likewise.
12 (*nabstd2_fpr): Likewise.
14 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
16 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
19 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
21 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
24 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
26 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
27 * builtins.c (default_libc_has_function): New.
28 (gnu_libc_has_function): Ditto.
29 (no_c99_libc_has_function): Ditto.
30 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
31 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
32 (fold_builtin_sincos): Likewise.
33 (fold_builtin_cexp): Likewise.
34 * builtins.def (DEF_C94_BUILTIN): Likewise.
35 (DEF_C99_BUILTIN): Likewise.
36 (DEF_C99_C90RES_BUILTIN): Likewise.
37 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
38 definitions to using this define.
39 * config/darwin-protos.h (darwin_libc_has_function): New.
40 * config/darwin.c: (darwin_libc_has_function: Ditto.
41 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
42 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
43 * config/darwin.h: Ditto.
44 * config/elfos.h: Ditto.
45 * config/freebsd.h: Ditto.
46 * config/i386/cygming.h: Ditto.
47 * config/i386/djgpp.h: Ditto.
48 * config/i386/i386-interix.h: Ditto.
49 * config/microblaze/microblaze.h: Ditto.
50 * config/mmix/mmix.h: Ditto.
51 * config/gnu-user.h: Ditto.
52 * config/ia64/hpux.h: Ditto.
53 * config/pa/pa-hpux.h: Ditto.
54 * config/pdp11/pdp11.h: Ditto.
55 * config/picochip/picochip.h: Ditto.
56 * config/linux.h: Ditto.
57 * config/netbsd.h: Ditto.
58 * config/openbsd.h: Ditto.
59 * config/rs6000/aix43.h: Ditto.
60 * config/rs6000/aix51.h: Ditto.
61 * config/rs6000/aix52.h: Ditto.
62 * config/rs6000/aix53.h: Ditto.
63 * config/rs6000/aix61.h: Ditto.
64 * config/rs6000/darwin.h: Ditto.
65 * config/rs6000/linux.h: Ditto.
66 * config/rs6000/linux64.h: Ditto.
67 * config/s390/tpf.h: Ditto.
68 * config/sol2-10.h: Ditto.
69 * config/sol2.h: Ditto.
70 * config/vms/vms.h: Ditto.
71 * config/vxworks.h: Ditto.
72 * config/linux-android.c (linux_android_libc_has_function):
73 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
74 * config/linux-protos.h (linux_android_libc_has_function):
76 * config/i386/i386.c (ix86_libc_has_function): New.
77 * config/i386/i386-protos.h
78 (ix86_libc_has_function): New declaration.
80 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
81 ("isinf<mode>2): Likewise.
82 * convert.c (convert_to_integer): Using new target hook
83 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
85 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
86 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
87 * coretypes.h (function_class): New enum for different
89 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
90 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
91 (TARGET_HAS_SINCOS): Likewise.
92 (TARGET_LIBC_HAS_FUNCTION): New.
93 * doc/tm.texi: Regenerated.
94 * targhooks.h (default_libc_has_function): New declaration.
95 (no_c99_libc_has_function): Ditto.
96 (gnu_libc_has_function): Ditto.
97 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
98 and TARGET_HAS_SINCOS.
100 2013-08-18 Jan Hubicka <jh@suse.cz>
102 * Makeifle-in (ipa-devirt.o): New.
103 (GTFILES): Add ipa-utils.h and ipa-devirt.c
104 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
105 (analyze_functions): Look into possible targets of polymorphic call.
106 * dumpfile.c (dump_files): Add type-inheritance dump.
107 * dumpfile.h (TDI_inheritance): New.
108 * ipa-devirt.c: New file.
109 * ipa-utils.h (odr_type_d): Forward declare.
110 (odr_type): New type.
111 (build_type_inheritance_graph): Declare.
112 (possible_polymorphic_call_targets): Declare and introduce inline
113 variant when only edge is pased.
114 (dump_possible_polymorphic_call_targets): Likewise.
115 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
116 * tree.c (type_in_anonymous_namespace_p): Break out from ...
117 (types_same_for_odr): ... here.
118 * tree.h (type_in_anonymous_namespace_p): Declare.
120 2013-08-18 Jakub Jelinek <jakub@redhat.com>
122 PR tree-optimization/58006
123 * tree-parloops.c (take_address_of): Don't ICE if get_name
125 (eliminate_local_variables_stmt): Remove clobber stmts.
127 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
129 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
130 error message is issued for an alias to undefined symbol.
132 2013-08-18 Jan Hubicka <jh@suse.cz>
134 * cgraph.c (cgraph_create_indirect_edge): Discover
135 polymorphic calls and record basic info into indirect_info.
136 * gimple-fold.c (gimple_fold_call): When doing BINFO based
137 devirtualization, ignore objc function calls.
138 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
139 call with no parm index info.
140 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
141 * tree.c (virtual_method_call_p): New function.
142 * tree.h (virtual_method_call_p): Declare.
144 2013-08-16 Jan Hubicka <jh@suse.cz>
147 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
149 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
151 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
154 2013-08-16 David Malcolm <dmalcolm@redhat.com>
156 * gengtype.c (type_for_name): Add special-case support for
157 locating types within the "gcc::" namespace.
158 (open_base_files): Emit a "using namespace gcc" directive.
160 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
163 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
164 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
166 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
167 array instead of each individual operand as a separate argument.
168 (emit_fusion_gpr_load): Likewise.
169 (expand_fusion_gpr_load): Add new function declaration.
171 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
172 signature to have the operands passed as an array, instead of as
173 separate arguments. Allow ZERO_EXTEND to be in the memory
174 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
175 depend on the register live/dead flags when peepholes are run.
176 (expand_fusion_gpr_load): New function to be called from the
177 peephole2 pass, to change the register that addis sets to be the
179 (emit_fusion_gpr_load): Change the calling signature to have the
180 operands passed as an array, instead of as separate arguments.
181 Allow ZERO_EXTEND to be in the memory address, and also
182 SIGN_EXTEND if -mpower8-fusion-sign.
184 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
186 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
187 adjust the register addis loads up in the peephole2 pass. Do not
188 depend on the register live/dead state when the peephole pass is
191 2013-08-16 David Malcolm <dmalcolm@redhat.com>
193 * gengtype.c (create_user_defined_type): Ensure that the kind
194 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
195 declaration is seen before the GTY((user)) marking.
197 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
200 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
202 2013-08-16 Jan Hubicka <jh@suse.cz>
204 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
205 arugment expected_type.
206 (gimple_fold_call): Use it.
207 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
208 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
209 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
210 (try_make_edge_direct_virtual_call): Likewise.
211 * tree.c (obj_type_ref_class): New.
212 * tree.h (obj_type_ref_class): Use it.
214 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
216 * sched-vis.c (rtl_slim_pp_initialized): Remove.
217 (rtl_slim_pp): Likewise.
218 (init_rtl_slim_pretty_print): Likewise.
219 (dump_value_slim): Don't call it. Use local pretty printer.
220 (dump_insn_slim): Likewise.
221 (dump_rtl_slim): Likewise.
222 (str_pattern_slim): Likewise.
223 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
226 2013-08-16 Jakub Jelinek <jakub@redhat.com>
228 PR tree-optimization/58164
229 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
230 walk gimple_goto_dest of GIMPLE_GOTO.
232 PR tree-optimization/58165
233 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
234 bi_call must be the last stmt in a bb, don't split_block, instead
235 use fallthru edge from it and give up if there is none.
236 Release conds vector when returning early.
238 2013-08-14 Xinliang David Li <davidxl@google.com>
240 * config/i386/i386.c (ix86_option_override_internal):
241 Remove unused variable and field.
243 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
246 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
247 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
248 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
249 and Linux, correct BLKmode alignment when 128-bit alignment is
250 required and compatibility flag is not set.
251 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
252 for zero-size arguments when compatibility flag is not set.
254 2013-08-14 Jakub Jelinek <jakub@redhat.com>
256 PR tree-optimization/58145
257 * tree-sra.c (build_ref_for_offset): If prev_base has
258 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
260 2013-08-14 Xinliang David Li <davidxl@google.com>
262 * config/i386/i386.c (ix86_option_override_internal):
263 Fix uninitialized variable error.
265 2013-08-14 Xinliang David Li <davidxl@google.com>
267 * config/i386/i386.opt: Define two new options.
268 * config/i386/x86-tune.def: Add arch selector field in macros.
269 * config/i386/i386.h: Adjust macro definition.
270 * config/i386/i386.c (ix86_option_override_internal):
272 (parse_mtune_ctrl_str): New function.
273 (set_ix86_tune_features): New function.
274 (ix86_function_specific_restore): Call the new helper function.
276 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
278 PR rtl-optimization/57662
279 * sel-sched.c (code_motion_process_successors): When the current insn
280 is removed after the recursive traversal, break from the loop.
281 Add comments and debug printouts.
283 2013-08-14 Jakub Jelinek <jakub@redhat.com>
284 Alexandre Oliva <aoliva@redhat.com>
287 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
288 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
289 there also UNSPEC_PLTOFF.
291 2013-08-14 Marek Polacek <polacek@redhat.com>
293 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
294 NUM_CONDITIONS bit positions.
296 2013-08-13 Cary Coutant <ccoutant@google.com>
298 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
299 (attr_checksum): Hash vector contents instead of pointer.
300 (attr_checksum_ordered): Likewise.
302 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
304 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
305 when Pmode != word_mode. Add length_address attribute.
306 (sse3_monitor_<mode>): Merge from sse3_monitor and
307 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
308 Pmode != word_mode. Update insn length attribute.
309 * config/i386/i386.c (ix86_option_override_internal): Update
310 ix86_gen_monitor selection for merged sse3_monitor insn.
312 2013-08-13 Julian Brown <julian@codesourcery.com>
314 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
315 perform invalid legitimization on greater-than-word-size modes for
318 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
320 * ira.c (setup_class_translate_array): Use aclass instead of cl
321 for classes not fully covered by allocno classes.
323 2013-08-13 Jakub Jelinek <jakub@redhat.com>
325 PR tree-optimization/57661
326 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
327 * tree-inline.c (tree_function_versioning): Initialize it.
328 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
329 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
330 that is not being copied.
333 * asan.c (instrument_strlen_call): Fix typo in comment.
334 Use char * type even for the lhs of POINTER_PLUS_EXPR.
336 2013-08-13 Steve Ellcey <sellcey@mips.com>
338 * config/mips/mips.md (prefetch): Use lw instead of ld on
339 loongson in 32bit mode.
341 2013-08-13 Nick Clifton <nickc@redhat.com>
343 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
345 2013-08-13 Jan Hubicka <jh@suse.cz>
347 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
348 introduced edge; fix typo in sanity check.
349 (cgraph_resolve_speculation): Export; improve diagnostic.
350 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
351 speculation at type mismatch.
352 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
353 (cgraph_resolve_speculation): Declare.
354 (symtab_can_be_discarded): New function.
355 * value-prof.c (gimple_ic_transform): Remove actual transform code.
356 * ipa-inline-transform.c (speculation_removed): New global var.
357 (clone_inlined_nodes): See if speculation can be removed.
358 (inline_call): If speculations was removed, we growths may not match.
359 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
360 (speculation_useful_p): New function.
361 (resolve_noninline_speculation): New function.
362 (inline_small_functions): Resolve useless speculations.
363 * ipa-inline.h (speculation_useful_p): Declare
364 * ipa.c (can_replace_by_local_alias): Simplify.
365 (ipa_profile): Produce speculative calls in non-lto, too;
366 add simple cost model; produce local aliases.
368 2013-08-13 David Malcolm <dmalcolm@redhat.com>
370 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
373 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
375 * config/i386/i386.c (ix86_function_versions): Use error + inform.
377 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
379 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
380 iterator instead of X87MODEF.
382 2013-08-12 Perez Read <netfirewall@gmail.com>
385 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
386 operand 0 for intel asm alternative.
387 (*movabs<mode>_2): Ditto for operand 1.
389 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
391 * config/aarch64/arm_none.h
392 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
394 2013-08-12 Nick Clifton <nickc@redhat.com>
396 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
398 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
400 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
401 expand for QI/HImode operand to produce more effictive code for
402 unsigned char(short) --> float(double) conversion.
404 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
406 * doc/invoke.texi: Mention that -ftls-model does not force the final
409 2013-08-12 Marek Polacek <polacek@redhat.com>
410 Marc Glisse <marc.glisse@inria.fr>
412 PR tree-optimization/57980
413 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
414 when creating -1 constant.
416 2013-08-10 Jan Hubicka <jh@suse.cz>
418 Workaround binutils PR14342.
419 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
420 (gimple_init_edge_profiler): Likewise.
421 (gimple_gen_ic_func_profiler): Likewise.
423 2013-08-09 Jan Hubicka <jh@suse.cz>
425 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
427 2013-08-09 Xinliang David Li <davidxl@google.com>
429 * config/i386/stringop.def: New file.
430 * config/i386/stringop.opt: New file.
431 * config/i386/i386-opts.h: Include stringopt.def.
432 * config/i386/i386.opt: Include stringopt.opt.
433 * config/i386/i386.c (ix86_option_override_internal):
434 Override default size based stringop inline strategies with options.
435 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
438 2013-08-09 Jan Hubicka <jh@suse.cz>
440 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
442 2013-08-09 Jan Hubicka <jh@suse.cz>
444 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
446 (dump_cgraph_node): Dump profile-id.
447 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
448 and common_target_probability.
449 * lto-cgraph.c (lto_output_edge): Stream common targets.
450 (lto_output_node): Stream profile ids.
451 (input_node): Stream profile ids.
452 (input_edge): Stream common targets.
453 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
454 * ipa.c: Include value-prof.h
455 (ipa_profile_generate_summary): Turn indirect call statement histograms
457 (ipa_profile): Turn common targets into speculative edges.
459 2013-08-09 Jan Hubicka <jh@suse.cz>
461 * cgraph.h (cgraph_node): Add profile_id.
462 * value-prof.c (cgraph_node_map): Turn into pointer_map.
463 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
464 (del_node_map): Update.
465 (find_func_by_funcdef_no): Replace by ...
466 (find_func_by_profile_id): ... this one.
467 (gimple_ic_transform): Do not remove useful histograms when
468 speculation is not done; dump info when indirect call removal
470 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
471 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
472 (__gcov_indirect_call_profiler_v2): .. this one.
473 * profile.h (init_node_map): Update.
474 * coverage.c (coverage_compute_profile_id): New function.
475 * coverage.h (coverage_compute_profile_id): Declare.
476 * tree-profile.c (init_ic_make_global_vars): Make
477 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
478 (gimple_init_edge_profiler): Update prototype of
479 __gcov_indirect_call_profiler.
480 (gimple_gen_ic_func_profiler): Simplify.
481 (tree_profiling): Use init_node_map
483 2013-08-09 Jan Hubicka <jh@suse.cz>
485 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
486 non-speculative refs.
487 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
488 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
489 (cgraph_set_call_stmt): Likewise.
490 (cgraph_create_edge_1): Fix release checking compilatoin;
492 (cgraph_free_edge): Free indirect info.
493 (cgraph_turn_edge_to_speculative): New function.
494 (cgraph_speculative_call_info): New function.
495 (cgraph_make_edge_direct): Return direct edge; handle speculation.
496 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
497 (dump_cgraph_node): Dump speculation.
498 (verify_edge_count_and_frequency): Accept speculative edges.
499 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
500 (verify_cgraph_node): Handle speculation.
501 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
502 (cgraph_set_call_stmt): Update prototype.
503 (cgraph_make_edge_direct): Update prototype.
504 (cgraph_speculative_call_info): Declare.
505 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
506 to change; update call of ipa_find_references.
507 * ipa-ref.c (ipa_record_reference): Fix return value; clear
508 lto_stmt_uid and speculative flags.
509 (ipa_dump_references): Dump speculation.
510 (ipa_clone_references): Clone speculative flag.
511 (ipa_clone_referring): Likewise.
512 (ipa_clone_ref): New function.
513 (ipa_find_reference): Look into lto_stmt_uids
514 (ipa_clear_stmts_in_references): Do not clear speculative calls.
515 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
516 (ipa_find_reference): Update declaration.
517 (ipa_clone_ref): Declare.
518 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
519 stream speculative flag.
520 (lto_output_ref): Stream statements uids and speculation.
521 (input_ref): Likewise.
522 (input_edge): Stream speuclation.
523 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
524 (cgraph_set_call_stmt_including_clones): Handle speculation.
525 * ipa-inline.c (heap_edge_removal_hook): New function.
526 (inline_small_functions): Register it.
527 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
528 also initialize refs.
529 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
531 (try_make_edge_direct_simple_call): Likewise.
532 (try_make_edge_direct_simple_call): Likewise.
533 (update_indirect_edges_after_inlining): Likewise.
534 (remove_described_reference): Look proper lto_stmt_uid.
535 (propagate_controlled_uses): Likewise.
536 (propagate_controlled_uses): Liekwise.
537 * tree-inline.c (copy_bb): Copy speculative edges.
538 (redirect_all_calls): New function.
539 (copy_cfg_body): Do redirection after loop info is updated.
540 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
542 2013-08-09 Jan Hubicka <jh@suse.cz>
544 * lto-streamer-out.c (output_function): Renumber PHIs.
545 * lto-streamer-in.c (input_function): Likewise.
547 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
549 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
550 (get_lane_unsigned): Likewise.
551 (dup_lane_scalar): Likewise.
552 (get_lane): enable for VALL.
553 * config/aarch64/aarch64-simd.md
554 (aarch64_dup_lane_scalar<mode>): Remove.
555 (aarch64_get_lane_signed<mode>): Likewise.
556 (aarch64_get_lane_unsigned<mode>): Likewise.
557 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
558 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
559 (aarch64_get_lane<mode>): Enable for all vector modes.
560 (aarch64_get_lanedi): Remove misleading constraints.
561 * config/aarch64/arm_neon.h
562 (__aarch64_vget_lane_any): Define.
563 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
564 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
565 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
566 * config/aarch64/iterators.md (VDQQH): New.
570 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
572 * configure.ac: Add GAS check for LEON instructions on SPARC.
573 * configure: Regenerate.
574 * config.in: Likewise.
575 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
577 * config/sparc/sparc.opt (LEON, LEON3): New masks.
578 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
580 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
581 (AS_LEON_FLAG): New macro.
582 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
583 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
584 Deal with LEON and LEON3 for the memory model.
585 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
586 (atomic_compare_and_swap<mode>_1): Likewise.
587 (*atomic_compare_and_swap<mode>_1): Likewise.
589 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
591 * config/arm/neon.md (vcond): Fix floating-point vector
592 comparisons against 0.
594 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
596 * lra-constraints.c (emit_spill_move): Remove assert.
597 (process_alt_operands): Add more debugging
598 output. Increase reject for spilling into memory. Decrease
599 reject for reloading scratch.
600 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
602 2013-08-08 Steve Ellcey <sellcey@mips.com>
604 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
605 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
606 micromips incompatible. Add nan2008.
607 (MULTILIB_DIRNAMES): Add nan2008.
608 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
609 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
610 and micromips incompatible. Add nan2008.
611 (MULTILIB_DIRNAMES): Add nan2008.
612 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
614 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
616 PR rtl-optimization/58079
617 * combine.c (combine_simplify_rtx): Avoid using SUBST if
618 simplify_comparison has widened a comparison with an integer.
620 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
622 * config/arm/neon.md (movmisalign<mode>): Disable when we
623 don't allow unaligned accesses.
624 (*movmisalign<mode>_neon_store): Likewise.
625 (*movmisalign<mode>_neon_load): Likewise.
626 (*movmisalign<mode>_neon_store): Likewise.
627 (*movmisalign<mode>_neon_load): Likewise.
629 2013-08-08 Jan Hubicka <jh@suse.cz>
631 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
632 (make_pass_rebuild_cgraph_edges): Also clear references.
633 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
634 * ipa-inline-transform.c (inline_transform): Remove all references
636 * cgraphunit.c (expand_function): Remove all references after
638 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
639 (ipa_find_reference): Rewrite to iterator.
640 (remove_stmt_references): Likewise.
641 (ipa_clear_stmts_in_references): New function.
642 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
643 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
645 * ipa-split.c (split_function): Remove references in split function.
647 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
650 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
651 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
653 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
656 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
657 suggested mode for the assignment isn't compatible with the
660 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
663 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
665 2013-08-07 Xinliang David Li <davidxl@google.com>
667 * config/i386/i386.opt: New option -mtune-ctrl=.
668 * config/i386/x86-tune.def: New file.
669 * config/i386/i386.h: include x86-tune.def.
670 * config/i386/i386.c (ix86_option_override_internal):
671 Parsing -mtune-ctrl= option and set tune features.
673 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
676 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
677 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
680 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
682 * rtl.h (update_alignments): Declare.
683 * final.c (grow_label_align): New function extracted from...
684 (shorten_branches): ...here. Call it.
685 (update_alignments): New function.
686 * reorg.c (sibling_labels): New variable.
687 (get_label_before): Add SIBLING parameter. If it is non-zero, push
688 the new label along with it onto the sibling_labels vector.
689 (fill_simple_delay_slots): Adjust call to get_label_before.
690 (fill_slots_from_thread): Likewise.
691 (relax_delay_slots): Likewise.
692 (make_return_insns): Likewise.
693 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
695 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
697 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
698 document its semantics.
699 (diagnostic_report_diagnostic): Adjust accordingly.
701 2013-08-07 David Malcolm <dmalcolm@redhat.com>
703 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
704 (sparc_option_override): ...and port to new C++ pass API.
705 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
707 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
709 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
711 2013-08-06 Caroline Tice <cmtice@google.com>
713 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
714 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
715 * tree-pass.h: Add pass_vtable_verify.
716 * varasm.c (assemble_variable): Add code to properly set the comdat
717 section and name for the .vtable_map_vars section.
718 (assemble_vtyv_preinit_initializer): New function.
719 (default_sectin_type_flags): Make sure .vtable_map_vars section has
721 * output.h: Add function decl for assemble_vtv_preinit_initializer.
722 * vtable-verify.c: New file.
723 * vtable-verify.h: New file.
724 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
725 initialiation levels.
726 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
727 * passes.def: Insert pass_vtable_verify.
728 * aclocal.m4: Reorder includes.
729 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
730 -fvtv-counts options.
731 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
732 as appropriate, if -fvtable-verify=... is used.
733 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
734 -fvtable-verify=... is used.
735 * Makefile.in (OBJS): Add vtable-verify.o to list.
736 (vtable-verify.o): Add new build rule.
737 (GTFILES): Add vtable-verify.c to list.
738 * common.opt (fvtable-verify=): New flag.
739 (vtv_priority): Values for fvtable-verify= flag.
740 (fvtv-counts): New flag.
741 (fvtv-debug): New flag.
742 * tree.h (save_vtable_map_decl): New extern function decl.
744 2013-08-07 David Malcolm <dmalcolm@redhat.com>
746 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
747 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
748 (pass_data_rl78_devirt): ...new pass_data instance and...
749 (make_pass_rl78_devirt): ...new function.
750 (rl78_asm_file_start): Port pass registration to new C++ API.
752 2013-08-07 David Malcolm <dmalcolm@redhat.com>
754 * coretypes.h (rtl_opt_pass): Add.
756 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
757 (epiphany_init): Port to new C++ pass API.
758 (epiphany_optimize_mode_switching): Likewise.
759 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
760 (pass_manager::get_pass_mode_switching): New.
761 (pass_manager::get_pass_peephole2): New.
762 * mode-switching.c (pass_mode_switching): Add clone method.
763 * recog.c (pass_peephole2): Add clone method.
764 (pass_split_all_insns): Add clone method.
766 2013-08-06 David Malcolm <dmalcolm@redhat.com>
768 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
769 (mips_option_override): ...here, porting to new C++ API for passes.
771 2013-08-06 Jan Hubicka <jh@suse.cz>
773 * cgraph.c (cgraph_get_body): New function based on lto.c
775 * cgraph.h (cgraph_get_body): Declare.
776 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
778 * cgraphunit.c (expand_function): Get body prior expanding.
779 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
780 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
782 * passes.c (do_per_function_toporder): Get body.
783 * tree-inline.c (expand_call_inline): Get body prior inlining it.
784 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
786 2013-08-06 Martin Jambor <mjambor@suse.cz>
789 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
790 is not re-finalized. Rename second parameter to no_collect.
792 2013-08-06 Martin Jambor <mjambor@suse.cz>
795 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
796 MEM_REF has proper alignment information.
798 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
801 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
803 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
804 rtx (*) (rtx, ...) with insn_gen_fn.
805 * genoutput.c (output_insn_data): Cast gen_? function pointers to
806 insn_gen_fn::stored_funcptr. Add initializer braces.
808 2013-08-05 David Malcolm <dmalcolm@redhat.com>
810 Rewrite how instances of passes are cloned to remove assumptions
811 about their sizes (thus allowing pass subclasses to have
812 additional data fields, albeit non-GC-managed ones at this point).
814 * passes.c (make_pass_instance): Now that passes have clone
815 methods, rewrite this function to eliminate XNEW and memcpy
816 calls that used hardcoded sizes. Since this function no longer
817 creates pass instances, rename it to...
818 (add_pass_instance): ...this. Document the old way that passes were
819 numbered and flagged, and rework this function to continue using it.
820 (next_pass_1): Add an initial_pass argument for use by
822 (position_pass): When adding multiple instances of a pass, use
823 the pass's clone method, rather than relying on the XNEW/memcpy
824 within the former make_pass_instance (now add_pass_instance).
825 (pass_manager::pass_manager): When invoking next_pass_1, also supply
826 the initial instance of the current pass within the pass manager.
828 2013-08-05 David Malcolm <dmalcolm@redhat.com>
830 This is the automated part of the conversion of passes from C
831 structs to C++ classes.
833 Patch autogenerated by refactor_passes.py from
834 https://github.com/davidmalcolm/gcc-refactoring-scripts
835 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
837 * asan.c (pass_asan): Convert from a global struct to a subclass of
838 gimple_opt_pass along with...
839 (pass_data_asan): ...new pass_data instance and...
840 (make_pass_asan): ...new function.
841 (pass_asan_O0): Convert from a global struct to a subclass of
842 gimple_opt_pass along with...
843 (pass_data_asan_O0): ...new pass_data instance and...
844 (make_pass_asan_O0): ...new function.
845 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
846 subclass of rtl_opt_pass along with...
847 (pass_data_inc_dec): ...new pass_data instance and...
848 (make_pass_inc_dec): ...new function.
849 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
850 a subclass of rtl_opt_pass along with...
851 (pass_data_reorder_blocks): ...new pass_data instance and...
852 (make_pass_reorder_blocks): ...new function.
853 (pass_duplicate_computed_gotos): Convert from a global struct to a
854 subclass of rtl_opt_pass along with...
855 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
856 (make_pass_duplicate_computed_gotos): ...new function.
857 (pass_partition_blocks): Convert from a global struct to a subclass of
858 rtl_opt_pass along with...
859 (pass_data_partition_blocks): ...new pass_data instance and...
860 (make_pass_partition_blocks): ...new function.
861 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
862 struct to a subclass of rtl_opt_pass along with...
863 (pass_data_branch_target_load_optimize1): ...new pass_data instance
865 (make_pass_branch_target_load_optimize1): ...new function.
866 (pass_branch_target_load_optimize2): Convert from a global struct to a
867 subclass of rtl_opt_pass along with...
868 (pass_data_branch_target_load_optimize2): ...new pass_data instance
870 (make_pass_branch_target_load_optimize2): ...new function.
871 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
872 of rtl_opt_pass along with...
873 (pass_data_jump): ...new pass_data instance and...
874 (make_pass_jump): ...new function.
875 (pass_jump2): Convert from a global struct to a subclass of
876 rtl_opt_pass along with...
877 (pass_data_jump2): ...new pass_data instance and...
878 (make_pass_jump2): ...new function.
879 * cfgexpand.c (pass_expand): Convert from a global struct to a
880 subclass of rtl_opt_pass along with...
881 (pass_data_expand): ...new pass_data instance and...
882 (make_pass_expand): ...new function.
883 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
884 of rtl_opt_pass along with...
885 (pass_data_free_cfg): ...new pass_data instance and...
886 (make_pass_free_cfg): ...new function.
887 (pass_into_cfg_layout_mode): Convert from a global struct to a
888 subclass of rtl_opt_pass along with...
889 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
890 (make_pass_into_cfg_layout_mode): ...new function.
891 (pass_outof_cfg_layout_mode): Convert from a global struct to a
892 subclass of rtl_opt_pass along with...
893 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
894 (make_pass_outof_cfg_layout_mode): ...new function.
895 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
896 struct to a subclass of gimple_opt_pass along with...
897 (pass_data_build_cgraph_edges): ...new pass_data instance and...
898 (make_pass_build_cgraph_edges): ...new function.
899 (pass_rebuild_cgraph_edges): Convert from a global struct to a
900 subclass of gimple_opt_pass along with...
901 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
902 (make_pass_rebuild_cgraph_edges): ...new function.
903 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
904 subclass of gimple_opt_pass along with...
905 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
907 (make_pass_remove_cgraph_callee_edges): ...new function.
908 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
909 struct to a subclass of rtl_opt_pass along with...
910 (pass_data_stack_adjustments): ...new pass_data instance and...
911 (make_pass_stack_adjustments): ...new function.
912 * combine.c (pass_combine): Convert from a global struct to a subclass
913 of rtl_opt_pass along with...
914 (pass_data_combine): ...new pass_data instance and...
915 (make_pass_combine): ...new function.
916 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
917 global struct to a subclass of rtl_opt_pass along with...
918 (pass_data_compare_elim_after_reload): ...new pass_data instance
920 (make_pass_compare_elim_after_reload): ...new function.
921 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
922 of rtl_opt_pass along with...
923 (pass_data_rtl_cprop): ...new pass_data instance and...
924 (make_pass_rtl_cprop): ...new function.
925 * cse.c (pass_cse): Convert from a global struct to a subclass of
926 rtl_opt_pass along with...
927 (pass_data_cse): ...new pass_data instance and...
928 (make_pass_cse): ...new function.
929 (pass_cse2): Convert from a global struct to a subclass of
930 rtl_opt_pass along with...
931 (pass_data_cse2): ...new pass_data instance and...
932 (make_pass_cse2): ...new function.
933 (pass_cse_after_global_opts): Convert from a global struct to a
934 subclass of rtl_opt_pass along with...
935 (pass_data_cse_after_global_opts): ...new pass_data instance and...
936 (make_pass_cse_after_global_opts): ...new function.
937 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
938 of rtl_opt_pass along with...
939 (pass_data_ud_rtl_dce): ...new pass_data instance and...
940 (make_pass_ud_rtl_dce): ...new function.
941 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
942 rtl_opt_pass along with...
943 (pass_data_fast_rtl_dce): ...new pass_data instance and...
944 (make_pass_fast_rtl_dce): ...new function.
945 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
946 a subclass of rtl_opt_pass along with...
947 (pass_data_df_initialize_opt): ...new pass_data instance and...
948 (make_pass_df_initialize_opt): ...new function.
949 (pass_df_initialize_no_opt): Convert from a global struct to a
950 subclass of rtl_opt_pass along with...
951 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
952 (make_pass_df_initialize_no_opt): ...new function.
953 (pass_df_finish): Convert from a global struct to a subclass of
954 rtl_opt_pass along with...
955 (pass_data_df_finish): ...new pass_data instance and...
956 (make_pass_df_finish): ...new function.
957 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
958 rtl_opt_pass along with...
959 (pass_data_rtl_dse1): ...new pass_data instance and...
960 (make_pass_rtl_dse1): ...new function.
961 (pass_rtl_dse2): Convert from a global struct to a subclass of
962 rtl_opt_pass along with...
963 (pass_data_rtl_dse2): ...new pass_data instance and...
964 (make_pass_rtl_dse2): ...new function.
965 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
966 subclass of rtl_opt_pass along with...
967 (pass_data_dwarf2_frame): ...new pass_data instance and...
968 (make_pass_dwarf2_frame): ...new function.
969 * except.c (pass_set_nothrow_function_flags): Convert from a global
970 struct to a subclass of rtl_opt_pass along with...
971 (pass_data_set_nothrow_function_flags): ...new pass_data instance
973 (make_pass_set_nothrow_function_flags): ...new function.
974 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
975 subclass of rtl_opt_pass along with...
976 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
978 (make_pass_convert_to_eh_region_ranges): ...new function.
979 * final.c (pass_compute_alignments): Convert from a global struct to a
980 subclass of rtl_opt_pass along with...
981 (pass_data_compute_alignments): ...new pass_data instance and...
982 (make_pass_compute_alignments): ...new function.
983 (pass_final): Convert from a global struct to a subclass of
984 rtl_opt_pass along with...
985 (pass_data_final): ...new pass_data instance and...
986 (make_pass_final): ...new function.
987 (pass_shorten_branches): Convert from a global struct to a subclass of
988 rtl_opt_pass along with...
989 (pass_data_shorten_branches): ...new pass_data instance and...
990 (make_pass_shorten_branches): ...new function.
991 (pass_clean_state): Convert from a global struct to a subclass of
992 rtl_opt_pass along with...
993 (pass_data_clean_state): ...new pass_data instance and...
994 (make_pass_clean_state): ...new function.
995 * function.c (pass_instantiate_virtual_regs): Convert from a global
996 struct to a subclass of rtl_opt_pass along with...
997 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
998 (make_pass_instantiate_virtual_regs): ...new function.
999 (pass_leaf_regs): Convert from a global struct to a subclass of
1000 rtl_opt_pass along with...
1001 (pass_data_leaf_regs): ...new pass_data instance and...
1002 (make_pass_leaf_regs): ...new function.
1003 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1004 subclass of rtl_opt_pass along with...
1005 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1007 (make_pass_thread_prologue_and_epilogue): ...new function.
1008 (pass_match_asm_constraints): Convert from a global struct to a
1009 subclass of rtl_opt_pass along with...
1010 (pass_data_match_asm_constraints): ...new pass_data instance and...
1011 (make_pass_match_asm_constraints): ...new function.
1012 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1013 subclass of rtl_opt_pass along with...
1014 (pass_data_rtl_fwprop): ...new pass_data instance and...
1015 (make_pass_rtl_fwprop): ...new function.
1016 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1017 rtl_opt_pass along with...
1018 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1019 (make_pass_rtl_fwprop_addr): ...new function.
1020 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1021 rtl_opt_pass along with...
1022 (pass_data_rtl_pre): ...new pass_data instance and...
1023 (make_pass_rtl_pre): ...new function.
1024 (pass_rtl_hoist): Convert from a global struct to a subclass of
1025 rtl_opt_pass along with...
1026 (pass_data_rtl_hoist): ...new pass_data instance and...
1027 (make_pass_rtl_hoist): ...new function.
1028 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1029 subclass of gimple_opt_pass along with...
1030 (pass_data_lower_cf): ...new pass_data instance and...
1031 (make_pass_lower_cf): ...new function.
1032 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1033 from a global struct to a subclass of gimple_opt_pass along with...
1034 (pass_data_strength_reduction): ...new pass_data instance and...
1035 (make_pass_strength_reduction): ...new function.
1036 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1037 of rtl_opt_pass along with...
1038 (pass_data_rtl_ifcvt): ...new pass_data instance and...
1039 (make_pass_rtl_ifcvt): ...new function.
1040 (pass_if_after_combine): Convert from a global struct to a subclass of
1041 rtl_opt_pass along with...
1042 (pass_data_if_after_combine): ...new pass_data instance and...
1043 (make_pass_if_after_combine): ...new function.
1044 (pass_if_after_reload): Convert from a global struct to a subclass of
1045 rtl_opt_pass along with...
1046 (pass_data_if_after_reload): ...new pass_data instance and...
1047 (make_pass_if_after_reload): ...new function.
1048 * init-regs.c (pass_initialize_regs): Convert from a global struct to
1049 a subclass of rtl_opt_pass along with...
1050 (pass_data_initialize_regs): ...new pass_data instance and...
1051 (make_pass_initialize_regs): ...new function.
1052 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1053 of ipa_opt_pass_d along with...
1054 (pass_data_ipa_cp): ...new pass_data instance and...
1055 (make_pass_ipa_cp): ...new function.
1056 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1057 global struct to a subclass of gimple_opt_pass along with...
1058 (pass_data_inline_parameters): ...new pass_data instance and...
1059 (make_pass_inline_parameters): ...new function.
1060 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1061 subclass of gimple_opt_pass along with...
1062 (pass_data_early_inline): ...new pass_data instance and...
1063 (make_pass_early_inline): ...new function.
1064 (pass_ipa_inline): Convert from a global struct to a subclass of
1065 ipa_opt_pass_d along with...
1066 (pass_data_ipa_inline): ...new pass_data instance and...
1067 (make_pass_ipa_inline): ...new function.
1068 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1069 struct to a subclass of gimple_opt_pass along with...
1070 (pass_data_local_pure_const): ...new pass_data instance and...
1071 (make_pass_local_pure_const): ...new function.
1072 (pass_ipa_pure_const): Convert from a global struct to a subclass of
1073 ipa_opt_pass_d along with...
1074 (pass_data_ipa_pure_const): ...new pass_data instance and...
1075 (make_pass_ipa_pure_const): ...new function.
1076 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1077 to a subclass of ipa_opt_pass_d along with...
1078 (pass_data_ipa_reference): ...new pass_data instance and...
1079 (make_pass_ipa_reference): ...new function.
1080 * ipa-split.c (pass_split_functions): Convert from a global struct to
1081 a subclass of gimple_opt_pass along with...
1082 (pass_data_split_functions): ...new pass_data instance and...
1083 (make_pass_split_functions): ...new function.
1084 (pass_feedback_split_functions): Convert from a global struct to a
1085 subclass of gimple_opt_pass along with...
1086 (pass_data_feedback_split_functions): ...new pass_data instance and...
1087 (make_pass_feedback_split_functions): ...new function.
1088 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1089 global struct to a subclass of simple_ipa_opt_pass along with...
1090 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1092 (make_pass_ipa_function_and_variable_visibility): ...new function.
1093 (pass_ipa_free_inline_summary): Convert from a global struct to a
1094 subclass of simple_ipa_opt_pass along with...
1095 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1096 (make_pass_ipa_free_inline_summary): ...new function.
1097 (pass_ipa_whole_program_visibility): Convert from a global struct to a
1098 subclass of ipa_opt_pass_d along with...
1099 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1101 (make_pass_ipa_whole_program_visibility): ...new function.
1102 (pass_ipa_profile): Convert from a global struct to a subclass of
1103 ipa_opt_pass_d along with...
1104 (pass_data_ipa_profile): ...new pass_data instance and...
1105 (make_pass_ipa_profile): ...new function.
1106 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1107 ipa_opt_pass_d along with...
1108 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1109 (make_pass_ipa_cdtor_merge): ...new function.
1110 * ira.c (pass_ira): Convert from a global struct to a subclass of
1111 rtl_opt_pass along with...
1112 (pass_data_ira): ...new pass_data instance and...
1113 (make_pass_ira): ...new function.
1114 (pass_reload): Convert from a global struct to a subclass of
1115 rtl_opt_pass along with...
1116 (pass_data_reload): ...new pass_data instance and...
1117 (make_pass_reload): ...new function.
1118 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1119 subclass of rtl_opt_pass along with...
1120 (pass_data_cleanup_barriers): ...new pass_data instance and...
1121 (make_pass_cleanup_barriers): ...new function.
1122 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1123 of rtl_opt_pass along with...
1124 (pass_data_loop2): ...new pass_data instance and...
1125 (make_pass_loop2): ...new function.
1126 (pass_rtl_loop_init): Convert from a global struct to a subclass of
1127 rtl_opt_pass along with...
1128 (pass_data_rtl_loop_init): ...new pass_data instance and...
1129 (make_pass_rtl_loop_init): ...new function.
1130 (pass_rtl_loop_done): Convert from a global struct to a subclass of
1131 rtl_opt_pass along with...
1132 (pass_data_rtl_loop_done): ...new pass_data instance and...
1133 (make_pass_rtl_loop_done): ...new function.
1134 (pass_rtl_move_loop_invariants): Convert from a global struct to a
1135 subclass of rtl_opt_pass along with...
1136 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1137 (make_pass_rtl_move_loop_invariants): ...new function.
1138 (pass_rtl_unswitch): Convert from a global struct to a subclass of
1139 rtl_opt_pass along with...
1140 (pass_data_rtl_unswitch): ...new pass_data instance and...
1141 (make_pass_rtl_unswitch): ...new function.
1142 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1143 subclass of rtl_opt_pass along with...
1144 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1146 (make_pass_rtl_unroll_and_peel_loops): ...new function.
1147 (pass_rtl_doloop): Convert from a global struct to a subclass of
1148 rtl_opt_pass along with...
1149 (pass_data_rtl_doloop): ...new pass_data instance and...
1150 (make_pass_rtl_doloop): ...new function.
1151 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1152 a subclass of rtl_opt_pass along with...
1153 (pass_data_lower_subreg): ...new pass_data instance and...
1154 (make_pass_lower_subreg): ...new function.
1155 (pass_lower_subreg2): Convert from a global struct to a subclass of
1156 rtl_opt_pass along with...
1157 (pass_data_lower_subreg2): ...new pass_data instance and...
1158 (make_pass_lower_subreg2): ...new function.
1159 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1160 struct to a subclass of ipa_opt_pass_d along with...
1161 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1162 (make_pass_ipa_lto_gimple_out): ...new function.
1163 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1164 of ipa_opt_pass_d along with...
1165 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1166 (make_pass_ipa_lto_finish_out): ...new function.
1167 * mode-switching.c (pass_mode_switching): Convert from a global struct
1168 to a subclass of rtl_opt_pass along with...
1169 (pass_data_mode_switching): ...new pass_data instance and...
1170 (make_pass_mode_switching): ...new function.
1171 * modulo-sched.c (pass_sms): Convert from a global struct to a
1172 subclass of rtl_opt_pass along with...
1173 (pass_data_sms): ...new pass_data instance and...
1174 (make_pass_sms): ...new function.
1175 * omp-low.c (pass_expand_omp): Convert from a global struct to a
1176 subclass of gimple_opt_pass along with...
1177 (pass_data_expand_omp): ...new pass_data instance and...
1178 (make_pass_expand_omp): ...new function.
1179 (pass_lower_omp): Convert from a global struct to a subclass of
1180 gimple_opt_pass along with...
1181 (pass_data_lower_omp): ...new pass_data instance and...
1182 (make_pass_lower_omp): ...new function.
1183 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1184 of gimple_opt_pass along with...
1185 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1186 (make_pass_diagnose_omp_blocks): ...new function.
1187 * passes.c (pass_early_local_passes): Convert from a global struct to
1188 a subclass of simple_ipa_opt_pass along with...
1189 (pass_data_early_local_passes): ...new pass_data instance and...
1190 (make_pass_early_local_passes): ...new function.
1191 (pass_all_early_optimizations): Convert from a global struct to a
1192 subclass of gimple_opt_pass along with...
1193 (pass_data_all_early_optimizations): ...new pass_data instance and...
1194 (make_pass_all_early_optimizations): ...new function.
1195 (pass_all_optimizations): Convert from a global struct to a subclass
1196 of gimple_opt_pass along with...
1197 (pass_data_all_optimizations): ...new pass_data instance and...
1198 (make_pass_all_optimizations): ...new function.
1199 (pass_all_optimizations_g): Convert from a global struct to a subclass
1200 of gimple_opt_pass along with...
1201 (pass_data_all_optimizations_g): ...new pass_data instance and...
1202 (make_pass_all_optimizations_g): ...new function.
1203 (pass_rest_of_compilation): Convert from a global struct to a subclass
1204 of rtl_opt_pass along with...
1205 (pass_data_rest_of_compilation): ...new pass_data instance and...
1206 (make_pass_rest_of_compilation): ...new function.
1207 (pass_postreload): Convert from a global struct to a subclass of
1208 rtl_opt_pass along with...
1209 (pass_data_postreload): ...new pass_data instance and...
1210 (make_pass_postreload): ...new function.
1211 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1212 subclass of rtl_opt_pass along with...
1213 (pass_data_gcse2): ...new pass_data instance and...
1214 (make_pass_gcse2): ...new function.
1215 * postreload.c (pass_postreload_cse): Convert from a global struct to
1216 a subclass of rtl_opt_pass along with...
1217 (pass_data_postreload_cse): ...new pass_data instance and...
1218 (make_pass_postreload_cse): ...new function.
1219 * predict.c (pass_profile): Convert from a global struct to a subclass
1220 of gimple_opt_pass along with...
1221 (pass_data_profile): ...new pass_data instance and...
1222 (make_pass_profile): ...new function.
1223 (pass_strip_predict_hints): Convert from a global struct to a subclass
1224 of gimple_opt_pass along with...
1225 (pass_data_strip_predict_hints): ...new pass_data instance and...
1226 (make_pass_strip_predict_hints): ...new function.
1227 * recog.c (pass_peephole2): Convert from a global struct to a subclass
1228 of rtl_opt_pass along with...
1229 (pass_data_peephole2): ...new pass_data instance and...
1230 (make_pass_peephole2): ...new function.
1231 (pass_split_all_insns): Convert from a global struct to a subclass of
1232 rtl_opt_pass along with...
1233 (pass_data_split_all_insns): ...new pass_data instance and...
1234 (make_pass_split_all_insns): ...new function.
1235 (pass_split_after_reload): Convert from a global struct to a subclass
1236 of rtl_opt_pass along with...
1237 (pass_data_split_after_reload): ...new pass_data instance and...
1238 (make_pass_split_after_reload): ...new function.
1239 (pass_split_before_regstack): Convert from a global struct to a
1240 subclass of rtl_opt_pass along with...
1241 (pass_data_split_before_regstack): ...new pass_data instance and...
1242 (make_pass_split_before_regstack): ...new function.
1243 (pass_split_before_sched2): Convert from a global struct to a subclass
1244 of rtl_opt_pass along with...
1245 (pass_data_split_before_sched2): ...new pass_data instance and...
1246 (make_pass_split_before_sched2): ...new function.
1247 (pass_split_for_shorten_branches): Convert from a global struct to a
1248 subclass of rtl_opt_pass along with...
1249 (pass_data_split_for_shorten_branches): ...new pass_data instance
1251 (make_pass_split_for_shorten_branches): ...new function.
1252 * ree.c (pass_ree): Convert from a global struct to a subclass of
1253 rtl_opt_pass along with...
1254 (pass_data_ree): ...new pass_data instance and...
1255 (make_pass_ree): ...new function.
1256 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1257 subclass of rtl_opt_pass along with...
1258 (pass_data_stack_regs): ...new pass_data instance and...
1259 (make_pass_stack_regs): ...new function.
1260 (pass_stack_regs_run): Convert from a global struct to a subclass of
1261 rtl_opt_pass along with...
1262 (pass_data_stack_regs_run): ...new pass_data instance and...
1263 (make_pass_stack_regs_run): ...new function.
1264 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1265 subclass of rtl_opt_pass along with...
1266 (pass_data_cprop_hardreg): ...new pass_data instance and...
1267 (make_pass_cprop_hardreg): ...new function.
1268 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1269 subclass of rtl_opt_pass along with...
1270 (pass_data_reginfo_init): ...new pass_data instance and...
1271 (make_pass_reginfo_init): ...new function.
1272 * regmove.c (pass_regmove): Convert from a global struct to a subclass
1273 of rtl_opt_pass along with...
1274 (pass_data_regmove): ...new pass_data instance and...
1275 (make_pass_regmove): ...new function.
1276 * regrename.c (pass_regrename): Convert from a global struct to a
1277 subclass of rtl_opt_pass along with...
1278 (pass_data_regrename): ...new pass_data instance and...
1279 (make_pass_regrename): ...new function.
1280 * reorg.c (pass_delay_slots): Convert from a global struct to a
1281 subclass of rtl_opt_pass along with...
1282 (pass_data_delay_slots): ...new pass_data instance and...
1283 (make_pass_delay_slots): ...new function.
1284 (pass_machine_reorg): Convert from a global struct to a subclass of
1285 rtl_opt_pass along with...
1286 (pass_data_machine_reorg): ...new pass_data instance and...
1287 (make_pass_machine_reorg): ...new function.
1288 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1289 of rtl_opt_pass along with...
1290 (pass_data_sched): ...new pass_data instance and...
1291 (make_pass_sched): ...new function.
1292 (pass_sched2): Convert from a global struct to a subclass of
1293 rtl_opt_pass along with...
1294 (pass_data_sched2): ...new pass_data instance and...
1295 (make_pass_sched2): ...new function.
1296 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1297 to a subclass of rtl_opt_pass along with...
1298 (pass_data_stack_ptr_mod): ...new pass_data instance and...
1299 (make_pass_stack_ptr_mod): ...new function.
1300 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1301 to a subclass of rtl_opt_pass along with...
1302 (pass_data_rtl_store_motion): ...new pass_data instance and...
1303 (make_pass_rtl_store_motion): ...new function.
1304 * tracer.c (pass_tracer): Convert from a global struct to a subclass
1305 of gimple_opt_pass along with...
1306 (pass_data_tracer): ...new pass_data instance and...
1307 (make_pass_tracer): ...new function.
1308 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1309 to a subclass of gimple_opt_pass along with...
1310 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1311 (make_pass_diagnose_tm_blocks): ...new function.
1312 (pass_lower_tm): Convert from a global struct to a subclass of
1313 gimple_opt_pass along with...
1314 (pass_data_lower_tm): ...new pass_data instance and...
1315 (make_pass_lower_tm): ...new function.
1316 (pass_tm_init): Convert from a global struct to a subclass of
1317 gimple_opt_pass along with...
1318 (pass_data_tm_init): ...new pass_data instance and...
1319 (make_pass_tm_init): ...new function.
1320 (pass_tm_mark): Convert from a global struct to a subclass of
1321 gimple_opt_pass along with...
1322 (pass_data_tm_mark): ...new pass_data instance and...
1323 (make_pass_tm_mark): ...new function.
1324 (pass_tm_edges): Convert from a global struct to a subclass of
1325 gimple_opt_pass along with...
1326 (pass_data_tm_edges): ...new pass_data instance and...
1327 (make_pass_tm_edges): ...new function.
1328 (pass_tm_memopt): Convert from a global struct to a subclass of
1329 gimple_opt_pass along with...
1330 (pass_data_tm_memopt): ...new pass_data instance and...
1331 (make_pass_tm_memopt): ...new function.
1332 (pass_ipa_tm): Convert from a global struct to a subclass of
1333 simple_ipa_opt_pass along with...
1334 (pass_data_ipa_tm): ...new pass_data instance and...
1335 (make_pass_ipa_tm): ...new function.
1336 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1337 subclass of gimple_opt_pass along with...
1338 (pass_data_call_cdce): ...new pass_data instance and...
1339 (make_pass_call_cdce): ...new function.
1340 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1341 subclass of gimple_opt_pass along with...
1342 (pass_data_build_cfg): ...new pass_data instance and...
1343 (make_pass_build_cfg): ...new function.
1344 (pass_split_crit_edges): Convert from a global struct to a subclass of
1345 gimple_opt_pass along with...
1346 (pass_data_split_crit_edges): ...new pass_data instance and...
1347 (make_pass_split_crit_edges): ...new function.
1348 (pass_warn_function_return): Convert from a global struct to a
1349 subclass of gimple_opt_pass along with...
1350 (pass_data_warn_function_return): ...new pass_data instance and...
1351 (make_pass_warn_function_return): ...new function.
1352 (pass_warn_function_noreturn): Convert from a global struct to a
1353 subclass of gimple_opt_pass along with...
1354 (pass_data_warn_function_noreturn): ...new pass_data instance and...
1355 (make_pass_warn_function_noreturn): ...new function.
1356 (pass_warn_unused_result): Convert from a global struct to a subclass
1357 of gimple_opt_pass along with...
1358 (pass_data_warn_unused_result): ...new pass_data instance and...
1359 (make_pass_warn_unused_result): ...new function.
1360 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1361 a subclass of gimple_opt_pass along with...
1362 (pass_data_merge_phi): ...new pass_data instance and...
1363 (make_pass_merge_phi): ...new function.
1364 * tree-complex.c (pass_lower_complex): Convert from a global struct to
1365 a subclass of gimple_opt_pass along with...
1366 (pass_data_lower_complex): ...new pass_data instance and...
1367 (make_pass_lower_complex): ...new function.
1368 (pass_lower_complex_O0): Convert from a global struct to a subclass of
1369 gimple_opt_pass along with...
1370 (pass_data_lower_complex_O0): ...new pass_data instance and...
1371 (make_pass_lower_complex_O0): ...new function.
1372 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1373 subclass of gimple_opt_pass along with...
1374 (pass_data_lower_eh): ...new pass_data instance and...
1375 (make_pass_lower_eh): ...new function.
1376 (pass_refactor_eh): Convert from a global struct to a subclass of
1377 gimple_opt_pass along with...
1378 (pass_data_refactor_eh): ...new pass_data instance and...
1379 (make_pass_refactor_eh): ...new function.
1380 (pass_lower_resx): Convert from a global struct to a subclass of
1381 gimple_opt_pass along with...
1382 (pass_data_lower_resx): ...new pass_data instance and...
1383 (make_pass_lower_resx): ...new function.
1384 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1385 of gimple_opt_pass along with...
1386 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1387 (make_pass_lower_eh_dispatch): ...new function.
1388 (pass_cleanup_eh): Convert from a global struct to a subclass of
1389 gimple_opt_pass along with...
1390 (pass_data_cleanup_eh): ...new pass_data instance and...
1391 (make_pass_cleanup_eh): ...new function.
1392 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1393 to a subclass of simple_ipa_opt_pass along with...
1394 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1395 (make_pass_ipa_lower_emutls): ...new function.
1396 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1397 a subclass of gimple_opt_pass along with...
1398 (pass_data_if_conversion): ...new pass_data instance and...
1399 (make_pass_if_conversion): ...new function.
1400 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1401 subclass of gimple_opt_pass along with...
1402 (pass_data_build_ssa): ...new pass_data instance and...
1403 (make_pass_build_ssa): ...new function.
1404 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1405 global struct to a subclass of gimple_opt_pass along with...
1406 (pass_data_loop_distribution): ...new pass_data instance and...
1407 (make_pass_loop_distribution): ...new function.
1408 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1409 subclass of gimple_opt_pass along with...
1410 (pass_data_mudflap_1): ...new pass_data instance and...
1411 (make_pass_mudflap_1): ...new function.
1412 (pass_mudflap_2): Convert from a global struct to a subclass of
1413 gimple_opt_pass along with...
1414 (pass_data_mudflap_2): ...new pass_data instance and...
1415 (make_pass_mudflap_2): ...new function.
1416 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1417 subclass of gimple_opt_pass along with...
1418 (pass_data_mudflap_1): ...new pass_data instance and...
1419 (make_pass_mudflap_1): ...new function.
1420 (pass_mudflap_2): Convert from a global struct to a subclass of
1421 gimple_opt_pass along with...
1422 (pass_data_mudflap_2): ...new pass_data instance and...
1423 (make_pass_mudflap_2): ...new function.
1424 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1425 gimple_opt_pass along with...
1426 (pass_data_nrv): ...new pass_data instance and...
1427 (make_pass_nrv): ...new function.
1428 (pass_return_slot): Convert from a global struct to a subclass of
1429 gimple_opt_pass along with...
1430 (pass_data_return_slot): ...new pass_data instance and...
1431 (make_pass_return_slot): ...new function.
1432 * tree-object-size.c (pass_object_sizes): Convert from a global struct
1433 to a subclass of gimple_opt_pass along with...
1434 (pass_data_object_sizes): ...new pass_data instance and...
1435 (make_pass_object_sizes): ...new function.
1436 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1437 global struct to a subclass of gimple_opt_pass along with...
1438 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1440 (make_pass_cleanup_cfg_post_optimizing): ...new function.
1441 (pass_fixup_cfg): Convert from a global struct to a subclass of
1442 gimple_opt_pass along with...
1443 (pass_data_fixup_cfg): ...new pass_data instance and...
1444 (make_pass_fixup_cfg): ...new function.
1445 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1446 (make_pass_mudflap_1): ...new function.
1447 (pass_mudflap_2): Replace declaration with that of...
1448 (make_pass_mudflap_2): ...new function.
1449 (pass_asan): Replace declaration with that of...
1450 (make_pass_asan): ...new function.
1451 (pass_asan_O0): Replace declaration with that of...
1452 (make_pass_asan_O0): ...new function.
1453 (pass_tsan): Replace declaration with that of...
1454 (make_pass_tsan): ...new function.
1455 (pass_tsan_O0): Replace declaration with that of...
1456 (make_pass_tsan_O0): ...new function.
1457 (pass_lower_cf): Replace declaration with that of...
1458 (make_pass_lower_cf): ...new function.
1459 (pass_refactor_eh): Replace declaration with that of...
1460 (make_pass_refactor_eh): ...new function.
1461 (pass_lower_eh): Replace declaration with that of...
1462 (make_pass_lower_eh): ...new function.
1463 (pass_lower_eh_dispatch): Replace declaration with that of...
1464 (make_pass_lower_eh_dispatch): ...new function.
1465 (pass_lower_resx): Replace declaration with that of...
1466 (make_pass_lower_resx): ...new function.
1467 (pass_build_cfg): Replace declaration with that of...
1468 (make_pass_build_cfg): ...new function.
1469 (pass_early_tree_profile): Replace declaration with that of...
1470 (make_pass_early_tree_profile): ...new function.
1471 (pass_cleanup_eh): Replace declaration with that of...
1472 (make_pass_cleanup_eh): ...new function.
1473 (pass_sra): Replace declaration with that of...
1474 (make_pass_sra): ...new function.
1475 (pass_sra_early): Replace declaration with that of...
1476 (make_pass_sra_early): ...new function.
1477 (pass_early_ipa_sra): Replace declaration with that of...
1478 (make_pass_early_ipa_sra): ...new function.
1479 (pass_tail_recursion): Replace declaration with that of...
1480 (make_pass_tail_recursion): ...new function.
1481 (pass_tail_calls): Replace declaration with that of...
1482 (make_pass_tail_calls): ...new function.
1483 (pass_tree_loop): Replace declaration with that of...
1484 (make_pass_tree_loop): ...new function.
1485 (pass_tree_loop_init): Replace declaration with that of...
1486 (make_pass_tree_loop_init): ...new function.
1487 (pass_lim): Replace declaration with that of...
1488 (make_pass_lim): ...new function.
1489 (pass_tree_unswitch): Replace declaration with that of...
1490 (make_pass_tree_unswitch): ...new function.
1491 (pass_predcom): Replace declaration with that of...
1492 (make_pass_predcom): ...new function.
1493 (pass_iv_canon): Replace declaration with that of...
1494 (make_pass_iv_canon): ...new function.
1495 (pass_scev_cprop): Replace declaration with that of...
1496 (make_pass_scev_cprop): ...new function.
1497 (pass_empty_loop): Replace declaration with that of...
1498 (make_pass_empty_loop): ...new function.
1499 (pass_record_bounds): Replace declaration with that of...
1500 (make_pass_record_bounds): ...new function.
1501 (pass_graphite): Replace declaration with that of...
1502 (make_pass_graphite): ...new function.
1503 (pass_graphite_transforms): Replace declaration with that of...
1504 (make_pass_graphite_transforms): ...new function.
1505 (pass_if_conversion): Replace declaration with that of...
1506 (make_pass_if_conversion): ...new function.
1507 (pass_loop_distribution): Replace declaration with that of...
1508 (make_pass_loop_distribution): ...new function.
1509 (pass_vectorize): Replace declaration with that of...
1510 (make_pass_vectorize): ...new function.
1511 (pass_slp_vectorize): Replace declaration with that of...
1512 (make_pass_slp_vectorize): ...new function.
1513 (pass_complete_unroll): Replace declaration with that of...
1514 (make_pass_complete_unroll): ...new function.
1515 (pass_complete_unrolli): Replace declaration with that of...
1516 (make_pass_complete_unrolli): ...new function.
1517 (pass_parallelize_loops): Replace declaration with that of...
1518 (make_pass_parallelize_loops): ...new function.
1519 (pass_loop_prefetch): Replace declaration with that of...
1520 (make_pass_loop_prefetch): ...new function.
1521 (pass_iv_optimize): Replace declaration with that of...
1522 (make_pass_iv_optimize): ...new function.
1523 (pass_tree_loop_done): Replace declaration with that of...
1524 (make_pass_tree_loop_done): ...new function.
1525 (pass_ch): Replace declaration with that of...
1526 (make_pass_ch): ...new function.
1527 (pass_ccp): Replace declaration with that of...
1528 (make_pass_ccp): ...new function.
1529 (pass_phi_only_cprop): Replace declaration with that of...
1530 (make_pass_phi_only_cprop): ...new function.
1531 (pass_build_ssa): Replace declaration with that of...
1532 (make_pass_build_ssa): ...new function.
1533 (pass_build_alias): Replace declaration with that of...
1534 (make_pass_build_alias): ...new function.
1535 (pass_build_ealias): Replace declaration with that of...
1536 (make_pass_build_ealias): ...new function.
1537 (pass_dominator): Replace declaration with that of...
1538 (make_pass_dominator): ...new function.
1539 (pass_dce): Replace declaration with that of...
1540 (make_pass_dce): ...new function.
1541 (pass_dce_loop): Replace declaration with that of...
1542 (make_pass_dce_loop): ...new function.
1543 (pass_cd_dce): Replace declaration with that of...
1544 (make_pass_cd_dce): ...new function.
1545 (pass_call_cdce): Replace declaration with that of...
1546 (make_pass_call_cdce): ...new function.
1547 (pass_merge_phi): Replace declaration with that of...
1548 (make_pass_merge_phi): ...new function.
1549 (pass_split_crit_edges): Replace declaration with that of...
1550 (make_pass_split_crit_edges): ...new function.
1551 (pass_pre): Replace declaration with that of...
1552 (make_pass_pre): ...new function.
1553 (pass_profile): Replace declaration with that of...
1554 (make_pass_profile): ...new function.
1555 (pass_strip_predict_hints): Replace declaration with that of...
1556 (make_pass_strip_predict_hints): ...new function.
1557 (pass_lower_complex_O0): Replace declaration with that of...
1558 (make_pass_lower_complex_O0): ...new function.
1559 (pass_lower_complex): Replace declaration with that of...
1560 (make_pass_lower_complex): ...new function.
1561 (pass_lower_vector): Replace declaration with that of...
1562 (make_pass_lower_vector): ...new function.
1563 (pass_lower_vector_ssa): Replace declaration with that of...
1564 (make_pass_lower_vector_ssa): ...new function.
1565 (pass_lower_omp): Replace declaration with that of...
1566 (make_pass_lower_omp): ...new function.
1567 (pass_diagnose_omp_blocks): Replace declaration with that of...
1568 (make_pass_diagnose_omp_blocks): ...new function.
1569 (pass_expand_omp): Replace declaration with that of...
1570 (make_pass_expand_omp): ...new function.
1571 (pass_expand_omp_ssa): Replace declaration with that of...
1572 (make_pass_expand_omp_ssa): ...new function.
1573 (pass_object_sizes): Replace declaration with that of...
1574 (make_pass_object_sizes): ...new function.
1575 (pass_strlen): Replace declaration with that of...
1576 (make_pass_strlen): ...new function.
1577 (pass_fold_builtins): Replace declaration with that of...
1578 (make_pass_fold_builtins): ...new function.
1579 (pass_stdarg): Replace declaration with that of...
1580 (make_pass_stdarg): ...new function.
1581 (pass_early_warn_uninitialized): Replace declaration with that of...
1582 (make_pass_early_warn_uninitialized): ...new function.
1583 (pass_late_warn_uninitialized): Replace declaration with that of...
1584 (make_pass_late_warn_uninitialized): ...new function.
1585 (pass_cse_reciprocals): Replace declaration with that of...
1586 (make_pass_cse_reciprocals): ...new function.
1587 (pass_cse_sincos): Replace declaration with that of...
1588 (make_pass_cse_sincos): ...new function.
1589 (pass_optimize_bswap): Replace declaration with that of...
1590 (make_pass_optimize_bswap): ...new function.
1591 (pass_optimize_widening_mul): Replace declaration with that of...
1592 (make_pass_optimize_widening_mul): ...new function.
1593 (pass_warn_function_return): Replace declaration with that of...
1594 (make_pass_warn_function_return): ...new function.
1595 (pass_warn_function_noreturn): Replace declaration with that of...
1596 (make_pass_warn_function_noreturn): ...new function.
1597 (pass_cselim): Replace declaration with that of...
1598 (make_pass_cselim): ...new function.
1599 (pass_phiopt): Replace declaration with that of...
1600 (make_pass_phiopt): ...new function.
1601 (pass_forwprop): Replace declaration with that of...
1602 (make_pass_forwprop): ...new function.
1603 (pass_phiprop): Replace declaration with that of...
1604 (make_pass_phiprop): ...new function.
1605 (pass_tree_ifcombine): Replace declaration with that of...
1606 (make_pass_tree_ifcombine): ...new function.
1607 (pass_dse): Replace declaration with that of...
1608 (make_pass_dse): ...new function.
1609 (pass_nrv): Replace declaration with that of...
1610 (make_pass_nrv): ...new function.
1611 (pass_rename_ssa_copies): Replace declaration with that of...
1612 (make_pass_rename_ssa_copies): ...new function.
1613 (pass_sink_code): Replace declaration with that of...
1614 (make_pass_sink_code): ...new function.
1615 (pass_fre): Replace declaration with that of...
1616 (make_pass_fre): ...new function.
1617 (pass_check_data_deps): Replace declaration with that of...
1618 (make_pass_check_data_deps): ...new function.
1619 (pass_copy_prop): Replace declaration with that of...
1620 (make_pass_copy_prop): ...new function.
1621 (pass_vrp): Replace declaration with that of...
1622 (make_pass_vrp): ...new function.
1623 (pass_uncprop): Replace declaration with that of...
1624 (make_pass_uncprop): ...new function.
1625 (pass_return_slot): Replace declaration with that of...
1626 (make_pass_return_slot): ...new function.
1627 (pass_reassoc): Replace declaration with that of...
1628 (make_pass_reassoc): ...new function.
1629 (pass_rebuild_cgraph_edges): Replace declaration with that of...
1630 (make_pass_rebuild_cgraph_edges): ...new function.
1631 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1632 (make_pass_remove_cgraph_callee_edges): ...new function.
1633 (pass_build_cgraph_edges): Replace declaration with that of...
1634 (make_pass_build_cgraph_edges): ...new function.
1635 (pass_local_pure_const): Replace declaration with that of...
1636 (make_pass_local_pure_const): ...new function.
1637 (pass_tracer): Replace declaration with that of...
1638 (make_pass_tracer): ...new function.
1639 (pass_warn_unused_result): Replace declaration with that of...
1640 (make_pass_warn_unused_result): ...new function.
1641 (pass_diagnose_tm_blocks): Replace declaration with that of...
1642 (make_pass_diagnose_tm_blocks): ...new function.
1643 (pass_lower_tm): Replace declaration with that of...
1644 (make_pass_lower_tm): ...new function.
1645 (pass_tm_init): Replace declaration with that of...
1646 (make_pass_tm_init): ...new function.
1647 (pass_tm_mark): Replace declaration with that of...
1648 (make_pass_tm_mark): ...new function.
1649 (pass_tm_memopt): Replace declaration with that of...
1650 (make_pass_tm_memopt): ...new function.
1651 (pass_tm_edges): Replace declaration with that of...
1652 (make_pass_tm_edges): ...new function.
1653 (pass_split_functions): Replace declaration with that of...
1654 (make_pass_split_functions): ...new function.
1655 (pass_feedback_split_functions): Replace declaration with that of...
1656 (make_pass_feedback_split_functions): ...new function.
1657 (pass_strength_reduction): Replace declaration with that of...
1658 (make_pass_strength_reduction): ...new function.
1659 (pass_ipa_lower_emutls): Replace declaration with that of...
1660 (make_pass_ipa_lower_emutls): ...new function.
1661 (pass_ipa_function_and_variable_visibility): Replace declaration with
1663 (make_pass_ipa_function_and_variable_visibility): ...new function.
1664 (pass_ipa_tree_profile): Replace declaration with that of...
1665 (make_pass_ipa_tree_profile): ...new function.
1666 (pass_early_local_passes): Replace declaration with that of...
1667 (make_pass_early_local_passes): ...new function.
1668 (pass_ipa_whole_program_visibility): Replace declaration with that
1670 (make_pass_ipa_whole_program_visibility): ...new function.
1671 (pass_ipa_lto_gimple_out): Replace declaration with that of...
1672 (make_pass_ipa_lto_gimple_out): ...new function.
1673 (pass_ipa_increase_alignment): Replace declaration with that of...
1674 (make_pass_ipa_increase_alignment): ...new function.
1675 (pass_ipa_inline): Replace declaration with that of...
1676 (make_pass_ipa_inline): ...new function.
1677 (pass_ipa_free_lang_data): Replace declaration with that of...
1678 (make_pass_ipa_free_lang_data): ...new function.
1679 (pass_ipa_free_inline_summary): Replace declaration with that of...
1680 (make_pass_ipa_free_inline_summary): ...new function.
1681 (pass_ipa_cp): Replace declaration with that of...
1682 (make_pass_ipa_cp): ...new function.
1683 (pass_ipa_reference): Replace declaration with that of...
1684 (make_pass_ipa_reference): ...new function.
1685 (pass_ipa_pure_const): Replace declaration with that of...
1686 (make_pass_ipa_pure_const): ...new function.
1687 (pass_ipa_pta): Replace declaration with that of...
1688 (make_pass_ipa_pta): ...new function.
1689 (pass_ipa_lto_finish_out): Replace declaration with that of...
1690 (make_pass_ipa_lto_finish_out): ...new function.
1691 (pass_ipa_tm): Replace declaration with that of...
1692 (make_pass_ipa_tm): ...new function.
1693 (pass_ipa_profile): Replace declaration with that of...
1694 (make_pass_ipa_profile): ...new function.
1695 (pass_ipa_cdtor_merge): Replace declaration with that of...
1696 (make_pass_ipa_cdtor_merge): ...new function.
1697 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1699 (make_pass_cleanup_cfg_post_optimizing): ...new function.
1700 (pass_init_datastructures): Replace declaration with that of...
1701 (make_pass_init_datastructures): ...new function.
1702 (pass_fixup_cfg): Replace declaration with that of...
1703 (make_pass_fixup_cfg): ...new function.
1704 (pass_expand): Replace declaration with that of...
1705 (make_pass_expand): ...new function.
1706 (pass_instantiate_virtual_regs): Replace declaration with that of...
1707 (make_pass_instantiate_virtual_regs): ...new function.
1708 (pass_rtl_fwprop): Replace declaration with that of...
1709 (make_pass_rtl_fwprop): ...new function.
1710 (pass_rtl_fwprop_addr): Replace declaration with that of...
1711 (make_pass_rtl_fwprop_addr): ...new function.
1712 (pass_jump): Replace declaration with that of...
1713 (make_pass_jump): ...new function.
1714 (pass_jump2): Replace declaration with that of...
1715 (make_pass_jump2): ...new function.
1716 (pass_lower_subreg): Replace declaration with that of...
1717 (make_pass_lower_subreg): ...new function.
1718 (pass_cse): Replace declaration with that of...
1719 (make_pass_cse): ...new function.
1720 (pass_fast_rtl_dce): Replace declaration with that of...
1721 (make_pass_fast_rtl_dce): ...new function.
1722 (pass_ud_rtl_dce): Replace declaration with that of...
1723 (make_pass_ud_rtl_dce): ...new function.
1724 (pass_rtl_dce): Replace declaration with that of...
1725 (make_pass_rtl_dce): ...new function.
1726 (pass_rtl_dse1): Replace declaration with that of...
1727 (make_pass_rtl_dse1): ...new function.
1728 (pass_rtl_dse2): Replace declaration with that of...
1729 (make_pass_rtl_dse2): ...new function.
1730 (pass_rtl_dse3): Replace declaration with that of...
1731 (make_pass_rtl_dse3): ...new function.
1732 (pass_rtl_cprop): Replace declaration with that of...
1733 (make_pass_rtl_cprop): ...new function.
1734 (pass_rtl_pre): Replace declaration with that of...
1735 (make_pass_rtl_pre): ...new function.
1736 (pass_rtl_hoist): Replace declaration with that of...
1737 (make_pass_rtl_hoist): ...new function.
1738 (pass_rtl_store_motion): Replace declaration with that of...
1739 (make_pass_rtl_store_motion): ...new function.
1740 (pass_cse_after_global_opts): Replace declaration with that of...
1741 (make_pass_cse_after_global_opts): ...new function.
1742 (pass_rtl_ifcvt): Replace declaration with that of...
1743 (make_pass_rtl_ifcvt): ...new function.
1744 (pass_into_cfg_layout_mode): Replace declaration with that of...
1745 (make_pass_into_cfg_layout_mode): ...new function.
1746 (pass_outof_cfg_layout_mode): Replace declaration with that of...
1747 (make_pass_outof_cfg_layout_mode): ...new function.
1748 (pass_loop2): Replace declaration with that of...
1749 (make_pass_loop2): ...new function.
1750 (pass_rtl_loop_init): Replace declaration with that of...
1751 (make_pass_rtl_loop_init): ...new function.
1752 (pass_rtl_move_loop_invariants): Replace declaration with that of...
1753 (make_pass_rtl_move_loop_invariants): ...new function.
1754 (pass_rtl_unswitch): Replace declaration with that of...
1755 (make_pass_rtl_unswitch): ...new function.
1756 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1757 (make_pass_rtl_unroll_and_peel_loops): ...new function.
1758 (pass_rtl_doloop): Replace declaration with that of...
1759 (make_pass_rtl_doloop): ...new function.
1760 (pass_rtl_loop_done): Replace declaration with that of...
1761 (make_pass_rtl_loop_done): ...new function.
1762 (pass_web): Replace declaration with that of...
1763 (make_pass_web): ...new function.
1764 (pass_cse2): Replace declaration with that of...
1765 (make_pass_cse2): ...new function.
1766 (pass_df_initialize_opt): Replace declaration with that of...
1767 (make_pass_df_initialize_opt): ...new function.
1768 (pass_df_initialize_no_opt): Replace declaration with that of...
1769 (make_pass_df_initialize_no_opt): ...new function.
1770 (pass_reginfo_init): Replace declaration with that of...
1771 (make_pass_reginfo_init): ...new function.
1772 (pass_inc_dec): Replace declaration with that of...
1773 (make_pass_inc_dec): ...new function.
1774 (pass_stack_ptr_mod): Replace declaration with that of...
1775 (make_pass_stack_ptr_mod): ...new function.
1776 (pass_initialize_regs): Replace declaration with that of...
1777 (make_pass_initialize_regs): ...new function.
1778 (pass_combine): Replace declaration with that of...
1779 (make_pass_combine): ...new function.
1780 (pass_if_after_combine): Replace declaration with that of...
1781 (make_pass_if_after_combine): ...new function.
1782 (pass_ree): Replace declaration with that of...
1783 (make_pass_ree): ...new function.
1784 (pass_partition_blocks): Replace declaration with that of...
1785 (make_pass_partition_blocks): ...new function.
1786 (pass_match_asm_constraints): Replace declaration with that of...
1787 (make_pass_match_asm_constraints): ...new function.
1788 (pass_regmove): Replace declaration with that of...
1789 (make_pass_regmove): ...new function.
1790 (pass_split_all_insns): Replace declaration with that of...
1791 (make_pass_split_all_insns): ...new function.
1792 (pass_fast_rtl_byte_dce): Replace declaration with that of...
1793 (make_pass_fast_rtl_byte_dce): ...new function.
1794 (pass_lower_subreg2): Replace declaration with that of...
1795 (make_pass_lower_subreg2): ...new function.
1796 (pass_mode_switching): Replace declaration with that of...
1797 (make_pass_mode_switching): ...new function.
1798 (pass_sms): Replace declaration with that of...
1799 (make_pass_sms): ...new function.
1800 (pass_sched): Replace declaration with that of...
1801 (make_pass_sched): ...new function.
1802 (pass_ira): Replace declaration with that of...
1803 (make_pass_ira): ...new function.
1804 (pass_reload): Replace declaration with that of...
1805 (make_pass_reload): ...new function.
1806 (pass_clean_state): Replace declaration with that of...
1807 (make_pass_clean_state): ...new function.
1808 (pass_branch_prob): Replace declaration with that of...
1809 (make_pass_branch_prob): ...new function.
1810 (pass_value_profile_transformations): Replace declaration with that
1812 (make_pass_value_profile_transformations): ...new function.
1813 (pass_postreload_cse): Replace declaration with that of...
1814 (make_pass_postreload_cse): ...new function.
1815 (pass_gcse2): Replace declaration with that of...
1816 (make_pass_gcse2): ...new function.
1817 (pass_split_after_reload): Replace declaration with that of...
1818 (make_pass_split_after_reload): ...new function.
1819 (pass_branch_target_load_optimize1): Replace declaration with that
1821 (make_pass_branch_target_load_optimize1): ...new function.
1822 (pass_thread_prologue_and_epilogue): Replace declaration with that
1824 (make_pass_thread_prologue_and_epilogue): ...new function.
1825 (pass_stack_adjustments): Replace declaration with that of...
1826 (make_pass_stack_adjustments): ...new function.
1827 (pass_peephole2): Replace declaration with that of...
1828 (make_pass_peephole2): ...new function.
1829 (pass_if_after_reload): Replace declaration with that of...
1830 (make_pass_if_after_reload): ...new function.
1831 (pass_regrename): Replace declaration with that of...
1832 (make_pass_regrename): ...new function.
1833 (pass_cprop_hardreg): Replace declaration with that of...
1834 (make_pass_cprop_hardreg): ...new function.
1835 (pass_reorder_blocks): Replace declaration with that of...
1836 (make_pass_reorder_blocks): ...new function.
1837 (pass_branch_target_load_optimize2): Replace declaration with that
1839 (make_pass_branch_target_load_optimize2): ...new function.
1840 (pass_leaf_regs): Replace declaration with that of...
1841 (make_pass_leaf_regs): ...new function.
1842 (pass_split_before_sched2): Replace declaration with that of...
1843 (make_pass_split_before_sched2): ...new function.
1844 (pass_compare_elim_after_reload): Replace declaration with that of...
1845 (make_pass_compare_elim_after_reload): ...new function.
1846 (pass_sched2): Replace declaration with that of...
1847 (make_pass_sched2): ...new function.
1848 (pass_stack_regs): Replace declaration with that of...
1849 (make_pass_stack_regs): ...new function.
1850 (pass_stack_regs_run): Replace declaration with that of...
1851 (make_pass_stack_regs_run): ...new function.
1852 (pass_df_finish): Replace declaration with that of...
1853 (make_pass_df_finish): ...new function.
1854 (pass_compute_alignments): Replace declaration with that of...
1855 (make_pass_compute_alignments): ...new function.
1856 (pass_duplicate_computed_gotos): Replace declaration with that of...
1857 (make_pass_duplicate_computed_gotos): ...new function.
1858 (pass_variable_tracking): Replace declaration with that of...
1859 (make_pass_variable_tracking): ...new function.
1860 (pass_free_cfg): Replace declaration with that of...
1861 (make_pass_free_cfg): ...new function.
1862 (pass_machine_reorg): Replace declaration with that of...
1863 (make_pass_machine_reorg): ...new function.
1864 (pass_cleanup_barriers): Replace declaration with that of...
1865 (make_pass_cleanup_barriers): ...new function.
1866 (pass_delay_slots): Replace declaration with that of...
1867 (make_pass_delay_slots): ...new function.
1868 (pass_split_for_shorten_branches): Replace declaration with that of...
1869 (make_pass_split_for_shorten_branches): ...new function.
1870 (pass_split_before_regstack): Replace declaration with that of...
1871 (make_pass_split_before_regstack): ...new function.
1872 (pass_convert_to_eh_region_ranges): Replace declaration with that
1874 (make_pass_convert_to_eh_region_ranges): ...new function.
1875 (pass_shorten_branches): Replace declaration with that of...
1876 (make_pass_shorten_branches): ...new function.
1877 (pass_set_nothrow_function_flags): Replace declaration with that of...
1878 (make_pass_set_nothrow_function_flags): ...new function.
1879 (pass_dwarf2_frame): Replace declaration with that of...
1880 (make_pass_dwarf2_frame): ...new function.
1881 (pass_final): Replace declaration with that of...
1882 (make_pass_final): ...new function.
1883 (pass_rtl_seqabstr): Replace declaration with that of...
1884 (make_pass_rtl_seqabstr): ...new function.
1885 (pass_release_ssa_names): Replace declaration with that of...
1886 (make_pass_release_ssa_names): ...new function.
1887 (pass_early_inline): Replace declaration with that of...
1888 (make_pass_early_inline): ...new function.
1889 (pass_inline_parameters): Replace declaration with that of...
1890 (make_pass_inline_parameters): ...new function.
1891 (pass_update_address_taken): Replace declaration with that of...
1892 (make_pass_update_address_taken): ...new function.
1893 (pass_convert_switch): Replace declaration with that of...
1894 (make_pass_convert_switch): ...new function.
1895 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1896 to a subclass of simple_ipa_opt_pass along with...
1897 (pass_data_ipa_tree_profile): ...new pass_data instance and...
1898 (make_pass_ipa_tree_profile): ...new function.
1899 * tree-sra.c (pass_sra_early): Convert from a global struct to a
1900 subclass of gimple_opt_pass along with...
1901 (pass_data_sra_early): ...new pass_data instance and...
1902 (make_pass_sra_early): ...new function.
1903 (pass_sra): Convert from a global struct to a subclass of
1904 gimple_opt_pass along with...
1905 (pass_data_sra): ...new pass_data instance and...
1906 (make_pass_sra): ...new function.
1907 (pass_early_ipa_sra): Convert from a global struct to a subclass of
1908 gimple_opt_pass along with...
1909 (pass_data_early_ipa_sra): ...new pass_data instance and...
1910 (make_pass_early_ipa_sra): ...new function.
1911 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1912 subclass of gimple_opt_pass along with...
1913 (pass_data_ccp): ...new pass_data instance and...
1914 (make_pass_ccp): ...new function.
1915 (pass_fold_builtins): Convert from a global struct to a subclass of
1916 gimple_opt_pass along with...
1917 (pass_data_fold_builtins): ...new pass_data instance and...
1918 (make_pass_fold_builtins): ...new function.
1919 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1920 subclass of gimple_opt_pass along with...
1921 (pass_data_copy_prop): ...new pass_data instance and...
1922 (make_pass_copy_prop): ...new function.
1923 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1924 global struct to a subclass of gimple_opt_pass along with...
1925 (pass_data_rename_ssa_copies): ...new pass_data instance and...
1926 (make_pass_rename_ssa_copies): ...new function.
1927 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1928 subclass of gimple_opt_pass along with...
1929 (pass_data_dce): ...new pass_data instance and...
1930 (make_pass_dce): ...new function.
1931 (pass_dce_loop): Convert from a global struct to a subclass of
1932 gimple_opt_pass along with...
1933 (pass_data_dce_loop): ...new pass_data instance and...
1934 (make_pass_dce_loop): ...new function.
1935 (pass_cd_dce): Convert from a global struct to a subclass of
1936 gimple_opt_pass along with...
1937 (pass_data_cd_dce): ...new pass_data instance and...
1938 (make_pass_cd_dce): ...new function.
1939 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1940 subclass of gimple_opt_pass along with...
1941 (pass_data_dominator): ...new pass_data instance and...
1942 (make_pass_dominator): ...new function.
1943 (pass_phi_only_cprop): Convert from a global struct to a subclass of
1944 gimple_opt_pass along with...
1945 (pass_data_phi_only_cprop): ...new pass_data instance and...
1946 (make_pass_phi_only_cprop): ...new function.
1947 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1948 subclass of gimple_opt_pass along with...
1949 (pass_data_dse): ...new pass_data instance and...
1950 (make_pass_dse): ...new function.
1951 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1952 a subclass of gimple_opt_pass along with...
1953 (pass_data_forwprop): ...new pass_data instance and...
1954 (make_pass_forwprop): ...new function.
1955 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1956 struct to a subclass of gimple_opt_pass along with...
1957 (pass_data_tree_ifcombine): ...new pass_data instance and...
1958 (make_pass_tree_ifcombine): ...new function.
1959 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1960 subclass of gimple_opt_pass along with...
1961 (pass_data_ch): ...new pass_data instance and...
1962 (make_pass_ch): ...new function.
1963 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1964 subclass of gimple_opt_pass along with...
1965 (pass_data_tree_loop): ...new pass_data instance and...
1966 (make_pass_tree_loop): ...new function.
1967 (pass_tree_loop_init): Convert from a global struct to a subclass of
1968 gimple_opt_pass along with...
1969 (pass_data_tree_loop_init): ...new pass_data instance and...
1970 (make_pass_tree_loop_init): ...new function.
1971 (pass_lim): Convert from a global struct to a subclass of
1972 gimple_opt_pass along with...
1973 (pass_data_lim): ...new pass_data instance and...
1974 (make_pass_lim): ...new function.
1975 (pass_tree_unswitch): Convert from a global struct to a subclass of
1976 gimple_opt_pass along with...
1977 (pass_data_tree_unswitch): ...new pass_data instance and...
1978 (make_pass_tree_unswitch): ...new function.
1979 (pass_predcom): Convert from a global struct to a subclass of
1980 gimple_opt_pass along with...
1981 (pass_data_predcom): ...new pass_data instance and...
1982 (make_pass_predcom): ...new function.
1983 (pass_vectorize): Convert from a global struct to a subclass of
1984 gimple_opt_pass along with...
1985 (pass_data_vectorize): ...new pass_data instance and...
1986 (make_pass_vectorize): ...new function.
1987 (pass_graphite): Convert from a global struct to a subclass of
1988 gimple_opt_pass along with...
1989 (pass_data_graphite): ...new pass_data instance and...
1990 (make_pass_graphite): ...new function.
1991 (pass_graphite_transforms): Convert from a global struct to a subclass
1992 of gimple_opt_pass along with...
1993 (pass_data_graphite_transforms): ...new pass_data instance and...
1994 (make_pass_graphite_transforms): ...new function.
1995 (pass_check_data_deps): Convert from a global struct to a subclass of
1996 gimple_opt_pass along with...
1997 (pass_data_check_data_deps): ...new pass_data instance and...
1998 (make_pass_check_data_deps): ...new function.
1999 (pass_iv_canon): Convert from a global struct to a subclass of
2000 gimple_opt_pass along with...
2001 (pass_data_iv_canon): ...new pass_data instance and...
2002 (make_pass_iv_canon): ...new function.
2003 (pass_scev_cprop): Convert from a global struct to a subclass of
2004 gimple_opt_pass along with...
2005 (pass_data_scev_cprop): ...new pass_data instance and...
2006 (make_pass_scev_cprop): ...new function.
2007 (pass_record_bounds): Convert from a global struct to a subclass of
2008 gimple_opt_pass along with...
2009 (pass_data_record_bounds): ...new pass_data instance and...
2010 (make_pass_record_bounds): ...new function.
2011 (pass_complete_unroll): Convert from a global struct to a subclass of
2012 gimple_opt_pass along with...
2013 (pass_data_complete_unroll): ...new pass_data instance and...
2014 (make_pass_complete_unroll): ...new function.
2015 (pass_complete_unrolli): Convert from a global struct to a subclass of
2016 gimple_opt_pass along with...
2017 (pass_data_complete_unrolli): ...new pass_data instance and...
2018 (make_pass_complete_unrolli): ...new function.
2019 (pass_parallelize_loops): Convert from a global struct to a subclass
2020 of gimple_opt_pass along with...
2021 (pass_data_parallelize_loops): ...new pass_data instance and...
2022 (make_pass_parallelize_loops): ...new function.
2023 (pass_loop_prefetch): Convert from a global struct to a subclass of
2024 gimple_opt_pass along with...
2025 (pass_data_loop_prefetch): ...new pass_data instance and...
2026 (make_pass_loop_prefetch): ...new function.
2027 (pass_iv_optimize): Convert from a global struct to a subclass of
2028 gimple_opt_pass along with...
2029 (pass_data_iv_optimize): ...new pass_data instance and...
2030 (make_pass_iv_optimize): ...new function.
2031 (pass_tree_loop_done): Convert from a global struct to a subclass of
2032 gimple_opt_pass along with...
2033 (pass_data_tree_loop_done): ...new pass_data instance and...
2034 (make_pass_tree_loop_done): ...new function.
2035 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2036 struct to a subclass of gimple_opt_pass along with...
2037 (pass_data_cse_reciprocals): ...new pass_data instance and...
2038 (make_pass_cse_reciprocals): ...new function.
2039 (pass_cse_sincos): Convert from a global struct to a subclass of
2040 gimple_opt_pass along with...
2041 (pass_data_cse_sincos): ...new pass_data instance and...
2042 (make_pass_cse_sincos): ...new function.
2043 (pass_optimize_bswap): Convert from a global struct to a subclass of
2044 gimple_opt_pass along with...
2045 (pass_data_optimize_bswap): ...new pass_data instance and...
2046 (make_pass_optimize_bswap): ...new function.
2047 (pass_optimize_widening_mul): Convert from a global struct to a
2048 subclass of gimple_opt_pass along with...
2049 (pass_data_optimize_widening_mul): ...new pass_data instance and...
2050 (make_pass_optimize_widening_mul): ...new function.
2051 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2052 subclass of gimple_opt_pass along with...
2053 (pass_data_phiopt): ...new pass_data instance and...
2054 (make_pass_phiopt): ...new function.
2055 (pass_cselim): Convert from a global struct to a subclass of
2056 gimple_opt_pass along with...
2057 (pass_data_cselim): ...new pass_data instance and...
2058 (make_pass_cselim): ...new function.
2059 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2060 subclass of gimple_opt_pass along with...
2061 (pass_data_phiprop): ...new pass_data instance and...
2062 (make_pass_phiprop): ...new function.
2063 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2064 subclass of gimple_opt_pass along with...
2065 (pass_data_pre): ...new pass_data instance and...
2066 (make_pass_pre): ...new function.
2067 (pass_fre): Convert from a global struct to a subclass of
2068 gimple_opt_pass along with...
2069 (pass_data_fre): ...new pass_data instance and...
2070 (make_pass_fre): ...new function.
2071 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2072 subclass of gimple_opt_pass along with...
2073 (pass_data_reassoc): ...new pass_data instance and...
2074 (make_pass_reassoc): ...new function.
2075 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2076 subclass of gimple_opt_pass along with...
2077 (pass_data_sink_code): ...new pass_data instance and...
2078 (make_pass_sink_code): ...new function.
2079 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2080 subclass of gimple_opt_pass along with...
2081 (pass_data_strlen): ...new pass_data instance and...
2082 (make_pass_strlen): ...new function.
2083 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2084 struct to a subclass of gimple_opt_pass along with...
2085 (pass_data_build_alias): ...new pass_data instance and...
2086 (make_pass_build_alias): ...new function.
2087 (pass_build_ealias): Convert from a global struct to a subclass of
2088 gimple_opt_pass along with...
2089 (pass_data_build_ealias): ...new pass_data instance and...
2090 (make_pass_build_ealias): ...new function.
2091 (pass_ipa_pta): Convert from a global struct to a subclass of
2092 simple_ipa_opt_pass along with...
2093 (pass_data_ipa_pta): ...new pass_data instance and...
2094 (make_pass_ipa_pta): ...new function.
2095 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2096 subclass of gimple_opt_pass along with...
2097 (pass_data_uncprop): ...new pass_data instance and...
2098 (make_pass_uncprop): ...new function.
2099 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2100 global struct to a subclass of gimple_opt_pass along with...
2101 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2102 (make_pass_late_warn_uninitialized): ...new function.
2103 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2104 to a subclass of gimple_opt_pass along with...
2105 (pass_data_init_datastructures): ...new pass_data instance and...
2106 (make_pass_init_datastructures): ...new function.
2107 (pass_early_warn_uninitialized): Convert from a global struct to a
2108 subclass of gimple_opt_pass along with...
2109 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2110 (make_pass_early_warn_uninitialized): ...new function.
2111 (pass_update_address_taken): Convert from a global struct to a
2112 subclass of gimple_opt_pass along with...
2113 (pass_data_update_address_taken): ...new pass_data instance and...
2114 (make_pass_update_address_taken): ...new function.
2115 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2116 struct to a subclass of gimple_opt_pass along with...
2117 (pass_data_release_ssa_names): ...new pass_data instance and...
2118 (make_pass_release_ssa_names): ...new function.
2119 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2120 subclass of gimple_opt_pass along with...
2121 (pass_data_stdarg): ...new pass_data instance and...
2122 (make_pass_stdarg): ...new function.
2123 * tree-switch-conversion.c (pass_convert_switch): Convert from a
2124 global struct to a subclass of gimple_opt_pass along with...
2125 (pass_data_convert_switch): ...new pass_data instance and...
2126 (make_pass_convert_switch): ...new function.
2127 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2128 to a subclass of gimple_opt_pass along with...
2129 (pass_data_tail_recursion): ...new pass_data instance and...
2130 (make_pass_tail_recursion): ...new function.
2131 (pass_tail_calls): Convert from a global struct to a subclass of
2132 gimple_opt_pass along with...
2133 (pass_data_tail_calls): ...new pass_data instance and...
2134 (make_pass_tail_calls): ...new function.
2135 * tree-vect-generic.c (pass_lower_vector): Convert from a global
2136 struct to a subclass of gimple_opt_pass along with...
2137 (pass_data_lower_vector): ...new pass_data instance and...
2138 (make_pass_lower_vector): ...new function.
2139 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2140 gimple_opt_pass along with...
2141 (pass_data_lower_vector_ssa): ...new pass_data instance and...
2142 (make_pass_lower_vector_ssa): ...new function.
2143 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2144 to a subclass of gimple_opt_pass along with...
2145 (pass_data_slp_vectorize): ...new pass_data instance and...
2146 (make_pass_slp_vectorize): ...new function.
2147 (pass_ipa_increase_alignment): Convert from a global struct to a
2148 subclass of simple_ipa_opt_pass along with...
2149 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2150 (make_pass_ipa_increase_alignment): ...new function.
2151 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2152 gimple_opt_pass along with...
2153 (pass_data_vrp): ...new pass_data instance and...
2154 (make_pass_vrp): ...new function.
2155 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2156 subclass of simple_ipa_opt_pass along with...
2157 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2158 (make_pass_ipa_free_lang_data): ...new function.
2159 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2160 gimple_opt_pass along with...
2161 (pass_data_tsan): ...new pass_data instance and...
2162 (make_pass_tsan): ...new function.
2163 (pass_tsan_O0): Convert from a global struct to a subclass of
2164 gimple_opt_pass along with...
2165 (pass_data_tsan_O0): ...new pass_data instance and...
2166 (make_pass_tsan_O0): ...new function.
2167 * var-tracking.c (pass_variable_tracking): Convert from a global
2168 struct to a subclass of rtl_opt_pass along with...
2169 (pass_data_variable_tracking): ...new pass_data instance and...
2170 (make_pass_variable_tracking): ...new function.
2171 * web.c (pass_web): Convert from a global struct to a subclass of
2172 rtl_opt_pass along with...
2173 (pass_data_web): ...new pass_data instance and...
2174 (make_pass_web): ...new function.
2175 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2176 declaration with that of...
2177 (make_pass_mode_switch_use): ...new function.
2178 (pass_resolve_sw_modes): Replace declaration with that of...
2179 (make_pass_resolve_sw_modes): ...new function.
2180 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2181 from a global struct to a subclass of rtl_opt_pass along with...
2182 (pass_data_mode_switch_use): ...new pass_data instance and...
2183 (make_pass_mode_switch_use): ...new function.
2184 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2185 from a global struct to a subclass of rtl_opt_pass along with...
2186 (pass_data_resolve_sw_modes): ...new pass_data instance and...
2187 (make_pass_resolve_sw_modes): ...new function.
2188 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2189 struct to a subclass of rtl_opt_pass along with...
2190 (pass_data_insert_vzeroupper): ...new pass_data instance and...
2191 (make_pass_insert_vzeroupper): ...new function.
2192 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2193 global struct to a subclass of rtl_opt_pass along with...
2194 (pass_data_work_around_errata): ...new pass_data instance and...
2195 (make_pass_work_around_errata): ...new function.
2196 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2197 struct to a subclass of rtl_opt_pass along with...
2198 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2199 (make_pass_mips_machine_reorg2): ...new function.
2201 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2203 * passes.c (pass_manager::operator new): New.
2205 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2207 Handwritten part of conversion of passes to C++ classes.
2209 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2210 (toplev.o): Add dep on PASS_MANAGER_H.
2211 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2212 of early local pases to reflect this moving from a global to a
2213 member of gcc::pass_manager.
2214 (cgraph_add_new_function): Likewise.
2215 * lto-cgraph.c (lto_output_node): Update for conversion of
2216 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2217 * passes.c (opt_pass::clone): New.
2218 (opt_pass::gate): New.
2219 (opt_pass::execute): New.
2220 (opt_pass::opt_pass): New.
2221 (pass_manager::execute_early_local_passes): New.
2222 (pass_manager::execute_pass_mode_switching): new.
2223 (finish_optimization_passes): Convert to...
2224 (pass_manager::finish_optimization_passes): ...this.
2225 (finish_optimization_passes): Update for conversion of passes to
2227 (register_dump_files_1): Use has_gate since we cannot portably
2228 check a vtable entry against NULL.
2229 (dump_one_pass): Likewise.
2230 (ipa_write_summaries_2): Likewise.
2231 (ipa_write_optimization_summaries_1): Likewise.
2232 (ipa_read_summaries_1): Likewise.
2233 (ipa_read_optimization_summaries_1): Likewise.
2234 (execute_ipa_stmt_fixups): Likewise.
2235 (pass_manager::pass_manager): Rewrite pass-creation, invoking
2236 pass-creation functions rather than wiring up globals, and
2237 storing the results in fields of pass_manager generated using
2239 (pass_manager::dump_profile_report): Update for conversion of
2240 passes to C++ classes.
2241 (pass_manager::execute_ipa_summary_passes): Likewise.
2242 (execute_one_ipa_transform_pass): Likewise.
2243 (execute_one_pass): Use has_gate and has_execute since we cannot
2244 portably check a vtable entry against NULL.
2245 * pass_manager.h (pass_manager::finish_optimization_passes): New.
2246 (pass_manager): Use pass-instances.def to add fields for the
2247 various pass instances.
2248 * toplev.c (finalize): Update for move of
2249 finish_optimization_passes to a method of gcc::pass_manager.
2250 * toplev.h (finish_optimization_passes): Move to method of class
2252 * tree-pass.h (struct pass_data): New.
2253 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2254 (opt_pass::gate): Convert to virtual function.
2255 (opt_pass::~opt_pass): New.
2256 (opt_pass::clone): New.
2257 (opt_pass::execute): Convert to virtual function.
2258 (opt_pass::opt_pass): New.
2259 (opt_pass::ctxt_): new.
2260 (gimple_opt_pass): Convert to subclass of opt_pass.
2261 (gimple_opt_pass::gimple_opt_pass): New.
2262 (rtl_opt_pass): Convert to subclass of opt_pass.
2263 (rtl_opt_pass::rtl_opt_pass): New.
2264 (ipa_opt_pass_d): Convert to subclass of opt_pass.
2265 (ipa_opt_pass_d::ipa_opt_pass_d): New.
2266 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2267 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2268 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2269 invocation of pass_mode_switching to reflect this moving from a
2270 global to a member of gcc::pass_manager.
2271 (ix86_option_override): Rework how pass_insert_vzeroupper is
2272 added to the pass_manager to reflect autogenerated changes.
2273 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2275 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
2277 PR rtl-optimization/57708
2278 * recog.c (peep2_find_free_register): Validate all regs in a
2281 2013-08-05 Jan Hubicka <jh@suse.cz>
2284 * cgraph.c (verify_cgraph_node): Accept local flags from other
2286 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2287 (function_and_variable_visibility): Likewise.
2288 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2290 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2292 * graph.c (init_graph_slim_pretty_print): Remove.
2293 (print_graph_cfg): Do not call it. Use local pretty printer.
2294 (start_graph_dump): Likewise.
2296 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2298 * gimple-pretty-print.c (buffer): Remove.
2299 (initialized): Likewise.
2300 (maybe_init_pretty_print): Likewise.
2301 (print_gimple_stmt): Do not call it. Use non-static local
2302 pretty_printer variable.
2303 (print_gimple_expr): Likewise.
2304 (print_gimple_seq): Likewise.
2305 (gimple_dump_bb): Likewise.
2307 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2309 * asan.c (asan_pp): Remove.
2310 (asan_pp_initialized): Likewise.
2311 (asan_pp_initialize): Likewise.
2312 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
2313 (asan_emit_stack_protection): Tidy. Use local pretty printer.
2314 (asan_add_global): Likewise.
2316 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
2318 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
2319 * diagnostic.h (diagnostic_flush_buffer): Adjust.
2320 * pretty-print.c (pp_formatted_text_data): Likewise.
2321 (pp_indent): Rename from pp_base_indent.
2322 (pp_format): Rename from pp_base_format.
2323 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2324 (pp_format_verbatim): Rename from pp_base_format_verbatim.
2325 (pp_flush): Rename from pp_base_flush.
2326 (pp_set_line_maximum_length): Rename from
2327 pp_base_set_line_maximum_length.
2328 (pp_clear_output_area): Rename from pp_base_clear_output_area.
2329 (pp_set_prefix): Rename from pp_base_set_prefix.
2330 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2331 (pp_emit_prefix): Rename from pp_base_emit_prefix.
2332 (pp_append_text): Rename from pp_base_append_text.
2333 (pp_formatted_text): Rename from pp_base_formatted_text.
2334 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2335 (pp_remaining_character_count_for_line): Rename from
2336 pp_base_remaining_character_count_for_line.
2337 (pp_newline): Rename from pp_base_newline.
2338 (pp_character): Rename from pp_base_character.
2339 (pp_string): Rename from pp_base_string.
2340 (pp_maybe_space): Rename from pp_base_maybe_space.
2341 * asan.c (asan_pp_string): Adjust.
2342 (asan_emit_stack_protection): Likewise.
2343 (asan_add_global): Likewise.
2344 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2345 * tree-mudflap.c (mf_varname_tree): Likewise.
2346 * tree-pretty-print.c (pp_tree_identifier): Rename from
2347 pp_base_tree_identifier.
2348 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2349 Declare as function.
2351 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2353 * pretty-print.h (pp_bar_bar): New.
2354 (pp_ampersand_ampersand): Likewise.
2355 (pp_less_equal): Likewise.
2356 (pp_greater_equal): Likewise.
2357 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2358 printer functions instead of pp_string or operators and punctuators.
2359 (dump_gimple_call): Likewise.
2360 (dump_gimple_omp_for): Likewise.
2361 (dump_gimple_transaction): Likewise.
2362 (dump_gimple_phi): Likewise.
2363 (pp_gimple_stmt_1): Likewise.
2364 * sched-vis.c (print_insn): Likewise.
2365 * tree-mudflap.c (mf_varname_tree): Likewise.
2366 * tree-pretty-print.c (dump_block_node): Likewise.
2367 (dump_generic_node): Likewise.
2369 2013-08-02 Jan Hubicka <jh@suse.cz>
2371 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2373 * lto-streamer-out.c (tree_is_indexable): Results decls and
2374 parm decls are not indexable.
2375 (DFS_write_tree_body): Do not follow args and results.
2376 (hash_tree): Likewise.
2377 (output_functions): Rearrange so struct function is needed
2378 only when real body is output; be able to also ouptut abstract
2379 functions; output DECL_ARGUMENTS and DECL_RESULT.
2380 (lto_output): When not in WPA, ale store abstract functions.
2381 (write_symbol): Do not care about RESULT_DECL.
2382 (output_symbol_p): Handle correctly sbtract decls.
2383 * lto-streamer-in.c (input_function): Rearrange so struct
2384 function can be NULL at entry; allow streaming of
2385 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2386 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2387 sanity check during LTO.
2388 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2389 RESULT_DECl and DECL_ARGUMENTS.
2390 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2393 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2395 * pretty-print.h (pp_underscore): New.
2397 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2398 printer functions instead of pp_character.
2399 (dump_binary_rhs): Likewise.
2400 (dump_ternary_rhs): Likewise.
2401 (dump_gimple_call_args): Likewise.
2402 (pp_points_to_solution): Likewise.
2403 (dump_gimple_call): Likewise.
2404 (dump_gimple_switch): Likewise.
2405 (dump_gimple_cond): Likewise.
2406 (dump_gimple_bind): Likewise.
2407 (dump_gimple_try): Likewise.
2408 (dump_gimple_omp_for): Likewise.
2409 (dump_gimple_omp_continue): Likewise.
2410 (dump_gimple_omp_single): Likewise.
2411 (dump_gimple_omp_sections): Likewise.
2412 (dump_gimple_omp_block): Likewise.
2413 (dump_gimple_omp_critical): Likewise.
2414 (dump_gimple_transaction): Likewise.
2415 (dump_gimple_asm): Likewise.
2416 (dump_gimple_phi): Likewise.
2417 (dump_gimple_omp_parallel): Likewise.
2418 (dump_gimple_omp_task): Likewise.
2419 (dump_gimple_omp_atomic_load): Likewise.
2420 (dump_gimple_omp_atomic_store): Likewise.
2421 (dump_gimple_mem_ops): Likewise.
2422 (pp_gimple_stmt_1): Likewise.
2423 (pp_cfg_jump): Likewise.
2424 (dump_implicit_edges): Likewise.
2425 (gimple_dump_bb_for_graph): Likewise.
2426 * graph.c (draw_cfg_node): Likewise.
2427 * langhooks.c (lhd_print_error_function): Likewise.
2428 * sched-vis.c (print_exp): Likewise.
2429 (print_value): Likewise.
2430 (print_pattern): Likewise.
2431 (print_insn): Likewise.
2432 (rtl_dump_bb_for_graph): Likewise.
2433 * tree-pretty-print.c (dump_function_declaration): Likewise.
2434 (dump_array_domain): Likewise.
2435 (dump_omp_clause): Likewise.
2436 (dump_location): Likewise.
2437 (dump_generic_node): Likewise.
2438 (print_struct_decl): Likewise.
2439 * diagnostic.c (diagnostic_show_locus): Use pp_space.
2441 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2443 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2444 candidate table when replacing a candidate statement.
2445 (replace_rhs_if_not_dup): Likewise.
2446 (replace_one_candidate): Likewise.
2448 2013-08-02 Jan Hubicka <jh@suse.cz>
2449 Martin Liska <marxin.liska@gmail.com>
2451 * cgraphunit.c (add_new_function): Fix logic when adding from
2453 (assemble_thunk): Rename to ...
2454 (expand_thunk); .. this one; export; get it working with
2455 general functions; make produced gimple valid.
2456 * cgraph.h (expand_thunk): Declare.
2458 2013-08-02 Jan Hubicka <jh@suse.cz>
2460 * ipa-cp.c (gather_context_independent_values): Use
2461 ipa_get_param_move_cost.
2462 (get_replacement_map): Remove PARAM; move parameter folding
2464 (create_specialized_node): Update.
2465 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2466 assert that we have gimple body; update move_cost.
2467 (count_formal_params): Assert that we have gimple body.
2468 (ipa_dump_param): New function.
2469 (ipa_alloc_node_params): Break out from ...
2470 (ipa_initialize_node_params): ... here.
2471 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2472 (ipa_write_node_info): Stream move costs.
2473 (ipa_read_node_info): Read move costs.
2474 (ipa_update_after_lto_read): Do not recompute node params.
2475 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2476 (ipa_get_param): Check we are not in WPA.
2477 (ipa_get_param_move_cost): New.
2478 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2479 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2480 parm numbers to be present.
2482 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2484 PR rtl-optimization/58048
2485 * lra-constraints.c (process_alt_operands): Don't check asm
2486 operand on register.
2488 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
2490 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2491 the implied StoreLoad barrier for atomic operations if before.
2493 2013-08-02 Jan Hubicka <jh@suse.cz>
2494 Martin Liska <marxin.liska@gmail.com>
2496 * cgraph.c (cgraph_function_body_availability): Do not check
2498 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2499 symtab_node_availability): Declare.
2500 * ipa.c (can_replace_by_local_alias): New.
2501 (function_and_variable_visibility): Use it.
2502 * symtab.c (symtab_for_node_and_aliases,
2503 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2505 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2507 PR rtl-optimization/57963
2508 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2509 (lra_constraints): Use them.
2511 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
2513 * config/arm/types.md (define_attr "type"): Add "load_acq"
2515 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2517 (cortex_a53_store1): Likewise.
2519 2013-08-01 Jan Hubicka <jh@suse.cz>
2521 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2522 partitions are not needed.
2524 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
2526 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2527 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2528 MAYBE_NON_Q_CLASS_P where appropriate.
2530 2013-08-01 Jan Hubicka <jh@suse.cz>
2532 * cgraph.h (release_function_body): Declare.
2533 * tree.c (free_lang_data_in_decl): Free, parameters and return values
2534 of unused delcarations.
2536 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2538 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2539 RTL form when subtracting a constant.
2541 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2543 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2544 Generate canonical plus rtx with negated immediate instead of minus
2546 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2548 2013-08-01 Jan Hubicka <jh@suse.cz>
2550 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2551 (cgraph_release_function_body): Likewise.
2552 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2553 * cgraph.h (cgrpah_node): Rename abstract_and_needed
2554 to used_as_abstract_origin.
2555 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2556 symbols used as abstract origins.
2557 * cgraphunit.c (analyze_functions): Update.
2558 * ipa.c (symtab_remove_unreachable_nodes): Recompute
2559 used_as_abstract_origin.
2560 * tree-inline.c (tree_function_versioning): Update
2561 used_as_abstract_origin; be ready for DECL_RESULT and
2562 DECL_ARGUMENTS to be NULL.
2564 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2565 for abstract functions.
2566 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2569 2013-08-01 Jan Hubicka <jh@suse.cz>
2571 * profile.c (compute_value_histograms): Fix thinko.
2573 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2575 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
2576 aarch-common-protos.h to extra_headers.
2577 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2578 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2579 * config/aarch64/t-aarch64 (aarch-common.o): Define.
2581 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2583 * config/aarch64/aarch64.md (define_attr "type"): Delete.
2584 Include "../arm/types.md". Define "type" attribute for all patterns.
2585 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2588 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2590 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2591 to support power8 load fusion.
2592 (fusion_gpr_mem_load): Likewise.
2594 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2596 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2597 declarations for power8 load fusion.
2598 (emit_fusion_gpr_load): Likewise.
2600 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2601 tuning for power8, turn on fusion mode by default. Turn on sign
2602 extending fusion mode if normal fusion mode is on, and we are at
2604 (fusion_gpr_load_p): New function, return true if we can fuse an
2605 addis instruction with a dependent load to a GPR.
2606 (emit_fusion_gpr_load): Emit the instructions for power8 load
2609 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2610 (VSX load fusion peepholes): New peepholes to fuse together an
2611 addi instruction with a VSX load instruction.
2613 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2614 peepholes to fuse an addis instruction with a load to a GPR base
2615 register. If we are supporting sign extending fusions, convert
2616 sign extending loads to zero extending loads and add an explicit
2619 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
2621 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
2622 aarch-common-protos.h to extra_headers.
2623 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2624 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2625 (arm_early_store_addr_dep): Likewise.
2626 (arm_no_early_alu_shift_dep): Likewise.
2627 (arm_no_early_alu_shift_value_dep): Likewise.
2628 (arm_no_early_mul_dep): Likewise.
2629 (arm_no_early_store_addr_dep): Likewise.
2630 (arm_mac_accumulator_is_mul_result): Likewise.
2631 (arm_mac_accumulator_is_result): Likewise.
2632 * config/arm/aarch-common.c: ... here. New file.
2633 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2635 (arm_early_store_addr_dep): Likewise.
2636 (arm_no_early_alu_shift_dep): Likewise.
2637 (arm_no_early_alu_shift_value_dep): Likewise.
2638 (arm_no_early_mul_dep): Likewise.
2639 (arm_no_early_store_addr_dep): Likewise.
2640 (arm_mac_accumulator_is_mul_result): Likewise.
2641 (arm_mac_accumulator_is_result): Likewise.
2642 * config/arm/aarch-common-protos.h: ... here. New file.
2643 * config/arm/t-arm (aarch-common.o): Define.
2645 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
2647 * config/arm/arm.md: Include new file "types.md".
2648 (define_attr "type"): Move from here to ...
2649 (define_attr "mul32"): Likewise.
2650 (define_attr "mul64"): Likewise.
2651 * config/arm/types.md: ... here. New file.
2653 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
2655 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2656 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2658 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2660 * gen-pass-instances.awk: Fix offset of substr().
2662 2013-07-31 David Malcolm <dmalcolm@redhat.com>
2664 * Makefile.in (pass-instances.def): New.
2665 (passes.o): Replace dependency on passes.def with one on
2668 * gen-pass-instances.awk: New.
2670 * passes.c (pass_manager::pass_manager): Use pass-instances.def
2671 rather than passes.def, updating local definition of NEXT_PASS
2672 macro to add an extra NUM parameter (currently unused).
2674 2013-07-30 David Malcolm <dmalcolm@redhat.com>
2676 * Makefile.in (PASS_MANAGER_H): New.
2677 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2678 (passes.o): Likewise.
2679 (statistics.o): Likewise.
2680 (cgraphunit.o): Likewise.
2681 (context.o): Depend on PASS_MANAGER_H.
2683 * pass_manager.h: New.
2685 * cgraphunit.c (cgraph_add_new_function): Update for moves
2686 of globals to fields of pass_manager.
2687 (analyze_function): Likewise.
2688 (expand_function): Likewise.
2689 (ipa_passes): Likewise.
2690 (compile): Likewise.
2692 * context.c (context::context): New.
2693 * context.h (context::context): New.
2694 (context::get_passes): New.
2695 (context::passes_): New.
2697 * lto-cgraph.c (input_node): Update for moves of globals to
2698 fields of pass_manager.
2700 * passes.c (all_passes): Remove, in favor of a field of the
2701 same name within the new class pass_manager.
2702 (all_small_ipa_passes): Likewise.
2703 (all_lowering_passes): Likewise.
2704 (all_regular_ipa_passes): Likewise.
2705 (all_late_ipa_passes): Likewise.
2706 (all_lto_gen_passes): Likewise.
2707 (passes_by_id): Likewise.
2708 (passes_by_id_size): Likewise.
2709 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2710 the new class pass_manager.
2711 (set_pass_for_id): Convert to...
2712 (pass_manager::set_pass_for_id): ...method.
2713 (get_pass_for_id): Convert to...
2714 (pass_manager::get_pass_for_id): ...method.
2715 (register_one_dump_file): Move body of implementation into...
2716 (pass_manager::register_one_dump_file): ...here.
2717 (register_dump_files_1): Convert to...
2718 (pass_manager::register_dump_files_1): ...method.
2719 (register_dump_files): Convert to...
2720 (pass_manager::register_dump_files): ...method.
2721 (create_pass_tab): Update for moves of globals to fields of
2723 (dump_passes): Move body of implementation into...
2724 (pass_manager::dump_passes): ...here.
2725 (register_pass): Move body of implementation into...
2726 (pass_manager::register_pass): ...here.
2727 (init_optimization_passes): Convert into...
2728 (pass_manager::pass_manager): ...constructor for new
2729 pass_manager class, and initialize the pass_lists array.
2730 (check_profile_consistency): Update for moves of globals to
2731 fields of pass_manager.
2732 (dump_profile_report): Move body of implementation into...
2733 (pass_manager::dump_profile_report): ...here.
2734 (ipa_write_summaries_1): Update for moves of pass lists from
2735 being globals to fields of pass_manager.
2736 (ipa_write_optimization_summaries): Likewise.
2737 (ipa_read_summaries): Likewise.
2738 (ipa_read_optimization_summaries): Likewise.
2739 (execute_all_ipa_stmt_fixups): Likewise.
2741 * statistics.c (statistics_fini): Update for moves of globals to
2742 fields of pass_manager.
2744 * toplev.c (general_init): Replace call to
2745 init_optimization_passes with construction of the pass_manager
2748 * tree-pass.h (all_passes): Remove, in favor of a field of the
2749 same name within the new class pass_manager.
2750 (all_small_ipa_passes): Likewise.
2751 (all_lowering_passes): Likewise.
2752 (all_regular_ipa_passes): Likewise.
2753 (all_lto_gen_passes): Likewise.
2754 (all_late_ipa_passes): Likewise.
2755 (passes_by_id): Likewise.
2756 (passes_by_id_size): Likewise.
2757 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2758 the new class pass_manager.
2759 (get_pass_for_id): Remove.
2761 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
2763 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2766 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
2768 * arm.md (mulhi3): New expand pattern.
2770 2013-07-30 Jan Hubicka <jh@suse.cz>
2771 Martin Liska <marxin.liska@gmail.com>
2773 * profile.c (compute_value_histograms): Do not ICE when
2774 there is mismatch only on some counters.
2776 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2778 PR rtl-optimization/57637
2779 * function.c (move_insn_for_shrink_wrap): Also check the
2780 GEN set of the LIVE problem for the liveness analysis
2781 if it exists, otherwise give up.
2783 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2785 PR tree-optimization/57993
2786 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2787 replaced statement in the candidate table.
2788 (phi_add_costs): Return infinite cost when the hidden basis does
2789 not dominate all phis on which the candidate is dependent.
2790 (replace_one_candidate): Record replaced statement in the
2793 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
2795 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2796 (ashlv2si3): New expander.
2797 (*ashlv2si3_i): New define_insn_and_split.
2798 * predicates.md (float_operation): Allow patterns with three
2801 PR rtl-optimization/58021
2802 * mode-switching.c (create_pre_exit): Always split off preceding
2803 insns if we are not at the basic block head.
2805 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
2807 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2808 (UCLIBC_DYNAMIC_LINKER): New macro.
2809 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2811 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2812 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
2814 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2815 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2816 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2817 for SF and DF modes. Use ieee_quad_format for TF mode.
2818 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2819 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2820 (mips_option_override): Handle `-mnan=legacy'.
2821 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2822 `-mabs=2008' and `-mnan=2008'.
2823 (OPTION_DEFAULT_SPECS): Add "nan" default.
2824 (ASM_SPEC): Handle `-mnan='.
2825 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2826 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2827 comment accordingly.
2828 (neg<mode>2): Likewise.
2829 * config/mips/mips.opt (mabs, mnan): New options.
2830 * doc/install.texi (Configuration): Document `--with-nan=' option.
2831 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2833 (MIPS Options): Document them.
2834 * config.gcc <mips*-*-*>: Handle `--with-nan='.
2835 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2836 * configure: Regenerate.
2837 * config.in: Regenerate.
2839 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
2841 * config/i386/i386.md (float post-reload splitters): Do not check
2842 for subregs of SSE registers.
2844 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
2845 H.J. Lu <hongjiu.lu@intel.com>
2849 * config/i386/i386.md (post-reload splitter
2850 to avoid partial SSE reg dependency stalls): New pattern.
2852 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
2854 * config/s390/s390.md ("movcc"): Swap load and store instructions.
2856 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
2858 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2859 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2861 2013-07-26 Cary Coutant <ccoutant@google.com>
2863 * dwarf2out.c (die_checksum_ordered): Don't include template
2864 instantiations in signature.
2865 (is_template_parameter): New function.
2866 (is_template_instantiation): New function.
2867 (generate_skeleton_bottom_up): Don't include template instantiations
2869 (generate_skeleton): Likewise.
2870 (break_out_comdat_types): Move recursive call to break out nested
2872 (prune_unused_types_mark_generic_parms_dies): Call
2873 is_template_parameter.
2875 2013-07-26 Ian Bolton <ian.bolton@arm.com>
2877 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2878 uses vector registers.
2879 * config/aarch64/iterators.md: Add attributes rtn and vas.
2881 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2882 Richard Earnshaw <richard.earnshaw@arm.com>
2884 * combine.c (simplify_comparison): Re-canonicalize operands
2886 * config/arm/arm.md (movcond_addsi): New splitter.
2888 2013-07-25 Sterling Augustine <saugustine@google.com>
2890 * dwarf2out.c (size_of_pubnames): Move code to...
2891 (include_pubname_in_output): ...here. New.
2892 (want_pubnames): Rearrange.
2893 (output_pubnames): Call include_pubname_in_output. Move assertion.
2895 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2897 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2899 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2902 * doc/extend.texi: Remove obsolete builtins. Fix
2903 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2905 2013-07-25 Jan Hubicka <jh@suse.cz>
2907 * cgraph.c (release_function_body): Break out from ...
2908 (cgraph_release_function_body): ... this one; also release DECL_RESULT
2910 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2911 old_tree in the map.
2912 (create_specialized_node): Update.
2913 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2915 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2916 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2917 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2918 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2921 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2923 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2924 addsi3_carryin_alt2_<optab>): Correct output template.
2926 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2928 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2929 Adjust for arm_restrict_it.
2930 Remove trailing whitespace.
2932 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
2934 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2935 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2937 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2939 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
2941 PR rtl-optimization/57960
2942 * lra-constraints.c (process_alt_operands): Use the right mode
2943 when checking strict_low.
2945 2013-07-25 Jan Hubicka <jh@suse.cz>
2947 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2948 * cgraph.c (cgraph_remove_node): Do not release function body
2949 when in cgraph streaming.
2950 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2951 in other partitions are not considered reachable; fix handling of
2954 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2956 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2958 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2963 * config/arm/arm.md ("*sibcall_insn): Replace use of
2964 Ss with US. Adjust output for v5 and v4t.
2965 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2967 * config/arm/constraints.md ("Ss"): Rename to US.
2969 2013-07-25 Terry Guo <terry.guo@arm.com>
2971 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2972 shift_add/shift_sub0/shift_sub1 RTXs.
2974 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
2975 Anton Blanchard <anton@au1.ibm.com>
2977 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2978 (altivec_vpks<VI_char>ss): Likewise.
2979 (altivec_vpks<VI_char>us): Likewise.
2980 (altivec_vpku<VI_char>us): Likewise.
2981 (altivec_vpku<VI_char>um): Likewise.
2983 2013-07-24 David Malcolm <dmalcolm@redhat.com>
2985 Introduce context class.
2987 * Makefile.in (CONTEXT_H): New.
2988 (OBJS): Add context.o.
2989 (toplev.o): Add CONTEXT_H to dependencies.
2992 * toplev.c (general_init): Create the singleton gcc::context instance.
2998 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
3000 PR rtl-optimization/57968
3001 * mode-switching.c (create_pre_exit): Allow instructions that
3002 don't set a return register to need a non-exit mode.
3004 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3005 Anton Blanchard <anton@au1.ibm.com>
3007 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3008 operands to vperm for little endian.
3009 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3010 of lvsl to create the control mask for a vperm for little endian.
3012 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3013 Anton Blanchard <anton@au1.ibm.com>
3015 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3016 two operands for little-endian.
3018 2013-07-23 Steve Ellcey <sellcey@mips.com>
3020 * config/mips/mips.c (mips_case_values_threshold): New.
3021 (TARGET_CASE_VALUES_THRESHOLD): Define.
3023 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3024 Anton Blanchard <anton@au1.ibm.com>
3026 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3027 selection of field for vector splat in little endian mode.
3029 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3031 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3032 expanders to rs6000.md.
3033 (ior<mode>3): Likewise.
3034 (and<mode>3): Likewise.
3035 (one_cmpl<mode>2): Likewise.
3036 (nor<mode>3): Likewise.
3037 (andc<mode>3): Likewise.
3038 (eqv<mode>3): Likewise.
3039 (nand<mode>3): Likewise.
3040 (orc<mode>3): Likewise.
3042 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3045 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3046 to split multi-word logical operations.
3047 (rs6000_split_logical_di): Likewise.
3048 (rs6000_split_logical): Likewise.
3050 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3051 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3052 and allow TImode operations in 32-bit.
3053 (vsx_and<mode>3_64bit): Likewise.
3054 (vsx_ior<mode>3_32bit): Likewise.
3055 (vsx_ior<mode>3_64bit): Likewise.
3056 (vsx_xor<mode>3_32bit): Likewise.
3057 (vsx_xor<mode>3_64bit): Likewise.
3058 (vsx_one_cmpl<mode>2_32bit): Likewise.
3059 (vsx_one_cmpl<mode>2_64bit): Likewise.
3060 (vsx_nor<mode>3_32bit): Likewise.
3061 (vsx_nor<mode>3_64bit): Likewise.
3062 (vsx_andc<mode>3_32bit): Likewise.
3063 (vsx_andc<mode>3_64bit): Likewise.
3064 (vsx_eqv<mode>3_32bit): Likewise.
3065 (vsx_eqv<mode>3_64bit): Likewise.
3066 (vsx_nand<mode>3_32bit): Likewise.
3067 (vsx_nand<mode>3_64bit): Likewise.
3068 (vsx_orc<mode>3_32bit): Likewise.
3069 (vsx_orc<mode>3_64bit): Likewise.
3071 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3072 logical types in GPRs.
3074 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3075 logical insns to rs6000.md, and allow TImode operations in
3077 (altivec_ior<mode>3): Likewise.
3078 (altivec_xor<mode>3): Likewise.
3079 (altivec_one_cmpl<mode>2): Likewise.
3080 (altivec_nor<mode>3): Likewise.
3081 (altivec_andc<mode>3): Likewise.
3083 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3084 attributes for moving the 128-bit logical operations into
3086 (BOOL_REGS_OUTPUT): Likewise.
3087 (BOOL_REGS_OP1): Likewise.
3088 (BOOL_REGS_OP2): Likewise.
3089 (BOOL_REGS_UNARY): Likewise.
3090 (BOOL_REGS_AND_CR0): Likewise.
3091 (one_cmpl<mode>2): Add support for DI logical operations on
3092 32-bit, splitting the operations to 32-bit.
3096 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3097 changes to combine the 32/64-bit code, allow logical operations on
3098 TI mode in 32-bit, and to use similar match_operator patterns like
3099 scalar mode uses. Combine the Altivec and VSX code for logical
3100 operations, and move it here.
3101 (ior<mode>3, 128-bit types): Likewise.
3102 (xor<mode>3, 128-bit types): Likewise.
3103 (one_cmpl<mode>3, 128-bit types): Likewise.
3104 (nor<mode>3, 128-bit types): Likewise.
3105 (andc<mode>3, 128-bit types): Likewise.
3106 (eqv<mode>3, 128-bit types): Likewise.
3107 (nand<mode>3, 128-bit types): Likewise.
3108 (orc<mode>3, 128-bit types): Likewise.
3109 (and<mode>3_internal): Likewise.
3110 (bool<mode>3_internal): Likewise.
3111 (boolc<mode>3_internal1): Likewise.
3112 (boolc<mode>3_internal2): Likewise.
3113 (boolcc<mode>3_internal1): Likewise.
3114 (boolcc<mode>3_internal2): Likewise.
3115 (eqv<mode>3_internal1): Likewise.
3116 (eqv<mode>3_internal2): Likewise.
3117 (one_cmpl1<mode>3_internal): Likewise.
3119 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3121 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3122 Rename flag_stack_usage to flag_stack_usage_info.
3124 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3126 * config/microblaze/sync.md: New file.
3127 * config/microblaze/microblaze.md: Include sync.md
3128 * config/microblaze/microblaze.c: Add print_operand 'y'.
3129 * config/microblaze/constraints.md: Add memory_contraint
3130 'Q' which is a single register.
3132 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
3134 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3136 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3138 * reload.c (find_reloads): Exit loop once we find this operand
3139 cannot be reloaded somehow for this alternative.
3141 * reload.c (find_reloads): Exit loop once we find a hard register.
3143 * rtlanal.c (computed_jump_p): Exit loop once we find label
3146 * i386.c (ix86_pad_returns): Exit loop after setting replace.
3148 * cfgloopmanip.c (remove_path): Exit loop after setting
3151 * gensupport.c (subst_dup): Avoid loop if code is not
3152 MATCH_DUP nor MATCH_OP_DUP.
3154 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
3156 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3158 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3160 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3161 true for SP_REGNUM if mode == ptr_mode.
3162 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3163 with value R0_REGNUM + 31.
3165 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3167 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3168 pad pointer-typed argument downward.
3170 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3172 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3173 and __ILP32__ when the ILP32 model is in use.
3175 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3177 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3178 (aarch64_load_symref_appropriately): In the case of
3179 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3180 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3181 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3182 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3183 if 'mode' doesn't equal to 'ptr_mode'.
3184 (aarch64_output_mi_thunk): Add an assertion on the alignment of
3185 'vcall_offset'; change to call aarch64_emit_move differently depending
3186 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3187 to calculate the upper bound of 'vcall_offset'.
3188 (aarch64_cannot_force_const_mem): Change to also return true if
3190 (aarch64_legitimize_reload_address): In the case of large
3191 displacements, add new local variable 'xmode' and an assertion
3192 based on it; change to use 'xmode' to generate the new rtx and
3194 (aarch64_asm_trampoline_template): Change to generate the template
3195 differently depending on TARGET_ILP32 or not; change to use
3196 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3197 (aarch64_trampoline_size): Removed.
3198 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3199 and replace immediate literals with it. Change to use 'ptr_mode'
3200 instead of 'DImode' and call convert_memory_address if the mode
3201 of 'fnaddr' doesn't equal to 'ptr_mode'.
3202 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3204 (aarch64_elf_asm_destructor): Likewise.
3205 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3206 on TARGET_ILP32 instead of aarch64_trampoline_size.
3207 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3208 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3209 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3210 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
3211 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3212 (storewb_pair<GPI:mode>_<P:mode>): ... this.
3213 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3214 depending on the value of 'mode'.
3215 (add_losym_<mode>): New.
3216 (ldr_got_small_<mode>): New, based on ldr_got_small.
3217 (ldr_got_small): Remove.
3218 (ldr_got_small_sidi): New.
3219 * config/aarch64/iterators.md (P): New.
3220 (PTR): Change to 'ptr_mode' in the condition.
3222 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3224 * config.gcc (aarch64*-*-*): Support --with-abi.
3225 (aarch64*-*-elf): Support --with-multilib-list.
3226 (aarch64*-*-linux*): Likewise.
3227 (supported_defaults): Add abi to aarch64*-*-*.
3228 * configure.ac: Mention AArch64 for --with-multilib-list.
3229 * configure: Re-generated.
3230 * config/aarch64/biarchilp32.h: New file.
3231 * config/aarch64/biarchlp64.h: New file.
3232 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3234 (MULTILIB_DEFAULTS): Ditto.
3235 (DRIVER_SELF_SPECS): Ditto.
3236 (ASM_SPEC): Update to also substitute -mabi.
3237 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3238 file whose name depends on -mabi= and -mbig-endian.
3239 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3241 (POINTER_SIZE): New define.
3242 (POINTERS_EXTEND_UNSIGNED): Ditto.
3243 (enum aarch64_abi_type): New enumeration tag.
3244 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3245 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3246 (TARGET_ILP32): New define.
3247 * config/aarch64/aarch64.opt (mabi): New.
3249 (ilp32, lp64): New values for -mabi.
3250 * config/aarch64/t-aarch64 (comma): New define.
3251 (MULTILIB_OPTIONS): Ditto.
3252 (MULTILIB_DIRNAMES): Ditto.
3253 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3254 * doc/invoke.texi: Document -mabi for AArch64.
3256 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
3258 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3260 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3261 Anton Blanchard <anton@au1.ibm.com>
3263 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3264 endianness when selecting field to splat.
3266 2013-07-22 Eric Christopher <echristo@gmail.com>
3268 * dwarf2out.c (die_odr_checksum): New function to use
3269 CHECKSUM_ macros and ULEB128 for DIE tag.
3270 (generate_type_signature): Use.
3272 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
3274 * config.gcc (sparc*-*-*): Accept leon3 processor.
3275 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3276 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3277 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3278 * config/sparc/sparc.opt (enum processor_type): Add leon3.
3279 (mfix-ut699): Adjust comment.
3280 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3281 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3282 (CPP_CPU_SPEC): Likewise.
3283 (ASM_CPU_SPEC): Likewise.
3284 * config/sparc/sparc.c (leon3_cost): New constant.
3285 (sparc_option_override): Add leon3 support.
3286 (mem_ref): New function.
3287 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3288 (sparc_do_work_around_errata): Look into the instruction in the delay
3289 slot and adjust accordingly. Add fix for the data cache nullify issues
3290 of the UT699. Change insertion position for the NOP.
3291 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3292 (leon3_load): New reservation.
3293 (leon_store): Bump latency to 2.
3294 (grfpu): New automaton.
3295 (grfpu_alu): New unit.
3296 (grfpu_ds): Likewise.
3297 (leon_fp_alu): Adjust.
3298 (leon_fp_mult): Delete.
3299 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3300 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3301 * config/sparc/sparc.md (cpu): Add leon3.
3302 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3304 (atomic_test_and_set): Likewise.
3307 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
3309 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
3310 default for R5900 targets.
3311 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3312 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3313 * config/mips/mips.c (mips_option_override): Report an error for
3314 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
3315 for -march=r5900 -mhard-float.
3317 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3319 * df-problems.c (can_move_insns_across): Exit loop once we
3320 find a non-fixed, non-global register.
3322 * ipa-pure-const.c (propagate_nothrow): Exit loop after
3325 * omega.c (omega_eliminate_red): Break after setting red_found.
3326 (omega_problem_has_red_equations): Similarly after setting found.
3327 (omega_query_variable): Similarly after setting coupled.
3329 2013-07-22 Marek Polacek <polacek@redhat.com>
3331 * gimplify.c: Don't include gimple.h twice.
3333 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3335 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3336 instead of TARGET_THUMB1.
3337 (Pz): New constraint.
3338 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3340 (compare_negsi_si): Likewise.
3341 (compare_addsi2_op0): Likewise.
3342 (compare_addsi2_op1): Likewise.
3343 (addsi3_carryin_<optab>): Likewise.
3344 (addsi3_carryin_alt2_<optab>): Likewise.
3345 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3346 for arm_restrict_it.
3347 (subsi3_carryin): Likewise.
3348 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3349 (minmax_arithsi): Disable for arm_restrict_it.
3350 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3351 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3352 (satsi_<SAT:code>_shift): Likewise.
3353 (arm_shiftsi3): Add alternative for 16-bit encoding.
3354 (arm32_movhf): Disable for arm_restrict_it.
3355 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3356 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3358 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
3360 * config/arm/arm.md (attribute "insn"): Delete.
3361 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3362 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3363 (not_shiftsi): Update for attribute change.
3364 (not_shiftsi_compare0): Likewise.
3365 (not_shiftsi_compare0_scratch): Likewise.
3366 (arm_one_cmplsi2): Likewise.
3367 (thumb1_one_cmplsi2): Likewise.
3368 (notsi_compare0): Likewise.
3369 (notsi_compare0_scratch): Likewise.
3370 (thumb1_movdi_insn): Likewise.
3371 (arm_movsi_insn): Likewise.
3372 (movhi_insn_arch4): Likewise.
3373 (movhi_bytes): Likewise.
3374 (arm_movqi_insn): Likewise.
3375 (thumb1_movqi_insn): Likewise.
3376 (arm32_movhf): Likewise.
3377 (thumb1_movhf): Likewise.
3378 (arm_movsf_soft_insn): Likewise.
3379 (thumb1_movsf_insn): Likewise.
3380 (thumb_movdf_insn): Likewise.
3381 (movsicc_insn): Likewise.
3382 (movsfcc_soft_insn): Likewise.
3383 (and_scc): Likewise.
3384 (cond_move): Likewise.
3385 (if_move_not): Likewise.
3386 (if_not_move): Likewise.
3387 (if_shift_move): Likewise.
3388 (if_move_shift): Likewise.
3389 (if_shift_shift): Likewise.
3390 (if_not_arith): Likewise.
3391 (if_arith_not): Likewise.
3392 (cond_move_not): Likewise.
3393 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3394 (neon_mov<mode>): Likewise.
3395 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3396 (thumb2_movsi_vfp): Likewise.
3397 (movsf_vfp): Likewise.
3398 (thumb2_movsf_vfp): Likewise.
3399 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3401 (cortexa7_older_only): Likewise.
3402 (cortexa7_younger): Likewise.
3403 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3404 (1020alu_shift_op): Likewise.
3405 (1020alu_shift_reg_op): Likewise.
3406 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3407 (alu_shift_op): Likewise.
3408 (alu_shift_reg_op): Likewise.
3409 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3410 (11_alu_shift_op): Likewise.
3411 (11_alu_shift_reg_op): Likewise.
3412 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3413 (9_alu_shift_reg_op): Likewise.
3414 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3416 (cortex_a15_alu_shift): Likewise.
3417 (cortex_a15_alu_shift_reg): Likewise.
3418 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3419 (cortex_a5_alu_shift): Likewise.
3420 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3422 (cortex_a53_alu_shift): Likewise.
3423 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3425 (cortex_a7_alu_reg): Likewise.
3426 (cortex_a7_alu_shift): Likewise.
3427 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3428 (cortex_a8_alu_shift): Likewise.
3429 (cortex_a8_alu_shift_reg): Likewise.
3430 (cortex_a8_mov): Likewise.
3431 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3432 (cortex_a9_dp_shift): Likewise.
3433 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3434 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3435 (cortex_r4_mov): Likewise.
3436 (cortex_r4_alu_shift): Likewise.
3437 (cortex_r4_alu_shift_reg): Likewise.
3438 * config/arm/fa526.md (526_alu_op): Update for attribute change.
3439 (526_alu_shift_op): Likewise.
3440 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3441 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3442 (626te_alu_shift_op): Likewise.
3443 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3444 (726te_alu_op): Likewise.
3445 (726te_alu_shift_op): Likewise.
3446 (726te_alu_shift_reg_op): Likewise.
3447 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3448 (mp626_alu_shift_op): Likewise.
3449 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3450 (pj4_alu_e1_conds): Likewise.
3451 (pj4_alu): Likewise.
3452 (pj4_alu_conds): Likewise.
3453 (pj4_shift): Likewise.
3454 (pj4_shift_conds): Likewise.
3455 (pj4_alu_shift): Likewise.
3456 (pj4_alu_shift_conds): Likewise.
3458 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3460 * config/arm/predicates.md (shiftable_operator_strict_it):
3462 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3463 Disable cond_exec version for arm_restrict_it.
3464 (thumb2_smaxsi3): Convert to generate cond_exec.
3465 (thumb2_sminsi3): Likewise.
3466 (thumb32_umaxsi3): Likewise.
3467 (thumb2_uminsi3): Likewise.
3468 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3469 (thumb2_neg_abssi2): Likewise.
3470 (thumb2_mov_scc): Add alternative for 16-bit encoding.
3471 (thumb2_movsicc_insn): Adjust alternatives.
3472 (thumb2_mov_negscc): Disable for arm_restrict_it.
3473 (thumb2_mov_negscc_strict_it): New pattern.
3474 (thumb2_mov_notscc_strict_it): New pattern.
3475 (thumb2_mov_notscc): Disable for arm_restrict_it.
3476 (thumb2_ior_scc): Likewise.
3477 (thumb2_ior_scc_strict_it): New pattern.
3478 (thumb2_cond_move): Adjust for arm_restrict_it.
3479 (thumb2_cond_arith): Disable for arm_restrict_it.
3480 (thumb2_cond_arith_strict_it): New pattern.
3481 (thumb2_cond_sub): Adjust for arm_restrict_it.
3482 (thumb2_movcond): Likewise.
3483 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3484 (thumb2_zero_extendhisi2_v6): Likewise.
3485 (thumb2_zero_extendqisi2_v6): Likewise.
3486 (orsi_notsi_si): Likewise.
3487 (orsi_not_shiftsi_si): Likewise.
3489 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
3491 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3492 instruction sequence is 1 byte shorter.
3494 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
3496 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3497 it is not needed after split.
3499 2013-07-20 Iain Sandoe <iain@codesourcery.com>
3502 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3503 second label for nonlocal goto receivers. Don't output pic base labels
3504 unless we're producing PIC; mark that action unreachable().
3505 (ix86_save_reg): If the function contains a nonlocal label, save the
3507 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3508 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3509 (update_pic_label_number_if_needed): New.
3510 (machopic_output_function_base_name): Adjust for nonlocal receiver
3512 (machopic_should_output_picbase_label): New.
3513 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3514 (nonlocal_goto_receiver): New insn and split.
3516 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
3518 * config/aarch64/aarch64-builtins.c
3519 (aarch64_fold_builtin): Fold abs in all modes.
3520 * config/aarch64/aarch64-simd-builtins.def
3521 (abs): Enable for all modes.
3522 * config/aarch64/arm_neon.h
3523 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3524 (vabs_f64): Add missing intrinsic.
3526 2013-07-19 Ian Bolton <ian.bolton@arm.com>
3528 * config/aarch64/arm_neon.h (vabs_s64): New function
3530 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
3533 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3534 * config/avr/avr.md (adjust_len): Add `round'.
3535 * config/avr/avr-protos.h (avr_out_round): New prototype.
3536 (avr_out_plus): Add `out_label' argument.
3537 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3538 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3539 Handle the case where `insn' is just a pattern.
3540 (avr_out_bitop): Handle the case where `insn' is just a pattern.
3541 (avr_out_round): New function.
3542 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3544 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
3546 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3547 Add check for flag_stack_usage to handle -fstack-usage support
3549 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
3551 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3552 interaction for new Power8 flags and VSX.
3554 2013-07-18 Sriraman Tallam <tmsriram@google.com>
3557 * tree-inline.c (expand_call_inline): Emit errors during
3558 early_inlining only if optimization is not turned on.
3560 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3564 * passes.c (init_optimization_passes): Move the construction of
3565 the pass hierarchy into a new passes.def file.
3567 * Makefile.in (passes.o): Add dependency on passes.def.
3569 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3571 * passes.c (init_optimization_passes): Introduce macros for
3572 constructing the tree of passes (INSERT_PASSES_AFTER,
3573 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3574 TERMINATE_PASS_LIST).
3576 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
3577 Wei Mi <wmi@google.com>
3579 PR rtl-optimization/57878
3580 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3582 (reload_pseudo_compare_func): Check nregs first for reload
3585 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3587 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3589 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
3591 * read-rtl.c (validate_const_int): Once an invalid character is
3592 seen, quit the loop.
3594 * gengtype.c (write_roots): Similarly once we find the "deletable"
3595 or "if_marked" option.
3597 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3599 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3600 "xtab" and "sat". Move value "clz" from here to ...
3601 (attriubte "type"): ... here.
3602 (satsi_<SAT:code>): Delete "insn" attribute.
3603 (satsi_<SAT:code>_shift): Likewise.
3604 (arm_zero_extendqisi2addsi): Likewise.
3605 (arm_extendqisi2addsi): Likewise.
3606 (clzsi2): Update for attribute changes.
3607 (rbitsi2): Likewise.
3608 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3610 (arm_usatsihi): Likewise.
3611 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3613 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3615 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3616 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
3617 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
3618 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
3619 in alphabetical order.
3620 (attribute "core_cycles"): Update for attribute changes.
3621 (arm_addsi3): Likewise.
3622 (addsi3_compare0): Likewise.
3623 (addsi3_compare0_scratch): Likewise.
3624 (addsi3_compare_op1): Likewise.
3625 (addsi3_compare_op2): Likewise.
3626 (compare_addsi2_op0): Likewise.
3627 (compare_addsi2_op1): Likewise.
3628 (addsi3_carryin_shift_<optab>): Likewise.
3629 (subsi3_carryin_shift): Likewise.
3630 (rsbsi3_carryin_shift): Likewise.
3631 (arm_subsi3_insn): Likewise.
3632 (subsi3_compare0): Likewise.
3633 (subsi3_compare): Likewise.
3634 (arm_andsi3_insn): Likewise.
3635 (thumb1_andsi3_insn): Likewise.
3636 (andsi3_compare0): Likewise.
3637 (andsi3_compare0_scratch): Likewise.
3638 (zeroextractsi_compare0_scratch
3639 (andsi_not_shiftsi_si): Likewise.
3640 (iorsi3_insn): Likewise.
3641 (iorsi3_compare0): Likewise.
3642 (iorsi3_compare0_scratch): Likewise.
3643 (arm_xorsi3): Likewise.
3644 (thumb1_xorsi3_insn): Likewise.
3645 (xorsi3_compare0): Likewise.
3646 (xorsi3_compare0_scratch): Likewise.
3647 (satsi_<SAT:code>_shift): Likewise.
3649 (arm_shiftsi3): Likewise.
3650 (shiftsi3_compare0): Likewise.
3651 (not_shiftsi): Likewise.
3652 (not_shiftsi_compare0): Likewise.
3653 (not_shiftsi_compare0_scratch): Likewise.
3654 (arm_one_cmplsi2): Likewise.
3655 (thumb_one_complsi2): Likewise.
3656 (notsi_compare0): Likewise.
3657 (notsi_compare0_scratch): Likewise.
3658 (thumb1_zero_extendhisi2): Likewise.
3659 (arm_zero_extendhisi2): Likewise.
3660 (arm_zero_extendhisi2_v6): Likewise.
3661 (arm_zero_extendhisi2addsi): Likewise.
3662 (thumb1_zero_extendqisi2): Likewise.
3663 (thumb1_zero_extendqisi2_v6): Likewise.
3664 (arm_zero_extendqisi2): Likewise.
3665 (arm_zero_extendqisi2_v6): Likewise.
3666 (arm_zero_extendqisi2addsi): Likewise.
3667 (thumb1_extendhisi2): Likewise.
3668 (arm_extendhisi2): Likewise.
3669 (arm_extendhisi2_v6): Likewise.
3670 (arm_extendqisi): Likewise.
3671 (arm_extendqisi_v6): Likewise.
3672 (arm_extendqisi2addsi): Likewise.
3673 (thumb1_extendqisi2): Likewise.
3674 (thumb1_movdi_insn): Likewise.
3675 (arm_movsi_insn): Likewise.
3676 (movsi_compare0): Likewise.
3677 (movhi_insn_arch4): Likewise.
3678 (movhi_bytes): Likewise.
3679 (arm_movqi_insn): Likewise.
3680 (thumb1_movqi_insn): Likewise.
3681 (arm32_movhf): Likewise.
3682 (thumb1_movhf): Likewise.
3683 (arm_movsf_soft_insn): Likewise.
3684 (thumb1_movsf_insn): Likewise.
3685 (movdf_soft_insn): Likewise.
3686 (thumb_movdf_insn): Likewise.
3687 (arm_cmpsi_insn): Likewise.
3688 (cmpsi_shiftsi): Likewise.
3689 (cmpsi_shiftsi_swp): Likewise.
3690 (arm_cmpsi_negshiftsi_si): Likewise.
3691 (movsicc_insn): Likewise.
3692 (movsfcc_soft_insn): Likewise.
3693 (arith_shiftsi): Likewise.
3694 (arith_shiftsi_compare0
3695 (arith_shiftsi_compare0_scratch
3696 (sub_shiftsi): Likewise.
3697 (sub_shiftsi_compare0
3698 (sub_shiftsi_compare0_scratch
3699 (and_scc): Likewise.
3700 (cond_move): Likewise.
3701 (if_plus_move): Likewise.
3702 (if_move_plus): Likewise.
3703 (if_move_not): Likewise.
3704 (if_not_move): Likewise.
3705 (if_shift_move): Likewise.
3706 (if_move_shift): Likewise.
3707 (if_shift_shift): Likewise.
3708 (if_not_arith): Likewise.
3709 (if_arith_not): Likewise.
3710 (cond_move_not): Likewise.
3711 (thumb1_ashlsi3): Set type attribute.
3712 (thumb1_ashrsi3): Likewise.
3713 (thumb1_lshrsi3): Likewise.
3714 (thumb1_rotrsi3): Likewise.
3715 (shiftsi3_compare0_scratch): Likewise.
3716 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3717 (neon_mov<mode>): Likewise.
3718 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3720 (thumb2_movsi_insn): Likewise.
3721 (thumb2_cmpsi_neg_shiftsi): Likewise.
3722 (thumb2_extendqisi_v6): Likewise.
3723 (thumb2_zero_extendhisi2_v6): Likewise.
3724 (thumb2_zero_extendqisi2_v6): Likewise.
3725 (thumb2_shiftsi3_short): Likewise.
3726 (thumb2_addsi3_compare0_scratch): Likewise.
3727 (orsi_not_shiftsi_si): Likewise.
3728 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3729 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3731 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3732 (1020alu_shift_op): Likewise.
3733 (1020alu_shift_reg_op): Likewise.
3734 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3735 (alu_shift_op): Likewise.
3736 (alu_shift_reg_op): Likewise.
3737 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3738 (11_alu_shift_op): Likewise.
3739 (11_alu_shift_reg_op): Likewise.
3740 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3741 (9_alu_shift_reg_op): Likewise.
3742 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3744 (cortex_a15_alu_shift): Likewise.
3745 (cortex_a15_alu_shift_reg): Likewise.
3746 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3748 (cortex_a5_alu_shift): Likewise.
3749 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3751 (cortex_a53_alu_shift): Likewise.
3752 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3754 (cortex_a7_alu_reg): Likewise.
3755 (cortex_a7_alu_shift): Likewise.
3756 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3758 (cortex_a8_alu_shift): Likewise.
3759 (cortex_a8_alu_shift_reg): Likewise.
3760 (cortex_a8_mov): Likewise.
3761 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3762 (cortex_a9_dp_shift): Likewise.
3763 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3765 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3767 (cortex_r4_mov): Likewise.
3768 (cortex_r4_alu_shift): Likewise.
3769 (cortex_r4_alu_shift_reg): Likewise.
3770 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3771 (526_alu_shift_op): Likewise.
3772 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3773 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3774 (626te_alu_shift_op): Likewise.
3775 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3776 (726te_alu_op): Likewise.
3777 (726te_alu_shift_op): Likewise.
3778 (726te_alu_shift_reg_op): Likewise.
3779 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3780 (mp626_alu_shift_op): Likewise.
3781 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3782 (pj4_alu_e1_conds): Likewise.
3783 (pj4_alu): Likewise.
3784 (pj4_alu_conds): Likewise.
3785 (pj4_shift): Likewise.
3786 (pj4_shift_conds): Likewise.
3787 (pj4_alu_shift): Likewise.
3788 (pj4_alu_shift_conds): Likewise.
3789 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3791 (cortexa7_older_only): Likewise.
3792 (cortexa7_younger): Likewise.
3794 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3796 * ipa-pure-const.c (generate_summary): Rename to...
3797 (pure_const_generate_summary): ... this.
3799 2013-07-17 Iain Sandoe <iain@codesourcery.com>
3801 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3803 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
3806 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3809 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3811 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3812 enabled without -march=zEC12.
3813 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3816 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3818 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3819 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3820 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3821 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3822 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3823 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3824 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3826 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3827 and ISA_HAS_NMADD3_NMSUB3.
3828 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3829 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3830 (nmsub4<mode>, nmsub3<mode>): Likewise.
3831 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3833 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3835 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3836 TARGET_MIPS5400 checking.
3838 2013-07-16 Jakub Jelinek <jakub@redhat.com>
3839 Peter Bergner <bergner@vnet.ibm.com>
3841 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3842 registers in the comment.
3843 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3844 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3845 rather than FIRST_PSEUDO_REGISTERS.
3847 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
3849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3850 enable extra ISA flags with TARGET_HTM.
3852 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3854 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3857 2013-07-15 Cong Hou <congh@google.com>
3859 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3860 in compare function for sorting.
3862 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
3864 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3865 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3866 * config/rs6000/rs6000.opt: Add -mhtm option.
3867 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3868 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3869 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3870 __HTM__ if the HTM instructions are available.
3871 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3872 htm_spr_reg_operand): New define_predicates.
3873 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3874 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3876 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3877 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3878 HTM builtin functions.
3879 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3880 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3881 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3882 (rs6000_builtin_mask_calculate): Likewise.
3883 (rs6000_option_override_internal): Likewise.
3884 (bdesc_htm): Add new HTM builtin support.
3885 (htm_spr_num): New function.
3886 (htm_spr_regno): Likewise.
3887 (rs6000_htm_spr_icode): Likewise.
3888 (htm_expand_builtin): Likewise.
3889 (htm_init_builtins): Likewise.
3890 (rs6000_expand_builtin): Add support for HTM builtin functions.
3891 (rs6000_init_builtins): Likewise.
3892 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3894 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3895 (TARGET_HTM, MASK_HTM): Define macros.
3896 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3897 (FIXED_REGISTERS): Likewise.
3898 (CALL_USED_REGISTERS): Likewise.
3899 (CALL_REALLY_USED_REGISTERS): Likewise.
3900 (REG_ALLOC_ORDER): Likewise.
3901 (enum reg_class): Likewise.
3902 (REG_CLASS_NAMES): Likewise.
3903 (REG_CLASS_CONTENTS): Likewise.
3904 (REGISTER_NAMES): Likewise.
3905 (ADDITIONAL_REGISTER_NAMES): Likewise.
3906 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3907 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3908 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3909 * config/rs6000/htm.md: New file.
3910 * config/rs6000/htmintrin.h: New file.
3911 * config/rs6000/htmxlintrin.h: New file.
3913 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
3915 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3916 Define SYMBOL_TINY_GOT, update comment.
3917 * config/aarch64/aarch64.c
3918 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3919 (aarch64_expand_mov_immediate): Likewise.
3920 (aarch64_print_operand): Likewise.
3921 (aarch64_classify_symbol): Likewise.
3922 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3923 (ldr_got_tiny): Define.
3925 2013-07-13 Tobias Grosser <tobias@grosser.es>
3927 PR tree-optimization/54094
3928 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3929 scheduling dimension for the parallelism check from the polyhedral
3930 information in the AST.
3931 * graphite-dependences.c (carries_deps): Do not assume the schedule is
3934 2013-07-13 Jason Merrill <jason@redhat.com>
3936 * print-tree.c (debug_vec_tree): Use debug_raw.
3937 (debug_raw (vec<tree, va_gc> &)): New.
3938 (debug_raw (vec<tree, va_gc> *)): New.
3939 * tree.h: Declare them.
3941 2013-07-13 Bin Cheng <bin.cheng@arm.com>
3943 * ifcvt.c (ifcvt_after_combine): New static variable.
3944 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3946 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
3947 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3948 rest_of_handle_if_after_reload): Pass new argument for if_convert.
3950 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
3952 * config/mips/mips.c (mips_expand_call): Remove empty statement.
3954 2013-07-12 Michael Matz <matz@suse.de>
3957 * convert.c (convert_to_real): Reject non-float inner types.
3959 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
3961 * config/aarch64/aarch64-protos.h
3962 (aarch64_simd_immediate_valid_for_move): Remove.
3963 * config/aarch64/aarch64.c (simd_immediate_info): New member.
3964 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3966 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3968 2013-07-11 Steve Ellcey <sellcey@mips.com>
3970 * config/mips/mips.c (mips_conditional_register_usage): Do not
3971 use t[0-7] registers in MIPS16 mode when optimizing for size.
3973 2013-07-11 Sriraman Tallam <tmsriram@google.com>
3975 * config/i386/i386.c (dispatch_function_versions): Fix array
3976 indexing of function_version_info to match actual_versions.
3978 2013-07-11 Teresa Johnson <tejohnson@google.com>
3980 * vec.h (struct va_gc): Move release out-of-line.
3981 (va_gc::release): Call ggc_free on released vec.
3983 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3985 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3986 Require GOT register as additional operand in UNSPEC.
3987 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3988 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3989 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3990 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3991 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3992 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3993 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3995 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
3998 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3999 name seen by assembler/linker rather if available.
4001 2013-07-11 Andreas Schwab <schwab@suse.de>
4003 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4005 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
4007 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4009 2013-07-10 Joseph Myers <joseph@codesourcery.com>
4011 * doc/tm.texi.in: Move hook documentation to ....
4012 * target.def: ... here.
4014 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4016 * doc/tm.texi: Regenerate.
4018 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
4021 * doc/invoke.texi: Document -Wconditionally-supported.
4023 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4026 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4029 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4032 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4033 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4034 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4035 Remove duplicate devices.
4036 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4037 * config/avr/t-multilib: Regenerate.
4038 * config/avr/avr-tables.opt: Regenerate.
4039 * doc/avr-mmcu.texi: Regenerate.
4041 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4044 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4046 2013-07-10 Graham Stott <graham.stott@btinternet.com>
4048 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4049 the cost of MULT when optimizing for size.
4051 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4053 * config/cr16/cr16-protos.h: Don't include target.h.
4055 2013-07-09 Joseph Myers <joseph@codesourcery.com>
4057 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4058 adjust register size for TDmode and TFmode for VSX registers.
4060 2013-07-08 Kai Tietz <ktietz@redhat.com>
4063 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4064 hook_bool_const_tree_true.
4066 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4068 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4069 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4070 * config/s390/s390.md: Define FPR*_REGNUM constants.
4071 Fix FPR2_REGNUM constant (18 -> 17).
4072 ("*trunc<BFP:mode><DFP_ALL:mode>2")
4073 ("*trunc<DFP_ALL:mode><BFP:mode>2")
4074 ("trunc<BFP:mode><DFP_ALL:mode>2")
4075 ("trunc<DFP_ALL:mode><BFP:mode>2")
4076 ("*extend<BFP:mode><DFP_ALL:mode>2")
4077 ("*extend<DFP_ALL:mode><BFP:mode>2")
4078 ("extend<BFP:mode><DFP_ALL:mode>2")
4079 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4082 2013-07-08 Graham Stott <graham.stott@btinternet.com>
4084 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4086 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4088 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4089 and cfun_fpr_bit_p to cfun_fpr_save_p.
4090 (s390_frame_area, s390_register_info, s390_frame_info)
4091 (s390_emit_prologue, s390_emit_epilogue)
4092 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4094 * config/s390/s390.h: Define *_REGNUM macros for floating point
4097 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
4099 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4101 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
4103 PR rtl-optimization/57786
4104 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4105 and break out of the loop when it is set to false.
4107 2013-07-08 Jakub Jelinek <jakub@redhat.com>
4110 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4111 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4112 (const_int 63)) 0)).
4113 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4114 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4115 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4117 PR rtl-optimization/57829
4118 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4119 mask bits outside of mode are just sign-extension from mode to HWI.
4121 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
4123 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4124 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4125 adjust_address instead of change_address to keep info about alignment.
4126 (emit_strmov): Remove.
4127 (emit_memmov): New function.
4128 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4129 (expand_movmem_epilogue): Likewise and return updated rtx for
4131 (expand_constant_movmem_prologue): Likewise and return updated rtx for
4132 destination and source.
4133 (decide_alignment): Refactor, handle vector_loop.
4134 (ix86_expand_movmem): Likewise.
4135 (ix86_expand_setmem): Likewise.
4136 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4138 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
4140 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4141 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4143 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4145 * config/i386/sse.md (sse_movlhps): Change alternative 3
4146 of operand 2 to "m".
4148 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4151 * config/i386/sse.md (iptr): New mode attribute.
4152 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4153 (<sse>_vm<plusminus_insn><mode>3): Ditto.
4154 (<sse>_vmmul<mode>3): Ditto.
4155 (<sse>_vmdiv<mode>3): Ditto.
4156 (sse_vmrcpv4sf2): Ditto.
4157 (<sse>_vmsqrt<mode>2): Ditto.
4158 (sse_vmrsqrtv4sf2): Ditto.
4159 (<sse>_vm<code><mode>3): Ditto.
4160 (avx_vmcmp<mode>3): Ditto.
4161 (<sse>_vmmaskcmp<mode>3): Ditto.
4162 (<sse>_comi): Ditto.
4163 (<sse>_ucomi): Ditto.
4164 (*xop_vmfrcz_<mode>): Ditto.
4165 (*fmai_fmadd_<mode>): Ditto.
4166 (*fmai_fmsub_<mode>): Ditto.
4167 (*fmai_fnmadd_<mode>): Ditto.
4168 (*fmai_fnmsub_<mode>): Ditto.
4169 (*fma4i_vmfmadd_<mode>): Ditto.
4170 (*fma4i_vmfmsub_<mode>): Ditto.
4171 (*fma4i_vmfnmadd_<mode>): Ditto.
4172 (*fma4i_vmfnmsub_<mode>): Ditto.
4173 (*xop_vmfrcz_<mode>): Ditto.
4174 (sse_cvtps2pi): Ditto.
4175 (sse_cvttps2pi): Ditto.
4176 (sse_cvtss2si): Ditto.
4177 (sse_cvtss2si_2): Ditto.
4178 (sse_cvtss2siq_2): Ditto.
4179 (sse_cvttss2si): Ditto.
4180 (sse_cvttss2siq): Ditto.
4181 (sse_cvtsd2si): Ditto.
4182 (sse_cvtsd2si_2): Ditto.
4183 (sse_cvtsd2siq_2): Ditto.
4184 (sse_cvttsd2si): Ditto.
4185 (sse_cvttsd2siq): Ditto.
4186 (sse_cvtsd2ss): Ditto.
4187 (sse_cvtss2sd): Ditto.
4188 (avx2_pbroadcast<mode>): Ditto.
4189 (avx2_pbroadcast<mode>_1): Ditto.
4190 (*avx_vperm_broadcast_v4sf): Ditto.
4192 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4193 (sse_movlhps): Ditto.
4194 (sse_storehps): Ditto.
4195 (sse_loadhps): Ditto.
4196 (sse_storelps): Ditto.
4197 (sse_loadlps): Ditto.
4198 (*vec_concatv4sf): Ditto.
4199 (*vec_interleave_highv2df): Ditto.
4200 (*vec_interleave_lowv2df): Ditto.
4201 (*vec_extractv2df_1_sse): Ditto.
4202 (*vec_extractv2df_0_sse): Ditto.
4203 (sse2_storelpd): Ditto.
4204 (sse2_loadlpd): Ditto.
4205 (sse2_movsd): Ditto.
4206 (*vec_concatv4si): Ditto.
4207 (vec_concatv2di): Ditto.
4209 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4210 for Intel asm dialect.
4211 (mmx_punpcklwd): Ditto.
4212 (mmx_punpckldq): Ditto.
4214 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4215 for intel assembler dialect.
4217 2013-07-06 Jakub Jelinek <jakub@redhat.com>
4220 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4221 for BUILT_IN_C{LZ,LRSB}*.
4222 * tree.h (CASE_INT_FN): Add FN##IMAX case.
4223 * tree-vrp.c (extract_range_basic): Handle
4224 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
4225 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4226 fall thru to code calling set_value*.
4227 * builtins.c (expand_builtin): Remove *IMAX cases.
4228 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4229 if width is bigger than 2*HWI.
4231 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
4233 PR rtl-optimization/55342
4234 * lra-int.h (lra_subreg_reload_pseudos): New.
4235 * lra.c: Add undoing optional reloads to the block diagram.
4236 (lra_subreg_reload_pseudos): New.
4237 (lra_optional_reload_pseudos): Change comments.
4238 (lra): Init and clear lra_subreg_reload_pseudos. Clear
4239 lra_optional_reload_pseudos after undo transformations.
4240 * lra-assigns.c (pseudo_prefix_title): New.
4241 (lra_setup_reg_renumber): Use it.
4242 (spill_for): Ditto. Check subreg reload pseudos too.
4243 (assign_by_spills): Consider subreg reload pseudos too.
4244 * lra-constraints.c (simplify_operand_subreg): Use
4245 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4246 (curr_insn_transform): Recognize and do optional reloads.
4247 (undo_optional_reloads): New.
4248 (lra_undo_inheritance): Call undo_optional_reloads.
4250 2013-07-05 Thomas Quinot <quinot@adacore.com>
4252 * tree-complex.c (expand_complex_operations_1): Fix typo.
4254 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
4256 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4257 (tune_params): New member 'const vec_costs'.
4258 * config/aarch64/aarch64.c (generic_vector_cost): New.
4259 (generic_tunings): New member 'generic_vector_cost'.
4260 (aarch64_builtin_vectorization_cost): New.
4261 (aarch64_add_stmt_cost): New.
4262 (TARGET_VECTORIZE_ADD_STMT_COST): New.
4263 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4265 2013-07-03 Jakub Jelinek <jakub@redhat.com>
4268 * config/i386/predicates.md (vsib_address_operand): Disallow
4269 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4271 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
4274 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4275 expand_builtin_setjmp_receiver.
4276 (expand_label): Adjust, call expand_builtin_setjmp_receiver
4277 with NULL for the label parameter.
4278 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4279 the frame-pointer. Adjust comments.
4280 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4281 only if LABEL is non-NULL.
4283 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
4285 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4286 (ARM_ABI_AAPCS64): Ditto.
4288 (ARM_DEFAULT_ABI): Ditto.
4290 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
4292 * config/aarch64/aarch64-builtins.c
4293 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4294 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4296 * config/aarch64/aarch64-simd.md
4297 (aarch64_ld1<VALL:mode>): New.
4298 (aarch64_st1<VALL:mode>): Likewise.
4299 * config/aarch64/arm_neon.h
4300 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4302 2013-07-02 Sriraman Tallam <tmsriram@google.com>
4304 * config/i386/i386.c (gate_insert_vzeroupper): Check if
4306 (ix86_option_override_internal):Turn on all -mavx target flags by
4307 default as they are dependent on AVX anyway.
4309 2013-07-02 Cary Coutant <ccoutant@google.com>
4311 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4313 (loc_checksum_ordered): Likewise.
4314 (hash_loc_operands): Remove inline keyword.
4316 2013-07-02 Jakub Jelinek <jakub@redhat.com>
4318 PR tree-optimization/57741
4319 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4320 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4321 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4322 Allow REAL_CST step_exprs if flag_associative_math.
4323 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4325 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4327 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4329 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4331 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4333 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4335 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4337 (iorsi3_insn): Likewise.
4338 (arm_xorsi3): Likewise.
4340 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
4342 * arm.md (attribute "wtype"): Delete. Move attribute values from here
4344 (attribute "type"): ... here, and prefix with "wmmx_".
4345 (attribute "core_cycles"): Update for attribute changes.
4346 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4347 (tbcstv4hi): Likewise.
4348 (tbcstv2si): Likewise.
4349 (iwmmxt_iordi3): Likewise.
4350 (iwmmxt_xordi3): Likewise.
4351 (iwmmxt_anddi3): Likewise.
4352 (iwmmxt_nanddi3): Likewise.
4353 (iwmmxt_arm_movdi): Likewise.
4354 (iwmmxt_movsi_insn): Likewise.
4355 (mov<mode>_internal): Likewise.
4356 (and<mode>3_iwmmxt): Likewise.
4357 (ior<mode>3_iwmmxt): Likewise.
4358 (xor<mode>3_iwmmxt): Likewise.
4359 (add<mode>3_iwmmxt): Likewise.
4360 (ssaddv8qi3): Likewise.
4361 (ssaddv4hi3): Likewise.
4362 (ssaddv2si3): Likewise.
4363 (usaddv8qi3): Likewise.
4364 (usaddv4hi3): Likewise.
4365 (usaddv2si3): Likewise.
4366 (sub<mode>3_iwmmxt): Likewise.
4367 (sssubv8qi3): Likewise.
4368 (sssubv4hi3): Likewise.
4369 (sssubv2si3): Likewise.
4370 (ussubv8qi3): Likewise.
4371 (ussubv4hi3): Likewise.
4372 (ussubv2si3): Likewise.
4373 (mulv4hi3_iwmmxt): Likewise.
4374 (smulv4hi3_highpart): Likewise.
4375 (umulv4hi3_highpart): Likewise.
4376 (iwmmxt_wmacs): Likewise.
4377 (iwmmxt_wmacsz): Likewise.
4378 (iwmmxt_wmacu): Likewise.
4379 (iwmmxt_wmacuz): Likewise.
4380 (iwmmxt_clrdi): Likewise.
4381 (iwmmxt_clrv8qi): Likewise.
4382 (iwmmxt_clr4hi): Likewise.
4383 (iwmmxt_clr2si): Likewise.
4384 (iwmmxt_uavgrndv8qi3): Likewise.
4385 (iwmmxt_uavgrndv4hi3): Likewise.
4386 (iwmmxt_uavgv8qi3): Likewise.
4387 (iwmmxt_uavgv4hi3): Likewise.
4388 (iwmmxt_tinsrb): Likewise.
4389 (iwmmxt_tinsrh): Likewise.
4390 (iwmmxt_tinsrw): Likewise.
4391 (iwmmxt_textrmub): Likewise.
4392 (iwmmxt_textrmsb): Likewise.
4393 (iwmmxt_textrmuh): Likewise.
4394 (iwmmxt_textrmsh): Likewise.
4395 (iwmmxt_textrmw): Likewise.
4396 (iwmxxt_wshufh): Likewise.
4397 (eqv8qi3): Likewise.
4398 (eqv4hi3): Likewise.
4399 (eqv2si3): Likewise.
4400 (gtuv8qi3): Likewise.
4401 (gtuv4hi3): Likewise.
4402 (gtuv2si3): Likewise.
4403 (gtv8qi3): Likewise.
4404 (gtv4hi3): Likewise.
4405 (gtv2si3): Likewise.
4406 (smax<mode>3_iwmmxt): Likewise.
4407 (umax<mode>3_iwmmxt): Likewise.
4408 (smin<mode>3_iwmmxt): Likewise.
4409 (umin<mode>3_iwmmxt): Likewise.
4410 (iwmmxt_wpackhss): Likewise.
4411 (iwmmxt_wpackwss): Likewise.
4412 (iwmmxt_wpackdss): Likewise.
4413 (iwmmxt_wpackhus): Likewise.
4414 (iwmmxt_wpackwus): Likewise.
4415 (iwmmxt_wpackdus): Likewise.
4416 (iwmmxt_wunpckihb): Likewise.
4417 (iwmmxt_wunpckihh): Likewise.
4418 (iwmmxt_wunpckihw): Likewise.
4419 (iwmmxt_wunpckilb): Likewise.
4420 (iwmmxt_wunpckilh): Likewise.
4421 (iwmmxt_wunpckilw): Likewise.
4422 (iwmmxt_wunpckehub): Likewise.
4423 (iwmmxt_wunpckehuh): Likewise.
4424 (iwmmxt_wunpckehuw): Likewise.
4425 (iwmmxt_wunpckehsb): Likewise.
4426 (iwmmxt_wunpckehsh): Likewise.
4427 (iwmmxt_wunpckehsw): Likewise.
4428 (iwmmxt_wunpckelub): Likewise.
4429 (iwmmxt_wunpckeluh): Likewise.
4430 (iwmmxt_wunpckeluw): Likewise.
4431 (iwmmxt_wunpckelsb): Likewise.
4432 (iwmmxt_wunpckelsh): Likewise.
4433 (iwmmxt_wunpckelsw): Likewise.
4434 (ror<mode>3): Likewise.
4435 (ashr<mode>3_iwmmxt): Likewise.
4436 (lshr<mode>3_iwmmxt): Likewise.
4437 (ashl<mode>3_iwmmxt): Likewise.
4438 (ror<mode>3_di): Likewise.
4439 (ashr<mode>3_di): Likewise.
4440 (lshr<mode>3_di): Likewise.
4441 (ashl<mode>3_di): Likewise.
4442 (iwmmxt_wmadds): Likewise.
4443 (iwmmxt_wmaddu): Likewise.
4444 (iwmmxt_tmia): Likewise.
4445 (iwmmxt_tmiaph): Likewise.
4446 (iwmmxt_tmiabb): Likewise.
4447 (iwmmxt_tmiatb): Likewise.
4448 (iwmmxt_tmiabt): Likewise.
4449 (iwmmxt_tmiatt): Likewise.
4450 (iwmmxt_tmovmskb): Likewise.
4451 (iwmmxt_tmovmskh): Likewise.
4452 (iwmmxt_tmovmskw): Likewise.
4453 (iwmmxt_waccb): Likewise.
4454 (iwmmxt_wacch): Likewise.
4455 (iwmmxt_waccw): Likewise.
4456 (iwmmxt_waligni): Likewise.
4457 (iwmmxt_walignr): Likewise.
4458 (iwmmxt_walignr0): Likewise.
4459 (iwmmxt_walignr1): Likewise.
4460 (iwmmxt_walignr2): Likewise.
4461 (iwmmxt_walignr3): Likewise.
4462 (iwmmxt_wsadb): Likewise.
4463 (iwmmxt_wsadh): Likewise.
4464 (iwmmxt_wsadbz): Likewise.
4465 (iwmmxt_wsadhz): Likewise.
4466 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4467 (iwmmxt_wabsdiffb): Likewise.
4468 (iwmmxt_wabsdiffh): Likewise.
4469 (iwmmxt_wabsdiffw): Likewise.
4470 (iwmmxt_waddsubhx): Likewise
4471 (iwmmxt_wsubaddhx): Likewise.
4472 (addc<mode>3): Likewise.
4473 (iwmmxt_avg4): Likewise.
4474 (iwmmxt_avg4r): Likewise.
4475 (iwmmxt_wmaddsx): Likewise.
4476 (iwmmxt_wmaddux): Likewise.
4477 (iwmmxt_wmaddsn): Likewise.
4478 (iwmmxt_wmaddun): Likewise.
4479 (iwmmxt_wmulwsm): Likewise.
4480 (iwmmxt_wmulwum): Likewise.
4481 (iwmmxt_wmulsmr): Likewise.
4482 (iwmmxt_wmulumr): Likewise.
4483 (iwmmxt_wmulwsmr): Likewise.
4484 (iwmmxt_wmulwumr): Likewise.
4485 (iwmmxt_wmulwl): Likewise.
4486 (iwmmxt_wqmulm): Likewise.
4487 (iwmmxt_wqmulwm): Likewise.
4488 (iwmmxt_wqmulmr): Likewise.
4489 (iwmmxt_wqmulwmr): Likewise.
4490 (iwmmxt_waddbhusm): Likewise.
4491 (iwmmxt_waddbhusl): Likewise.
4492 (iwmmxt_wqmiabb): Likewise.
4493 (iwmmxt_wqmiabt): Likewise.
4494 (iwmmxt_wqmiatb): Likewise.
4495 (iwmmxt_wqmiatt): Likewise.
4496 (iwmmxt_wqmiabbn): Likewise.
4497 (iwmmxt_wqmiabtn): Likewise.
4498 (iwmmxt_wqmiatbn): Likewise.
4499 (iwmmxt_wqmiattn): Likewise.
4500 (iwmmxt_wmiabb): Likewise.
4501 (iwmmxt_wmiabt): Likewise.
4502 (iwmmxt_wmiatb): Likewise.
4503 (iwmmxt_wmiatt): Likewise.
4504 (iwmmxt_wmiabbn): Likewise.
4505 (iwmmxt_wmiabtn): Likewise.
4506 (iwmmxt_wmiatbn): Likewise.
4507 (iwmmxt_wmiattn): Likewise.
4508 (iwmmxt_wmiawbb): Likewise.
4509 (iwmmxt_wmiawbt): Likewise.
4510 (iwmmxt_wmiawtb): Likewise.
4511 (iwmmxt_wmiawtt): Likewise.
4512 (iwmmxt_wmiawbbn): Likewise.
4513 (iwmmxt_wmiawbtn): Likewise.
4514 (iwmmxt_wmiawtbn): Likewise.
4515 (iwmmxt_wmiawttn): Likewise.
4516 (iwmmxt_wmerge): Likewise.
4517 (iwmmxt_tandc<mode>3): Likewise.
4518 (iwmmxt_torc<mode>3): Likewise.
4519 (iwmmxt_torvsc<mode>3): Likewise.
4520 (iwmmxt_textrc<mode>3): Likewise.
4521 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4522 (wmmxt_pack): Likewise.
4523 (wmmxt_mult_c1): Likewise.
4524 (wmmxt_mult_c2): Likewise.
4525 (wmmxt_alu_c1): Likewise.
4526 (wmmxt_alu_c2): Likewise.
4527 (wmmxt_alu_c3): Likewise.
4528 (wmmxt_transfer_c1): Likewise.
4529 (wmmxt_transfer_c2): Likewise.
4530 (wmmxt_transfer_c3): Likewise.
4531 (marvell_f_iwmmxt_wstr): Likewise.
4532 (marvell_f_iwmmxt_wldr): Likewise.
4534 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
4536 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4538 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4541 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4542 * lra-constraints.c (need_for_split_p): Check call used hard regs
4543 living through calls.
4545 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4546 call used regs for call insn.
4548 2013-06-28 Jakub Jelinek <jakub@redhat.com>
4551 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4552 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4554 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
4556 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4558 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4560 * lra-constraints.c (need_for_split_p): Check call used hard regs
4561 living through calls.
4563 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4566 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4567 to tie with any other modes. Eliminate Altivec vector mode tests,
4568 since these are a subset of ALTIVEC or VSX vector modes. Simplify
4569 code, to return 0 if testing MODE2 for a condition, if we've
4570 already tested MODE1 for the same condition.
4572 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4574 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4577 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4579 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4580 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4582 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4584 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4586 (aarch64_symbolic_constant_p): Remove.
4587 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4588 static. Fix line length and white space.
4589 (aarch64_symbolic_constant_p): Remove.
4590 * config/aarch64/predicates.md (aarch64_valid_symref):
4591 Use aarch64_classify_symbol_expression.
4593 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4595 * config/arm/constraints.md (Ts): New constraint.
4596 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4598 (compare_scc): Use "Ts" constraint for operand 0.
4599 (ior_scc_scc): Likewise.
4600 (and_scc_scc): Likewise.
4601 (and_scc_scc_nodom): Likewise.
4602 (ior_scc_scc_cmp): Likewise for operand 7.
4603 (and_scc_scc_cmp): Likewise.
4604 * config/arm/thumb2.md (thumb2_movsi_insn):
4605 Add alternatives for 16-bit encodings.
4606 (thumb2_movhi_insn): Likewise.
4607 (thumb2_movsicc_insn): Likewise.
4608 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
4609 (thumb2_negscc): Use "Ts" constraint.
4610 Move mvn instruction outside cond_exec block.
4611 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4612 for 16-bit encodings.
4614 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4616 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4618 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4619 (mulsi3subsi): Likewise.
4620 (mulsidi3adddi): Likewise.
4621 (mulsidi3_v6): Likewise.
4622 (umulsidi3_v6): Likewise.
4623 (umulsidi3adddi_v6): Likewise.
4624 (smulsi3_highpart_v6): Likewise.
4625 (umulsi3_highpart_v6): Likewise.
4626 (mulhisi3tb): Likewise.
4627 (mulhisi3bt): Likewise.
4628 (mulhisi3tt): Likewise.
4629 (maddhisi4): Likewise.
4630 (maddhisi4tb): Likewise.
4631 (maddhisi4tt): Likewise.
4632 (maddhidi4): Likewise.
4633 (maddhidi4tb): Likewise.
4634 (maddhidi4tt): Likewise.
4635 (zeroextractsi_compare0_scratch): Likewise.
4636 (insv_zero): Likewise.
4637 (insv_t2): Likewise.
4638 (anddi_notzesidi_di): Likewise.
4639 (anddi_notsesidi_di): Likewise.
4640 (andsi_notsi_si): Likewise.
4641 (iordi_zesidi_di): Likewise.
4642 (xordi_zesidi_di): Likewise.
4643 (andsi_iorsi3_notsi): Likewise.
4645 (smax_m1): Likewise.
4647 (not_shiftsi): Likewise.
4648 (unaligned_loadsi): Likewise.
4649 (unaligned_loadhis): Likewise.
4650 (unaligned_loadhiu): Likewise.
4651 (unaligned_storesi): Likewise.
4652 (unaligned_storehi): Likewise.
4653 (extv_reg): Likewise.
4654 (extzv_t2): Likewise.
4656 (udivsi3): Likewise.
4657 (arm_zero_extendhisi2addsi): Likewise.
4658 (arm_zero_extendqisi2addsi): Likewise.
4659 (compareqi_eq0): Likewise.
4660 (arm_extendhisi2_v6): Likewise.
4661 (arm_extendqisi2addsi): Likewise.
4662 (arm_movt): Likewise.
4663 (thumb2_ldrd): Likewise.
4664 (thumb2_ldrd_base): Likewise.
4665 (thumb2_ldrd_base_neg): Likewise.
4666 (thumb2_strd): Likewise.
4667 (thumb2_strd_base): Likewise.
4668 (thumb2_strd_base_neg): Likewise.
4669 (arm_negsi2): Add alternative for 16-bit encoding.
4670 (arm_one_cmplsi2): Likewise.
4672 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4674 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4675 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4676 (movdfcc): Likewise.
4677 * config/arm/vfp.md (*thumb2_movsf_vfp):
4678 Disable predication for arm_restrict_it.
4679 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4680 (*thumb2_movdfcc_vfp): Likewise.
4681 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4682 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4683 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4684 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4685 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4686 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4687 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4688 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4689 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4690 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4691 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4692 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4693 Disable predication for arm_restrict_it.
4695 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
4697 * config/i386/bmiintrin.h (_bextr_u32): New.
4698 (_bextr_u64): Ditto.
4700 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
4702 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4703 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4704 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4705 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4706 * config/mips/n32-elf.h: ...this new file.
4708 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
4711 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4712 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4714 2013-06-27 Catherine Moore <clm@codesourcery.com>
4716 * config/mips/mips-tables.opt: Regenerate.
4717 * config/mips/mips-cpus.def: Add m14ke and m14kec.
4718 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4719 * doc/invoke.texi: Add -m14kc.
4721 2013-06-27 Jakub Jelinek <jakub@redhat.com>
4724 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4725 constraints of operand 1 and 2.
4728 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4729 to match RTL canonicalization. Swap predicates and
4730 constraints of operand 1 and 2.
4732 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
4734 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4735 Process OP_INOUT regs for splitting too.
4737 2013-06-27 Jakub Jelinek <jakub@redhat.com>
4739 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4740 decl before the loop, initialize to NULL.
4741 (vectorizable_load): Initialize ptr_incr to NULL.
4743 2013-06-27 Martin Jambor <mjambor@suse.cz>
4746 * ipa-ref.h (ipa_maybe_record_reference): Declare.
4747 * ipa-ref.c (ipa_maybe_record_reference): New function.
4748 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4749 * ipa-cp.c (create_specialized_node): Record potential references from
4751 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4753 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
4755 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4756 parameter 'mode' of type 'enum machine_mode mode'; change to pass
4757 'mode' to force_reg.
4758 (aarch64_add_offset): Update calls to aarch64_force_temporary.
4759 (aarch64_expand_mov_immediate): Likewise.
4761 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
4763 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4764 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4766 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4768 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4769 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4770 (struct machine_function): Add tbegin_p.
4771 (s390_canonicalize_comparison): Fold CC mode compares to
4772 conditional jump if possible.
4773 (s390_emit_jump): Return the emitted jump.
4774 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4775 Handle CCRAWmode compares.
4776 (s390_option_override): Default to -mhtm if available.
4777 (s390_reg_clobbered_rtx): Handle floating point regs as well.
4778 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4779 FPRs instead of df_regs_ever_live_p.
4780 (s390_optimize_nonescaping_tx): New function.
4781 (s390_init_frame_layout): Extend clobbered_regs array to cover
4783 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4784 (s390_expand_tbegin): New function.
4785 (enum s390_builtin): New enum definition.
4786 (code_for_builtin): New array definition.
4787 (s390_init_builtins): New function.
4788 (s390_expand_builtin): New function.
4789 (TARGET_INIT_BUILTINS): Define.
4790 (TARGET_EXPAND_BUILTIN): Define.
4791 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4792 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4793 (s390_alc_comparison): Likewise.
4794 * config/s390/s390-modes.def: Add CCRAWmode.
4795 * config/s390/s390.h (processor_flags): Add PF_TX.
4796 (TARGET_CPU_HTM): Define macro.
4797 (TARGET_HTM): Define macro.
4798 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4799 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4800 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4801 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4802 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4803 (TBEGIN_MASK, TBEGINC_MASK): New constants.
4804 ("*cc_to_int"): Move up.
4805 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4806 constants other than 0.
4807 ("*ccraw_to_int"): New insn and splitter definition.
4808 ("tbegin", "tbegin_nofloat", "tbegin_retry")
4809 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4810 ("tx_assist"): New expander.
4811 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4812 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4813 * config/s390/s390.opt: Add -mhtm option.
4814 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4815 * config/s390/htmxlintrin.h: New file.
4816 * config/s390/htmintrin.h: New file.
4817 * config/s390/s390intrin.h: New file.
4818 * doc/extend.texi: Document htm builtins.
4819 * config.gcc: Add the new header files to extra_headers.
4821 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
4823 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4824 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4826 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4827 Pat Haugen <pthaugen@us.ibm.com>
4828 Peter Bergner <bergner@vnet.ibm.com>
4830 * config/rs6000/power8.md: New.
4831 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4832 setting for power8 entry.
4833 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4834 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4835 test for Power4/Power5 only.
4836 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4838 (force_new_group): Adjust comment.
4839 * config/rs6000/rs6000.md: Include power8.md.
4841 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
4843 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4844 * config/arm/arm-protos.h (arm_max_conditional_execute): New
4846 (tune_params): Update comment.
4847 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4848 (arm_max_conditional_execute): New function.
4849 (thumb2_final_prescan_insn): Use max_insn_skipped and
4850 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4852 2013-06-25 Jakub Jelinek <jakub@redhat.com>
4854 PR tree-optimization/57705
4855 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4856 SSA_NAME step, provided that it is not defined inside the loop.
4857 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4858 (get_initial_def_for_induction): Handle SSA_NAME IV step.
4860 2013-06-25 Martin Jambor <mjambor@suse.cz>
4863 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4864 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4866 (ipa_note_param_call): Initialize member_ptr flag.
4867 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4868 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4869 (ipa_write_indirect_edge_info): Stream member_ptr flag.
4870 (ipa_read_indirect_edge_info): Likewise.
4872 2013-06-25 Richard Biener <rguenther@suse.de>
4875 * passes.c (init_optimization_passes): Move pass_fold_builtins
4876 and pass_dce earlier with -Og.
4878 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
4880 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4881 <BIT_FIELD_REF>: Remove trailing TAB.
4882 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4885 2013-06-24 Martin Jambor <mjambor@suse.cz>
4887 PR tree-optimization/57358
4888 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4889 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4890 (ipa_analyze_params_uses): Generate pessimistic info when true.
4892 2013-06-24 Martin Jambor <mjambor@suse.cz>
4894 PR tree-optimization/57539
4895 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4896 global.inlined_to of the new node to it. All callers changed.
4897 * ipa-inline-transform.c (clone_inlined_nodes): New variable
4898 inlining_into, pass it to cgraph_clone_node.
4899 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4900 ipa_free_edge_args_substructures.
4901 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4902 rdesc linked list. Do not assert rdesc edges have inlined caller.
4903 Assert we have found an rdesc in the rdesc list.
4905 2013-06-24 Richard Biener <rguenther@suse.de>
4907 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4908 (pointer_set_lookup): Declare.
4909 (class pointer_map): New template class implementing a
4910 generic pointer to T map.
4911 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4912 pointer_map<T>::contains, pointer_map<T>::insert,
4913 pointer_map<T>::traverse): New functions.
4914 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4915 (pointer_set_lookup): New function.
4916 (pointer_set_contains): Use pointer_set_lookup.
4917 (pointer_set_insert): Likewise.
4918 (insert_aux): Remove.
4919 (struct pointer_map_t): Embed a pointer_set_t.
4920 (pointer_map_create): Adjust.
4921 (pointer_map_destroy): Likewise.
4922 (pointer_map_contains): Likewise.
4923 (pointer_map_insert): Likewise.
4924 (pointer_map_traverse): Likewise.
4925 * tree-streamer.h (struct streamer_tree_cache_d): Use a
4926 pointer_map<unsigned> instead of a pointer_map_t.
4927 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4928 (streamer_tree_cache_lookup): Likewise.
4929 (streamer_tree_cache_create): Likewise.
4930 (streamer_tree_cache_delete): Likewise.
4931 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4932 pointer_map<unsigned> instead of a pointer_map_t.
4933 (lto_init_tree_ref_encoder): Adjust.
4934 (lto_destroy_tree_ref_encoder): Likewise.
4935 * lto-section-out.c (lto_output_decl_index): Likewise.
4936 (lto_record_function_out_decl_state): Likewise.
4937 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4939 2013-06-24 Richard Biener <rguenther@suse.de>
4941 PR tree-optimization/57488
4942 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4944 2013-06-24 Alan Modra <amodra@gmail.com>
4946 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4947 (gen_easy_altivec_constant): Likewise.
4948 * config/rs6000/predicates.md (easy_vector_constant_add_self,
4949 easy_vector_constant_msb): Likewise.
4951 2013-06-23 Jakub Jelinek <jakub@redhat.com>
4954 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4955 add missing return true.
4957 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
4960 * config/sh/predicates.md (general_extend_operand): Invoke
4961 general_movsrc_operand for memory operands.
4962 (general_movsrc_operand): Allow reg+reg addressing, do not use
4963 general_operand for memory operands.
4965 2013-06-23 Sriraman Tallam <tmsriram@google.com>
4967 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
4968 when current target options does not apply.
4969 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4970 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4971 * config/i386/bmiintrin.h: Pass appropriate target
4972 attributes to header.
4973 * config/i386/mmintrin.h: Ditto.
4974 * config/i386/nmmintrin.h: Ditto.
4975 * config/i386/avx2intrin.h: Ditto.
4976 * config/i386/fxsrintrin.h: Ditto.
4977 * config/i386/tbmintrin.h: Ditto.
4978 * config/i386/xsaveintrin.h: Ditto.
4979 * config/i386/f16cintrin.h: Ditto.
4980 * config/i386/xtestintrin.h: Ditto.
4981 * config/i386/xsaveoptintrin.h: Ditto.
4982 * config/i386/bmi2intrin.h: Ditto.
4983 * config/i386/lzcntintrin.h: Ditto.
4984 * config/i386/smmintrin.h: Ditto.
4985 * config/i386/wmmintrin.h: Ditto.
4986 * config/i386/x86intrin.h: Remove all header include guards.
4987 * config/i386/prfchwintrin.h: Ditto.
4988 * config/i386/pmmintrin.h: Ditto.
4989 * config/i386/tmmintrin.h: Ditto.
4990 * config/i386/xmmintrin.h: Ditto.
4991 * config/i386/popcntintrin.h: Ditto.
4992 * config/i386/rdseedintrin.h: Ditto.
4993 * config/i386/ammintrin.h: Ditto.
4994 * config/i386/emmintrin.h: Ditto.
4995 * config/i386/immintrin.h: Remove all header include guards.
4996 * config/i386/fma4intrin.h: Ditto.
4997 * config/i386/lwpintrin.h: Ditto.
4998 * config/i386/xopintrin.h: Ditto.
4999 * config/i386/ia32intrin.h: Ditto.
5000 * config/i386/avxintrin.h: Ditto.
5001 * config/i386/rtmintrin.h: Ditto.
5002 * config/i386/fmaintrin.h: Ditto.
5003 * config/i386/mm3dnow.h: Ditto.
5005 2013-06-22 Sriraman Tallam <tmsriram@google.com>
5007 * common/config/i386/i386-common.c: Handle LZCNT.
5009 2013-06-22 Andi Kleen <ak@linux.intel.com>
5011 * doc/extend.texi: Use __atomic_store_n instead of
5012 __atomic_store in HLE example.
5014 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
5016 * config/sh/sh.c: Remove <cstdlib> workaround.
5018 2013-06-21 Andi Kleen <ak@linux.intel.com>
5020 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
5022 2013-06-21 Andi Kleen <ak@linux.intel.com>
5024 * doc/extend.texi: Document that __atomic_clear and
5025 __atomic_test_and_set should only be used with bool.
5027 2013-06-20 Jan Hubicka <jh@suse.cz>
5029 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5031 * tree.c (decls_same_for_odr): New function.
5032 (same_for_edr): New function.
5033 (types_same_for_odr): New function.
5034 (get_binfo_at_offset): Use it.
5035 * tree.h (types_same_for_odr): Declare.
5037 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
5038 Jason Merrill <jason@redhat.com>
5040 * system.h: Include <cstdlib> as well as <stdlib.h>.
5042 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
5045 * config/i386/i386.c (construct_container): Report error if
5046 long double is used with disabled x87 float returns.
5048 2013-06-20 Jan Hubicka <jh@suse.cz>
5050 * lto-cgraph.c (input_symtab): Do not set cgraph state.
5052 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
5054 PR rtl-optimization/57425
5055 PR rtl-optimization/57569
5056 * alias.c (write_dependence_p): Remove parameters mem_mode and
5057 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
5058 Changed all callers.
5059 (canon_anti_dependence): Get comments and semantics in sync.
5060 Add parameter mem_canonicalized. Changed all callers.
5061 * rtl.h (canon_anti_dependence): Update prototype.
5063 2013-06-20 Richard Biener <rguenther@suse.de>
5065 * data-streamer-in.c (streamer_read_uhwi): Optimize single
5066 byte case, inline streamer_read_uchar and defer section
5069 2013-06-20 Richard Biener <rguenther@suse.de>
5071 PR tree-optimization/57584
5072 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5073 SSA names into the expanded expression that take part in
5074 abnormal coalescing.
5076 2013-06-19 Sharad Singhai <singhai@google.com>
5078 * gcov.c (print_usage): Handle new option.
5079 (process_args): Ditto.
5080 (get_gcov_intermediate_filename): New function.
5081 (output_intermediate_file): New function.
5082 (output_gcov_file): New function
5083 (generate_results): Handle new option.
5084 (release_function): Relase demangled name.
5085 (read_graph_file): Handle demangled name.
5086 (output_lines): Ditto.
5087 * doc/gcov.texi: Document gcov intermediate format.
5089 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
5092 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5093 (lra_emit_add): Use the functions. Add comment about Y as an
5096 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
5099 * collect2.c (collect_atexit): New.
5100 (collect_exit): Delete.
5101 (main): Register collect_atexit with atexit.
5102 (collect_wait): Change collect_exit to exit.
5104 * collect2.h (collect_exit): Delete.
5105 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5107 2013-06-19 Wei Mi <wmi@google.com>
5109 PR rtl-optimization/57518
5110 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5111 if regno is used in paradoxical subreg.
5112 (update_equiv_regs): Check pdx_subregs[regno] before
5113 set a reg to be equivalent with a mem.
5115 2013-06-19 Matthias Klose <doko@ubuntu.com>
5118 * file-find.h (find_a_file): Add a mode parameter.
5119 * file-find.c (find_a_file): Likewise.
5120 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5121 with X_OK for the executables.
5122 * collect2.c (main): Call find_a_file with X_OK.
5124 2013-06-19 Steve Ellcey <sellcey@mips.com>
5127 * config/mips/mips.md (casesi_internal_mips16_<mode>):
5128 Use NEXT_INSN instead of next_real_insn.
5130 2013-06-19 Jan Hubicka <jh@suse.cz>
5132 * cgraph.h (const_value_known_p): Replace by ...
5133 (ctor_for_folding): .. this one.
5134 * cgraphunit.c (process_function_and_variable_attributes): Use it.
5135 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5136 * expr.c (expand_expr_real_1): Likewise.
5137 (string_constant): Likewise.
5138 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5139 * ipa.c (process_references): Likewise.
5140 (symtab_remove_unreachable_nodes): Likewise.
5141 * ipa-inline-analysis.c (param_change_prob): Likewise.
5142 * gimple-fold.c (canonicalize_constructor_val): Likewise.
5143 (get_base_constructor): Likwise.
5144 * varpool.c (varpool_remove_node): Likewise.
5145 (varpool_remove_initializer): LIkewise.
5146 (dump_varpool_node): LIkwise.
5147 (const_value_known_p): Rewrite to ...
5148 (ctor_for_folding): ... this one.
5150 2013-06-19 Jakub Jelinek <jakub@redhat.com>
5153 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5154 PERSONALITY in $PATH derived prefixes.
5156 2013-06-19 Jeff Law <law@redhat.com>
5158 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5161 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5162 (simplify_bitwise_binary): Use it to simpify certain binary ops on
5165 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
5167 * config/arm/vfp.md: Move VFP instruction classification documentation
5169 * config/arm/arm.md: ... here. Update instruction classification
5172 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
5174 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5175 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5176 pattern. Use more efficient sequences on ARMv5 and Thumb2.
5178 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
5181 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5182 with NEXT_INSN. Use tablejump_p to check for jump table data
5185 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
5188 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5189 that now in C++ the value is correct per the C++ standards.
5191 2013-06-19 Richard Biener <rguenther@suse.de>
5193 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5196 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5199 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5202 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5203 with next_active_insn.
5205 2013-06-18 Sriraman Tallam <tmsriram@google.com>
5207 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5208 functions are inlined during failures to flag an error.
5209 * tree-inline.c (expand_call_inline): Allow the error to be flagged
5210 in early inline pass.
5212 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
5214 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5217 2013-06-18 Julian Brown <julian@codesourcery.com>
5219 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5220 Permit virtual register pre-reload if !strict.
5221 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5223 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5225 * config/arm/neon.md (movmisalign<mode>): Use
5226 neon_perm_struct_or_reg_operand instead of
5227 neon_struct_or_register_operand.
5228 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5229 neon_permissive_struct_operand instead of neon_struct_operand.
5230 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5231 neon_vector_mem_operand.
5232 * config/arm/predicates.md (neon_struct_operand): Adjust call to
5233 neon_vector_mem_operand.
5234 (neon_permissive_struct_operand): New.
5235 (neon_struct_or_register_operand): Rename to...
5236 (neon_perm_struct_or_reg_operand): This. Adjust call to
5237 neon_vector_mem_operand.
5239 2013-06-18 Richard Biener <rguenther@suse.de>
5241 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5242 * lto-streamer.h: Include pointer-set.h.
5243 (struct lto_decl_slot): Remove.
5244 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5245 Remove next_index entry.
5246 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5247 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5248 (lto_init_tree_ref_encoder): Adjust.
5249 (lto_destroy_tree_ref_encoder): Likewise.
5250 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5251 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5252 (lto_output_decl_index): Adjust.
5253 (lto_new_out_decl_state): Likewise.
5254 (lto_record_function_out_decl_state): Likewise.
5255 * lto-streamer-out.c (copy_function): Likewise.
5257 2013-06-18 Richard Biener <rguenther@suse.de>
5259 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5260 * cgraphunit.c: Include cfgloop.h.
5261 (init_lowered_empty_function): Initialize the loop tree.
5262 (assemble_thunk): Insert new BBs into loops.
5264 2013-06-18 Richard Biener <rguenther@suse.de>
5266 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5267 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5268 the map from cache entry to cache index optional.
5269 (streamer_tree_cache_replace_tree): Adjust accordingly.
5270 (streamer_tree_cache_append): Likewise.
5271 (streamer_tree_cache_delete): Likewise.
5272 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5273 streamer cache map from cache entry to cache index.
5274 * lto-streamer-out.c (create_output_block): Adjust.
5276 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
5278 * config/arm/arm.md (attribute "insn"): Move multiplication and
5279 division attributes to...
5280 (attribute "type"): ... here. Remove mult.
5281 (attribute "mul32"): New attribute.
5282 (attribute "mul64"): Add umaal.
5283 (*arm_mulsi3): Update attributes.
5284 (*arm_mulsi3_v6): Likewise.
5285 (*thumb_mulsi3): Likewise.
5286 (*thumb_mulsi3_v6): Likewise.
5287 (*mulsi3_compare0): Likewise.
5288 (*mulsi3_compare0_v6): Likewise.
5289 (*mulsi_compare0_scratch): Likewise.
5290 (*mulsi_compare0_scratch_v6): Likewise.
5291 (*mulsi3addsi): Likewise.
5292 (*mulsi3addsi_v6): Likewise.
5293 (*mulsi3addsi_compare0): Likewise.
5294 (*mulsi3addsi_compare0_v6): Likewise.
5295 (*mulsi3addsi_compare0_scratch): Likewise.
5296 (*mulsi3addsi_compare0_scratch_v6): Likewise.
5297 (*mulsi3subsi): Likewise.
5298 (*mulsidi3adddi): Likewise.
5299 (*mulsi3addsi_v6): Likewise.
5300 (*mulsidi3adddi_v6): Likewise.
5301 (*mulsidi3_nov6): Likewise.
5302 (*mulsidi3_v6): Likewise.
5303 (*umulsidi3_nov6): Likewise.
5304 (*umulsidi3_v6): Likewise.
5305 (*umulsidi3adddi): Likewise.
5306 (*umulsidi3adddi_v6): Likewise.
5307 (*smulsi3_highpart_nov6): Likewise.
5308 (*smulsi3_highpart_v6): Likewise.
5309 (*umulsi3_highpart_nov6): Likewise.
5310 (*umulsi3_highpart_v6): Likewise.
5311 (mulhisi3): Likewise.
5312 (*mulhisi3tb): Likewise.
5313 (*mulhisi3bt): Likewise.
5314 (*mulhisi3tt): Likewise.
5315 (maddhisi4): Likewise.
5316 (*maddhisi4tb): Likewise.
5317 (*maddhisi4tt): Likewise.
5318 (maddhidi4): Likewise.
5319 (*maddhidi4tb): Likewise.
5320 (*maddhidi4tt): Likewise.
5322 (udivsi3): Likewise.
5323 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5324 (thumb2_mulsi_short_compare0): Likewise.
5325 (thumb2_mulsi_short_compare0_scratch): Likewise.
5326 * config/arm/arm1020e.md (1020mult1): Update attribute change.
5327 (1020mult2): Likewise.
5328 (1020mult3): Likewise.
5329 (1020mult4): Likewise.
5330 (1020mult5): Likewise.
5331 (1020mult6): Likewise.
5332 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5334 (cortex_a15_mult64): Likewise.
5335 (cortex_a15_sdiv): Likewise.
5336 (cortex_a15_udiv): Likewise.
5337 * config/arm/arm1026ejs.md (mult1): Update attribute change.
5343 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5344 (pj4_ir_div): Likewise.
5345 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5346 (11_mult2): Likewise.
5347 (11_mult3): Likewise.
5348 (11_mult4): Likewise.
5349 (11_mult5): Likewise.
5350 (11_mult6): Likewise.
5351 (11_mult7): Likewise.
5352 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5353 (cortex_a8_mla): Likewise.
5354 (cortex_a8_mull): Likewise.
5355 (cortex_a8_smulwy): Likewise.
5356 (cortex_a8_smlald): Likewise.
5357 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5358 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5359 (cortex_r4_mul_3): Likewise.
5360 (cortex_r4_mla_4): Likewise.
5361 (cortex_r4_mla_3): Likewise.
5362 (cortex_r4_smlald): Likewise.
5363 (cortex_r4_mull): Likewise.
5364 (cortex_r4_sdiv): Likewise.
5365 (cortex_r4_udiv): Likewise.
5366 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5367 (cortex_a7_idiv): Likewise.
5368 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5369 (9_mult2): Likewise.
5370 (9_mult3): Likewise.
5371 (9_mult4): Likewise.
5372 (9_mult5): Likewise.
5373 (9_mult6): Likewise.
5374 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5375 (cortex_a53_sdiv): Likewise.
5376 (cortex_a53_udiv): Likewise.
5377 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5378 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5379 (mp626_mult2): Likewise.
5380 (mp626_mult3): Likewise.
5381 (mp626_mult4): Likewise.
5382 * config/arm/fa526.md (526_mult1): Update attribute change.
5383 (526_mult2): Likewise.
5384 * config/arm/arm-generic.md (mult): Update attribute change.
5385 (mult_ldsched_strongarm): Likewise.
5386 (mult_ldsched): Likewise.
5387 (multi_cycle): Likewise.
5388 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5389 * config/arm/fa606te.md (606te_mult1): Update attribute change.
5390 (606te_mult2): Likewise.
5391 (606te_mult3): Likewise.
5392 (606te_mult4): Likewise.
5393 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5394 (cortex_a9_mac16): Likewise.
5395 (cortex_a9_multiply): Likewise.
5396 (cortex_a9_mac): Likewise.
5397 (cortex_a9_multiply_long): Likewise.
5398 * config/arm/fa626te.md (626te_mult1): Update attribute change.
5399 (626te_mult2): Likewise.
5400 (626te_mult3): Likewise.
5401 (626te_mult4): Likewise.
5403 2013-06-18 Richard Biener <rguenther@suse.de>
5406 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5408 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5411 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5412 with next_active_insn.
5414 2013-06-18 Alan Modra <amodra@gmail.com>
5416 * config/rs6000/rs6000.h (enum data_align): New.
5417 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5418 (DATA_ABI_ALIGNMENT): Define.
5419 (CONSTANT_ALIGNMENT): Correct comment.
5420 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5421 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5423 2013-06-17 David Malcolm <dmalcolm@redhat.com>
5425 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5426 ATTRIBUTE_UNUSED marking.
5428 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
5430 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5431 alternative and update.
5432 (aarch64_dup_lanedi): Delete.
5433 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5434 * config/aarch64/aarch64-simd-builtins.def: Update.
5436 2013-06-17 Richard Biener <rguenther@suse.de>
5438 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5439 (lto_input_scc): Declare.
5440 (lto_input_tree_1): Likewise.
5441 (struct lto_stats_d): Add num_tree_bodies_output and
5442 num_pickle_refs_output.
5443 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5444 (lto_read_tree_1): Split out from ...
5445 (lto_read_tree): ... this.
5446 (lto_input_scc): New function.
5447 (lto_input_tree_1): Split out from ...
5448 (lto_input_tree): ... this. Handle LTO_tree_scc.
5449 (lto_data_in_create): Create the streamer cache without hashes.
5450 * lto-streamer-out.c (create_output_block): Create the streamer
5451 cache with hashes when not doing WPA.
5452 (lto_write_tree_1): Split out from ...
5453 (lto_write_tree): ... this.
5454 (get_symbol_initial_value): New function.
5455 (lto_output_tree_1): Split out from ...
5456 (lto_output_tree): ... this. Write trees as series of SCCs
5457 using a DFS walk via DFS_write_tree.
5458 (struct sccs, struct scc_entry): New types.
5459 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5460 (DFS_write_tree_body): New function.
5461 (DFS_write_tree): Likewise.
5462 (hash_tree): Likewise.
5463 (scc_entry_compare): Likewise.
5464 (hash_scc): Likewise.
5465 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5466 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5467 TREE_CHAIN as regular reference.
5468 (streamer_read_integer_cst): Remove.
5469 (streamer_get_pickled_tree): Adjust.
5470 * tree-streamer-out.c (streamer_write_chain): Disable streaming
5471 of DECL_EXTERNALs in BLOCK_VARS for now.
5472 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5474 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5475 Add hash value argument and record that if hashes are recorded
5477 (streamer_tree_cache_insert_1): Adjust.
5478 (streamer_tree_cache_insert): Likewise.
5479 (streamer_tree_cache_insert_at): Rename to ...
5480 (streamer_tree_cache_replace_tree): ... this and adjust.
5481 (streamer_tree_cache_append): Adjust.
5482 (record_common_node): Likewise.
5483 (streamer_tree_cache_create): Add argument whether to
5484 record hash values together with trees.
5485 (streamer_tree_cache_delete): Adjust.
5486 * tree-streamer.h (struct streamer_tree_cache_d): Add
5488 (streamer_read_integer_cst): Remove.
5489 (streamer_tree_cache_insert): Adjust.
5490 (streamer_tree_cache_append): Likewise.
5491 (streamer_tree_cache_insert_at): Rename to ...
5492 (streamer_tree_cache_replace_tree): ... this and adjust.
5493 (streamer_tree_cache_create): Add argument whether to record hashes.
5494 (streamer_tree_cache_get): Rename to ...
5495 (streamer_tree_cache_get_tree): ... this.
5496 (streamer_tree_cache_get_hash): New function.
5497 * tree.c (cache_integer_cst): New function.
5498 * tree.h (cache_integer_cst): Declare.
5499 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5500 * lto-symtab.c (lto_varpool_replace_node): Only release
5501 DECL_INITIAL of non-prevailing decls.
5502 * varpool.c (varpool_remove_initializer): Do not release
5503 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5505 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5507 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5508 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5509 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5510 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5511 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5512 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5513 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5514 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5515 instead of TARGET_64BIT.
5516 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5517 Require ISA_HAS_<D>DIV.
5519 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
5521 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5522 (mips*-*-linux*): Move default with_llsc setting to where other
5524 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5526 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5527 Likewise. Remove default with_tune setting. Move default float
5528 setting to its own block. Handle with_llsc in the same block as above.
5530 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
5532 PR rtl-optimization/57425
5533 PR rtl-optimization/57569
5534 * alias.c (write_dependence_p): Add new parameters mem_mode,
5535 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
5536 Changed all callers.
5537 (canon_anti_dependence): New function.
5538 * cse.c (check_dependence): Use canon_anti_dependence.
5539 * cselib.c (cselib_invalidate_mem): Likewise.
5540 * rtl.h (canon_anti_dependence): Declare.
5542 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5544 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5545 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5546 ".set mips3" for 64-bit targets.
5548 2013-06-15 Dehao Chen <dehao@google.com>
5550 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5551 * gimple-low.c (gimple_check_call_matching_types): Likewise.
5552 (gimple_check_call_args): Likewise.
5553 * value-prof.c (check_ic_target): Likewise.
5554 * ipa-inline.c (early_inliner): Likewise.
5555 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5556 * cgraph.c (cgraph_create_edge_1): Likewise.
5557 (cgraph_make_edge_direct): Likewise.
5559 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5562 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5563 rs6000_output_move_128bit to handle emitting quad memory
5564 operations. Set attribute length to 8 bytes.
5566 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
5568 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5570 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5571 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5572 (aarch64_<su>mlsl<mode>): Likewise.
5574 2013-06-14 Mike Stump <mikestump@comcast.net>
5576 * Makefile.in (TARGET_H): Add insn-codes.h.
5578 2013-06-14 Alan Modra <amodra@gmail.com>
5582 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5583 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
5584 bitfield expansion when EXPAND_MEMORY.
5585 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5587 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5590 test for clearing quad memory on 32-bit later.
5592 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5594 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5595 (fold_negate_expr): Likewise.
5596 (fold_real_zero_addition_p): Handle vectors.
5597 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5599 2013-06-14 Alan Modra <amodra@gmail.com>
5601 * varasm.c (force_const_mem): Revert 2013-06-07 change.
5603 2013-06-13 Jan Hubicka <jh@suse.cz>
5605 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5606 Local comdats are not externally visible.
5607 * symtab.c (dump_symtab_base): Dump externally visible.
5608 (verify_symtab_base): Verify back links in the symtab hash.
5610 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5612 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5613 CONVERT_EXPR as equal nodes.
5615 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5617 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5619 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5621 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5622 Generalize to complex and vector.
5623 * tree.c (build_all_ones_cst): New function.
5624 * tree.h (build_all_ones_cst): Declare it.
5626 2013-06-13 Alan Modra <amodra@gmail.com>
5628 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5629 * config/rs6000/rs6000.md (signbittf2): New insn.
5630 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5631 (abstf2_internal, cmptf_internal2): Likewise.
5632 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5634 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5635 Pat Haugen <pthaugen@us.ibm.com>
5636 Peter Bergner <bergner@vnet.ibm.com>
5638 * config/rs6000/rs6000.c (emit_load_locked): Add support for
5639 power8 byte, half-word, and quad-word atomic instructions.
5640 (emit_store_conditional): Likewise.
5641 (rs6000_expand_atomic_compare_and_swap): Likewise.
5642 (rs6000_expand_atomic_op): Likewise.
5644 * config/rs6000/sync.md (larx): Add new modes for power8.
5646 (AINT): New mode iterator to include TImode as well as normal
5647 integer modes on power8.
5648 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5649 that VSX registers are not considered. Use AINT mode iterator
5650 instead of INT1 to allow inclusion of quad word atomic operations
5652 (load_locked<mode>): Likewise.
5653 (store_conditional<mode>): Likewise.
5654 (atomic_compare_and_swap<mode>): Likewise.
5655 (atomic_exchange<mode>): Likewise.
5656 (atomic_nand<mode>): Likewise.
5657 (atomic_fetch_<fetchop_name><mode>): Likewise.
5658 (atomic_nand_fetch<mode>): Likewise.
5659 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5661 (ATOMIC): On power8, add QImode, HImode modes.
5662 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5663 modes that promote to SImode.
5664 (load_lockedti): Convert TImode arguments to PTImode, so that we
5665 get a guaranteed even/odd register pair.
5666 (load_lockedpti): Likewise.
5667 (store_conditionalti): Likewise.
5668 (store_conditionalpti): Likewise.
5670 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5671 atomic load/store instructions.
5674 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
5676 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5678 (insn_count): New attribute, with most cases extracted from...
5679 (length): ...here. Redefine most cases in terms of insn_count.
5680 (single_insn): Delete.
5681 (can_delay): Use insn_count to check for single instructions.
5682 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5683 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5684 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5685 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5686 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5687 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5688 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5689 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5690 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5691 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5692 rather than "length".
5693 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
5694 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5695 Use "insn_count" rather than "length".
5696 * config/mips/mips-dsp.md
5697 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5698 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5701 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
5703 PR tree-optimization/57361
5704 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5706 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
5708 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5710 (aarch64_simd_combine<mode>): New instruction expansion.
5711 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5713 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5714 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5716 2013-06-12 Jan Hubicka <jh@suse.cz>
5718 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5719 decl has when in streaming stage.
5720 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5721 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5723 2013-06-12 Roland Stigge <stigge@antcom.de>
5726 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5728 2013-06-12 Jakub Jelinek <jakub@redhat.com>
5730 PR tree-optimization/57537
5731 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5732 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5734 2013-06-12 Richard Biener <rguenther@suse.de>
5736 * data-streamer.h (streamer_write_char_stream): CSE
5737 obs->current_pointer.
5738 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5739 streamer_write_char_stream manually and optimize the resulting loop.
5740 (streamer_write_hwi_stream): Likewise.
5742 2013-06-12 Jan Hubicka <jh@suse.cz>
5744 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5745 * cgraph.h (varpool_create_empty_node): Declare.
5746 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5748 * symtab.c (symtab_unregister_node): Be lax about missin entries
5750 (symtab_get_node): Update comment.
5751 * varpool.c (varpool_create_empty_node): Break out from ...
5752 (varpool_node_for_decl): ... here.
5753 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5755 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
5757 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5758 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5759 part. Use straight-line flow at the end.
5760 <COMPONENT_REF>: Remove superfluous else.
5761 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5763 2013-06-12 Jakub Jelinek <jakub@redhat.com>
5766 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5767 target hook even for !TREE_PUBLIC decls. If no resolution info
5768 is available, return false for common and external decls.
5770 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
5772 * config/rl78/constraints.md (U): New constraint.
5773 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5776 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
5779 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5780 to allow returning address to AT_PLATFORM name.
5782 2013-06-11 Jan Hubicka <jh@suse.cz>
5784 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5785 * cgraph.h (symtab_node_base): Add weakref flag.
5786 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5787 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5788 (output_weakrefs): Use weakref flag.
5789 * fold-const.c (simple_operand_p): Handle WEAK.
5790 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5791 * ipa.c (varpool_externally_visible_p): Drop weakref.
5792 (function_and_variable_visibility): Update comment; fix weakref
5793 sanity checks; do not clear DECL_WEAK on them.
5794 * lto-cgraph.c (lto_output_node): update.
5795 (lto_output_varpool_node): Update.
5796 (input_overwrite_node): Update.
5797 (input_node): Update.
5798 (input_varpool_node): Update.
5799 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5800 (lto_symtab_merge_symbols): Add sanity check.
5801 (lto_symtab_prevailing_decl): Do not special case weakrefs.
5802 * passes.c (rest_of_decl_compilation): Set static flag, too.
5803 * symtab.c (dump_symtab_base): Dump weakref.
5804 (verify_symtab_base): Sanity check weakrefs.
5805 (symtab_make_decl_local): Remove duplicated code.
5806 (symtab_alias_ultimate_target): Simplify.
5807 * varpool.c (varpool_create_variable_alias): Set weakref flag.
5809 2013-06-11 DJ Delorie <dj@redhat.com>
5811 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5812 (rl78_unwind_word_mode): New.
5814 2013-06-11 David Malcolm <dmalcolm@redhat.com>
5816 * final.c (debug_prefix_maps): Make static.
5818 2013-06-11 David Malcolm <dmalcolm@redhat.com>
5820 * function.c (initial_trampoline): Remove stray copy.
5822 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
5824 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5826 2013-06-11 Martin Jambor <mjambor@suse.cz>
5828 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5829 within bounds at the beginning of the function.
5831 2013-06-11 Alan Modra <amodra@gmail.com>
5833 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
5835 (get_named_section): Don't NULL !DECL_P decl.
5837 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
5839 * doc/invoke.texi (core-avx2): Document.
5841 (atom): Updated with MOVBE.
5843 2013-06-11 Richard Biener <rguenther@suse.de>
5845 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5847 2013-06-11 Anton Blanchard <anton@samba.org>
5849 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5850 correct shift value in little-endian mode.
5852 2013-06-11 Jakub Jelinek <jakub@redhat.com>
5855 * varasm.c (get_variable_align): Move #endif to the right place.
5857 2013-06-10 Cary Coutant <ccoutant@google.com>
5859 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5860 for hash so that hash table traversal order is deterministic.
5862 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5863 Pat Haugen <pthaugen@us.ibm.com>
5864 Peter Bergner <bergner@vnet.ibm.com>
5866 * config/rs6000/vector.md (GPR move splitter): Do not split moves
5867 of vectors in GPRS if they are direct moves or quad word load or
5870 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5872 (direct_move_p): Likewise.
5873 (quad_load_store_p): Likewise.
5875 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5876 classes into bins based on the physical register type.
5877 (reg_class_to_reg_type): Likewise.
5878 (IS_STD_REG_TYPE): Likewise.
5879 (IS_FP_VECT_REG_TYPE): Likewise.
5880 (reload_fpr_gpr): Arrays to determine what insn to use if we can
5881 use direct move instructions.
5882 (reload_gpr_vsx): Likewise.
5883 (reload_vsx_gpr): Likewise.
5884 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5885 information that is a simplification of register classes. Also
5886 precalculate direct move reload helpers.
5887 (direct_move_p): New function to return true if the operation can
5888 be done as a direct move instruciton.
5889 (quad_load_store_p): New function to return true if the operation
5890 is a quad memory operation.
5891 (rs6000_legitimize_address): If quad memory, only allow register
5892 indirect for TImode addresses.
5893 (rs6000_legitimate_address_p): Likewise.
5894 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5895 (rs6000_reload_register_type): Likewise.
5896 (register_to_reg_type): Return register type.
5897 (rs6000_secondary_reload_simple_move): New helper function for
5898 secondary reload and secondary memory needed to identify anything
5899 that is a simple move, and does not need reloading.
5900 (rs6000_secondary_reload_direct_move): New helper function for
5901 secondary reload to identify cases that can be done with several
5902 instructions via the direct move instructions.
5903 (rs6000_secondary_reload_move): New helper function for secondary
5904 reload to identify moves between register types that can be done.
5905 (rs6000_secondary_reload): Add support for quad memory operations
5906 and for direct move.
5907 (rs6000_secondary_memory_needed): Likewise.
5908 (rs6000_debug_secondary_memory_needed): Change argument names.
5909 (rs6000_output_move_128bit): New function to return the move to
5910 use for 128-bit moves, including knowing about the various
5911 limitations of quad memory operations.
5913 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5914 memory operations. call rs6000_output_move_128bit for the actual
5915 instruciton(s) to generate.
5916 (vsx_movti_64bit): Likewise.
5918 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5919 (UNSPEC_P8V_MTVSRWZ): Likewise.
5920 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5921 (UNSPEC_P8V_MTVSRD): Likewise.
5922 (UNSPEC_P8V_XXPERMDI): Likewise.
5923 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5924 (UNSPEC_FUSION_GPR): Likewise.
5925 (FMOVE128_GPR): New iterator for direct move.
5926 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5929 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5930 loads and direct move instructions.
5931 (zero_extendsidi2_lfiwzx): Likewise.
5932 (extendsidi2_lfiwax): Likewise.
5933 (extendsidi2_nocell): Likewise.
5934 (floatsi<mode>2_lfiwax): Likewise.
5936 (floatunssi<mode>2_lfiwzx): Likewise.
5938 (fix_trunc<mode>_stfiwx): Likewise.
5939 (fixuns_trunc<mode>_stfiwx): Likewise.
5940 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5941 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5942 (parity<mode>2_cmpb): Set length/type attr.
5943 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5944 for 'mr.' to fast_compare.
5945 (bpermd_<mode>): Change type attr to popcnt.
5946 (p8_fmrgow_<mode>): New insns for power8 direct move support.
5947 (p8_mtvsrwz_1): Likewise.
5948 (p8_mtvsrwz_2): Likewise.
5949 (reload_fpr_from_gpr<mode>): Likewise.
5950 (p8_mtvsrd_1): Likewise.
5951 (p8_mtvsrd_2): Likewise.
5952 (p8_xxpermdi_<mode>): Likewise.
5953 (reload_vsx_from_gpr<mode>): Likewise.
5954 (reload_vsx_from_gprsf): Likewise.
5955 (p8_mfvsrd_3_<mode>): LIkewise.
5956 (reload_gpr_from_vsx<mode>): Likewise.
5957 (reload_gpr_from_vsxsf): Likewise.
5958 (p8_mfvsrd_4_disf): Likewise.
5959 (multi-word GPR splits): Do not split direct moves or quad memory
5962 2013-06-10 David Malcolm <dmalcolm@redhat.com>
5964 * tree-into-ssa.c (interesting_blocks): Make static.
5966 2013-06-10 Jakub Jelinek <jakub@redhat.com>
5969 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5970 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5971 Use DATA_ABI_ALIGNMENT for that case instead if defined.
5972 (get_variable_align): New function.
5973 (get_variable_section, emit_bss, emit_common,
5974 assemble_variable_contents, place_block_symbol): Use
5975 get_variable_align instead of DECL_ALIGN.
5976 (assemble_noswitch_variable): Add align argument, use it
5977 instead of DECL_ALIGN.
5978 (assemble_variable): Adjust caller. Use get_variable_align
5979 instead of DECL_ALIGN.
5980 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5982 (DATA_ABI_ALIGNMENT): Define.
5983 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5984 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
5985 opt is false, only return the psABI mandated alignment increase.
5986 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5987 (DATA_ABI_ALIGNMENT): ... this.
5988 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5989 (DATA_ABI_ALIGNMENT): ... this.
5990 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5991 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5992 (DATA_ABI_ALIGNMENT): ... this.
5993 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5994 * doc/tm.texi: Regenerated.
5996 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
5998 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5999 cmp_code to construct REG_EQUAL note.
6001 2013-06-09 Jakub Jelinek <jakub@redhat.com>
6004 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6005 that operands[2] doesn't overlap with operands[0].
6007 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
6008 Jan Hubicka <jh@suse.cz>
6010 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6011 hack to mark symbols as used.
6013 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
6015 PR rtl-optimization/57559
6016 * lra-constraints.c (process_alt_operands): Don't discourage
6017 memory with known offset for offsetable memory constraint.
6018 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6020 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
6022 * varasm.c (struct oc_local_state): Reorder fields.
6023 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6024 and adjust accordingly.
6025 (output_constructor): Reorder initialization code and adjust call to
6026 output_constructor_bitfield.
6028 2013-06-07 Jan Hubicka <jh@suse.cz>
6030 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6032 2013-06-07 David Malcolm <dmalcolm@redhat.com>
6034 * tree-object-size.c (unknown): Make const.
6036 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6038 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6039 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6040 for last alternative in the cpu_facility attribute.
6042 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6045 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6046 (xordi3): Change operand 2 constraint to arm_xordi_operand.
6047 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6048 * config/arm/constraints.md (Dg): New constraint.
6049 * config/arm/neon.md (xordi3_neon): Remove.
6050 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6051 * config/arm/predicates.md (arm_xordi_operand): New predicate.
6053 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6055 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6056 Clean up alternatives.
6058 2013-06-07 Alan Modra <amodra@gmail.com>
6060 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6063 2013-06-07 Alan Modra <amodra@gmail.com>
6065 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6067 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6069 * config/arm/constraints.md (Df): New constraint.
6070 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6071 Correct length attribute for last two alternatives.
6073 2013-06-07 Alan Modra <amodra@gmail.com>
6075 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6076 override user -mfp-in-toc.
6077 (offsettable_ok_by_alignment): Consider just the current access
6078 rather than the whole object, unless BLKmode. Handle
6079 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6080 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6081 for -mcmodel=medium.
6082 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6083 override user -mfp-in-toc or -msum-in-toc. Default to
6084 -mno-fp-in-toc for -mcmodel=medium.
6086 2013-06-06 DJ Delorie <dj@redhat.com>
6088 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6089 TARGET_VALID_POINTER_MODE.
6091 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6092 Pat Haugen <pthaugen@us.ibm.com>
6093 Peter Bergner <bergner@vnet.ibm.com>
6095 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6096 Document new power8 builtins.
6098 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6099 condition code register, to allow 128-bit logical operations to be
6100 done in the VSX or GPR registers.
6101 (nor<mode>3): Use the canonical form for nor.
6102 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6103 vclz*, and vpopcnt* vector instructions.
6104 (nand<mode>3): Likewise.
6105 (orc<mode>3): Likewise.
6106 (clz<mode>2): LIkewise.
6107 (popcount<mode>2): Likewise.
6109 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6110 that only the GPRs are recognized.
6112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6113 support for new power8 builtins.
6115 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6117 (xscvdpspn): Likewise.
6123 (vpopcnt): Likewise.
6124 (vpopcntb): Likewise.
6125 (vpopcnth): Likewise.
6126 (vpopcntw): Likewise.
6127 (vpopcntd): Likewise.
6132 (eqv_v16qi3): Likewise.
6133 (eqv_v8hi3): Likewise.
6134 (eqv_v4si3): Likewise.
6135 (eqv_v2di3): Likewise.
6136 (eqv_v4sf3): Likewise.
6137 (eqv_v2df3): Likewise.
6139 (nand_v16qi3): Likewise.
6140 (nand_v8hi3): Likewise.
6141 (nand_v4si3): Likewise.
6142 (nand_v2di3): Likewise.
6143 (nand_v4sf3): Likewise.
6144 (nand_v2df3): Likewise.
6146 (orc_v16qi3): Likewise.
6147 (orc_v8hi3): Likewise.
6148 (orc_v4si3): Likewise.
6149 (orc_v2di3): Likewise.
6150 (orc_v4sf3): Likewise.
6151 (orc_v2df3): Likewise.
6153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6154 allow power8 quad mode in 64-bit.
6155 (rs6000_builtin_vectorized_function): Add support to vectorize
6156 ISA 2.07 count leading zeros, population count builtins.
6157 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6158 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6159 (builtin_function_type): Add vgbbd builtin function which takes an
6161 (altivec_expand_vec_perm_const): Add support for new power8 merge
6164 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6165 that does not include TImdoe for use with 32-bit.
6166 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6168 (UNSPEC_VSX_CVDPSPN): Likewise.
6169 (vsx_xscvdpspn): Likewise.
6170 (vsx_xscvspdpn): Likewise.
6171 (vsx_xscvdpspn_scalar): Likewise.
6172 (vsx_xscvspdpn_directmove): Likewise.
6173 (vsx_and<mode>3): Split logical operations into 32-bit and
6174 64-bit. Add support to do logical operations on TImode as well as
6175 VSX vector types. Allow logical operations to be done in either
6176 VSX registers or in general purpose registers in 64-bit mode. Add
6177 splitters if GPRs were used. For AND, add clobber of CCmode to
6178 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
6180 (vsx_and<mode>3_32bit): Likewise.
6181 (vsx_and<mode>3_64bit): Likewise.
6182 (vsx_ior<mode>3): Likewise.
6183 (vsx_ior<mode>3_32bit): Likewise.
6184 (vsx_ior<mode>3_64bit): Likewise.
6185 (vsx_xor<mode>3): Likewise.
6186 (vsx_xor<mode>3_32bit): Likewise.
6187 (vsx_xor<mode>3_64bit): Likewise.
6188 (vsx_one_cmpl<mode>2): Likewise.
6189 (vsx_one_cmpl<mode>2_32bit): Likewise.
6190 (vsx_one_cmpl<mode>2_64bit): Likewise.
6191 (vsx_nor<mode>3): Likewise.
6192 (vsx_nor<mode>3_32bit): Likewise.
6193 (vsx_nor<mode>3_64bit): Likewise.
6194 (vsx_andc<mode>3): Likewise.
6195 (vsx_andc<mode>3_32bit): Likewise.
6196 (vsx_andc<mode>3_64bit): Likewise.
6197 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6198 and xxlorc instructions.
6199 (vsx_eqv<mode>3_64bit): Likewise.
6200 (vsx_nand<mode>3_32bit): Likewise.
6201 (vsx_nand<mode>3_64bit): Likewise.
6202 (vsx_orc<mode>3_32bit): Likewise.
6203 (vsx_orc<mode>3_64bit): Likewise.
6205 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6207 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6209 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6210 (p8_vmrgow): Likewise.
6211 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6212 GPRs to be split under VSX.
6213 (p8v_clz<mode>2): Add power8 count leading zero support.
6214 (p8v_popcount<mode>2): Add power8 population count support.
6215 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6218 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6221 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6223 (vec_nand): Likewise.
6224 (vec_vclz): Likewise.
6225 (vec_vclzb): Likewise.
6226 (vec_vclzd): Likewise.
6227 (vec_vclzh): Likewise.
6228 (vec_vclzw): Likewise.
6229 (vec_vgbbd): Likewise.
6230 (vec_vmrgew): Likewise.
6231 (vec_vmrgow): Likewise.
6232 (vec_vpopcnt): Likewise.
6233 (vec_vpopcntb): Likewise.
6234 (vec_vpopcntd): Likewise.
6235 (vec_vpopcnth): Likewise.
6236 (vec_vpopcntw): Likewise.
6238 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6240 PR rtl-optimization/57468
6241 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6244 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6246 PR rtl-optimization/57459
6247 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6248 type when setting live regs.
6250 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6252 * config/s390/s390.opt (mlra): New option.
6253 * config/s390/s390.c (s390_decompose_address): Check displacement
6254 for all registers for LRA.
6255 (s390_secondary_reload): Don't used secondary reloads for LRA.
6256 (s390_lra_p): New function.
6257 (TARGET_LRA_P): Define.
6258 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6259 of attribute cpu_facility to zarch for the last alternative.
6260 (*cmpmem_short): Ditto.
6262 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
6264 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6265 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
6266 (arm_expand_prologue): Likewise.
6268 2013-06-06 Teresa Johnson <tejohnson@google.com>
6271 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6272 as this is now done by redirect_edge_and_branch_force.
6273 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6274 barriers, and fix interaction with splitting.
6275 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6276 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6277 reflect changes made in the routine.
6278 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6279 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6280 since this is called in cfglayout mode, and replace partition fixup
6281 with assert as that is now done by force_nonfallthru_and_redirect.
6282 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6283 already be marked with region crossing note.
6284 (insert_section_boundary_note): Make non-static, gate on flag
6285 has_bb_partition, rewrite to also check for multiple partitions.
6286 (rest_of_handle_reorder_blocks): Remove call to
6287 insert_section_boundary_note, now done later during free_cfg.
6288 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6289 * bb-reorder.h (insert_section_boundary_note): Declare.
6290 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6291 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6292 invoke insert_section_boundary_note.
6293 (try_redirect_by_replacing_jump): Remove unnecessary
6294 check for region crossing note.
6295 (fixup_partition_crossing): New function.
6296 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6297 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6298 in non-cfglayout mode.
6299 (force_nonfallthru_and_redirect): Fixup partition boundaries,
6300 remove old code that tried to do this. Emit barrier correctly
6301 when we are in cfglayout mode.
6302 (last_bb_in_partition): New function.
6303 (rtl_split_edge): Correctly fixup partition boundaries.
6304 (commit_one_edge_insertion): Remove old code that tried to
6305 fixup region crossing edge since this is now handled in
6306 split_block, and set up insertion point correctly since
6307 block may now end in a jump.
6308 (verify_hot_cold_block_grouping): Guard against checking when not in
6309 linearized RTL mode.
6310 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6312 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6313 rtl_verify_flow_info, so not called in cfglayout mode.
6314 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6315 (fixup_reorder_chain): Remove old code that attempted to fixup region
6316 crossing note as this is now handled in force_nonfallthru_and_redirect.
6317 (duplicate_insn_chain): Don't duplicate switch section notes.
6318 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6320 * basic-block.h (emit_barrier_after_bb): Declare.
6322 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6324 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6325 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6326 arm_usatsihi): Adjust alternatives for arm_restrict_it.
6328 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6330 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6332 * config/arm/ldmstm.md: Regenerate.
6334 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6336 * config/arm/sync.md (atomic_loaddi_1):
6337 Disable predication for arm_restrict_it.
6338 (arm_load_exclusive<mode>): Likewise.
6339 (arm_load_exclusivesi): Likewise.
6340 (arm_load_exclusivedi): Likewise.
6341 (arm_load_acquire_exclusive<mode>): Likewise.
6342 (arm_load_acquire_exclusivesi): Likewise.
6343 (arm_load_acquire_exclusivedi): Likewise.
6344 (arm_store_exclusive<mode>): Likewise.
6345 (arm_store_exclusive<mode>): Likewise.
6346 (arm_store_release_exclusivedi): Likewise.
6347 (arm_store_release_exclusive<mode>): Likewise.
6349 2013-06-06 Richard Biener <rguenther@suse.de>
6351 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6353 (lto_tag_is_tree_code_p): Adjust.
6354 (lto_tag_is_gimple_code_p): Likewise.
6355 (lto_gimple_code_to_tag): Likewise.
6356 (lto_tag_to_gimple_code): Likewise.
6357 (lto_tree_code_to_tag): Likewise.
6358 (lto_tag_to_tree_code): Likewise.
6359 * data-streamer.h (streamer_write_hwi_in_range): Use
6360 uhwi streaming to stream the normalized range.
6361 (streamer_read_hwi_in_range): Likewise.
6363 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6365 * config/arm/arm.md (enabled_for_depr_it): New attribute.
6366 (predicable_short_it): Likewise.
6367 (predicated): Likewise.
6368 (enabled): Handle above.
6369 (define_cond_exec): Set predicated attribute to yes.
6371 2013-06-05 Mike Stump <mikestump@comcast.net>
6373 * gdbinit.in (__FUNCTION__): Add.
6375 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
6377 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6378 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6380 2013-06-05 Jan Hubicka <jh@suse.cz>
6382 * varasm.c (mark_decl_referenced): Revert the removal until targets
6385 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
6387 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6388 instead of mark_decl_referenced.
6390 2013-06-05 Jan Hubicka <jh@suse.cz>
6392 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6393 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6394 and symtab_used_from_object_file_p.
6395 (cgraph_make_node_local_1): Clear forced_by_abi.
6396 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6397 * cgraph.h (symtab_node_base): Add forced_by_abi.
6398 (decide_is_variable_needed): Remove.
6399 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6400 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6401 (decide_is_symbol_needed): ... this one; handle symbols in general;
6402 always analyze virtuals; honnor forced_by_abi.
6403 (cgraph_finalize_function): Update.
6404 (varpool_finalize_decl): Update.
6405 (symbol_defined_and_needed): Remove.
6406 (analyze_functions): Update.
6407 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6408 output_refs, input_overwrite_node): Handle forced_by_abi.
6409 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6410 (address_taken_from_non_vtable_p): ... this one.
6411 (comdat_can_be_unshared_p_1): New function.
6412 (cgraph_comdat_can_be_unshared_p): Rename to ...
6413 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6414 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6415 (function_and_variable_visibility): Clear forced_by_abi as needed.
6416 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6417 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6418 * symtab.c (dump_symtab_base): Dump forced_by_abi.
6419 * varpool.c (decide_is_variable_needed): Remove.
6421 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6423 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6424 (arm_option_override): Override arm_restrict_it where appropriate.
6425 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6426 * config/arm/arm.opt (mrestrict-it): New command-line option.
6427 * doc/invoke.texi: Document -mrestrict-it.
6429 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6431 * tsan.c (tsan_atomic_table): Make const.
6433 2013-06-05 Richard Biener <rguenther@suse.de>
6435 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6436 index associated with the tree we are supposed to replace.
6437 * tree-streamer-out.c (pack_ts_base_value_fields): Output
6438 TREE_ASM_WRITTEN as zero for everything but SSA names.
6440 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6442 * tree-ssa-structalias.c (call_stmt_vars): Make static.
6444 2013-06-04 Jan Hubicka <jh@suse.cz>
6446 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6447 (input_node, input_varpool_node): Handle correctly external same
6449 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6450 nodes at ltrans stage.
6452 2013-06-04 Jan Hubicka <jh@suse.cz>
6454 * ipa-inline.c (update_caller_keys): Fix availability test.
6455 (update_callee_keys): Likewise.
6456 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6457 to follow ELF standard.
6459 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
6461 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6462 (mips64r5900el-*-elf*): New configurations.
6463 * config/mips/mips-cpus.def (r5900): New processor.
6464 * config/mips/mips-tables.opt: Regenerate.
6465 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6466 (mips_issue_rate): Handle PROCESSOR_R5900.
6467 (mips_reorg_process_insns): Force reorder mode for the R5900.
6468 * config/mips/mips.h (TARGET_MIPS5900): Define.
6469 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6471 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6473 * config/mips/mips.md (processor): Add r5900.
6474 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6476 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6478 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6479 into function to generate MOVI instruction.
6480 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6481 (aarch64_preferred_simd_mode): Turn into wrapper.
6482 (aarch64_output_scalar_simd_mov_immediate): New function.
6483 * config/aarch64/aarch64-protos.h: Add prototype for above.
6485 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6487 * config/aarch64/aarch64.c (simd_immediate_info): Remove
6488 element_char member.
6489 (sizetochar): Return signed char.
6490 (aarch64_simd_valid_immediate): Remove elchar and other
6491 unnecessary variables.
6492 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6493 Calculate element_char as required.
6494 * config/aarch64/aarch64-protos.h: Update and move prototype
6495 for aarch64_output_simd_mov_immediate.
6496 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6499 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6501 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6502 information completed by aarch64_simd_valid_immediate.
6503 (aarch64_legitimate_constant_p): Update arguments.
6504 (aarch64_simd_valid_immediate): Work with struct rather than many
6506 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6507 (aarch64_simd_make_constant): Update arguments.
6508 (aarch64_output_simd_mov_immediate): Work with struct rather than
6509 many pointers. Output immediate directly rather than as operand.
6510 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6512 * config/aarch64/constraints.md (Dn): Update arguments.
6514 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6516 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6518 (aarch64_simd_immediate_valid_for_move): Remove.
6519 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6520 (aarch64_simd_make_constant): Update call.
6521 (aarch64_output_simd_mov_immediate): Update call.
6522 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6524 * config/aarch64/constraints.md (Dn): Update call.
6526 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6528 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6529 return type to bool for prototype.
6530 (aarch64_legitimate_constant_p): Check for true instead of not -1.
6531 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6532 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6534 2013-06-04 Catherine Moore <clm@codesourcery.com>
6536 * config/mips/mips.opt (meva): New.
6537 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6538 (ASM_SPEC): Handle -meva.
6539 * doc/invoke.texi (meva): Document.
6541 2013-06-04 Alan Modra <amodra@gmail.com>
6543 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6546 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6548 * rtl.def: Add extra fourth optional field to define_cond_exec.
6549 * gensupport.c (process_one_cond_exec): Process attributes from
6551 * doc/md.texi: Document fourth field in define_cond_exec.
6553 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
6555 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6556 out the processing order as in store_bit_field_1.
6558 2013-06-04 Jan Hubicka <jh@suse.cz>
6561 * cgraphunit.c (cgraph_process_same_body_aliases): Create
6562 non-VAR_DECL node if it does not exist yet.
6564 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
6566 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6567 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6568 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6569 target_cpu_default setting.
6571 2013-06-03 Teresa Johnson <tejohnson@google.com>
6573 * dumpfile.c (opt_info_switch_p): Change -fopt-info
6574 default to -fopt-info=optimized instead of all.
6575 * doc/invoke.texi: Ditto.
6576 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6577 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6578 (execute_vect_slp): Emit BB vectorization success under
6579 MSG_OPTIMIZED_LOCATIONS.
6580 * tree-vect-slp.c (vect_slp_transform_bb): Change
6581 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6582 * tree-vect-loop.c (vect_transform_loop): Ditto.
6584 2013-06-03 Jason Merrill <jason@redhat.com>
6587 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6588 Use TARGET_EXPR for C++.
6590 2013-06-03 Jakub Jelinek <jakub@redhat.com>
6592 PR rtl-optimization/57268
6593 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6594 if DEBUG_INSN_P (insn).
6597 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
6599 PR rtl-optimization/57268
6600 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6601 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6603 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6605 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6606 (ix86_avoid_lea_for_addr): Likewise.
6607 (exact_dependency_1): Likewise.
6608 (ix86_adjust_cost): Likewise.
6609 (swap_top_of_ready_list): Fix formatting and !reload_completed check
6611 (do_reorder_for_imul): Fix typo, formatting and
6612 !reload_completed check removed.
6613 (ix86_sched_reorder): Fix typo and formatting.
6614 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6617 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
6619 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6621 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
6623 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6624 <STRING_CST>: Likewise.
6625 <VECTOR_CST>: Likewise.
6627 2013-06-01 Janus Weil <janus@gcc.gnu.org>
6628 Mikael Morin <mikael@gcc.gnu.org>
6630 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6631 * config.in: Regenerated.
6632 * configure: Regenerated.
6634 2013-06-01 Jan Hubicka <jh@suse.cz>
6637 * cgraphunit.c (compile): When weakref is not supported,
6638 set up transparent aliases before final output pass.
6639 * varasm.c (assemble_alias): Do not try to do it here.
6641 2013-06-01 Jan Hubicka <jh@suse.cz>
6644 * passes.c (for_per_function): Skip unanalyzed functions.
6646 2013-06-01 Jan Hubicka <jh@suse.cz>
6648 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6649 (lto_symtab_merge_symbols_1): ... this one.
6650 (lto_symtab_merge_cgraph_nodes): Rename to ...
6651 (lto_symtab_merge_symbols): ... this one; simplify.
6652 * cgraph.c (same_body_aliases_done): Rename to ...
6653 (cpp_implicit_aliases_done): ... this one.
6654 (cgraph_create_function_alias): Update.
6655 (cgraph_same_body_alias): Update.
6656 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6657 (verify_edge_corresponds_to_fndecl): Simplify.
6658 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6659 (cgraph_node): Remove same_body_alias.
6660 (varpool_node): Remove alias_of and extra_name_alias.
6661 (same_body_aliases_done): Rename to ..
6662 (cpp_implicit_aliases_done): ... this one.
6663 (symtab_alias_ultimate_target): Add default parameter.
6664 (symtab_resolve_alias): New function.
6665 (fixup_same_cpp_alias_visibility): Declare.
6666 (cgraph_function_node): Add default parameter.
6667 (cgraph_node_asm_name): Likewise.
6668 (cgraph_function_or_thunk_node): Add default parameter; do
6669 not ICE when it is NULL.
6670 (varpool_variable_node): Likewise.
6671 * tree-emutls.c (create_emultls_var): Update.
6672 (ipa_lower_emutls): Update.
6673 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6674 (cgraph_reset_node): Reset alias info.
6675 (cgraph_finalize_function): Update.
6676 (fixup_same_cpp_alias_visibility): Move to symtab.c.
6677 (analyze_function): Simplify.
6678 (cgraph_process_same_body_aliases): Simplify.
6679 (analyze_functions): Fixup same body aliases.
6680 (handle_alias_pairs): Simplify.
6681 (assemble_thunk): Update.
6682 (assemble_thunks_and_aliases): Update.
6683 (output_weakrefs): Rewrite.
6684 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6685 (lto_output_varpool_node): Likewise.
6686 (compute_ltrans_boundary): Remve assert.
6687 (get_alias_symbol): New functoin.
6688 (input_node): Rewrite alias handling.
6689 (input_varpool_node): Likewise.
6690 * ipa-pure-const.c (propagate_pure_const): Fix formating.
6691 * ipa.c (process_references): Handle weakrefs correctly.
6692 (symtab_remove_unreachable_nodes): Likewise.
6693 * trans-mem.c (get_cg_data): Update.
6694 (ipa_tm_create_version_alias): Update.
6695 (ipa_tm_execute): Update.
6696 * symtab.c (dump_symtab_base): Dump aliases.
6697 (verify_symtab_base): Verify aliases.
6698 (symtab_node_availability): New function.
6699 (symtab_alias_ultimate_target): Simplify.
6700 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6701 handle all the fixup cases.
6702 (symtab_resolve_alias): New function.
6703 * passes.c (ipa_write_summaries): Handle weakrefs.
6704 * varpool.c (varpool_analyze_node): Simplify.
6705 (assemble_aliases): Update.
6706 (varpool_create_variable_alias): Simplify.
6707 (varpool_extra_name_alias): Simplify.
6708 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6709 (lto_symtab_merge_symbols): ... this one.
6711 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
6714 PR rtl-optimization/57268
6715 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6716 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6718 2013-06-01 Tobias Burnus <burnus@net-b.de>
6721 2013-05-31 Tobias Burnus <burnus@net-b.de>
6724 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6727 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
6729 PR rtl-optimization/57268
6730 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6731 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6733 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
6735 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6736 unordered comparison operators when -fno-trapping-math is in effect
6738 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6739 and implement unordered comparison operators properly on the e500.
6741 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
6743 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6744 for constant scalar integers.
6745 (simplify_relational_operation_1): Likewise.
6747 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6749 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6750 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
6753 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6754 Igor Zamyatin <igor.zamyatin@intel.com>
6756 Silvermont (SLM) architecture performance tuning.
6757 * config/i386/i386.h (enum ix86_tune_indices): Add
6758 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6759 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6761 * config/i386/i386.c (initial_ix86_tune_features)
6762 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6763 (ix86_lea_outperforms): Handle Silvermont tuning.
6764 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6766 (ix86_use_lea_for_mov): Likewise.
6767 (ix86_avoid_lea_for_addr): Likewise.
6768 (ix86_lea_for_add_ok): Likewise.
6769 (exact_dependency_1): New function.
6770 (exact_store_load_dependency): Likewise.
6771 (ix86_adjust_cost): Handle Silvermont tuning.
6772 (do_reoder_for_imul): Likewise.
6773 (swap_top_of_ready_list): New function.
6774 (ix86_sched_reorder): Changed to handle Silvermont tuning.
6776 * config/i386/i386.md (peepholes that split memory operand in fp
6779 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
6781 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6782 Remove un-necessary braces.
6784 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
6786 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6787 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6789 2013-05-31 Tobias Burnus <burnus@net-b.de>
6792 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6794 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6797 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6798 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6799 * config/arm/neon.md (iordi3_neon): Remove.
6800 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6801 * config/arm/predicates.md (imm_for_neon_logic_operand):
6802 Move to earlier in the file.
6803 (neon_logic_op2): Likewise.
6804 (arm_iordi_operand_neon): New predicate.
6806 2013-05-31 Richard Biener <rguenther@suse.de>
6808 PR tree-optimization/57478
6809 PR tree-optimization/57453
6810 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6813 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
6815 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6816 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6818 2013-05-30 Tobias Burnus <burnus@net-b.de>
6819 Thomas Koenig <tkoenig@gcc.gnu.org>
6822 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6823 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6825 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
6827 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6829 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
6831 * target.def (register_usage_leveling_p): New hook.
6832 * targhooks.c (default_register_usage_leveling_p): New.
6833 * targhooks.h (default_register_usage_leveling_p): New prototype.
6834 * lra-assigns.c (register_usage_leveling_p): Use the hook.
6835 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6836 * doc/tm.texi: Update.
6837 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6839 2013-05-30 Ian Bolton <ian.bolton@arm.com>
6841 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6842 (*insv_reg<mode>): New define_insn.
6844 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
6846 PR rtl-optimization/57439
6847 * postreload.c (move2add_valid_value_p): Check that we have
6848 a zero subreg_regno_offset when accessing the register in
6851 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6852 Igor Zamyatin <igor.zamyatin@intel.com>
6854 Silvermont (SLM) architecture pipeline model, tuning and
6856 * config.gcc: Add slm config options and target.
6858 * config/i386/slm.md: New.
6860 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6862 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
6864 (ix86_target_macros_internal): Likewise.
6866 * gcc/config/i386/i386.c (slm_cost): New cost.
6867 (m_SLM): New macro flag.
6868 (initial_ix86_tune_features): Set m_SLM.
6869 (x86_accumulate_outgoing_args): Likewise.
6870 (x86_arch_always_fancy_math_387): Likewise.
6871 (processor_target_table): Add slm cost.
6872 (cpu_names): Add slm cpu name.
6873 (x86_option_override_internal): Set SLM ISA.
6874 (ix86_issue_rate): New case PROCESSOR_SLM.
6875 (ia32_multipass_dfa_lookahead): Likewise.
6876 (fold_builtin_cpu): Add slm.
6878 * config/i386/i386.h (TARGET_SLM): New target macro.
6879 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6880 (processor_type): Add PROCESSOR_SLM.
6882 * config/i386/i386.md (cpu): Add new value "slm".
6883 (slm.md): Include slm.md.
6885 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
6886 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6888 * config/arm/arm-protos.h: Add and update function protos.
6889 * config/arm/arm.c (use_simple_return_p): New added.
6890 (thumb2_expand_return): Check simple_return flag.
6891 * config/arm/arm.md: Add simple_return and conditional simple_return.
6892 * config/arm/iterators.md: Add iterator for return and simple_return.
6894 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6896 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6897 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6898 (arm_emit_vfp_multi_reg_pop): Likewise.
6899 (thumb2_emit_ldrd_pop): Likewise.
6900 (arm_expand_epilogue): Add misc REG_CFA notes.
6901 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6903 2013-05-29 Lawrence Crowl <crowl@google.com>
6905 * config/arm/t-arm: Update for below.
6907 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6908 Change type to hash_table. Update dependent calls and types.
6910 * config/i386/t-cygming: Update for below.
6912 * config/i386/t-interix: Update for below.
6914 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6915 Change type to hash_table. Update dependent calls and types.
6916 (i386_find_on_wrapper_list::wrappers): Likewise.
6918 * config/ia64/t-ia64: Update for below.
6920 * config/ia64/ia64.c (bundle_state_table):
6921 Change type to hash_table. Update dependent calls and types.
6923 * config/mips/mips.c (mips_reorg_process_insns::htab):
6924 Change type to hash_table. Update dependent calls and types.
6926 * config/sol2.c (solaris_comdat_htab):
6927 Change type to hash_table. Update dependent calls and types.
6929 * config/t-sol2: Update for above.
6931 2013-05-29 Teresa Johnson <tejohnson@google.com>
6933 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6934 functions are not yet marked as defined.
6936 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6937 Pat Haugen <pthaugen@us.ibm.com>
6938 Peter Bergner <bergner@vnet.ibm.com>
6940 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6944 (vrotl<mode>3): Likewise.
6945 (vashl<mode>3): Likewise.
6946 (vlshr<mode>3): Likewise.
6947 (vashr<mode>3): Likewise.
6949 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6950 support for power8 V2DI builtins.
6952 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6953 power8 V2DI builtins.
6954 (vupkhsw): Likewise.
6955 (vupklsw): Likewise.
6956 (vaddudm): Likewise.
6961 (vpkudum): Likewise.
6962 (vpksdss): Likewise.
6963 (vpkudus): Likewise.
6964 (vpksdus): Likewise.
6969 (vsubudm): Likewise.
6970 (vcmpequd): Likewise.
6971 (vcmpgtsd): Likewise.
6972 (vcmpgtud): Likewise.
6973 (vcmpequd_p): Likewise.
6974 (vcmpgtsd_p): Likewise.
6975 (vcmpgtud_p): Likewise.
6976 (vupkhsw): Likewise.
6977 (vupklsw): Likewise.
6978 (vaddudm): Likewise.
6983 (vpksdss): Likewise.
6984 (vpksdus): Likewise.
6985 (vpkudum): Likewise.
6986 (vpkudus): Likewise.
6991 (vsubudm): Likewise.
6993 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6994 support for power8 V2DI instructions.
6996 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6997 power8 V2DI instructions. Combine pack and unpack insns to use an
6998 iterator for each mode. Check whether a particular mode supports
6999 Altivec instructions instead of just checking TARGET_ALTIVEC.
7000 (UNSPEC_VPKUWUM): Likewise.
7001 (UNSPEC_VPKSHSS): Likewise.
7002 (UNSPEC_VPKSWSS): Likewise.
7003 (UNSPEC_VPKUHUS): Likewise.
7004 (UNSPEC_VPKSHUS): Likewise.
7005 (UNSPEC_VPKUWUS): Likewise.
7006 (UNSPEC_VPKSWUS): Likewise.
7007 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7008 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7009 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7010 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7011 (UNSPEC_VUPKHSB): Likewise.
7012 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7013 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7014 (UNSPEC_VUPKHSH): Likewise.
7015 (UNSPEC_VUPKLSB): Likewise.
7016 (UNSPEC_VUPKLSH): Likewise.
7018 (VI_char): Likewise.
7019 (VI_scalar): Likewise.
7020 (VI_unit): Likewise.
7022 (VP_small): Likewise.
7023 (VP_small_lc): Likewise.
7024 (VU_char): Likewise.
7025 (add<mode>3): Likewise.
7026 (altivec_vaddcuw): Likewise.
7027 (altivec_vaddu<VI_char>s): Likewise.
7028 (altivec_vadds<VI_char>s): Likewise.
7029 (sub<mode>3): Likewise.
7030 (altivec_vsubcuw): Likewise.
7031 (altivec_vsubu<VI_char>s): Likewise.
7032 (altivec_vsubs<VI_char>s): Likewise.
7033 (altivec_vavgs<VI_char>): Likewise.
7034 (altivec_vcmpbfp): Likewise.
7035 (altivec_eq<mode>): Likewise.
7036 (altivec_gt<mode>): Likewise.
7037 (altivec_gtu<mode>): Likewise.
7038 (umax<mode>3): Likewise.
7039 (smax<mode>3): Likewise.
7040 (umin<mode>3): Likewise.
7041 (smin<mode>3): Likewise.
7042 (altivec_vpkuhum): Likewise.
7043 (altivec_vpkuwum): Likewise.
7044 (altivec_vpkshss): Likewise.
7045 (altivec_vpkswss): Likewise.
7046 (altivec_vpkuhus): Likewise.
7047 (altivec_vpkshus): Likewise.
7048 (altivec_vpkuwus): Likewise.
7049 (altivec_vpkswus): Likewise.
7050 (altivec_vpks<VI_char>ss): Likewise.
7051 (altivec_vpks<VI_char>us): Likewise.
7052 (altivec_vpku<VI_char>us): Likewise.
7053 (altivec_vpku<VI_char>um): Likewise.
7054 (altivec_vrl<VI_char>): Likewise.
7055 (altivec_vsl<VI_char>): Likewise.
7056 (altivec_vsr<VI_char>): Likewise.
7057 (altivec_vsra<VI_char>): Likewise.
7058 (altivec_vsldoi_<mode>): Likewise.
7059 (altivec_vupkhsb): Likewise.
7060 (altivec_vupkhs<VU_char>): Likewise.
7061 (altivec_vupkls<VU_char>): Likewise.
7062 (altivec_vupkhsh): Likewise.
7063 (altivec_vupklsb): Likewise.
7064 (altivec_vupklsh): Likewise.
7065 (altivec_vcmpequ<VI_char>_p): Likewise.
7066 (altivec_vcmpgts<VI_char>_p): Likewise.
7067 (altivec_vcmpgtu<VI_char>_p): Likewise.
7068 (abs<mode>2): Likewise.
7069 (vec_unpacks_hi_v16qi): Likewise.
7070 (vec_unpacks_hi_v8hi): Likewise.
7071 (vec_unpacks_lo_v16qi): Likewise.
7072 (vec_unpacks_hi_<VP_small_lc>): Likewise.
7073 (vec_unpacks_lo_v8hi): Likewise.
7074 (vec_unpacks_lo_<VP_small_lc>): Likewise.
7075 (vec_pack_trunc_v8h): Likewise.
7076 (vec_pack_trunc_v4si): Likewise.
7077 (vec_pack_trunc_<mode>): Likewise.
7079 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7081 (vec_vmaxsd): Likewise.
7082 (vec_vmaxud): Likewise.
7083 (vec_vminsd): Likewise.
7084 (vec_vminud): Likewise.
7085 (vec_vpksdss): Likewise.
7086 (vec_vpksdus): Likewise.
7087 (vec_vpkudum): Likewise.
7088 (vec_vpkudus): Likewise.
7089 (vec_vrld): Likewise.
7090 (vec_vsld): Likewise.
7091 (vec_vsrad): Likewise.
7092 (vec_vsrd): Likewise.
7093 (vec_vsubudm): Likewise.
7094 (vec_vupkhsw): Likewise.
7095 (vec_vupklsw): Likewise.
7097 2013-05-29 Jan Hubicka <jh@suse.cz>
7099 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7100 flags; reorder rest of fields in more consistent way.
7101 (varpool_node): Remove analyzed, finalized and alias.
7102 (cgraph_ndoe): Likewise.
7103 (symtab_alias_ultimate_target): New function.
7104 (cgraph_function_node): Move offline.
7105 (cgraph_reset_node): Declare.
7106 (cgraph_comdat_can_be_unshared_p): Remove.
7107 (varpool_remove_initializer): Declare.
7108 (varpool_first_defined_variable, varpool_next_defined_variable
7109 cgraph_first_defined_function, cgraph_next_defined_function): Update.
7110 (cgraph_function_with_gimple_body_p): Update.
7111 (varpool_all_refs_explicit_p): Update.
7112 (symtab_alias_target): New function.
7113 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7114 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7115 (cgraph_function_or_thunk_node): Simplify using
7116 symtab_alias_ultimate_target.
7117 (varpool_variable_node): Likewise.
7118 * cgraph.c (cgraph_create_function_alias): Update.
7119 (cgraph_add_thunk): Update.
7120 (cgraph_remove_node): Update.
7121 (dump_cgraph_node): Do not dump removed flags.
7122 (cgraph_function_body_availability): Update.
7123 (cgraph_propagate_frequency): Update.
7124 (verify_cgraph_node): Check sanity of local flag.
7125 (cgraph_function_node): Move here from cgraph.h; revamp for
7126 cgraph_function_or_thunk_node.
7127 * lto-symtab.c (lto_varpool_replace_node): Update.
7128 (lto_symtab_resolve_can_prevail_p): Update.
7129 (lto_symtab_merge_cgraph_nodes): Update.
7130 * ipa-cp.c (determine_versionability, initialize_node_lattices,
7131 propagate_constants_accross_call, devirtualization_time_bonus,
7132 ipcp_propagate_stage): Update.
7133 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7134 * ipa-inline-transform.c (clone_inlined_nodes,
7135 preserve_function_body_p): Update.
7136 * ipa-reference.c (propagate): Update.
7137 (write_node_summary_p): Update.
7138 * toplev.c (wrapup_global_declaration_2): Update.
7139 * cgraphunit.c (cgraph_analyze_function): Rename to ...
7140 (analyze_function) ... this one.
7141 (cgraph_process_new_functions): Update.
7142 (cgraph_reset_node): Export.
7143 (cgraph_finalize_function): Update.
7144 (cgraph_add_new_function): Update.
7145 (process_function_and_variable_attributes): Update.
7146 (varpool_finalize_decl): Update.
7147 (symbol_finalized): Remove.
7148 (symbol_finalized_and_needed): Rename to ...
7149 (symbol_defined_and_needed): ... update.
7150 (cgraph_analyze_functions): Update.
7151 (handle_alias_pairs): Update.
7152 (mark_functions_to_output): Update.
7153 (assemble_thunk): Update.
7154 (output_in_order): Update.
7155 (output_weakrefs): Update.
7156 (finalize_compilation_unit): Update.
7157 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7158 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7159 input_node, input_varpool_node): Update.
7160 * dbxout.c (dbxout_expand_expr): Update.
7161 * cgraphclones.c (cgraph_clone_node): Update.
7162 (cgraph_copy_node_for_versioning): Update.
7163 (cgraph_materialize_clone): Update.
7164 (cgraph_materialize_all_clones): Update.
7165 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7166 propagate_pure_const, propagate_nothrow): Update.
7167 * lto-streamer-out.c (lto_output, write_symbol): Update.
7168 * ipa-utils.c (ipa_reverse_postorder): Update.
7169 * ipa-inline.c (can_inline_edge_p): Update.
7170 (update_caller_keys, ipa_inline): Update.
7171 * dwarf2out.c (reference_to_unused,
7172 premark_types_used_by_global_vars_helper): Update.
7173 * tree-eh.c (tree_could_trap_p): Update.
7174 * ipa-split.c (consider_split, execute_split_functions): Update.
7175 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7176 has_addr_references_p): Update; move ahead in file for better
7178 (process_references): Simplify.
7179 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7181 (cgraph_comdat_can_be_unshared_p): Make static.
7182 (cgraph_externally_visible_p): Update.
7183 (varpool_externally_visible_p): Update.
7184 (function_and_variable_visibility): Update.
7185 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7186 ipa_tm_mark_force_output_node): Update.
7187 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7188 estimate_edge_devirt_benefit, inline_generate_summary,
7189 inline_write_summary): Update.
7190 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7191 * ipa-prop.c (ipa_compute_jump_functions): Update.
7192 (ipa_print_node_params, ipa_prop_read_section,
7193 ipa_update_after_lto_read, read_replacements_section): Update.
7194 * varasm.c (mark_decl_referenced): Update.
7195 (assemble_alias, dump_tm_clone_pairs): Update.
7196 * tree-inline.c (copy_bb): Update.
7197 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7199 * symtab.c (dump_symtab_base): Print new flags.
7200 (verify_symtab_base): Verify new flags.
7201 (symtab_alias_ultimate_target): New function.
7202 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7203 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7205 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7207 * i386.c (ix86_get_function_versions_dispatcher,
7208 ix86_generate_version_dispatcher_body): Update.
7209 (fold_builtin_cpu): Use varpool_add_new_variable.
7210 * varpool.c (varpool_remove_initializer): Break out from ...
7211 (varpool_remove_node): ... this one.
7212 (dump_varpool_node, varpool_node_for_asm,
7213 cgraph_variable_initializer_availability, varpool_analyze_node,
7214 varpool_assemble_decl, varpool_remove_unreferenced_decls,
7215 varpool_finalize_named_section_flags, varpool_create_variable_alias):
7218 2013-05-29 Jan Hubicka <jh@suse.cz>
7220 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7222 2013-05-29 Easwaran Raman <eraman@google.com>
7224 PR tree-optimization/57442
7225 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7226 when control exits the main loop.
7228 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
7230 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7232 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7233 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7234 * rx/t-rx: Add rx100 under multi library matches option for nofpu
7237 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7239 PR tree-optimization/57441
7240 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7241 Don't limit size of incr_vec to number of candidates.
7243 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
7245 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7246 and mips16 directories.
7247 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7248 (MULTILIB_DIRNAMES): Ditto.
7249 (MULTILIB_EXCEPTIONS): Add new exceptions.
7250 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7251 (MULTILIB_DIRNAMES): Ditto.
7252 (MULTILIB_EXCEPTIONS): Add new exceptions.
7254 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7255 Marcus Shawcroft <marcus.shawcroft@arm.com>
7257 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7258 SYMBOL_TINY_ABSOLUTE.
7259 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7260 SYMBOL_TINY_ABSOLUTE.
7261 (aarch64_expand_mov_immediate): Likewise.
7262 (aarch64_classify_symbol): Likewise.
7263 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7264 Permit SYMBOL_TINY_ABSOLUTE.
7265 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7267 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7268 Marcus Shawcroft <marcus.shawcroft@arm.com>
7270 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7271 Refactor if/switch. Replace gcc_assert with if.
7273 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
7275 * config/i386/i386.c (initial_ix86_tune_features): Enable
7276 FP Reassociation for AMD bdver1 and bdver2.
7278 2013-05-29 Martin Jambor <mjambor@suse.cz>
7280 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7281 and IMAGPART_EXPR do not occur within other handled_components.
7283 2013-05-29 Richard Biener <rguenther@suse.de>
7285 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7286 access on whether the use is in the BB we currently try to
7288 (vect_bb_vectorization_profitable_p): Pass the BB we currently
7289 vectorize to vect_bb_slp_scalar_cost.
7291 2013-05-29 Richard Biener <rguenther@suse.de>
7293 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7294 computing scalar cost offsetted by stmts that are kept live
7296 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7297 for computation of scalar cost.
7299 2013-05-28 Steve Ellcey <sellcey@mips.com>
7301 * config/mips/mips-cpus.def (mips32r2): Change processor type.
7303 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
7305 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7306 array notation built-in reduction functions.
7307 * doc/passes.texi (Passes): Added documentation about changes done
7309 * doc/invoke.texi (C Dialect Options): Added documentation about
7310 the -fcilkplus flag.
7311 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7312 (BUILTINS_DEF): Depend on cilkplus.def.
7313 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
7314 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7315 * cilkplus.def: New file.
7317 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
7319 PR rtl-optimization/57439
7320 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7322 2013-05-28 Easwaran Raman <eraman@google.com>
7324 PR tree-optimization/57337
7325 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7326 (find_insert_point): Correctly identify the insertion point
7327 when two statements with the same UID is compared.
7329 2013-05-28 Richard Biener <rguenther@suse.de>
7331 PR tree-optimization/56787
7332 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7333 from the list of data references.
7334 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7336 (vect_analyze_loop_operations): Likewise.
7337 (vect_transform_loop): Remove clobbers.
7339 2013-05-28 Martin Jambor <mjambor@suse.cz>
7341 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7342 and REALPART_EXPRs have scalar type.
7344 2013-05-28 Richard Biener <rguenther@suse.de>
7346 PR tree-optimization/57411
7347 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7349 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7350 virtual operand propagation.
7352 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7354 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7355 destination register for bmasksi_vis.
7356 (vector_init_bshuffle): Likewise.
7357 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7359 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7361 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7362 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7363 mode if the instruction isn't available in the original mode.
7364 * config/sparc/sparc.opt (mfix-ut699): New option.
7365 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7366 (divdf3): Turn into expander.
7367 (divdf3_nofix): New insn.
7368 (divdf3_fix): Likewise.
7369 (divsf3): Disable if -mfix-ut699.
7370 (sqrtdf2): Turn into expander.
7371 (sqrtdf2_nofix): New insn.
7372 (sqrtdf2_fix): Likewise.
7373 (sqrtsf2): Disable if -mfix-ut699.
7375 2013-05-27 Richard Biener <rguenther@suse.de>
7378 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7379 block for the new loop.
7381 2013-05-27 Richard Biener <rguenther@suse.de>
7383 PR tree-optimization/57343
7384 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7385 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7386 (number_of_iterations_cond): Do not build the folded tree.
7388 2013-05-27 Richard Biener <rguenther@suse.de>
7392 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7393 OEP_CONSTANT_ADDRESS_OF retained.
7395 PR tree-optimization/57417
7396 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7398 (set_ssa_val_to): Compare addresses using
7399 get_addr_base_and_unit_offset.
7401 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
7403 PR rtl-optimization/56833
7404 * postreload.c (move2add_record_mode): New function.
7405 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7406 (move2add_use_add2_insn): Use move2add_record_sym_value.
7407 (move2add_use_add3_insn): Likewise.
7408 (reload_cse_move2add): Use move2add_valid_value_p and
7409 move2add_record_mode. Invalidate call-clobbered and REG_INC
7410 affected regs by setting reg_mode to VOIDmode.
7411 (move2add_note_store): Don't pretend the inside of a SUBREG is
7412 the actual destination. Invalidate single/leading registers by
7413 setting reg_mode to VOIDmode.
7414 Use move2add_record_sym_value, move2add_valid_value_p and
7415 move2add_record_mode.
7417 2013-05-27 Richard Biener <rguenther@suse.de>
7419 PR tree-optimization/57396
7420 * tree-affine.c (double_int_constant_multiple_p): Properly
7421 return false for val == 0 and div != 0.
7423 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7425 * config/mips/mips.h: Use #elif in preprocessor conditions.
7427 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7430 * config/mips/constraints.md (kl): New constraint.
7431 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7432 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
7433 constraint for operand 0. Split after CSE for MIPS16. Emit a move
7435 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7437 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7440 * config/mips/mips.c (mips_can_inline_p): New function.
7441 (TARGET_CAN_INLINE_P): Define.
7443 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
7445 * sched-int.h (ds_t, dw_t): Make unsigned int.
7446 Fix documentation that describes how all the ds_t bits are used.
7447 Reserve the last bit for delayed-branch scheduling.
7448 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7449 (BITS_PER_DEP_WEAK): Fix definition and documentation.
7450 (gen_dep_weak_1): Remove prototype.
7451 * sched-deps.c (get_dep_weak_1): Make static.
7452 * target.def (speculate_insn, needs_block_p, gen_spec_check,
7453 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7454 * doc/tm.texi: Regenerate.
7455 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7457 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
7460 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7462 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
7463 Sandra Loosemore <sandra@codesourcery.com>
7465 * config.gcc (powerpc-*): Allow native for with-cpu.
7467 2013-05-24 Jeff Law <law@redhat.com>
7469 PR tree-optimization/57124
7470 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7471 conversion feeding a condition if the range has an overflow
7472 if -fstrict-overflow. Add warnings for when we do make the
7475 2013-05-24 Dehao Chen <dehao@google.com>
7477 * tree-cfg.c (locus_discrim_map): Fix the typo.
7478 (locus_discrim_hasher): Likewise.
7479 (locus_discrim_hasher::hash): Likewise.
7480 (locus_discrim_hasher::equal): Likewise.
7482 2013-05-24 Martin Jambor <mjambor@suse.cz>
7484 PR tree-optimization/57294
7485 * cgraph.h (ipa_record_stmt_references): Declare.
7486 * cgraphbuild.c (ipa_record_stmt_references): New function.
7487 (build_cgraph_edges): Use ipa_record_stmt_references.
7488 (rebuild_cgraph_edges): Likewise.
7489 (cgraph_rebuild_references): Likewise.
7490 * ipa-prop.c (ipa_modify_call_arguments): Discard references
7491 associated with the old statement and build references from the
7492 newly built statements.
7493 * ipa-ref.c (ipa_remove_stmt_references): New function.
7494 * ipa-ref.h (ipa_remove_stmt_references): Declare.
7496 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
7498 * lra-constraints.c (emit_spill_move): Use smaller mode for
7500 (check_and_process_move): Consider mem-reg moves for secondary
7502 (curr_insn_transform): Don't lose insns emitted before for
7503 secondary memory moves.
7504 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
7505 reg set up in the current insn.
7507 2013-05-24 Dehao Chen <dehao@google.com>
7509 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7511 (locus_descrim_hasher::equal): Likewise.
7512 (build_gimple_cfg): New discriminator assignment algorithm.
7513 (make_edges): Likewise.
7514 (next_discriminator_for_locus): Likewise.
7515 (same_line_p): Likewise.
7516 (assign_discriminators): Likewise.
7517 (make_cond_expr_edges): Likewise.
7518 (make_gimple_switch_edges): Likewise.
7519 (make_goto_expr_edges): Likewise.
7520 (make_gimple_asm_edges): Likewise.
7522 2013-05-24 Ian Bolton <ian.bolton@arm.com>
7524 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7525 X format specifier to only display bottom 16 bits.
7526 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7527 immediate to match for operand 2, since it will be masked.
7529 2013-05-24 Richard Biener <rguenther@suse.de>
7531 PR tree-optimization/57287
7532 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7533 all SSA names that occur in abnormal PHIs.
7535 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
7538 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7539 that index is not negative.
7541 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
7543 PR rtl-optimization/55177
7544 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7545 (simplify_byte_swapping_operation): New.
7546 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7547 (simplify_relational_operation_1): Deal with BSWAP.
7549 2013-05-23 Richard Henderson <rth@redhat.com>
7552 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7553 (ix86_reorg): Call it.
7555 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
7558 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7559 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7560 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7562 2013-05-23 Christian Bruel <christian.bruel@st.com>
7565 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7567 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7568 Marcus Shawcroft <marcus.shawcroft@arm.com>
7570 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7571 * config/aarch64/constraints.md (Usa): Remove.
7572 * doc/md.texi (AArch64 Usa): Remove.
7574 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7575 Marcus Shawcroft <marcus.shawcroft@arm.com>
7577 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7578 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7579 * config/aarch64/predicates.md (aarch64_const_address): Remove.
7580 (aarch64_mov_operand): Use aarch64_mov_operand_p.
7582 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
7584 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7585 instruction (AdvSIMD).
7586 * config/aarch64/aarch64-builtins.c
7587 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7588 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7590 2013-05-23 Martin Jambor <mjambor@suse.cz>
7593 * tree.h (contains_bitfld_component_ref_p): Declare.
7594 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7595 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
7597 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7598 not access a bit-field. Assert all final offsets are byte-aligned.
7600 2013-05-23 Richard Biener <rguenther@suse.de>
7602 PR tree-optimization/57380
7603 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7604 least one invariant or re-used load.
7605 * passes.c (init_optimization_passes): Move pass_phiprop before
7608 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
7610 * config/aarch64/aarch64-simd.md
7611 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7613 2013-05-23 Richard Biener <rguenther@suse.de>
7616 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7617 OEP_CONSTANT_ADDRESS_OF retained.
7619 2013-05-23 Jakub Jelinek <jakub@redhat.com>
7622 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7623 don't lower unit. Handle unit not being always BITS_PER_WORD.
7625 2013-05-23 Richard Biener <rguenther@suse.de>
7627 PR rtl-optimization/57341
7628 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7629 instead of true_dependence.
7631 2013-05-22 David Malcolm <dmalcolm@redhat.com>
7633 * bb-reorder.c (branch_threshold): Make const.
7634 (exec_threshold): Ditto.
7636 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7637 Pat Haugen <pthaugen@us.ibm.com>
7638 Peter Bergner <bergner@vnet.ibm.com>
7640 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7641 documentation for the power8 crypto builtins.
7643 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7645 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7646 macros for defining power8 builtin functions.
7647 (BU_P8V_AV_2): Likewise.
7648 (BU_P8V_AV_P): Likewise.
7649 (BU_P8V_VSX_1): Likewise.
7650 (BU_P8V_OVERLOAD_1): Likewise.
7651 (BU_P8V_OVERLOAD_2): Likewise.
7652 (BU_CRYPTO_1): Likewise.
7653 (BU_CRYPTO_2): Likewise.
7654 (BU_CRYPTO_3): Likewise.
7655 (BU_CRYPTO_OVERLOAD_1): Likewise.
7656 (BU_CRYPTO_OVERLOAD_2): Likewise.
7657 (XSCVSPDP): Fix typo, point to the correct instruction.
7658 (VCIPHER): Add power8 crypto builtins.
7659 (VCIPHERLAST): Likewise.
7660 (VNCIPHER): Likewise.
7661 (VNCIPHERLAST): Likewise.
7662 (VPMSUMB): Likewise.
7663 (VPMSUMH): Likewise.
7664 (VPMSUMW): Likewise.
7665 (VPERMXOR_V2DI): Likewise.
7666 (VPERMXOR_V4SI: Likewise.
7667 (VPERMXOR_V8HI: Likewise.
7668 (VPERMXOR_V16QI: Likewise.
7669 (VSHASIGMAW): Likewise.
7670 (VSHASIGMAD): Likewise.
7672 (VPERMXOR): Likewise.
7673 (VSHASIGMA): Likewise.
7675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7676 __CRYPTO__ if the crypto instructions are available.
7677 (altivec_overloaded_builtins): Add support for overloaded power8
7680 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7681 support for power8 crypto builtins.
7682 (builtin_function_type): Likewise.
7683 (altivec_init_builtins): Add support for builtins that take vector
7684 long long (V2DI) arguments.
7686 * config/rs6000/crypto.md: New file, define power8 crypto
7689 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7690 Pat Haugen <pthaugen@us.ibm.com>
7691 Peter Bergner <bergner@vnet.ibm.com>
7693 * doc/invoke.texi (Option Summary): Add power8 options.
7694 (RS/6000 and PowerPC Options): Likewise.
7696 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7697 constraints.md instead of rs6000.h. Reorder w* constraints. Add
7698 wm, wn, wr documentation.
7700 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7701 registers if direct move instructions are enabled.
7702 (wn): New constraint for no registers.
7703 (wq): New constraint for quad word even GPR registers.
7704 (wr): New constraint if 64-bit instructions are enabled.
7705 (wv): New constraint if power8 vector instructions are enabled.
7706 (wQ): New constraint for quad word memory locations.
7708 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7709 constraint for 0..15 for crypto instructions.
7710 (gpc_reg_operand): If VSX allow registers in VSX registers as well
7711 as GPR and floating point registers.
7712 (int_reg_operand): New predicate to match only GPR registers.
7713 (base_reg_operand): New predicate to match base registers.
7714 (quad_int_reg_operand): New predicate to match even GPR registers
7715 for quad memory operations.
7716 (vsx_reg_or_cint_operand): New predicate to allow vector logical
7717 operations in both GPR and VSX registers.
7718 (quad_memory_operand): New predicate for quad memory operations.
7719 (reg_or_indexed_operand): New predicate for direct move support.
7721 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7722 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7723 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7724 (POWERPC_MASKS): Add power8 options.
7725 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7728 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7729 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7731 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7732 (-mpower8-fusion): New power8 options.
7733 (-mpower8-fusion-sign): Likewise.
7734 (-mpower8-vector): Likewise.
7735 (-mcrypto): Likewise.
7736 (-mdirect-move): Likewise.
7737 (-mquad-memory): Likewise.
7739 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7741 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7743 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7744 (rs6000_debug_vector_unit): Add p8_vector.
7745 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7746 definitions. Also print fusion state.
7747 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7748 (rs6000_builtin_mask_calculate): Add power8 builtin support.
7749 (rs6000_option_override_internal): Add support for power8.
7750 (rs6000_common_init_builtins): Add debugging for skipped builtins
7752 (rs6000_adjust_cost): Add power8 support.
7753 (rs6000_issue_rate): Likewise.
7754 (insn_must_be_first_in_group): Likewise.
7755 (insn_must_be_last_in_group): Likewise.
7756 (force_new_group): Likewise.
7757 (rs6000_register_move_cost): Likewise.
7758 (rs6000_opt_masks): Likewise.
7760 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7761 power8 capable assembler, default to power7 options.
7762 (TARGET_DIRECT_MOVE): Likewise.
7763 (TARGET_CRYPTO): Likewise.
7764 (TARGET_P8_VECTOR): Likewise.
7765 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7766 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7767 (VECTOR_MEM_P8_VECTOR_P): Likewise.
7768 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7769 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7770 (TARGET_XSCVDPSPN): Likewise.
7771 (TARGET_XSCVSPDPN): Likewsie.
7772 (TARGET_SYNC_HI_QI): Likewise.
7773 (TARGET_SYNC_TI): Likewise.
7774 (MASK_CRYPTO): Likewise.
7775 (MASK_DIRECT_MOVE): Likewise.
7776 (MASK_P8_FUSION): Likewise.
7777 (MASK_P8_VECTOR): Likewise.
7778 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7779 temporary used by some of the direct move instructions to get two FP
7780 temporary registers does not force creation of a stack frame.
7781 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7782 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7783 that any VSX registers are tieable, even if they are also an
7784 Altivec vector mode.
7785 (r6000_reg_class_enum): Add wm, wr, wv constraints.
7786 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7787 (RS6000_BTM_CRYPTO): Likewise.
7788 (RS6000_BTM_COMMON): Likewise.
7790 * config/rs6000/rs6000.md (cpu attribute): Add power8.
7791 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7792 (enum rs6000_vector): Add power8 vector support.
7794 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7798 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7799 (any_sibcall_could_use_r3): this and handle indirect calls.
7800 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7802 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7804 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7806 2013-05-22 Richard Biener <rguenther@suse.de>
7809 * profile.c (branch_prob): Do not split blocks that are
7810 abnormally receiving from ECF_RETURNS_TWICE functions.
7812 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
7814 * recog.c (offsettable_address_addr_space_p): Fix calculation of
7815 address mode. Move pointer mode initialization to the same place.
7817 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
7819 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7820 while it has any effect.
7822 2013-05-21 Easwaran Raman <eraman@google.com>
7824 PR tree-optimization/57322
7825 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7826 UID of the statement added to the BB to be 1.
7828 2013-05-21 Jakub Jelinek <jakub@redhat.com>
7830 PR tree-optimization/57331
7831 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7832 of conversion from pointer type to integral type with integer.
7834 2013-05-21 Martin Jambor <mjambor@suse.cz>
7837 * ipa-prop.c (ipa_read_node_info): Process param_used and
7838 controlled_uses in the same order as when writing.
7840 2013-05-21 Magnus Granberg <baldrick@free.fr>
7843 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7845 2013-05-21 Richard Biener <rguenther@suse.de>
7847 PR tree-optimization/57318
7848 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7849 estimate stmts with side-effects as likely eliminated.
7851 2013-05-21 Richard Biener <rguenther@suse.de>
7853 PR tree-optimization/57330
7854 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7855 preserve the call stmts fntype.
7857 2013-05-21 Richard Biener <rguenther@suse.de>
7859 PR tree-optimization/57303
7860 * tree-ssa-sink.c (statement_sink_location): Improve killing
7861 stmt detection and properly handle self-assignments.
7863 2013-05-21 Christian Bruel <christian.bruel@st.com>
7865 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7866 spanning registers. LEAF_REG_REMAP is supported only for contiguous
7867 registers. Set register size out of the PARALLEL loop.
7869 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
7872 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7873 (*fmasf4, *fmasf4_media): New insns.
7875 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7877 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7878 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7879 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7880 (mips_idiv_insns): Update the comments to say that the returned
7881 instruction counts are in units of BASE_INSN_LENGTH.
7882 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7883 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
7884 using 2 rather than 4 as the length of indirect MIPS16 and
7885 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
7886 length of a NOP. Don't divide MIPS16 lengths by 2.
7887 (mips16_split_long_branches): Assume a branch is long if the
7888 length is greater than 4 rather than 8.
7889 * config/mips/mips.md (length): Give MIPS16 lengths directly,
7890 rather than multiplying them by 2. Multiply instruction counts
7891 by BASE_INSN_LENGTH rather than 4.
7892 (*jump_mips16, tls_get_tp_mips16_<mode>)
7893 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7895 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7897 * config/mips/mips.md (extended_mips16): Remove branch case.
7898 (length): Remove duplicated extended_mips16 test.
7900 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7902 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7904 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
7906 * recog.h (Recog_data): Rename to...
7907 (recog_data_d): ...this.
7908 (recog_data): Update accordingly.
7909 * recog.c (recog_data): Likewise.
7910 * reload.c (save_recog_data): Likewise.
7911 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7912 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7914 2013-05-17 Julian Brown <julian@codesourcery.com>
7916 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7917 found in a REG_EQUAL note, invalidate it.
7919 2013-05-17 Easwaran Raman <eraman@google.com>
7921 * tree-ssa-reassoc.c (find_insert_point): New function.
7922 (insert_stmt_after): Likewise.
7923 (get_def_stmt): Likewise.
7924 (ensure_ops_are_available): Likewise.
7925 (not_dominated_by): Likewise.
7926 (rewrite_expr_tree): Do not move statements beyond what is
7927 necessary. Remove call to swap_ops_for_binary_stmt...
7928 (reassociate_bb): ... and move it here.
7929 (build_and_add_sum): Assign UIDs for new statements.
7930 (linearize_expr): Likewise.
7931 (do_reassoc): Renumber gimple statement UIDs.
7933 2013-05-17 Jan Hubicka <jh@suse.cz>
7935 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7937 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7938 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7940 (output_weakrefs): Update.
7942 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
7943 Martin Jambor <mjambor@suse.cz>
7946 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7947 value that corresponds to the given aggval is found in values vector.
7949 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
7951 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7952 sse, sse2, sse3, ssse3 and sse4a flags to options.
7954 2013-05-17 David Malcolm <dmalcolm@redhat.com>
7956 * gengtype-state.c: (s_expr_writer): New class, to handle
7957 prettifying of output layout of s-expressions.
7958 (state_writer): New class, to write out gtype.state.
7959 (state_written_type_count): Move this variable into member data of
7961 (s_expr_writer::s_expr_writer): New code: constructor for new class
7962 (state_writer::state_writer(): ditto
7963 (s_expr_writer::write_new_line): New function
7964 (s_expr_writer::write_any_indent): ditto
7965 (s_expr_writer::begin_s_expr): ditto
7966 (s_expr_writer::end_s_expr): ditto
7967 (write_state_fileloc): convert to method of state_writer...
7968 (state_writer:: write_state_fileloc): ...and use methods of
7969 s_expr_writer to write indentation into the gtype.state output file
7970 to visually represent the hierarchical structure of the list
7972 (write_state_fields): ditto, renaming to...
7973 (state_writer::write_state_fields)
7974 (write_state_a_string): ditto, renaming to...
7975 (state_writer::write_state_a_string)
7976 (write_state_string_option): ditto, renaming to...
7977 (state_writer::write_state_string_option)
7978 (write_state_type_option): ditto, renaming to...
7979 (state_writer::write_state_type_option)
7980 (write_state_nested_option): ditto, renaming to...
7981 (state_writer::write_state_nested_option)
7982 (write_state_option): ditto, renaming to...
7983 (state_writer::write_state_option)
7984 (write_state_options): ditto, renaming to...
7985 (state_writer::write_state_options)
7986 (write_state_lang_bitmap): ditto, renaming to...
7987 (state_writer::write_state_lang_bitmap)
7988 (write_state_version): ditto, renaming to...
7989 (state_writer::write_state_version)
7990 (write_state_scalar_type): ditto, renaming to...
7991 (state_writer::write_state_scalar_type)
7992 (write_state_string_type): ditto, renaming to...
7993 (state_writer::write_state_string_type)
7994 (write_state_undefined_type): ditto, renaming to...
7995 (state_writer::write_state_undefined_type)
7996 (write_state_struct_union_type): ditto, renaming to...
7997 (state_writer::write_state_struct_union_type)
7998 (write_state_struct_type): ditto, renaming to...
7999 (state_writer::write_state_struct_type)
8000 (write_state_user_struct_type): ditto, renaming to...
8001 (state_writer::write_state_user_struct_type)
8002 (write_state_lang_struct_type): ditto, renaming to...
8003 (state_writer::write_state_lang_struct_type)
8004 (write_state_param_struct_type): ditto, renaming to...
8005 (state_writer::write_state_param_struct_type)
8006 (write_state_pointer_type): ditto, renaming to...
8007 (state_writer::write_state_pointer_type)
8008 (write_state_array_type): ditto, renaming to...
8009 (state_writer::write_state_array_type)
8010 (write_state_gc_used): ditto, renaming to...
8011 (state_writer::write_state_gc_used)
8012 (write_state_common_type_content): ditto, renaming to...
8013 (state_writer::write_state_common_type_content)
8014 (write_state_type): ditto, renaming to...
8015 (state_writer::write_state_type)
8016 (write_state_pair_list): ditto, renaming to...
8017 (state_writer::write_state_pair_list)
8018 (write_state_pair): ditto, renaming to...
8019 (state_writer::write_state_pair)
8020 (write_state_typedefs): ditto, renaming to...
8021 (state_writer::write_state_typedefs)
8022 (write_state_structures): ditto, renaming to...
8023 (state_writer::write_state_structures)
8024 (write_state_param_structs): ditto, renaming to...
8025 (state_writer::write_state_param_structs)
8026 (write_state_variables): ditto, renaming to...
8027 (state_writer::write_state_variables)
8028 (write_state_srcdir): ditto, renaming to...
8029 (state_writer::write_state_srcdir)
8030 (write_state_files_list): ditto, renaming to...
8031 (state_writer::write_state_files_list)
8032 (write_state_languages): ditto, renaming to...
8033 (state_writer::write_state_languages)
8034 (write_state): create a state_writer instance and use it when
8035 writing out the state file
8037 2013-05-17 Mike Stump <mikestump@comcast.net>
8039 PR rtl-optimization/57304
8040 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8041 accessing DF_REF_REAL_LOC.
8043 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8045 PR rtl-optimization/57281
8046 PR rtl-optimization/57300
8047 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8048 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8049 what the other splitter did if the registers are dead.
8051 2013-05-17 Richard Biener <rguenther@suse.de>
8053 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8056 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8058 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8061 2013-05-17 Marek Polacek <polacek@redhat.com>
8063 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8064 length when doing non-zero store of storing '\0' to '\0'.
8066 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8068 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8069 vect_external_def oprnd1 with loop_vinfo, try to emit
8070 optional cast, negation and and stmts on the loop preheader
8071 edge instead of into the pattern def seq.
8073 PR tree-optimization/57051
8074 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8075 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8077 2013-05-16 Nick Clifton <nickc@redhat.com>
8079 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8080 (rl78_is_naked_func): New function.
8081 (rl78_expand_prologue): Skip prologue generation for naked functions.
8082 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8083 * doc/extend.texi (naked): Add RL78 to the list of processors
8084 that supports this attribute.
8086 2013-05-16 Jeff Law <law@redhat.com>
8088 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8090 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8092 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8093 cache parameters using detect_caches_amd also for CYRIX,
8094 NSC and TM2 signatures.
8096 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8097 Dzianis Kahanovich <mahatma@eu.by>
8101 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8102 VIA/Centaur processors and determine their cache parameters
8103 using detect_caches_amd.
8105 2013-05-16 Teresa Johnson <tejohnson@google.com>
8107 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8108 (rtl_verify_edges): New function.
8109 (rtl_verify_bb_insns): Ditto.
8110 (rtl_verify_bb_pointers): Ditto.
8111 (rtl_verify_bb_insn_chain): Ditto.
8112 (rtl_verify_fallthru): Ditto.
8113 (rtl_verify_bb_layout): Ditto.
8114 (rtl_verify_flow_info_1): Outline checks into new functions.
8115 (rtl_verify_flow_info): Ditto.
8117 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
8119 * cfghooks.c (copy_bbs): Add update_dominance argument.
8120 * cfghooks.h (copy_bbs): Update prototype.
8121 * tree-cfg.c (gimple_duplicate_sese_region):
8122 Add update_dominance argument.
8123 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8124 * tree-ssa-loop-ch.c (copy_loop_headers): Update
8125 gimple_duplicate_sese_region call.
8126 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8127 Update copy_bbs call.
8128 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8129 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8131 2013-05-16 Jakub Jelinek <jakub@redhat.com>
8133 * tree-vectorizer.h (NUM_PATTERNS): Increment.
8134 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8135 vect_recog_rotate_pattern.
8136 (vect_recog_rotate_pattern): New function.
8138 2013-05-16 Jason Merrill <jason@redhat.com>
8140 * Makefile.in (LLINKER): New variable.
8141 (mostlyclean): Remove link mutex.
8142 * configure.ac: Handle --enable-link-mutex.
8143 * lock-and-run.sh: New script.
8145 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8148 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8151 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8153 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8155 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
8157 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8158 * config/arm/arm.c (next_consecutive_mem): New function.
8159 (gen_movmem_ldrd_strd): Likewise.
8160 * config/arm/arm.md (movmemqi): Update condition and code.
8161 (unaligned_loaddi, unaligned_storedi): New patterns.
8163 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8165 * config.gcc: Obsolete *-*-solaris2.9*.
8166 * doc/install.texi (Specific, *-*-solaris2*): Document it.
8168 2013-05-16 Richard Biener <rguenther@suse.de>
8170 * passes.c (init_optimization_passes): Move pass_parallelize_loops
8171 earlier, after GRAPHITE transforms and IV canonicalization.
8173 2013-05-16 Jakub Jelinek <jakub@redhat.com>
8175 * omp-low.c (extract_omp_for_data): For collapsed loops,
8176 if at least one of the loops is known at compile time to
8177 iterate zero times, set count to 0.
8178 (expand_omp_regimplify_p): New function.
8179 (expand_omp_for_generic): For collapsed loops, if at least
8180 one of the loops isn't known to iterate at least once,
8181 add runtime check with setting count to 0.
8182 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8183 For unsigned types if it isn't known at compile time that
8184 the loop will iterate at least once, add runtime check to bypass
8185 the whole loop if initial condition isn't true.
8187 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
8189 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8191 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
8194 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8195 transformations to avoid an infinite loop.
8197 2013-05-16 Marek Polacek <polacek@redhat.com>
8199 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8201 2013-05-15 Leif Ekblad <leif@rdos.net>
8203 * config/i386/i386.c (ix86_decompose_address): Use
8204 DEFAULT_TLS_SEG_REG to access TLS segment register.
8205 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8206 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8207 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8209 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
8212 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8213 sibling calls to functions that would normally be lazily bound,
8214 unless $gp is call-clobbered.
8216 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
8218 * config/i386/i386.c (ix86_option_override_internal): Update
8219 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
8220 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8221 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8222 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8224 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8226 2013-05-15 Andreas Schwab <schwab@suse.de>
8228 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8229 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
8232 2013-05-15 Teresa Johnson <tejohnson@google.com>
8234 * loop-unroll.c (report_unroll_peel): Check decision before
8235 emitting unroll/peel message.
8237 2013-05-15 Teresa Johnson <tejohnson@google.com>
8239 * function.h (has_bb_partition): New rtl_data flag.
8240 (bb_reorder_complete): Ditto.
8241 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8242 instead of flag_reorder_blocks_and_partition.
8243 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8244 with some enhancements.
8245 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8246 * bb-reorder.c (connect_traces): Check for has_bb_partition
8247 instead of flag_reorder_blocks_and_partition.
8248 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8249 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8250 verify_hot_cold_block_grouping.
8251 (partition_hot_cold_basic_blocks): Set has_bb_partition.
8253 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8256 * config/arm/predicates.md (call_insn_operand): New predicate.
8257 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
8258 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8259 if insn is not a tail call.
8260 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8262 * config/arm/arm.h (enum reg_class): New caller save register class.
8263 (REG_CLASS_NAMES): Likewise.
8264 (REG_CLASS_CONTENTS): Likewise.
8265 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8268 2013-05-15 Richard Biener <rguenther@suse.de>
8270 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8271 of MSG_OPTIMIZED_LOCATIONS.
8272 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8273 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8275 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8276 of MSG_OPTIMIZED_LOCATIONS.
8277 (execute_vect_slp): Likewise.
8278 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8279 (vect_create_cond_for_alias_checks): Likewise.
8280 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8281 (vect_recog_widen_mult_pattern): Likewise.
8282 (vect_recog_widen_sum_pattern): Likewise.
8283 (vect_recog_over_widening_pattern): Likewise.
8284 (vect_recog_widen_shift_pattern): Likewise.
8285 (vect_recog_vector_vector_shift_pattern): Likewise.
8286 (vect_recog_divmod_pattern): Likewise.
8287 (vect_recog_mixed_size_cond_pattern): Likewise.
8288 (vect_recog_bool_pattern): Likewise.
8289 (vect_pattern_recog_1): Likewise.
8291 2013-05-15 Martin Jambor <mjambor@suse.cz>
8293 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8294 non-functions to builtin_unreachable.
8295 * ipa-inline-transform.c (inline_call): Do not assert estimates were
8296 correct when new direct edges were discovered.
8298 2013-05-15 Martin Jambor <mjambor@suse.cz>
8300 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8301 header, print symbol order instead of node uid, print more information
8302 about indirect edge targets.
8303 (ipa_make_edge_direct_to_target): Print symbol order instead of node
8305 (ipa_make_edge_direct_to_target): Likewise.
8306 (remove_described_reference): Likewise.
8307 (propagate_controlled_uses): Likewise.
8308 (ipa_print_node_params): Also print symbol order.
8309 (ipcp_transform_function): Print symbol order instead of node uids.
8310 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8311 (cgraph_get_create_real_symbol_node): Likewise.
8312 * ipa-cp.c (print_lattice): Likewise.
8313 (print_all_lattices): Likewise.
8314 (determine_versionability): Likewise.
8315 (initialize_node_lattices): Likewise.
8316 (estimate_local_effects): Likewise.
8317 (update_profiling_info): Likewise.
8318 (create_specialized_node): Likewise.
8319 (perhaps_add_new_callers): Likewise.
8320 (decide_about_value): Likewise.
8321 (decide_whether_version_node): Likewise.
8322 (identify_dead_nodes): Likewise.
8323 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8324 (dump_inline_summary): Likewise.
8325 (estimate_node_size_and_time): Likewise.
8326 (inline_analyze_function): Likewise.
8327 * ipa-inline.c (report_inline_failed_reason): Likewise.
8328 (want_early_inline_function_p): Likewise.
8329 (edge_badness): Likewise.
8330 (update_edge_key): Likewise.
8331 (inline_small_functions): Likewise. Add dumping of order to two other
8333 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8334 instead of node uids.
8335 (propagate_pure_const): Likewise.
8336 (propagate_pure_const): Likewise.
8337 * ipa-utils.c (dump_cgraph_node_set): Likewise.
8338 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8339 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8341 * tree-pretty-print.c (dump_function_header): Likewise.
8342 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8343 Print symbol order instead of node uids.
8345 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8347 * config/s390/s390.c (s390_register_move_cost): Don't impose the
8348 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8350 2013-05-15 Richard Biener <rguenther@suse.de>
8352 PR tree-optimization/57275
8353 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8354 return value for fail to do runtime alias checks for gather loads.
8356 2013-05-15 Jan Hubicka <jh@suse.cz>
8360 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8361 weakrefs are not external.
8362 (lto_symtab_merge_decls): Fix thinko when dealing with
8363 non-lto_symtab decls.
8364 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8365 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8366 * varpool.c (dump_varpool_node): Dump more flags.
8368 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8370 * config/i386/i386.c (processor_alias_table): Add instruction
8371 FSGSBASE for AMD bdver3 architecture.
8373 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8375 * tree.c (warn_deprecated_use): Print file:line using locus color.
8376 * diagnostic.c (diagnostic_report_current_module): Print file:line
8377 and file:line:column using locus color.
8379 2013-05-14 Mike Stump <mikestump@comcast.net>
8381 * gdbinit.in: Add __null.
8383 2013-05-14 Mike Stump <mikestump@comcast.net>
8385 * recog.h: Rename struct recog_data to Recog_data.
8386 * recog.c: Likewise.
8387 * reload.c (can_reload_into): Likewise.
8388 * config/picochip/picochip.c: Likewise.
8390 2013-05-14 Mike Stump <mikestump@comcast.net>
8392 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8394 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
8396 * resource.h (struct resources): Remove unch_memory member.
8397 (CLEAR_RESOURCE): Don't clear unch_memory.
8398 * resource.c (mark_referenced_resources): Don't set it.
8399 (mark_set_resources): Likewise.
8400 (mark_target_live_regs): Don't clear it.
8401 (init_resource_info): Likewise.
8402 * reorg.c (resource_conflicts_p): Don't compare it.
8403 (redundant_insn): Don't set it.
8405 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8407 * emit-rtl.c (next_label): Remove unused function.
8408 (skip_consecutive_labels, link_cc0_insns): Move to ...
8409 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8410 only place where these functions are used, and make them static.
8412 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8414 * fold-const.c (fold_negate_expr): Handle vectors.
8415 (fold_truth_not_expr): Make it static.
8416 (fold_invert_truthvalue): New static function.
8417 (invert_truthvalue_loc): Handle vectors. Do not call
8418 fold_truth_not_expr directly.
8419 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8420 <TRUTH_NOT_EXPR>: Do not cast to boolean.
8421 (fold_comparison): Handle vector constants.
8422 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8423 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8424 * tree.h (fold_truth_not_expr): Remove declaration.
8426 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
8428 * config/aarch64/aarch64-simd.md
8429 (aarch64_vcond_internal<mode>): Rename to...
8430 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8431 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8432 float modes. Clarify all iterator modes.
8433 (vcond<mode><mode>): Use new name for vcond expanders.
8434 (vcond<v_cmp_result><mode>): Likewise.
8435 (vcondu<mode><mode>: Likewise.
8436 * config/aarch64/iterators.md (VDQF_COND): New.
8438 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8441 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8442 variable for the shift amount. Check that we shift by non-negative
8445 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
8448 * config/arm/arm.h (EPILOGUE_USES): Only return true
8449 for LR_REGNUM after epilogue_completed.
8451 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
8453 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8456 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8459 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8460 and Solaris 11+/x86 with gld.
8461 * configure: Regenerate.
8463 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8465 * expmed.c (expand_shift_1): Canonicalize rotates by
8466 constant bitsize / 2 to bitsize - 1.
8467 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8468 case ROTATERT>: Likewise.
8471 2013-05-10 Jakub Jelinek <jakub@redhat.com>
8473 * config/i386/i386.md (rotateinv): New code attr.
8474 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8475 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8476 roll $31, %eax, etc.
8478 2013-05-14 Richard Biener <rguenther@suse.de>
8481 * tree-eh.c (sink_clobbers): Give up for successors with
8482 multiple predecessors and no virtual uses.
8484 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
8486 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8487 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8489 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8492 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8493 the case when both op0 and op1 have VOIDmode.
8495 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8497 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8498 in multiply-accumulate mode.
8500 2013-05-13 Guozhi Wei <carrot@google.com>
8502 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8504 2013-05-13 Kai Tietz <ktietz@redhat.com>
8507 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8508 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8509 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8510 * config/i386/i386.c (ix86_option_override_internal): Likewise.
8511 (ix86_expand_prologue): Likewise.
8512 (ix86_expand_split_stack_prologue): Likewise.
8513 (legitimate_pic_address_disp_p): Likewise.
8514 (legitimize_pic_address): Likewise.
8515 (legitimize_tls_address): Likewise.
8516 (legitimize_pe_coff_symbol): Likewise.
8517 (output_pic_addr_const): Likewise.
8518 (construct_plt_address): Likewise.
8519 (ix86_expand_call): Likewise.
8520 (x86_output_mi_thunk): Likewise.
8521 (x86_function_profiler): Likewise.
8523 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
8525 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8526 similar switch cases.
8527 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8528 (aarch64_simd_mov_to_<mode>low): Delete.
8529 (aarch64_simd_mov_to_<mode>high): Delete.
8530 (move_lo_quad_<mode>): Add w<-r alternative.
8531 (aarch64_simd_move_hi_quad_<mode>): Likewise.
8532 (aarch64_simd_mov_from_*): Update type attribute.
8533 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8536 2013-05-13 Jan Hubicka <jh@suse.cz>
8538 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8539 * config/i386/i386.c (ix86_compute_frame_layout,
8540 ix86_expand_epilogue, emit_i387_cw_initialization,
8541 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8542 ix86_local_alignment): Fix use of size/speed predicates.
8544 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8546 PR tree-optimization/45216
8547 PR tree-optimization/57157
8548 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8549 the (-Y) & (B - 1) variant if OP is |.
8550 * expmed.c (expand_shift_1): For rotations by const0_rtx just
8551 return shifted. Use (-op1) & (prec - 1) as other_amount
8552 instead of prec - op1.
8554 2013-05-13 Martin Jambor <mjambor@suse.cz>
8557 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8558 (ipa_constant_data): New type.
8559 (ipa_jump_func): Use ipa_constant_data to hold information about
8560 constant jump functions.
8561 (ipa_get_jf_constant): Adjust to jump function type changes.
8562 (ipa_get_jf_constant_rdesc): New function.
8563 (ipa_param_descriptor): New field controlled_uses.
8564 (ipa_get_controlled_uses): New function.
8565 (ipa_set_controlled_uses): Likewise.
8566 * ipa-ref.h (ipa_find_reference): Declare.
8567 * ipa-prop.c (ipa_cst_ref_desc): New type.
8568 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8570 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
8571 New parameter cs. Adjust all callers.
8572 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8573 (remove_described_reference): New function.
8574 (jfunc_rdesc_usable): Likewise.
8575 (try_make_edge_direct_simple_call): Decrement controlled use count,
8576 attempt to remove reference if it hits zero.
8577 (combine_controlled_uses_counters): New function.
8578 (propagate_controlled_uses): Likewise.
8579 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8580 (ipa_edge_duplication_hook): Duplicate reference descriptions.
8581 (ipa_print_node_params): Print described use counter.
8582 (ipa_write_jump_function): Adjust to jump function type changes.
8583 (ipa_read_jump_function): New parameter CS, pass it to
8584 ipa_set_jf_constant. Adjust caller.
8585 (ipa_write_node_info): Stream controlled use count
8586 (ipa_read_node_info): Likewise.
8587 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8589 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8590 count. Remove cloning-added reference if it reaches zero.
8591 * ipa-ref.c (ipa_find_reference): New function.
8593 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8595 * config/i386/i386.c (processor_target_table): Modified default
8596 alignment values for AMD BD and BT architectures.
8598 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
8600 * tree-vect-generic.c (uniform_vector_p): Move ...
8601 * tree.c (uniform_vector_p): ... here.
8602 * tree.h (uniform_vector_p): Declare it.
8603 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8606 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8608 PR tree-optimization/57230
8609 * tree-ssa-strlen.c (handle_char_store): Record length for
8610 array store from STRING_CST.
8612 PR tree-optimization/57230
8613 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8616 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
8618 * config/epiphany/epiphany.c (epiphany_init): Check size of
8619 NUM_MODES_FOR_MODE_SWITCHING.
8620 (epiphany_expand_prologue):
8621 Remove CONFIG_REGNUM initial value handling code.
8622 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8623 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8624 (emit_set_fp_mode, epiphany_mode_after): Likewise.
8625 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8626 Don't return 1 for FP_MODE_NONE.
8627 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8628 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8629 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8630 * config/epiphany/epiphany.md (save_config): New pattern.
8632 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
8634 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8636 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
8638 * config/i386/i386.md (memory): Handle sseishft1.
8639 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8640 (*vec_extractv2di_1): Ditto.
8642 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
8644 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8647 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
8649 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8650 Add mthumb/march=armv7-a multilib.
8651 Add mthumb/march=armv7-r multilib.
8652 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8654 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
8656 * config/v850/t-rtems: Add more multilibs.
8658 2013-05-10 Richard Biener <rguenther@suse.de>
8660 PR tree-optimization/57214
8661 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8662 not propagate from SSA names that occur in abnormal PHI nodes.
8664 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
8666 * stor-layout.c (element_precision): New function.
8667 * machmode.h (element_precision): Declare it.
8668 * tree.c (build_minus_one_cst): New function.
8669 (element_precision): Likewise.
8670 * tree.h (build_minus_one_cst): Declare new function.
8671 (element_precision): Likewise.
8672 * fold-const.c (operand_equal_p): Use element_precision.
8673 (fold_binary_loc): Handle vector types.
8674 * convert.c (convert_to_integer): Use element_precision.
8675 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8678 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
8680 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8681 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8682 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8683 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8684 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8685 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8686 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8687 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8688 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8689 (Uuw8): New constraints.
8690 (Usb4): Move into alphabetical order.
8691 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8692 (sd8_operand, ub8_operand, uw8_operand): New predicates.
8693 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8694 previously unnamed patterns.
8695 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8696 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8697 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8698 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8699 of set_attr_alternative/if_then_else. Use extended_mips16 instead
8700 of specific lengths.
8702 2013-05-10 Jakub Jelinek <jakub@redhat.com>
8704 * config/i386/i386.md (rotateinv): New code attr.
8705 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8706 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8707 roll $31, %eax, etc.
8709 PR tree-optimization/45216
8710 PR tree-optimization/57157
8711 * tree-ssa-forwprop.c (simplify_rotate): New function.
8712 (ssa_forward_propagate_and_combine): Call it.
8714 2013-05-10 Richard Biener <rguenther@suse.de>
8716 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8717 disable peeling when we version for aliasing.
8718 (vector_alignment_reachable_p): Honor explicit user alignment.
8719 (vect_supportable_dr_alignment): Likewise.
8720 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8721 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8722 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8723 then peeling to arrange for the cost-model check to come first.
8725 2013-05-10 Alan Modra <amodra@gmail.com>
8727 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8728 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8729 * configure: Regenerate.
8731 2013-05-10 Alan Modra <amodra@gmail.com>
8734 * varasm.c (default_elf_select_section): Move !DECL_P check..
8735 (get_named_section): ..to here before calling get_section_name.
8737 (default_section_type_flags): Add DECL_P check.
8738 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8739 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8741 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
8743 * config/epiphany/epiphany.c (epiphany_expand_prologue):
8744 When using gen_stack_adjust_str with a register offset, add a
8745 REG_FRAME_RELATED_EXPR note.
8747 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
8749 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8750 (*vec_extractv4si_zext_mem): Ditto.
8751 (*vec_extractv2di): Add 0->x and x->x alternatives.
8752 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8753 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8755 2013-05-09 Jason Merrill <jason@redhat.com>
8757 N3639 C++1y VLA support
8758 * gimplify.c (gimplify_vla_decl): Don't touch an existing
8761 * tree.c (build_constructor_va): New.
8762 * tree.h: Declare it.
8764 2013-05-09 Martin Jambor <mjambor@suse.cz>
8767 * gimple-fold.c (canonicalize_constructor_val): Call
8768 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8770 2013-05-09 Jan Hubicka <jh@suse.cz>
8771 Richard Biener <rguenther@suse.de>
8774 * symtab.c (symtab_make_decl_local): Do not add private names.
8776 2013-05-09 Jan Hubicka <jh@suse.cz>
8779 * symtab.c (insert_to_assembler_name_hash): Handle clones.
8780 (unlink_from_assembler_name_hash): Likewise.
8781 (symtab_prevail_in_asm_name_hash, symtab_register_node,
8782 symtab_unregister_node, symtab_initialize_asm_name_hash,
8783 change_decl_assembler_name): Update.
8785 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
8787 * config/aarch64/aarch64.md: New movtf split.
8788 (*movtf_aarch64): Update.
8789 (aarch64_movdi_tilow): Handle TF modes and rename to
8790 aarch64_movdi_<mode>low.
8791 (aarch64_movdi_tihigh): Handle TF modes and rename to
8792 aarch64_movdi_<mode>high
8793 (aarch64_movtihigh_di): Handle TF modes and rename to
8794 aarch64_mov<mode>high_di
8795 (aarch64_movtilow_di): Handle TF modes and rename to
8796 aarch64_mov<mode>low_di
8797 (aarch64_movtilow_tilow): Remove spurious whitespace.
8798 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8800 (aarch64_print_operand): Update.
8802 2013-05-09 Alan Modra <amodra@gmail.com>
8804 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8806 * configure: Regenerate.
8808 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
8810 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8811 splitter preparation statements.
8812 * config/i386/sse.md (*vec_extract* splitters): Ditto.
8813 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8816 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8818 * gimple-ssa-strength-reduction.c (count_candidates): Change
8819 return value to int.
8820 (analyze_candidates_and_replace): Change type of length to int.
8822 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
8824 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8825 (*vec_extract<mode>): Use VI12_128 mode iterator.
8826 (*vec_extract<mode>_mem): Ditto.
8827 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8830 2013-05-08 Diego Novillo <dnovillo@google.com>
8835 2012-08-17 Diego Novillo <dnovillo@google.com>
8838 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8839 * config.in: Regenerate.
8840 * configure: Regenerate.
8841 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8843 2013-05-08 Jan Hubicka <jh@suse.cz>
8846 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8847 * cgraph.h (symtab_node_base): Add unique_name.
8848 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8849 input_overwrite_node, input_varpool_node): Stream unique_name.
8850 * cgraphclones.c (cgraph_create_virtual_clone,
8851 cgraph_function_versioning): Set unique_name.
8852 * ipa.c (function_and_variable_visibility): Set unique_name.
8854 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8856 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8857 (alloc_cand_and_find_basis): Restrict conditional candidate
8858 processing to CAND_MULTs.
8860 2013-05-08 Jan Hubicka <jh@suse.cz>
8863 lto-symtab.c (lto_symtab_symbol_p): New function.
8864 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8865 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8866 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8867 Skip static symbols.
8869 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
8871 PR tree-optimization/57200
8872 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8873 Only call inform if the preceding warning_at returns true.
8875 2013-05-07 Han Shen <shenhan@google.com>
8877 * cfgexpand.c (record_or_union_type_has_array_p): New function.
8878 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8879 * common.opt (fstack-protector-strong): New option.
8880 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8881 * doc/invoke.texi (Optimization Options): Document
8882 "-fstack-protector-strong".
8883 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8885 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
8887 * config/mips/mips.c (mips_machine_reorg2): Return 0.
8889 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
8891 * ira.c (update_equiv_regs): Add insn having equiv memory even if
8892 it is not lhs of the insn.
8893 (setup_reg_equiv): Remove insn having equiv memory which it is not
8895 * lra-constraints.c (process_address): Try to improve generation
8896 code for address base + disp.
8897 (lra_constraints): Make correct the code for checking insn setting
8898 up backward equivalence. Remove insn only if it is in the init
8900 * lra-eliminations.c (update_reg_eliminate): Change return value.
8901 (lra_eliminate): Use the result.
8903 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
8905 * config/i386/sse.md (ssescalarnummask): New mode attribute.
8906 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8907 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8908 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8909 register target operands.
8910 (*vec_extractv8hi_sse2): New pattern.
8911 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8912 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8913 (*vec_extract<mode>_mem): New insn and split pattern.
8915 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
8917 * config/arm/arm.c (arm_asan_shadow_offset): New function.
8918 (TARGET_ASAN_SHADOW_OFFSET): Define.
8919 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8920 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8922 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8924 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8925 (incr_vec_index): Return -1 if increment not found.
8926 (create_add_on_incoming_edge): Assert if increment not found.
8927 (record_increment): Limit number of increments recorded.
8928 (all_phi_incrs_profitable): Return false if an increment not found.
8929 (replace_profitable_candidates): Don't process increments that were
8931 (analyze_candidates_and_replace): Limit size of incr_vec.
8933 2013-05-07 Richard Biener <rguenther@suse.de>
8935 * calls.c (special_function_p): setjmp-like functions are leaf.
8936 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8937 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8939 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
8941 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8942 (aarch64_simd_mov<mode>): New expander.
8943 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8944 (aarch64_simd_mov_to_<mode>high): Likewise.
8945 (aarch64_simd_mov_from_<mode>low): Likewise.
8946 (aarch64_simd_mov_from_<mode>high): Likewise.
8947 (aarch64_dup_lane<mode>): Update.
8948 (aarch64_dup_lanedi): New instruction pattern.
8949 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8950 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8952 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8954 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8955 (replace_mult_candidate): Remove unnecessary argument; remove
8956 unnecessary parameter from call to introduce_cast_before_cand.
8957 (replace_unconditional_candidate): Remove unnecessary parameter
8958 from call to replace_mult_candidate.
8959 (replace_conditional_candidate): Likewise.
8960 (insert_initializers): Use make_temp_ssa_name.
8961 (introduce_cast_before_cand): Remove unnecessary argument; use
8963 (replace_one_candidate): Remove unnecessary argument; remove
8964 unnecessary parameter from calls to introduce_cast_before_cand.
8965 (replace_profitable_candidates): Remove unnecessary parameters
8966 from calls to replace_one_candidate.
8968 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8970 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8971 phi def as possibly hiding a basis for a CAND_ADD whose operands
8972 have been commuted in the analysis.
8973 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8975 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
8977 * config/aarch64/aarch64.md
8978 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8979 shift value between 0-4.
8981 2013-05-07 Richard Biener <rguenther@suse.de>
8983 * double-int.h (rshift): New overload.
8984 * double-int.c (rshift): New function.
8985 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8986 (create_reference_ops_from_ref): Remove.
8987 (vn_reference_insert): Use shared ops for constructing the
8988 reference and copy it.
8990 2013-05-07 Richard Biener <rguenther@suse.de>
8993 * tree-eh.c (sink_clobbers): Properly propagate
8994 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8996 2013-05-07 Jakub Jelinek <jakub@redhat.com>
8998 PR tree-optimization/57149
8999 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9000 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9001 collect_phi_def_edges, execute_late_warn_uninitialized): Use
9002 uninit_undefined_value_p instead of ssa_undefined_value_p.
9005 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9006 for modifier == EXPAND_INITIALIZER.
9008 2013-05-07 Anton Blanchard <anton@samba.org>
9010 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9011 for powerpc64 little endian.
9012 * configure: Regenerate.
9014 2013-05-06 Graham Stott <grahams@btinternet.com>
9016 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9017 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9018 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9019 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9021 2013-05-06 Graham Stott <grahams@btinternet.com>
9023 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9024 codes which allow non-lvalues.
9026 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
9028 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9029 components are all 1s.
9030 (integer_minus_onep): New function.
9031 * tree.h (integer_minus_onep): Declare it.
9032 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9033 integer_minus_onep instead of integer_all_onesp.
9035 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9038 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9039 variations of these patterns.
9041 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9043 * config/i386/i386.md (isa): Add x64_sse4 member.
9044 (enabled): Handle x64_sse4.
9045 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9046 instruction for 64bit SSE4_1 targets. Update insn attributes.
9047 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9048 instruction for SSE4_1 targets. Update insn attributes.
9049 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9050 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9051 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9053 (*vec_extractv4si): Rename from *sse4_1_pextrd.
9054 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9055 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9057 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9060 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9062 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
9064 * final.c (do_assembler_dialects): Don't handle curly braces and
9065 vertical bar escaped by % as dialect delimiters.
9066 (output_asm_insn): Print curly braces and vertical bar if escaped
9067 by % and ASSEMBLER_DIALECT defined.
9068 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9069 * doc/tm.texi: Regenerated.
9071 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
9073 * config/mips/mips.c: Include tree-pass.h.
9074 (mips_reorg): Split in pre- and post-dbr_schedule parts.
9075 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9076 (pass_mips_machine_reorg2): New machine specific pass.
9077 (insert_pass_mips_machine_reorg2): New pass plugin definition.
9078 (mips_option_override): Register the new pass.
9079 * rtl.h (cleanup_barriers): Remove prototype.
9080 (dbr_schedule): Likewise.
9081 * jump.c (cleanup_barriers): Make static.
9082 * reorg.c (dbr_schedule): Likewise.
9084 2013-05-06 Richard Biener <rguenther@suse.de>
9086 PR tree-optimization/57185
9087 * tree-parloops.c (add_field_for_reduction): Handle anonymous
9090 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9093 * config/i386/i386.c (add_parameter_dependencies): Add dependence
9094 between "first_arg" and "insn", not "last" and "insn".
9096 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9098 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9099 (find_candidates_in_block): Re-enable slsr_process_phi.
9100 (create_phi_basis): Fix double counting of candidate adjustment.
9102 2013-05-06 Richard Biener <rguenther@suse.de>
9105 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9106 the edge is also fallthru, preserve it and just clear the
9108 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9109 also complex, preserve that and just clear the fallthru flag.
9110 * tree-inline.c (update_ssa_across_abnormal_edges): Also
9111 update virtual operands.
9113 2013-05-06 Alan Modra <amodra@gmail.com>
9115 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9116 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9117 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9118 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9119 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9120 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9122 2013-05-06 Alan Modra <amodra@gmail.com>
9124 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9125 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
9126 (DEFAULT_ASM_ENDIAN): Define.
9127 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9128 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9129 Update -K PIC clause from sysv4.h.
9130 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9131 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9133 2013-05-06 Alan Modra <amodra@gmail.com>
9135 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9136 twice for little-endian.
9137 (ashrdi3_no_power, ashrdi3): Support little-endian.
9139 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9142 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9143 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9145 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9146 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9147 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9150 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
9151 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
9153 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9154 * config/pa/pa.opt: Make mbig-switch a no-op.
9155 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9156 (CASE_VECTOR_MODE): Always return SImode.
9157 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9158 for the !TARGET_BIG_SWITCH case.
9159 * config/pa/pa-linux.h: Likewise.
9160 * config/pa/pa-openbsd.h: Likewise.
9161 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9162 * config/pa/pa.md (short_jump): Remove define_insn.
9163 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9164 (casesi0): Remove define_insn.
9165 (type): Remove btable_branch.
9166 (pa_combine_type): Likewise.
9167 (in_nullified_branch_delay): Likewise.
9168 (in_call_delay): Likewise.
9169 (define_delay): Likewise.
9170 (define_insn_reservation "Z3"): Likewise.
9171 (define_insn_reservation "Z4"): Likewise.
9172 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9173 (pa_adjust_insn_length): Remove adjustment for btable branches.
9174 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9177 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
9179 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9180 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9181 Add m->r,x alternatives.
9182 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9183 splitters using SWI48x mode iterator.
9184 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
9185 TARGET_64BIT. Add m->x alternative.
9186 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9187 Add o->x alternative. Enable for TARGET_SSE.
9188 (sse_storeq): Remove expander.
9189 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9190 with memory input operand.
9191 (*vec_extractv2di_1 splitter): New.
9192 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9193 * config/i386/i386.md (ssevecmodelower): New mode attribute.
9195 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
9197 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
9198 (INT_LOWPART): Delete.
9199 (extract_MB): Adjust.
9200 (extract_ME): Adjust.
9201 (print_operand): Adjust.
9203 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
9205 * config/rs6000/predicates.md (reg_or_add_cint_operand,
9206 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9207 (reg_or_logical_cint_operand, easy_fp_constant,
9208 logical_const_operand): Delete "CONST_DOUBLE" case.
9209 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9210 "HOST_BITS_PER_WIDE_INT == 64" test.
9211 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
9212 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9213 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9214 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9216 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9217 CONST_DOUBLE DImode/VOIDmode case.
9218 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9219 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
9220 CONST_DOUBLE VOIDmode case.
9221 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9222 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9223 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9224 Delete CONST_DOUBLE case.
9225 (splitters for mov FMOVE64 const_double): Delete
9226 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
9227 "HOST_BITS_PER_WIDE_INT >= 64" test.
9228 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9230 (mov DI const_double): Delete.
9232 2013-05-04 Jakub Jelinek <jakub@redhat.com>
9234 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9235 on op shows all bits zero in mode of a lowpart subreg, return zero.
9237 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9240 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9241 to save TFmode registers and DImode to save TImode registers for
9242 caller save operations.
9243 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9244 mark being partially clobbered since they only use the first
9247 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9248 and TDmode only use the upper 64-bits of each VSX register.
9250 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9252 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9253 (find_candidates_in_block): Disable slsr_process_phi.
9255 2013-05-03 Guozhi Wei <carrot@google.com>
9257 * coverage.c (coverage_obj_init): Move the construction of gcov
9259 (build_init_ctor): ... here.
9261 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9263 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9264 (slsr_cand_d): Redefine def_phi.
9265 (stride_status, phi_adjust_status, count_phis_status): New enums.
9266 (find_phi_def): New.
9267 (find_basis_for_base_expr): New.
9268 (find_basis_for_candidate): Handle hidden bases.
9269 (alloc_cand_and_find_basis): Handle phi candidates.
9270 (slsr_process_phi): New.
9271 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9272 (create_mul_imm_cand): Likewise.
9273 (create_add_ssa_cand): Exclude phi base candidates.
9274 (create_add_imm_cand): Likewise.
9275 (slsr_process_cast): Likewise.
9276 (slsr_process_copy): Likewise.
9277 (find_candidates_in_block): Handle phi candidates.
9278 (dump_candidate): Likewise.
9279 (unconditional_cands): Delete.
9280 (unconditional_cands_with_known_stride_p): Delete.
9281 (phi_dependent_cand_p): New.
9282 (cand_increment): Handle phi-dependent candidates.
9283 (replace_dependent): Delete.
9284 (replace_mult_candidate): New.
9285 (replace_unconditional_candidate): New.
9286 (incr_vec_index): Move to avoid forward reference.
9287 (create_add_on_incoming_edge): New.
9288 (create_phi_basis): New.
9289 (replace_dependents): Delete.
9290 (replace_conditional_candidate): New.
9291 (phi_add_costs): New.
9292 (replace_uncond_cands_and_profitable_phis): New.
9293 (record_increment): Handle phi adjustments.
9294 (record_phi_increments): New.
9295 (record_increments): Handle phi adjustments.
9296 (phi_incr_cost): New.
9297 (lowest_cost_path): Handle phis.
9298 (total_savings): Likewise.
9299 (analyze_increments): Likewise.
9300 (ncd_with_phi): New.
9301 (ncd_of_cand_and_phis): New.
9302 (nearest_common_dominator_for_cands): Handle phi increments.
9303 (all_phi_incrs_profitable): New.
9304 (replace_profitable_candidates): Handle phi-dependent candidates.
9305 (analyze_candidates_and_replace): Likewise.
9307 2013-05-03 Teresa Johnson <tejohnson@google.com>
9310 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9311 do not exceed REG_BR_PROB_BASE.
9313 2013-05-03 Jeff Law <law@redhat.com>
9315 PR tree-optimization/57144
9316 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9317 operand of the condition will bit into the new type when eliminating
9318 a cast feeding a condition.
9320 2013-05-03 Jakub Jelinek <jakub@redhat.com>
9322 PR rtl-optimization/57130
9323 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9324 of COMPARE as in_code to the recursive call if needed.
9326 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
9328 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9329 (enabled): Handle new members.
9330 * config/i386/sse.md (*vec_concatv2si): Merge from
9331 *vec_concatv2si_sse2 and vec_concatv2si_sse.
9332 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9334 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
9336 PR tree-optimization/57027
9337 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9338 for fnms opportunity, check we got the prerequisite kind
9339 of tree / gimple before using accessor functions.
9341 2013-05-03 Richard Biener <rguenther@suse.de>
9343 * double-int.h (lshift): New overload without precision
9345 (operator *=, operator +=, operator -=): Move ...
9346 * double-int.c (operator *=, operator +=, operator -=): ... here
9347 and implement more efficiently.
9348 (mul_double_with_sign): Remove.
9349 (lshift_double): Adjust to take unsinged shift argument, push
9350 dispatching code to callers.
9351 (mul_double_wide_with_sign): Add early out for callers that
9352 are not interested in high parts or overflow.
9353 (lshift): New function.
9354 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9356 (lrotate, rrotate): Use logical shifts.
9357 * expr.c (get_inner_reference): Use lshift.
9358 * fixed-value.c (do_fixed_divide): Likewise.
9359 * tree-dfa.c (get_ref_base_and_extent): Likewise.
9360 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9361 (indirect_refs_may_alias_p): Likewise.
9362 (stmt_kills_ref_p_1): Likewise.
9364 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9366 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9368 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9370 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9371 scalar form of FABD instruction.
9373 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
9375 * lra-constraints.c (process_alt_operands): Add checking alt
9376 number to choose the best alternative.
9378 2013-05-02 Richard Biener <rguenther@suse.de>
9380 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9381 bitmap and its handling.
9382 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9384 2013-05-02 Richard Biener <rguenther@suse.de>
9387 * tree-inline.c (copy_loops): Properly handle removed loops.
9388 (copy_cfg_body): Mark destination loops for fixup if source
9391 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
9394 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9395 generating simple_return for naked functions.
9397 2013-05-02 Martin Jambor <mjambor@suse.cz>
9400 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9401 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9403 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9404 ipa_agg_replacement_value structures.
9405 (known_aggs_to_agg_replacement_list): Likewise.
9406 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9407 (read_agg_replacement_chain): Likewise.
9408 (ipcp_transform_function): Also check that by_ref flags match.
9410 2013-05-02 Richard Biener <rguenther@suse.de>
9412 * graphds.h (struct graph): Add obstack member.
9413 * graphds.c (new_graph): Initialize obstack and allocate
9415 (add_edge): Allocate edge from the obstack.
9416 (free_graph): Free the obstack instead of all edges and vertices.
9418 2013-05-02 Teresa Johnson <tejohnson@google.com>
9420 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9422 * cfg.c (update_bb_profile_for_threading): Ditto.
9423 * tree-inline.c (copy_bb): Ditto.
9424 (copy_edges_for_bb): Ditto.
9425 (initialize_cfun): Ditto.
9426 (copy_cfg_body): Ditto.
9427 (expand_call_inline): Ditto.
9428 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9429 (estimate_node_size_and_time): Ditto.
9430 (inline_merge_summary): Ditto.
9431 * cgraphclones.c (cgraph_clone_edge): Ditto.
9432 (cgraph_clone_node): Ditto.
9433 * sched-rgn.c (compute_dom_prob_ps): Ditto.
9434 (compute_trg_info): Ditto.
9436 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9438 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9439 S reg when fp attribute set.
9440 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9442 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9444 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9446 (*and_one_cmplsi3_compare0_uxtw): Likewise.
9447 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9448 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9450 2013-05-02 Richard Biener <rguenther@suse.de>
9452 * tree-scalar-evolution.c (scev_info_hasher): Remove.
9453 (struct instantiate_cache_entry): New type.
9454 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9455 (struct instantiate_cache_type): New type.
9456 (set_instantiated_value, get_instantiated_value): Remove.
9457 (get_instantiated_value_entry): New function.
9458 (instantiate_scev_name): Use the new cache and adjust.
9459 (instantiate_scev_poly): Adjust.
9460 (instantiate_scev_binary): Likewise.
9461 (instantiate_array_ref): Likewise.
9462 (instantiate_scev_convert): Likewise.
9463 (instantiate_scev_not): Likewise.
9464 (instantiate_scev_3): Likewise.
9465 (instantiate_scev_2): Likewise.
9466 (instantiate_scev_r): Likewise.
9467 (instantiate_scev): Likewise.
9468 (resolve_mixers): Likewise.
9470 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
9473 * lra-constraints.c (best_small_class_operands_num): Remove.
9474 (process_alt_operands): Remove small_class_operands_num. Take
9475 small classes operands into losers and only if the operand is not
9476 matched. Modify debugging output.
9477 (curr_insn_transform): Remove best_small_class_operands_num.
9480 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9482 * config/aarch64/aarch64-builtins.c
9483 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9484 * config/aarch64/aarch64-simd-builtins.def
9485 (reduc_splus_): Add new modes.
9486 (reduc_uplus_): New.
9487 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9488 (reduc_uplus_v4sf): Likewise.
9489 (reduc_splus_v4sf): Likewise.
9490 (aarch64_addv<mode>): Likewise.
9491 (reduc_uplus_<mode>): Likewise.
9492 (reduc_splus_<mode>): Likewise.
9493 (aarch64_addvv2di): Likewise.
9494 (reduc_uplus_v2di): Likewise.
9495 (reduc_splus_v2di): Likewise.
9496 (aarch64_addvv2si): Likewise.
9497 (reduc_uplus_v2si): Likewise.
9498 (reduc_splus_v2si): Likewise.
9499 (reduc_<sur>plus_<mode>): New.
9500 (reduc_<sur>plus_v2di): Likewise.
9501 (reduc_<sur>plus_v2si): Likewise.
9502 (reduc_<sur>plus_v4sf): Likewise.
9503 (aarch64_addpv4sf): Likewise.
9504 * config/aarch64/arm_neon.h
9505 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9506 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9507 add UNSPEC_SADDV, UNSPEC_UADDV.
9509 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9511 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9513 * config/aarch64/arm_neon.h
9514 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9516 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9518 * config/aarch64/aarch64-builtins
9519 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9521 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9523 * config/aarch64/aarch64-simd-builtins.def
9525 (reduc_smin_): Likewise.
9526 (reduc_umax_): Likewise.
9527 (reduc_umin_): Likewise.
9528 (reduc_smax_nan_): Likewise.
9529 (reduc_smin_nan_): Likewise.
9532 (smax): Update for V2SF, V4SF and V2DF modes.
9535 (smin_nan): Likewise.
9536 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9537 (<su><maxmin><mode>3): ...This, refactor.
9538 (s<maxmin><mode>3): New.
9539 (<maxmin_uns><mode>3): Likewise.
9540 (reduc_<maxmin_uns>_<mode>): Refactor.
9541 (reduc_<maxmin_uns>_v4sf): Likewise.
9542 (reduc_<maxmin_uns>_v2si): Likewise.
9543 (aarch64_<fmaxmin><mode>: Remove.
9544 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9546 (vmin<q>_f<32,64>): Likewise.
9547 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9549 (su): Add mappings for smax, smin, umax, umin.
9551 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9552 (FMAXMIN): Rename as...
9553 (FMAXMIN_UNS): ...This.
9555 (fmaxminv): Likewise.
9556 (fmaxmin): Likewise.
9558 (maxmin_uns_op): Likewise.
9560 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9562 * config/aarch64/arm_neon.h
9563 (vac<ge, gt><sd>_f<32, 64>): Rename to...
9564 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9565 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9567 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9569 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9570 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9572 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9574 * config/aarch64/aarch64-simd.md
9575 (vcond<mode>_internal): Handle special cases for constant masks.
9576 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9577 (vcondu<mode><mode>): Likewise.
9578 (vcond<v_cmp_result><mode>): New.
9580 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9582 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9583 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9584 * config/aarch64/aarch64-simd-builtins.def
9585 (cmeq): Update to BUILTIN_VALLDI.
9590 * config/aarch64/arm_neon.h
9591 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9592 to builtins or C as appropriate.
9594 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9596 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9598 (cmhi): Rename to...
9600 * config/aarch64/aarch64-simd.md
9601 (simd_mode): Add SF.
9602 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9603 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9604 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9605 (cstore<mode>_neg): ...This.
9606 * config/aarch64/iterators.md
9608 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9610 (UCOMPARISONS): Likewise.
9611 (optab): Add missing comparisons.
9619 (V_cmp_result): Add DF, SF modes.
9620 (v_cmp_result): Likewise.
9623 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9625 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
9627 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9628 define_insn to define_insn_and_split.
9629 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9630 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9631 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9632 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9633 (thumb2_negscc): Likewise.
9635 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
9637 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9639 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
9641 * config/arm/thumb2.md: Remove trailing whitespaces.
9643 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
9645 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9646 Use gen_int_mode rather than GEN_INT.
9648 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
9650 * value-prof.c (stream_in_histogram_value): Remove the strayed
9653 2013-04-30 Richard Biener <rguenther@suse.de>
9656 * cfghooks.c (split_edge): Properly check for the loop latch edge.
9658 2013-04-30 Richard Biener <rguenther@suse.de>
9661 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9663 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
9665 PR rtl-optimization/56957
9666 PR rtl-optimization/57105
9667 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9668 variable. Use just INSN_UID for determining whether an insn
9669 should be only disconnected from the insn stream.
9670 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9672 2013-04-30 Jakub Jelinek <jakub@redhat.com>
9674 PR tree-optimization/57104
9675 * tsan.c (instrument_expr): Don't instrument accesses to
9676 DECL_HARD_REGISTER VAR_DECLs.
9678 2013-04-30 Richard Biener <rguenther@suse.de>
9680 * function.h (loops_for_fn): New inline function.
9681 (set_loops_for_fn): Likewise.
9682 * cfgloop.h (place_new_loop): Add struct function parameter.
9683 (get_loop): Likewise.
9684 (get_loops): Likewise.
9685 (number_of_loops): Likewise.
9687 (fel_init): Likewise.
9688 * cfg.c (get_loop_copy): Adjust.
9689 * cfgloop.c (flow_loops_dump): Likewise.
9690 (record_loop_exits): Likewise.
9691 (verify_loop_structure): Likewise.
9692 * cfgloopanal.c (mark_irreducible_loops): Likewise.
9693 (estimate_reg_pressure_cost): Likewise.
9694 (mark_loop_exit_edges): Likewise.
9695 * cfgloopmanip.c (place_new_loop): Likewise.
9696 (add_loop): Likewise.
9697 (duplicate_loop): Likewise.
9698 * graph.c (draw_cfg_nodes): Likewise.
9699 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9700 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9701 (extract_affine_chrec): Likewise.
9702 (build_scop_iteration_domain): Likewise.
9703 * graphite.c (graphite_initialize): Likewise.
9704 * ira-build.c (create_loop_tree_nodes): Likewise.
9705 (more_one_region_p): Likewise.
9706 (rebuild_regno_allocno_maps): Likewise.
9707 (mark_loops_for_removal): Likewise.
9708 (mark_all_loops_for_removal): Likewise.
9709 (remove_unnecessary_regions): Likewise.
9710 (ira_build): Likewise.
9711 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9712 * loop-init.c (fix_loop_structure): Likewise.
9713 (gate_rtl_move_loop_invariants): Likewise.
9714 (gate_rtl_unswitch): Likewise.
9715 (gate_rtl_unroll_and_peel_loops): Likewise.
9716 (rtl_doloop): Likewise.
9717 * lto-streamer-in.c (input_cfg): Likewise.
9718 * lto-streamer-out.c (output_cfg): Likewise.
9719 * modulo-sched.c (sms_schedule): Likewise.
9720 * predict.c (tree_estimate_probability): Likewise.
9721 (tree_estimate_probability_driver): Likewise.
9722 (estimate_loops): Likewise.
9723 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9724 (move_sese_region_to_fn): Likewise.
9725 (debug_loop_num): Likewise.
9726 * tree-chrec.c (chrec_evaluate): Likewise.
9727 (hide_evolution_in_other_loops_than_loop): Likewise.
9728 (chrec_component_in_loop_num): Likewise.
9729 (reset_evolution_in_loop): Likewise.
9730 (evolution_function_is_invariant_rec_p): Likewise.
9731 * tree-if-conv.c (main_tree_if_conversion): Likewise.
9732 * tree-inline.c (copy_loops): Likewise.
9733 (copy_cfg_body): Likewise.
9734 (tree_function_versioning): Likewise.
9735 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9736 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9738 (add_to_evolution_1): Likewise.
9739 (scev_const_prop): Likewise.
9740 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9741 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9742 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9743 (tree_ssa_lim_initialize): Likewise.
9744 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9745 (verify_loop_closed_ssa): Likewise.
9746 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
9747 (tree_ssa_loop_im): Likewise.
9748 (tree_ssa_loop_unswitch): Likewise.
9749 (tree_vectorize): Likewise.
9750 (check_data_deps): Likewise.
9751 (tree_ssa_loop_ivcanon): Likewise.
9752 (tree_ssa_loop_bounds): Likewise.
9753 (tree_complete_unroll): Likewise.
9754 (tree_complete_unroll_inner): Likewise.
9755 (tree_parallelize_loops): Likewise.
9756 (tree_ssa_loop_prefetch): Likewise.
9757 (tree_ssa_loop_ivopts): Likewise.
9758 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9759 * tree-vectorizer.c (vectorize_loops): Likewise.
9761 2013-04-29 Mike Frysinger <vapier@gentoo.org>
9763 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9764 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9765 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9766 with EABI_LINK_SPEC.
9768 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9771 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9774 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
9777 * lra-constraints.c (process_alt_operands): Discourage a bit more
9778 using memory for pseudos. Print cost dump for alternatives.
9779 Modify cost values for conflicts with early clobbers.
9780 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9782 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9785 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9787 2013-04-29 Ian Bolton <ian.bolton@arm.com>
9789 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9791 (movdi_aarch64): Support LDR/STR from/to D register.
9793 2013-04-29 Ian Bolton <ian.bolton@arm.com>
9795 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9796 or higher by default.
9798 2013-04-29 Richard Biener <rguenther@suse.de>
9801 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9802 even if not adding abnormal edges for calls that can make
9805 2013-04-29 Richard Biener <rguenther@suse.de>
9808 * tree-cfg.c (move_stmt_op): Fix condition under which to update
9810 (move_stmt_r): Remove redundant checking.
9812 2013-04-29 Teresa Johnson <tejohnson@google.com>
9815 * basic-block.h (apply_scale): New function.
9816 (apply_probability): Use apply_scale.
9817 * gimple-streamer-in.c (input_bb): Ditto.
9818 * lto-streamer-in.c (input_cfg): Ditto.
9819 * lto-cgraph.c (merge_profile_summaries): Ditto.
9820 * tree-optimize.c (execute_fixup_cfg): Ditto.
9821 * tree-inline.c (copy_bb): Update comment to use apply_scale.
9822 (copy_edges_for_bb): Ditto.
9823 (copy_cfg_body): Ditto.
9825 2013-04-29 Tom de Vries <tom@codesourcery.com>
9827 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9828 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9829 (tail_merge_optimize): Handle current_loops == NULL.
9831 2013-04-26 Jeff Law <law@redhat.com>
9833 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9834 (simplify_cond_using_ranges): Generalize code to simplify
9835 COND_EXPRs where one argument is a constant and the other
9836 is an SSA_NAME created by an integral type conversion.
9838 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9840 * config/arm/arm.md (store_minmaxsi): Use only when
9841 optimize_insn_for_size_p.
9843 2013-04-29 Christian Bruel <christian.bruel@st.com>
9846 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9848 2013-04-29 Richard Biener <rguenther@suse.de>
9851 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9852 loop tree make sure to schedule a fixup for the child as well.
9853 (expand_omp_for_generic): Properly add loops.
9854 (expand_omp_for_static_nochunk): Likewise.
9855 (expand_omp_for_static_chunk): Likewise.
9856 (expand_omp_for): For the degenerate case fixup loops.
9857 (expand_omp_sections): Fix default bb placement in loops.
9858 (expand_omp_atomic_pipeline): Properly add loops.
9860 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9862 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9864 2013-04-29 Tom de Vries <tom@codesourcery.com>
9866 * tree-ssa-tail-merge.c: Update header comment.
9868 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9870 * config/aarch64/arm_neon.h
9871 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9872 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9873 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9874 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9875 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9876 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9877 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9878 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9880 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9882 * config/aarch64/aarch64-simd.md
9883 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9884 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9885 fix_trunc, fixuns_trunc.
9886 (ftrunc<VDQF:mode>2): New.
9887 * config/aarch64/iterators.md (optab): Add fix, fixuns.
9888 (fix_trunc_optab): New.
9890 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9892 * config/aarch64/aarch64-builtins.c
9893 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9894 iceilf, lround, iroundf.
9896 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9899 * config/i386/i386.h (enum ix86_tune_indices)
9900 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9901 New, split from X86_TUNE_INTER_UNIT_MOVES.
9902 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9903 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9904 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9905 (TARGET_INTER_UNIT_MOVES): Remove.
9906 * config/i386/i386.c (initial_ix86_tune_features): Update.
9907 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9908 (ix86_expand_convert_uns_didf_sse): Use
9909 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9910 (ix86_expand_vector_init_one_nonzero): Ditto.
9911 (ix86_expand_vector_init_interleave): Ditto.
9912 (inline_secondary_memory_needed): Return true for moves from SSE class
9913 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9914 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9915 * config/i386/constraints.md (Yi, Ym): Depend on
9916 TARGET_INTER_UNIT_MOVES_TO_VEC.
9917 (Yj, Yn): New constraints.
9918 * config/i386/i386.md (*movdi_internal): Change constraints of
9919 operand 1 from Yi to Yj and from Ym to Yn.
9920 (*movsi_internal): Ditto.
9921 (*movdf_internal): Ditto.
9922 (*movsf_internal): Ditto.
9923 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9924 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9925 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9926 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9927 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9928 * config/i386/sse.md (movdi_to_sse): Ditto.
9929 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9930 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9931 TARGET_INTER_UNIT_MOVES.
9932 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9933 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9934 instead of TARGET_INTER_UNIT_MOVES.
9935 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9936 operand 1 from Yi to Yj and from Ym to Yn.
9938 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9940 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9941 (float_truncate_hi_): Likewise.
9942 (float_extend_lo_): Likewise.
9943 (float_truncate_lo_): Likewise.
9944 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9945 (aarch64_float_extend_lo_v2df): Likewise.
9946 (vec_unpacks_hi_v4sf): Likewise.
9947 (aarch64_float_truncate_lo_v2sf): Likewise.
9948 (aarch64_float_truncate_hi_v4sf): Likewise.
9949 (vec_pack_trunc_v2df): Likewise.
9950 (vec_pack_trunc_df): Likewise.
9952 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9954 * config/aarch64/aarch64-builtins.c
9955 (aarch64_fold_builtin): Fold float conversions.
9956 * config/aarch64/aarch64-simd-builtins.def
9957 (floatv2si, floatv4si, floatv2di): New.
9958 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9959 * config/aarch64/aarch64-simd.md
9960 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9961 * config/aarch64/iterators.md (FLOATUORS): New.
9962 (optab): Add float, floatuns.
9963 (su_optab): Likewise.
9965 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9967 * config/aarch64/aarch64-builtins.c
9968 (aarch64_builtin_vectorized_function): Use new names for
9970 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9971 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9972 (fcvtzu): Split as...
9973 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9974 (fcvtas): Split as...
9975 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9976 (fcvtau): Split as...
9977 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9978 (fcvtps): Split as...
9979 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9980 (fcvtpu): Split as...
9981 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9982 (fcvtms): Split as...
9983 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9984 (fcvtmu): Split as...
9985 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9986 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9987 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9988 (lfrintnusf, lfrintnudf): Likewise.
9989 * config/aarch64/aarch64-simd.md
9990 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9992 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9993 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9994 (fcvt_pattern): Likewise.
9996 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9998 * config/aarch64/aarch64-simd.md
9999 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10000 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10002 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10004 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10005 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10006 (vrnd<a,m,n,p>_f32): Implement using builtins.
10007 (vrnd<i,x><q>_f<32, 64>): New.
10009 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10011 * config/aarch64/aarch64-builtins.c
10012 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10013 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10014 (frintz): Rename to...
10016 (frintp): Rename to...
10018 (frintm): Rename to...
10020 (frinti): Rename to...
10021 (nearbyint): ...this.
10022 (frintx): Rename to...
10024 (frinta): Rename to...
10026 * config/aarch64/aarch64-simd.md
10027 (aarch64_frint<frint_suffix><mode>): Delete.
10028 (<frint_pattern><mode>2): Convert to insn.
10029 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10030 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10031 (frint_pattern): Likewise.
10032 (frint_suffix): Likewise.
10034 2013-04-29 Richard Biener <rguenther@suse.de>
10036 PR tree-optimization/57081
10037 * loop-init.c: Include tree-flow.h.
10038 (loop_optimizer_finalize): Free number of iteration estimates.
10039 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10041 2013-04-29 Jakub Jelinek <jakub@redhat.com>
10043 PR tree-optimization/57083
10044 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10045 non-singleton shift count range, zero extend low_bound for uns case.
10047 * config/i386/predicates.md (general_vector_operand): New predicate.
10048 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10049 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10050 if they aren't nonimmediate operands. If their original values
10051 satisfy const_vector_equal_evenodd_p, don't shift them.
10052 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10053 predicates. For the SSE4.1 case force operands[{1,2}] into registers
10054 if not nonimmediate_operand.
10055 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10056 instead of register_operand.
10057 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10059 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
10061 * stor-layout.c (finalize_size_functions): Allocate a structure and
10062 reset cfun before dumping the functions.
10064 2013-04-27 Jakub Jelinek <jakub@redhat.com>
10066 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10069 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10070 use xop_pmacsdqh if uns_p.
10071 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10072 the immediate rotate count.
10074 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
10076 * rtl.h (struct rtx_def): Add comment for field jump.
10077 (LRA_SUBREG_P): New macro.
10078 * recog.c (register_operand): Check LRA_SUBREG_P.
10079 * lra.c (lra): Add note at the end of RTL code. Align non-empty
10081 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10082 (lra_final_code_change): Skip subreg change for operators.
10083 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10084 if there are no operand changes.
10085 * lra-constraints.c (curr_insn_set): New.
10086 (match_reload): Set LRA_SUBREG_P.
10087 (emit_spill_move): Ditto.
10088 (check_and_process_move): Use curr_insn_set. Process only single
10089 set insns. Don't initialize sec_mem_p and change_p.
10090 (simplify_operand_subreg): Use LRA_SUBREG_P.
10091 (reg_in_class_p): New function.
10092 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
10093 of #ifdef. Add code to remove cycling.
10094 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10095 non-null disp. Reload inner instead of disp when base and index
10096 are null. Try to put lo_sum into register.
10097 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10098 (check_and_process_move): Move code for move cost check to
10099 simple_move_p. Remove equiv_substitution.
10100 (simple_move_p): New function.
10101 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
10102 curr_insn_set. Call check_and_process_move only for single set
10103 insns. Use the new function. Move call of check_and_process_move
10104 after operand equiv substitution and address process.
10106 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10109 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10110 with nonlocal goto receivers or returns twice calls, ignore
10111 unininitialized values from abnormal edges to nl goto receiver
10112 or returns twice call.
10114 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10116 PR tree-optimization/57051
10117 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10118 and VEC_RSHIFT_EXPR if shift count is a multiple of element
10121 2013-04-26 Richard Biener <rguenther@suse.de>
10123 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10124 (expand_omp_taskreg): Likewise. Mark loops for fixup.
10125 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10126 (fixup_loop_arrays_after_move): New function.
10127 (move_sese_region_to_fn): Properly outline the loop tree parts
10128 of the SESE region.
10130 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
10132 * config/i386/i386.md (type, unit): Fix long lines.
10134 2013-04-26 Richard Biener <rguenther@suse.de>
10136 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10137 (lto-streamer-out.o): Likewise.
10138 * cfgloop.c (init_loops_structure): Export, add struct function
10139 argument and adjust.
10140 (flow_loops_find): Adjust.
10141 * cfgloop.h (enum loop_estimation): Add EST_LAST.
10142 (init_loops_structure): Declare.
10143 * lto-streamer-in.c: Include cfgloop.h.
10144 (input_cfg): Input the loop tree.
10145 * lto-streamer-out.c: Include cfgloop.h.
10146 (output_cfg): Output the loop tree.
10147 (output_struct_function_base): Do not drop PROP_loops.
10149 2013-03-26 Richard Biener <rguenther@suse.de>
10151 * tree-cfg.c (execute_build_cfg): Build the loop tree.
10152 (pass_build_cfg): Provide PROP_loops.
10153 (move_sese_region_to_fn): Remove loops that are outlined into fn
10155 * tree-inline.c: Include cfgloop.h.
10156 (initialize_cfun): Do not drop PROP_loops.
10157 (copy_loops): New function.
10158 (copy_cfg_body): Copy loop structure.
10159 (tree_function_versioning): Initialize destination loop tree.
10160 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10161 (pass_parallelize_loops): Do IL verification.
10162 * loop-init.c (loop_optimizer_init): Fixup loops if required.
10163 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10164 the CFG make sure we fixup loops as well.
10165 * tree-ssa-tail-merge.c: Include cfgloop.h.
10166 (replace_block_by): When merging loop latches mark loops for fixup.
10167 * lto-streamer-out.c (output_struct_function_base): Drop
10168 PROP_loops for now.
10169 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10170 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10171 * ipa-split.c: Include cfgloop.h.
10172 (split_function): Add the new return block to the loop tree root.
10173 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10174 whether we have removed the forwarder block.
10175 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10176 * cfgloop.h (place_new_loop): Declare.
10177 * cfgloopmanip.c (place_new_loop): Export.
10178 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10179 (tree-switch-conversion.o): Likewise.
10180 (tree-complex.o): Likewise.
10181 (tree-inline.o): Likewise.
10182 (tree-ssa-tailmerge.o): Likewise.
10183 (ipa-split.o): Likewise.
10184 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10185 (tree-ssa-copy.o): Likewise.
10186 * tree-switch-conversion.c: Include cfgloop.h
10187 (process_switch): If we emit a bit-test cascade, schedule loops
10189 * tree-complex.c: Include cfgloop.h.
10190 (expand_complex_div_wide): Properly add new basic-blocks to loops.
10191 * asan.c: Include cfgloop.h.
10192 (create_cond_insert_point): Properly add new basic-blocks to
10193 loops, schedule loop fixup.
10194 * cfgloop.c (verify_loop_structure): Check that looks are not
10196 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10198 (expand_omp_for_generic): Likewise.
10199 (expand_omp_sections): Likewise.
10200 (expand_omp_atomic_pipeline): Schedule loops for fixup.
10201 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10202 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10203 is initialized, not when loops are present.
10204 * tree-parloops.c (parallelize_loops): Remove checking here.
10205 * passes.c (init_optimization_passes): Schedule a copy-propagation
10206 pass before complete unrolling of inner loops.
10208 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10210 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10211 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10212 (colorize_init): Add argument to _WIN32 version.
10213 * toplev.c: Include diagnostic-color.h.
10214 (process_options): Default to -fdiagnostics-color=auto if
10215 GCC_COLORS env var is in the environment.
10216 * common.opt (fdiagnostics-color=): Add Var and Init.
10217 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10218 env var is in the environment, the default is auto rather than never.
10220 * diagnostic.h (file_name_as_prefix): Add context argument.
10221 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
10222 the string as locus.
10223 * langhooks.c (lhd_print_error_function): Adjust caller.
10225 2013-04-25 Lawrence Crowl <crowl@google.com>
10227 * var-tracking.c (shared_hash_def::htab):
10228 Change type to hash_table. Update dependent calls and types.
10230 2013-04-25 Lawrence Crowl <crowl@google.com>
10232 * Makefile.in: Update as needed below.
10234 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10235 Move declaration to after the type's method definitons.
10237 * attribs.c (htab_t scoped_attributes::attribute_hash):
10238 Change type to hash_table. Update dependent calls and types.
10240 * bitmap.c (htab_t bitmap_desc_hash):
10241 Change type to hash_table. Update dependent calls and types.
10243 * cselib.c (htab_t cselib_hash_table):
10244 Change type to hash_table. Update dependent calls and types.
10246 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10247 (hash_string_slot_node): Move implementation into lto-streamer.h
10248 struct string_slot_hasher.
10249 (eq_string_slot_node): Likewise.
10251 * data-streamer-out.c: Update output_block::string_hash_table
10252 dependent calls and types.
10254 * dwarf2cfi.c (htab_t trace_index):
10255 Change type to hash_table. Update dependent calls and types.
10257 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10258 Change type to hash_table. Update dependent calls and types.
10259 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10260 (htab_t optimize_external_refs::map): Likewise.
10261 (htab_t output_comp_unit::extern_map): Likewise.
10262 (htab_t output_comdat_type_unit::extern_map): Likewise.
10263 (htab_t output_macinfo::macinfo_htab): Likewise.
10264 (htab_t optimize_location_lists::htab): Likewise.
10265 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10267 * except.c (htab_t ehspec_hash_type):
10268 Change type to hash_table. Update dependent calls and types.
10269 (assign_filter_values::ttypes): Likewise.
10270 (assign_filter_values::ehspec): Likewise.
10271 (sjlj_assign_call_site_values::ar_hash): Likewise.
10272 (convert_to_eh_region_ranges::ar_hash): Likewise.
10274 * gcse.c (htab_t pre_ldst_table):
10275 Change type to hash_table. Update dependent calls and types.
10277 * ggc-common.c (htab_t saving_htab):
10278 Change type to hash_table. Update dependent calls and types.
10279 (htab_t loc_hash): Likewise.
10280 (htab_t ptr_hash): Likewise.
10281 (call_count): Rename ggc_call_count.
10282 (call_alloc): Rename ggc_call_alloc.
10283 (loc_descriptor): Rename make_loc_descriptor.
10284 (add_statistics): Rename ggc_add_statistics.
10286 * ggc-common.c (saving_htab):
10287 Change type to hash_table. Update dependent calls and types.
10289 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10290 (push_gimplify_context): Likewise.
10291 (pop_gimplify_context): Likewise.
10292 (struct gimple_temp_hash_elt): Added.
10293 (struct gimplify_hasher): Likewise.
10294 (struct gimplify_ctx.temp_htab):
10295 Change type to hash_table. Update dependent calls and types.
10297 * gimple-fold.c: Include gimplify-ctx.h.
10299 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10300 Change type to hash_table. Update dependent calls and types.
10301 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10302 avoid potential global name collision.
10304 * gimplify.c: Include gimplify-ctx.h.
10305 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10306 (htab_t gimplify_ctx::temp_htab):
10307 Update dependent calls and types for new type hash_table.
10308 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10309 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10311 * gimplify-ctx.h: New.
10312 (struct gimple_temp_hash_elt): Move from gimplify.c.
10313 (class gimplify_hasher): New.
10314 (struct gimplify_ctx): Move from gimple.h.
10315 (htab_t gimplify_ctx::temp_htab):
10316 Change type to hash_table. Update dependent calls and types.
10318 * graphite-clast-to-gimple.c: Include graphite-htab.h.
10319 (htab_t ivs_params::newivs_index):
10320 Change type to hash_table. Update dependent calls and types.
10321 (htab_t ivs_params::params_index): Likewise.
10322 (htab_t print_generated_program::params_index): Likewise.
10323 (htab_t gloog::newivs_index): Likewise.
10324 (htab_t gloog::params_index): Likewise.
10326 * graphite.c: Include graphite-htab.h.
10327 4htab_t graphite_transform_loops::bb_pbb_mapping):
10328 Change type to hash_table. Update dependent calls and types.
10330 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10331 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10332 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10334 * graphite-dependences.c: Include graphite-htab.h.
10335 (loop_is_parallel_p): Change hash table type of parameter.
10337 * graphite-htab.h: New.
10338 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10339 (extern find_pbb_via_hash): Move from graphite-poly.h.
10340 (extern loop_is_parallel_p): Move from graphite-poly.h.
10341 (extern get_loop_body_pbbs): Move from graphite-poly.h.
10343 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10344 (extern loop_is_parallel_p): Move to graphite-htab.h.
10345 (extern get_loop_body_pbbs): Move to graphite-htab.h.
10347 * haifa-sched.c (htab_t delay_htab):
10348 Change type to hash_table. Update dependent calls and types.
10349 (htab_t delay_htab_i2): Likewise.
10351 * ira-color.c (htab_t allocno_hard_regs_htab):
10352 Change type to hash_table. Update dependent calls and types.
10354 * ira-costs.c (htab_t cost_classes_htab):
10355 Change type to hash_table. Update dependent calls and types.
10357 * loop-invariant.c (htab_t merge_identical_invariants::eq):
10358 Change type to hash_table. Update dependent calls and types.
10360 * loop-iv.c (htab_t bivs):
10361 Change type to hash_table. Update dependent calls and types.
10363 * loop-unroll.c (htab_t opt_info::insns_to_split):
10364 Change type to hash_table. Update dependent calls and types.
10365 (htab_t opt_info::insns_with_var_to_expand): Likewise.
10367 * lto-streamer.h (struct string_slot): Move from data-streamer.h
10368 (struct string_slot_hasher): New.
10369 (htab_t output_block::string_hash_table):
10370 Change type to hash_table. Update dependent calls and types.
10372 * lto-streamer-in.c (freeing_string_slot_hasher): New.
10373 (htab_t file_name_hash_table):
10374 Change type to hash_table. Update dependent calls and types.
10376 * lto-streamer-out.c: Update output_block::string_hash_table dependent
10379 * lto-streamer.c (htab_t tree_htab):
10380 Change type to hash_table. Update dependent calls and types.
10382 * omp-low.c: Include gimplify-ctx.h.
10384 * passes.c (htab_t name_to_pass_map):
10385 Change type to hash_table. Update dependent calls and types.
10386 (pass_traverse): Rename to passes_pass_traverse.
10388 * plugin.c (htab_t event_tab):
10389 Change type to hash_table. Update dependent calls and types.
10391 * postreload-gcse.c (htab_t expr_table):
10392 Change type to hash_table. Update dependent calls and types.
10393 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10395 * sese.c (debug_rename_map_1): Make extern.
10396 (htab_t copy_bb_and_scalar_dependences::rename_map):
10397 Change type to hash_table. Update dependent calls and types.
10399 * sese.h (extern debug_rename_map): Move to .c file.
10401 * store-motion.c (htab_t store_motion_mems_table):
10402 Change type to hash_table. Update dependent calls and types.
10404 * trans-mem.c (htab_t tm_new_mem_hash):
10405 Change type to hash_table. Update dependent calls and types.
10407 * tree-browser.c (htab_t TB_up_ht):
10408 Change type to hash_table. Update dependent calls and types.
10410 * tree-cfg.c (htab_t discriminator_per_locus):
10411 Change type to hash_table. Update dependent calls and types.
10413 * tree-complex.c: Include tree-hasher.h
10414 (htab_t complex_variable_components):
10415 Change type to hash_table. Update dependent calls and types.
10417 * tree-eh.c (htab_t finally_tree):
10418 Change type to hash_table. Update dependent calls and types.
10420 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10421 struct int_tree_hasher.
10422 (extern int_tree_map_eq): Likewise.
10423 (uid_decl_map_hash): Removed.
10424 (extern decl_tree_map_eq): Likewise.
10426 * tree-hasher.h: New.
10427 (struct int_tree_hasher): New.
10428 (typedef int_tree_htab_type): New.
10430 * tree-inline.c: Include gimplify-ctx.h.
10432 * tree-mudflap.c: Include gimplify-ctx.h.
10434 * tree-parloops.c: Include tree-hasher.h.
10435 (htab_t eliminate_local_variables_stmt::decl_address):
10436 Change type to hash_table. Update dependent calls and types.
10437 (htab_t separate_decls_in_region::decl_copies): Likewise.
10439 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10440 Change type to hash_table. Update dependent calls and types.
10442 * tree-sra.c (candidates):
10443 Change type to hash_table. Update dependent calls and types.
10445 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10447 (int_tree_map_hash): Likewise.
10449 * tree-ssa-dom.c (htab_t avail_exprs):
10450 Change type to hash_table. Update dependent calls and types.
10452 * tree-ssa-live.c (var_map_base_init::tree_to_index):
10453 Change type to hash_table. Update dependent calls and types.
10455 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10456 Change type to hash_table. Update dependent calls and types.
10458 * tree-ssa-phiopt.c (seen_ssa_names):
10459 Change type to hash_table. Update dependent calls and types.
10461 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10462 Change type to hash_table. Update dependent calls and types.
10464 * tree-ssa-uncprop.c (equiv):
10465 Change type to hash_table. Update dependent calls and types.
10467 2013-04-25 Jakub Jelinek <jakub@redhat.com>
10469 PR rtl-optimization/57003
10470 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10471 call note_stores with kill_clobbered_value callback again after
10472 killing regs_invalidated_by_call.
10474 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10476 * config/aarch64/aarch64-simd.md
10477 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10478 (aarch64_simd_bsl<mode>): Likewise.
10479 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10481 2013-04-25 Marek Polacek <polacek@redhat.com>
10483 PR tree-optimization/57066
10484 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10486 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
10488 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10490 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10492 * config/aarch64/aarch64-builtins.c
10493 (aarch64_fold_builtin): New.
10494 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10495 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10496 * config/aarch64/aarch64-simd-builtins.def (abs): New.
10497 * config/aarch64/arm_neon.h
10498 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10500 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10501 Tejas Belagod <tejas.belagod@arm.com>
10503 * config/aarch64/aarch64-builtins.c
10504 (aarch64_gimple_fold_builtin): New.
10505 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10506 * config/aarch64/aarch64-simd-builtins.def (addv): New.
10507 * config/aarch64/aarch64-simd.md (addpv4sf): New.
10508 (addvv4sf): Update.
10509 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10511 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10513 * config/aarch64/aarch64.md
10514 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10516 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10518 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10519 (*ngcsi_uxtw): New pattern.
10521 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10522 Julian Brown <julian@codesourcery.com>
10524 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10525 (TB_DREG): Add T_V4HF.
10526 (v4hf_UP): New macro.
10527 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10528 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10529 Handle initialisation of V4HF. Adjust initialisation of reinterpret
10531 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10532 (arm_vector_mode_supported_p): Handle V4HF.
10533 (arm_mangle_map): Handle V4HFmode.
10534 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10535 * config/arm/arm_neon_builtins.def: Add entries for
10536 vcvtv4hfv4sf, vcvtv4sfv4hf.
10537 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10538 (neon_vcvtv4hfv4sf): Likewise.
10539 * config/arm/neon-gen.ml: Handle half-precision floating point
10541 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10542 * config/arm/arm_neon.h: Regenerate.
10543 * config/arm/neon.ml (type elts): Add F16.
10544 (type vectype): Add T_float16x4, T_floatHF.
10545 (type vecmode): Add V4HF.
10546 (type features): Add Requires_FP_bit feature.
10547 (elt_width): Handle F16.
10548 (elt_class): Likewise.
10549 (elt_of_class_width): Likewise.
10550 (mode_of_elt): Refactor.
10551 (type_for_elt): Handle F16, fix error messages.
10552 (vectype_size): Handle T_float16x4.
10553 (vcvt_sh): New function.
10554 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10555 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10556 (string_of_mode): Handle V4HF.
10557 * doc/arm-neon-intrinsics.texi: Regenerate.
10559 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10561 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10562 format specifier in 'X' case.
10564 2013-04-25 Alan Modra <amodra@gmail.com>
10567 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10568 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10569 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10570 Repeat for many other rotate/shift and mask patterns using subregs.
10571 Name lshiftrt insns.
10572 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10573 on WORDS_BIG_ENDIAN.
10575 2013-04-25 Alan Modra <amodra@gmail.com>
10577 * config.gcc: Support little-endian powerpc-linux targets.
10578 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10579 (LINK_OS_LINUX_SPEC): Define.
10580 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10581 Preserve MASK_LITTLE_ENDIAN.
10582 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10583 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10584 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10585 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10586 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10587 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10588 Correct fp word order for little-endian. Don't shift toc entries
10589 smaller than a word for little-endian.
10590 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10591 (bswapdi2 splits): Correct low-part subreg for little-endian.
10592 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10593 low/high where such is correct only for be.
10594 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10595 little-endian for -mcall-aixdesc.
10597 2013-04-25 Alan Modra <amodra@gmail.com>
10599 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10600 replace_equiv_address_nv.
10602 2013-04-25 Alan Modra <amodra@gmail.com>
10604 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10606 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10609 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10610 * rtl.h (struct rtx_def): ...
10612 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10614 PR rtl-optimizations/57046
10615 * lra-constraints (split_reg): Set up lra_risky_transformations_p
10616 for multi-reg splits.
10618 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
10620 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10622 2013-04-24 Sterling Augustine <saugustine@google.com>
10624 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10625 (comp_dir_string, debug_str_dwo_section): New.
10626 (DEBUG_STR_DWO_SECTION): Rename to ...
10627 (DEBUG_DWO_STR_SECTION): ... this.
10628 (DEBUG_NORM_STR_SECTION): Delete.
10629 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10630 (DEBUG_STR_DWO_SECTION_FLAGS): New.
10631 (find_AT_string): Move most logic to ...
10632 (find_AT_string_in_table): ... here. New.
10633 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10634 add_skeleton_AT_string. Delete logic.
10635 (output_skeleton_debug_sections): Remove call to
10636 add_top_level_skeleton_die_attrs.
10637 (add_comp_dir_attribute): Move logic to comp_dir_string.
10638 (dwarf2out_init): Initialize debug_str_dwo_section.
10639 (output_indirect_string): Call find_string_form.
10640 (output_indirect_strings): Rewrite.
10641 (prune_unused_types): Empty skeleton_debug_str_hash.
10642 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10643 (dwarf2out_finish): Call output_indirect_strings.
10645 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
10647 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10649 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10651 * rtl.h (struct rtx_def): Add comment for field jump.
10652 (LRA_SUBREG_P): New macro.
10653 * recog.c (register_operand): Check LRA_SUBREG_P.
10654 * lra.c (lra): Add note at the end of RTL code. Align non-empty
10656 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10657 (lra_final_code_change): Skip subreg change for operators.
10658 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10659 if there are no operand changes.
10660 * lra-constraints.c (curr_insn_set): New.
10661 (match_reload): Set LRA_SUBREG_P.
10662 (emit_spill_move): Ditto.
10663 (check_and_process_move): Use curr_insn_set. Process only single
10664 set insns. Don't initialize sec_mem_p and change_p.
10665 (simplify_operand_subreg): Use LRA_SUBREG_P.
10666 (reg_in_class_p): New function.
10667 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
10668 of #ifdef. Add code to remove cycling.
10669 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10670 non-null disp. Reload inner instead of disp when base and index
10671 are null. Try to put lo_sum into register.
10672 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10673 (check_and_process_move): Move code for move cost check to
10674 simple_move_p. Remove equiv_substitution.
10675 (simple_move_p): New function.
10676 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
10677 curr_insn_set. Call check_and_process_move only for single set
10678 insns. Use the new function. Move call of check_and_process_move
10679 after operand equiv substitution and address process.
10681 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
10683 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10684 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10685 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10687 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
10689 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10691 2013-04-24 Marek Polacek <polacek@redhat.com>
10693 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10694 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10695 (select_loops_exit_conditions): Likewise.
10696 (number_of_iterations_for_all_loops): Likewise.
10697 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10698 (scev_analysis): Likewise.
10700 2013-04-02 Catherine Moore <clm@codesourcery.com>
10701 Chao-ying Fu <fu@mips.com>
10703 * config/mips/micromips.md (jraddiusp): New pattern.
10704 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10705 instruction if possible.
10707 2013-04-24 Alan Modra <amodra@gmail.com>
10709 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10711 2013-04-24 Julian Brown <julian@codesourcery.com>
10712 Chung-Lin Tang <cltang@codesourcery.com>
10714 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10715 dependency behavior in enumeration type DIE generation. Add TODO note
10716 to comments about future DW_FORM_sdata/udata re-work of related code.
10718 2013-04-23 Lawrence Crowl <crowl@google.com>
10720 * Makefile.in: Update as needed below.
10722 * hash-table.h (class hash_table):
10723 Correct many methods with parameter types compare_type to the correct
10724 value_type. (Correct code was unlikely to notice the change.)
10725 (hash_table::elements_with_deleted) New.
10726 (class hashtable::iterator): New.
10727 (hashtable::begin()): New.
10728 (hashtable::end()): New.
10729 (FOR_EACH_HASH_TABLE_ELEMENT): New.
10731 * statistics.c (statistics_hashes):
10732 Change type to hash_table. Update dependent calls and types.
10734 * tree-into-ssa.c (var_infos):
10735 Change type to hash_table. Update dependent calls and types.
10737 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10738 Change type to hash_table. Update dependent calls and types.
10740 * tree-ssa-loop-im.c (struct mem_ref.refs):
10741 Change type to hash_table. Update dependent calls and types.
10743 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10744 Change type to hash_table. Update dependent calls and types.
10746 * tree-ssa-sccvn.c (vn_tables_s::nary):
10747 Change type to hash_table. Update dependent calls and types.
10748 (vn_tables_s::phis): Likewise.
10749 (vn_tables_s::references): Likewise.
10751 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10752 (vn_reference_eq): Update parameter and return types.
10754 * tree-ssa-structalias.c (pointer_equiv_class_table):
10755 Change type to hash_table. Update dependent calls and types.
10756 (location_equiv_class_table): Likewise.
10758 * tree-vect-data-refs.c: Consequential changes for making
10759 peeling a hash_table.
10761 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10762 (destroy_loop_vec_info): Dependent hash_table update.
10764 * tree-vectorizer.h (peeling_htab):
10765 Change type to hash_table. Update dependent calls and types.
10767 2013-04-23 Shiva Chen <shiva0217@gmail.com>
10769 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10770 to check the register content is equal or not.
10771 * lra-constraints.c (match_reload): Use lra_assign_reg_val
10772 to assign register content record.
10773 * lra-eliminations.c (update_reg_eliminate): Use
10774 lra_update_reg_val_offset to update register content offset.
10775 * lra-int.h (struct lra_reg): Add offset member.
10776 (lra_reg_val_equal_p): New static inline function.
10777 (lra_update_reg_val_offset): New static inline function.
10778 (lra_assign_reg_val): New static inline function.
10779 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10780 to assign register content record.
10781 (initialize_lra_reg_info_element): Initial offset to zero.
10783 2013-04-23 Catherine Moore <clm@codesourcery.com>
10785 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10786 operands. Record compression.
10788 2013-04-23 Xinliang David Li <davidxl@google.com>
10790 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10792 2013-04-23 Richard Biener <rguenther@suse.de>
10794 PR middle-end/57036
10795 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10796 parameter, only add abnormal goto edges from the copied body
10797 if the call could perform abnormal gotos.
10798 (copy_cfg_body): Adjust.
10800 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
10802 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10804 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
10806 * coretypes.h (gimple_stmt_iterator): Add struct to make
10809 2013-04-23 Richard Biener <rguenther@suse.de>
10811 PR tree-optimization/57026
10812 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10813 from SSA names occuring in abnormal PHI nodes.
10815 2013-04-22 Andi Kleen <ak@linux.intel.com>
10817 * lto/lto.c (print_lto_report_1): Fix LTO report names.
10819 2013-04-22 Andi Kleen <ak@linux.intel.com>
10821 * lto/lto.c (print_lto_report_1): Declare early.
10822 (read_cgraph_and_symbols): Call print_lto_report_1 early.
10824 2013-04-22 Andi Kleen <ak@linux.intel.com>
10826 * common.opt (-flto-report-wpa): Add.
10827 * doc/invoke.texi (-flto-report-wpa): Add.
10828 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10831 2013-04-22 Xinliang David Li <davidxl@google.com>
10833 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10834 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10835 * Makefile.in: New dependency
10837 David Daney <ddaney.cavm@gmail.com>
10839 * configure.ac (gcc_cv_as_micromips_support): Use the
10840 --fatal-warnings option.
10841 * configure: Regenerate.
10843 2013-04-22 Marek Polacek <polacek@redhat.com>
10846 * tsan.c (instrument_expr): Don't instrument expression
10847 in case its size is zero.
10849 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
10853 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
10855 * config/alpha/alpha.c (TARGET_LRA_P): New define.
10857 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10859 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10860 (gimple_stmt_iterator): New typedef.
10861 * gimple.h (gimple_stmt_iterator): Rename to...
10862 (gimple_stmt_iterator_d): ... This.
10863 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10864 trees be valid for GIMPLE and GENERIC.
10865 (TARGET_GIMPLE_FOLD_BUILTIN): New.
10866 * gimple-fold.c (gimple_fold_call): Call target hook
10867 gimple_fold_builtin.
10868 * hooks.c (hook_bool_gsiptr_false): New.
10869 * hooks.h (hook_bool_gsiptr_false): New.
10870 * target.def (fold_stmt): New.
10871 * doc/tm.texi: Regenerate.
10873 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
10876 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10877 a set sp if no stack realignment.
10879 2013-04-22 Nick Clifton <nickc@redhat.com>
10881 * config.gcc (tilegx-linux): Extend extra_objs rather than
10883 (tilepro-linux): Likewise.
10885 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10887 * config/aarch64/aarch64-builtins.c
10889 (CF0, CF1, CF2, CF3, CF4, CF10): New.
10890 (VAR<1-12>): Add MAP parameter.
10891 (BUILTIN_*): Likewise.
10892 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10893 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10894 (aarch64_ushl_n<mode>): Likewise.
10895 (aarch64_sshr_n<mode>): Likewise.
10896 (aarch64_ushr_n<mode>): Likewise.
10897 (aarch64_<maxmin><mode>): Likewise.
10898 (aarch64_sqrt<mode>): Likewise.
10899 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10900 (vshr<q>_n_*): Likewise.
10902 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10904 * config/aarch64/aarch64-builtins.c
10905 (aarch64_simd_builtin_type_mode): Handle SF types.
10907 (BUILTIN_GPF): Define.
10908 (aarch64_init_simd_builtins): Handle SF types.
10909 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10910 (frecps): Likewise.
10911 (frecpx): Likewise.
10912 * config/aarch64/aarch64-simd.md
10913 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10914 (aarch64_frecpe<mode>): New.
10915 (aarch64_frecps<mode>): Likewise.
10916 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10917 (v8type): Add frecp<esx>.
10918 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10919 (aarch64_frecps<mode>): Likewise.
10920 * config/aarch64/iterators.md (FRECP): New.
10921 (frecp_suffix): Likewise.
10922 * config/aarch64/arm_neon.h
10923 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10925 2013-04-22 Christian Bruel <christian.bruel@st.com>
10928 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10929 (REG_CLASS_NAMES): Idem.
10930 (REG_CLASS_CONTENTS): Idem.
10931 (REGCLASS_HAS_FP_REG): Idem.
10932 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10933 (sh_conditional_register_usage): Idem.
10935 2013-04-21 Jeff Law <law@redhat.com>
10937 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10938 (ssa_forward_propagate_and_combine): Use it.
10940 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10942 * lra.c: Update the flow chart diagram.
10944 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10946 PR rtl-optimization/56847
10947 * lra-constraints.c (process_alt_operands): Discourage alternative
10948 with non-matche doffsettable memory constraint fro memory with
10951 2013-04-19 Richard Biener <rguenther@suse.de>
10953 PR tree-optimization/56982
10954 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10956 * gimplify.c (gimplify_call_expr): Notice special calls.
10957 (gimplify_modify_expr): Likewise.
10958 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10959 abnormal control flow receivers.
10960 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10961 in the same way as cfun->has_nonlocal_labels.
10962 (gimple_purge_dead_abnormal_call_edges): Likewise.
10963 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10964 receivers start a basic-block.
10966 2013-04-19 Richard Biener <rguenther@suse.de>
10968 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10970 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
10971 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10972 (SLP_TREE_LOAD_PERMUTATION): Add.
10973 (vect_transform_slp_perm_load): Adjust prototype.
10974 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10975 (vect_free_slp_instance): Likewise.
10976 (vect_create_new_slp_node): Likewise.
10977 (vect_supported_slp_permutation_p): Remove.
10978 (vect_slp_rearrange_stmts): Adjust.
10979 (vect_supported_load_permutation_p): Likewise. Inline
10980 vect_supported_slp_permutation_p here.
10981 (vect_analyze_slp_instance): Compute load permutations per
10982 slp node instead of per instance.
10983 (vect_get_slp_defs): Adjust.
10984 (vect_transform_slp_perm_load): Likewise.
10985 (vect_schedule_slp_instance): Remove redundant code.
10986 (vect_schedule_slp): Remove hack for PR56270, add it ...
10987 * tree-vect-stmts.c (vectorizable_load): ... here, do not
10988 CSE loads for SLP. Adjust.
10990 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10992 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10993 spelling in two comments.
10995 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10998 * config/arm/arm.c (load_multiple_sequence): Require SP
10999 as base register for loads if SP is in the register list.
11001 2013-04-19 Martin Jambor <mjambor@suse.cz>
11003 PR tree-optimization/56718
11004 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11005 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11006 and made public. Adjusted all callers.
11007 (ipa_intraprocedural_devirtualization): New function.
11008 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11009 (ipa_intraprocedural_devirtualization): Likewise.
11010 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11012 2013-04-19 Richard Biener <rguenther@suse.de>
11014 PR tree-optimization/57000
11015 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11017 2013-04-19 Terry Guo <terry.guo@arm.com>
11019 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11021 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
11022 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11023 (cortex_m4_fmacs): Use new reservations.
11024 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11026 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
11028 PR rtl-optimization/56999
11029 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11031 (lra_coalesce): Remove split_origin_bitmap and related code.
11032 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
11033 ranges if necessary.
11035 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
11037 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11039 (ix86_expand_call): Remove clobbered_registers array and use
11040 x86_64_ms_sysv_extra_clobbered_registers instead.
11041 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11043 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11045 * config/i386/i386.md (*call_rex64_ms_sysv): Use
11046 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
11047 (*call_value_rex64_ms_sysv): Ditto.
11049 2013-04-18 Cary Coutant <ccoutant@google.com>
11051 * dwarf2out.c (output_pubnames): Check die_perennial_p of
11052 parent instead of die_mark.
11054 2013-04-18 Diego Novillo <dnovillo@google.com>
11056 * gimple.c (create_gimple_tmp): New.
11057 (get_expr_type): New.
11058 (build_assign): New.
11059 (build_type_cast): New.
11060 * gimple.h (enum ssa_mode): Define.
11061 (gimple_seq_set_location): New.
11062 * asan.c (build_check_stmt): Change some gimple_build_* calls
11063 to use build_assign and build_type_cast.
11065 2013-04-18 Richard Biener <rguenther@suse.de>
11067 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11068 handle negative step. Remove redundant checks.
11069 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11070 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11071 for negative step and grouped loads fail to vectorize.
11073 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11075 * emit-rtl.c (reset_insn_used_flags): New function.
11076 (reset_all_used_flags): Use it.
11077 (verify_insn_sharing): New function.
11078 (verify_rtl_sharing): Fix verification for SEQUENCEs.
11080 2013-04-18 Jakub Jelinek <jakub@redhat.com>
11082 PR tree-optimization/56984
11083 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11084 and (x >> M) >= N don't register any assertion if N << M is the
11087 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11089 * lower-subreg.c (resolve_simple_move): If called self-recursive,
11090 do not delete_insn insns that have not yet been emitted, only
11091 unlink them with remove_insn.
11092 * df-scan.c (df_insn_delete): Revert r197492.
11094 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
11096 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11097 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11099 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11101 * config/arm/arm.md (movsicc_insn): Convert define_insn into
11102 define_insn_and_split.
11103 (and_scc,ior_scc,negscc): Likewise.
11104 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11106 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11108 * config/arm/arm.c (use_return_insn): Return 0 for targets that
11109 can benefit from using a sequence of LDRD instructions in epilogue
11110 instead of a single LDM instruction.
11112 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
11115 * doc/extend.texi: Fix typo.
11117 2013-04-17 Richard Biener <rguenther@suse.de>
11119 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11120 (vect_build_slp_tree): ... here.
11121 (vect_build_slp_tree_1): Compute which stmts of the SLP group
11122 match. Remove special-casing of mismatched complex loads.
11123 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11124 re-try the match with swapped commutative operands.
11125 (vect_supported_load_permutation_p): Remove special-casing of
11126 mismatched complex loads.
11127 (vect_analyze_slp_instance): Adjust.
11129 2013-04-17 Richard Biener <rguenther@suse.de>
11131 PR rtl-optimization/56921
11132 * cfgloop.h (struct loop): Add simple_loop_desc member.
11133 (struct niter_desc): Mark with GTY(()).
11134 (simple_loop_desc): Do not use aux field but simple_loop_desc.
11135 * loop-iv.c (get_simple_loop_desc): Likewise.
11136 (free_simple_loop_desc): Likewise.
11139 2013-04-16 Richard Biener <rguenther@suse.de>
11141 PR rtl-optimization/56921
11142 * loop-init.c (pass_rtl_move_loop_invariants): Add
11143 TODO_do_not_ggc_collect to todo_flags_finish.
11144 (pass_rtl_unswitch): Same.
11145 (pass_rtl_unroll_and_peel_loops): Same.
11146 (pass_rtl_doloop): Same.
11148 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
11150 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11151 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11152 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11154 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11155 * tree-streamer.c (record_common_node): Adjust reference in comment.
11157 2013-04-17 Terry Guo <terry.guo@arm.com>
11159 * config/arm/cortex-m4.md: Add a new bypass.
11161 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11163 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11165 (*subs_<optab><mode>_multp2): New pattern.
11166 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11167 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11169 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11171 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11172 (*subs_mul_imm_<mode>): New pattern.
11174 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
11177 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11178 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
11179 (vsx_movti_32bit): Change j->wa to O->wa.
11181 2013-04-16 Richard Biener <rguenther@suse.de>
11183 PR rtl-optimization/56921
11184 * loop-init.c (pass_rtl_move_loop_invariants): Add
11185 TODO_do_not_ggc_collect to todo_flags_finish.
11186 (pass_rtl_unswitch): Same.
11187 (pass_rtl_unroll_and_peel_loops): Same.
11188 (pass_rtl_doloop): Same.
11190 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
11192 * config/arm/arm.c (emit_multi_reg_push): New declaration
11193 for an existing function.
11194 (arm_emit_strd_push): New function.
11195 (arm_expand_prologue): Used here.
11196 (arm_emit_ldrd_pop): New function.
11197 (arm_expand_epilogue): Used here.
11198 (arm_get_frame_offsets): Update condition.
11199 (arm_emit_multi_reg_pop): Add a special case for load of a single
11200 register with writeback.
11202 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
11204 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11207 2013-04-16 Richard Biener <rguenther@suse.de>
11209 PR tree-optimization/56756
11210 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11211 (first_mem_ref_loc): New.
11212 (execute_sm): Place the load temporarily before a previous
11213 access instead of in the latch edge to ensure its SSA dependencies
11214 are defined at points dominating the load.
11216 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
11218 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11219 correct fix by moving header and footer insn to the footer of
11220 the merged basic block. Clear BB_END of the merged-away block.
11222 PR middle-end/43631
11223 * emit-rtl.c (make_note_raw): New function.
11224 (link_insn_into_chain): New static inline function.
11225 (add_insn): Use it.
11226 (add_insn_before, add_insn_after): Factor insn chain linking code...
11227 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11228 using link_insn_into_chain.
11229 (note_outside_basic_block_p): New helper function for emit_note_after
11230 and emit_note_before.
11231 (emit_note_after): Use nobb variant of add_insn_after if the note
11232 should not be contained in a basic block.
11233 (emit_note_before): Use nobb variant of add_insn_before if the note
11234 should not be contained in a basic block.
11235 (emit_note_copy): Use make_note_raw.
11236 (emit_note): Likewise.
11237 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11238 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11239 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11240 the moved barrier the tail of the basic block it follows.
11241 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11243 2013-04-15 Jakub Jelinek <jakub@redhat.com>
11245 PR tree-optimization/56962
11246 * gimple-ssa-strength-reduction.c (record_increment): Only set
11247 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11248 either rhs1 or rhs2 is equal to c->base_expr.
11250 2013-04-15 Richard Biener <rguenther@suse.de>
11252 PR tree-optimization/56933
11253 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11255 (GROUP_READ_WRITE_DEPENDENCE): Remove.
11256 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11257 * tree-vect-data-refs.c (vect_analyze_group_access): Move
11258 dependence check ...
11259 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11261 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11262 GROUP_READ_WRITE_DEPENDENCE.
11264 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11266 * emit-rtl.c (reset_all_used_flags): New function.
11267 (verify_rtl_sharing): Call reset_all_used_flags before and after
11268 performing the checks.
11270 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11272 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11273 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11274 * config/arm/constraints.md (De): New constraint.
11275 * config/arm/neon.md (anddi3_neon): Delete.
11276 (neon_vand<mode>): Expand to standard anddi3 pattern.
11277 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11278 Move earlier in the file.
11279 (neon_inv_logic_op2): Likewise.
11280 (arm_anddi_operand_neon): New predicate.
11282 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11284 * configure.ac (gcc_cv_ld_as_needed): Set
11285 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11286 Use -z ignore, -z record on *-*-solaris2*.
11287 (HAVE_LD_AS_NEEDED): Update comment.
11288 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11289 * configure: Regenerate.
11290 * config.in: Regenerate.
11291 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11292 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11293 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11294 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11295 equivalents. Fix markup.
11296 * doc/tm.texi: Regenerate.
11298 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
11300 * config/i386/i386.opt: New option mstack-protector-guard=.
11301 * config/i386/i386-opts.h: Add enum stack_protector_guard.
11302 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11303 TARGET_SSP_TLS_GUARD.
11304 * config/i386/i386.c (ix86_option_override_internal): Set
11305 ix86_stack_protector_guard.
11306 * config/i386/i386.md (stack_protect_set): Enable for
11307 TARGET_SSP_TLS_GUARD only.
11308 (stack_protect_set_<mode>): Ditto.
11309 (stack_protect_test): Ditto.
11310 (stack_protect_test_<mode>): Ditto.
11311 * doc/invoke.texi (i386 Option): Document.
11313 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
11316 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11317 (S_MODES): Set H_MODE bit.
11318 (SF_MODES): Set only S_MODE and SF_MODE bits.
11319 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11320 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11321 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11322 <MODE_FLOAT>: Likewise.
11324 2013-04-15 Joey Ye <joey.ye@arm.com>
11326 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11328 2013-04-15 Joey Ye <joey.ye@arm.com>
11330 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11332 (thumb_far_jump_used_p): Count instruction size and set
11335 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
11337 * reorg.c (fill_simple_delay_slots): Reindent block of code.
11338 * resource.c (mark_target_live_regs): Reformat conditional block.
11340 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11342 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11343 notes, they are emitted only just before final.
11344 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11346 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11348 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11349 * cfgrtl.c (delete_insn): Call it here instead.
11350 * lra-spills.c (lra_final_code_change): Use delete_insn.
11351 * haifa-sched.c (sched_remove_insn): Likewise.
11352 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11353 returning to the nop pool.
11354 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11355 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
11357 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11359 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11360 * doc/tm.texi: Regenerated.
11362 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
11364 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11367 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11369 * df-core.c (df_find_def): Compare register numbers.
11370 (df_find_use): Likewise.
11372 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
11375 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11376 lra_in_progress for return.
11378 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11380 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11381 define_insn into define_insn_and_split and emit movsicc patterns.
11383 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11385 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11387 2013-04-12 Richard Biener <rguenther@suse.de>
11389 * tree-pass.h (TODO_do_not_ggc_collect): New.
11390 * passes.c (execute_one_ipa_transform_pass): Honor
11391 TODO_do_not_ggc_collect.
11392 (execute_one_pass): Likewise.
11395 2013-04-10 Richard Biener <rguenther@suse.de>
11397 * passes.c (init_optimization_passes): Remove reload pass.
11398 * ira.c (do_reload): Merge into ...
11400 (rest_of_handle_reload): Remove.
11401 (pass_reload): Likewise.
11402 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11403 of reload for vzeroupper pass placement.
11405 2013-04-12 Jakub Jelinek <jakub@redhat.com>
11407 PR tree-optimization/56918
11408 PR tree-optimization/56920
11409 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11410 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
11411 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
11412 use wide_mul_with_sign method.
11414 2013-04-12 Richard Biener <rguenther@suse.de>
11416 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11417 not be considered a gimple constant.
11419 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
11421 * fold-const.c (const_binop): Handle vector shifts by a scalar.
11422 (fold_binary_loc): Call const_binop also for mixed vector-scalar
11425 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
11426 Jakub Jelinek <jakub@redhat.com>
11428 * opts.c: Include diagnostic-color.h.
11429 (common_handle_option): Handle OPT_fdiagnostics_color_.
11430 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11431 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11432 (diagnostic-color.o): New.
11433 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11434 (diagnostic_color_rule): New enum.
11435 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11436 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11437 the location string.
11438 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11439 either NULL, or color kind.
11440 * diagnostic-color.c: New file.
11441 * diagnostic-color.h: New file.
11442 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11444 * doc/invoke.texi (-fdiagnostics-color): Document.
11445 * pretty-print.h (pp_show_color): Define.
11446 (struct pretty_print_info): Add show_color field.
11447 * diagnostic.c: Include diagnostic-color.h.
11448 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11449 macros. Colorize error:, warning: etc. strings and also the location
11451 (diagnostic_show_locus): Colorize the caret line.
11452 * pretty-print.c: Include diagnostic-color.h.
11453 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
11454 inside of %< %> quotes or quoted through q format modifier.
11456 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11458 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11460 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11462 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11463 code in CC_NZ mode.
11464 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11467 2013-04-11 Marek Polacek <polacek@redhat.com>
11469 PR tree-optimization/48184
11470 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11472 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
11474 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11475 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11476 (skip_simple_arithmetic): Tidy up.
11477 * tree.h (skip_simple_constant_arithmetic): Declare.
11479 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11481 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11483 2013-04-11 Richard Biener <rguenther@suse.de>
11485 * tree-vect-loop.c (get_initial_def_for_induction): Properly
11486 generate vector constants.
11488 2013-04-11 Richard Biener <rguenther@suse.de>
11490 PR tree-optimization/56878
11491 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11492 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11494 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11495 Prefer to align the DR with the most invariant base address.
11497 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11499 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11502 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
11504 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11505 floating-point vector comparisons against 0.
11507 2013-04-11 Jakub Jelinek <jakub@redhat.com>
11509 PR tree-optimization/56899
11510 * fold-const.c (extract_muldiv_1): Apply distributive law
11511 only if TYPE_OVERFLOW_WRAPS (ctype).
11513 2013-04-11 Bin Cheng <bin.cheng@arm.com>
11516 * ira-costs.c (scan_one_insn): Check whether the source rtx of
11517 loading has side effect.
11519 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
11521 * config/sparc/sparc.c: Include tree-pass.h.
11522 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11523 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
11524 head of file. Change return type. Split off gate function.
11525 (sparc_gate_work_around_errata): New function.
11526 (pass_work_around_errata): New pass definition.
11527 (insert_pass_work_around_errata) New pass insert definition to
11528 insert pass_work_around_errata just after delayed-branch scheduling.
11529 (sparc_option_override): Insert the pass.
11530 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11532 2013-04-10 David S. Miller <davem@davemloft.net>
11534 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11535 or -mcpu=hypersparc.
11537 * target.def (cstore_mode): New hook.
11538 * target.h: Include insn-codes.h
11539 * targhooks.c: Likewise.
11540 (default_cstore_mode): New function.
11541 * targhooks.h: Declare it.
11542 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11543 * doc/tm.texi: Rebuild.
11544 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11545 target hook, rather than inspecting the insn_data.
11546 * config/sparc/sparc.c (sparc_cstore_mode): New function.
11547 (TARGET_CSTORE_MODE): Redefine.
11548 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11550 * config/sparc/predicates.md (cstore_result_operand): New special
11552 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11553 Use it for operand 0.
11554 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11555 (*snesi_special): Likewise.
11556 (*snesi_zero): Likewise.
11557 (*seqsi_zero): Likewise.
11558 (*sltu_insn): Likewise.
11559 (*sgeu_insn): Likewise.
11560 (*seqdi_special): Make operand 0 and comparison operation be of
11562 (*snedi_special): Likewise.
11563 (*snedi_special_vis3): Likewise.
11564 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11565 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11566 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11567 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11568 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11569 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11570 (*sltu_extend_sp64): Likewise.
11571 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11572 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11573 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11574 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11575 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11577 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
11579 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11580 (aarch64_start_file): Use the new function.
11582 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11583 Jason Merrill <jason@redhat.com>
11585 * common.opt: Add -gdwarf.
11586 * opts.c (common_handle_option): Handle it.
11587 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11589 2013-04-10 Richard Biener <rguenther@suse.de>
11591 * passes.c (execute_todo): Do not call ggc_collect conditional here.
11592 (execute_one_ipa_transform_pass): But unconditionally here.
11593 (execute_one_pass): And here.
11594 (init_optimization_passes): Remove reload pass.
11595 * tree-pass.h (TODO_ggc_collect): Remove.
11596 (pass_reload): Likewise.
11597 * ira.c (do_reload): Merge into ...
11599 (rest_of_handle_reload): Remove.
11600 (pass_reload): Likewise.
11601 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11602 of reload for vzeroupper pass placement.
11603 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11604 and todo_flags_finish of all passes.
11606 2013-04-10 Richard Biener <rguenther@suse.de>
11608 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11609 first_const_oprnd field, rename first_def_type to first_op_type.
11610 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11611 (vect_get_and_check_slp_defs): Always use the type of the
11612 operand. Allow mixed vect_external_def, vect_constant_def types.
11613 (vect_get_constant_vectors): Handle mixed vect_external_def,
11614 vect_constant_def types.
11616 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
11618 PR tree-optimization/55524
11619 * tree-ssa-math-opts.c
11620 (convert_mult_to_fma): Don't use an fms construct
11621 when we don't have an fms operation, but fnma, and it looks
11622 likely that we'll be able to use the latter.
11624 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11626 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11628 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11629 inline fail caused by overwritable functions.
11631 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
11633 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11634 unnecessary bits in the constant power of two case.
11636 2013-04-10 Richard Biener <rguenther@suse.de>
11638 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11639 broken code swapping operands.
11640 (vect_build_slp_tree): Do not compute load permutations here.
11641 (vect_analyze_slp_instance): Compute load permutations here,
11642 after building the SLP tree.
11644 2013-04-09 Christian Bruel <christian.bruel@st.com>
11646 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11647 of next/prev_real_insn.
11649 2013-04-09 Jan Hubicka <jh@suse.cz>
11651 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11652 Drop aliased parameter.
11653 (function_and_variable_visibility): Do not handle alias pairs.
11654 * cgraph.c (varpool_externally_visible_p): Update prototype.
11655 * varpool.c (varpool_add_new_variable): Update.
11657 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11659 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11661 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
11663 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11665 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11667 2013-04-09 Marek Polacek <polacek@redhat.com>
11669 PR tree-optimization/48762
11670 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11672 2013-04-09 Richard Biener <rguenther@suse.de>
11674 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11676 (vect_build_slp_tree): Likewise.
11677 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11678 calculating the cost of a SLP instance.
11679 (vect_analyze_slp_instance): Use it from here, after building
11682 2013-04-09 Jakub Jelinek <jakub@redhat.com>
11684 PR middle-end/56883
11685 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11686 expand_omp_for_static_chunk): Use simple_p = true in
11687 force_gimple_operand_gsi calls when assigning to addressable decls.
11689 2013-04-09 Jeff Law <law@redhat.com>
11691 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11692 when the boolean was created by converting a wider object which
11693 had a boolean range.
11695 2013-04-09 Richard Biener <rguenther@suse.de>
11697 * tree-vectorizer.h (slp_void_p): Remove.
11698 (slp_tree): Typedef before _slp_tree declaration.
11699 (struct _slp_tree): Use a vector of slp_tree as children.
11700 (vect_get_place_in_interleaving_chain): Remove.
11701 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11703 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11705 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11706 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11707 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11708 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11709 Use slp_node instead of slp_void_p and adjust.
11711 2013-04-09 Richard Biener <rguenther@suse.de>
11713 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11714 work that is not necessary.
11716 2013-04-09 Jakub Jelinek <jakub@redhat.com>
11718 PR tree-optimization/56854
11719 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11720 forward into clobber stmts if it would change MEM_REF lhs into
11723 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
11725 * tree.c (type_hash_lookup, type_hash_add): Make static.
11726 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11728 2013-04-09 Richard Biener <rguenther@suse.de>
11730 * tree.h (unsave_expr_now): Remove.
11731 * tree-inline.c (mark_local_for_remap_r): Remove.
11732 (unsave_expr_1): Likewise.
11733 (unsave_r): Likewise.
11734 (unsave_expr_now): Likewise.
11735 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11736 (propagate_tree_value): Likewise.
11738 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
11740 * doc/rtl.texi (sequence): Rewrite documentation to match the
11741 current use of SEQUENCE rtl objects.
11742 * rtl.def (SEQUENCE): Likewise.
11744 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11745 Update documentation.
11746 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11747 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11749 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11751 2013-04-08 Teresa Johnson <tejohnson@google.com>
11753 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11754 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11756 (estimate_edge_size_and_time): Add comment to suggest using rounding
11758 (estimate_node_size_and_time): Ditto.
11759 (remap_edge_change_prob): Use helper rounding divide methods.
11760 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11761 (gimple_mod_pow2_value_transform): Ditto.
11762 (gimple_mod_subtract_transform): Ditto.
11763 (gimple_ic_transform): Ditto.
11764 (gimple_stringops_transform): Ditto.
11765 * stmt.c (conditional_probability): Ditto.
11766 (emit_case_dispatch_table): Ditto.
11767 * lto-cgraph.c (merge_profile_summaries): Ditto.
11768 * tree-optimize.c (execute_fixup_cfg): Ditto.
11769 * cfgcleanup.c (try_forward_edges): Ditto.
11770 * cfgloopmanip.c (scale_loop_profile): Ditto.
11772 (duplicate_loop_to_header_edge): Ditto.
11773 (lv_adjust_loop_entry_edge): Ditto.
11774 * tree-vect-loop.c (vect_transform_loop): Ditto.
11775 * profile.c (compute_branch_probabilities): Ditto.
11776 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11777 * lto-streamer-in.c (input_cfg): Ditto.
11778 * gimple-streamer-in.c (input_bb): Ditto.
11779 * ipa-cp.c (update_profiling_info): Ditto.
11780 (update_specialized_profile): Ditto.
11781 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11782 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11784 * sched-rgn.c (compute_dom_prob_ps): Ditto.
11785 (compute_trg_info): Ditto.
11786 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11787 (purge_dead_edges): Ditto.
11788 * loop-unswitch.c (unswitch_loop): Ditto.
11789 * cgraphclones.c (cgraph_clone_edge): Ditto.
11790 (cgraph_clone_node): Ditto.
11791 * tree-inline.c (copy_bb): Ditto.
11792 (copy_edges_for_bb): Ditto.
11793 (initialize_cfun): Ditto.
11794 (copy_cfg_body): Ditto.
11795 (expand_call_inline): Ditto.
11797 2013-04-08 Kai Tietz <ktietz@redhat.com>
11799 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11800 TARGET_CYGWIN64 by TARGET_64BIT.
11802 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11804 * config/epiphany/epiphany.md (GPR_1): New constant.
11805 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11806 * config/epiphany/epiphany.c (gen_compare_reg):
11807 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11808 is already in place.
11809 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11810 Don't require being called during rtl expansion; If y operlaps r0,
11812 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11813 (epiphany_expand_epilogue): Likewise.
11815 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11816 Don't use CC_FPmode for ORDERED / UNORDERED.
11817 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11819 * config/epiphany/constraints.md (CnL): New constraint.
11820 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11821 * config/epiphany/predicates.md (add_operand): Allow 1024.
11823 * config/epiphany/epiphany.md (logical_op): New code iterator.
11824 (op_mnc): New code attribute.
11825 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11826 (mov_f+1, mov_f+2): New peephole2 patterns.
11828 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11829 (cstoresi4): Also allow re-use of zero result when doing a NE
11830 comparison to a non-zero operand.
11831 Use (clobber (scratch)) for first insn if the gpr output is not needed.
11833 * config/epiphany/epiphany.md (<insn_opname>v2si3):
11834 Use gen_addsi3_i / gen_subsi3_i.
11836 2013-04-08 Jakub Jelinek <jakub@redhat.com>
11840 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11841 contain anything but clobbers, at most one __builtin_stack_restore,
11842 optionally debug stmts and final resx, and if it has at least one
11843 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
11844 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11845 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11846 which isn't defaut definition, remove them.
11847 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11848 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11849 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11850 with MEM_REF LHS with SSA_NAME address.
11852 2013-04-08 Jeff Law <law@redhat.com>
11854 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11856 2013-04-08 Richard Biener <rguenther@suse.de>
11858 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11860 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11861 determined vector type.
11862 (vect_analyze_data_refs): Likewise.
11863 (vect_get_new_vect_var): Adjust.
11864 (vect_create_destination_var): Preserve SSA name versions.
11865 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11866 not dump anything here.
11868 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11870 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11871 Add member lr_slot_known.
11872 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11874 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11875 Remove code that sets lr_slot_offset according to what a previous
11876 version of epiphany_emit_save_restore used to do.
11877 (epiphany_emit_save_restore): When doing an lr save or restore,
11878 set/verify lr_slot_known and lr_slot_offset.
11880 2013-04-08 Xinyu Qi <xyqi@marvell.com>
11883 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11886 2013-04-08 Richard Biener <rguenther@suse.de>
11888 * alias.c (find_base_term): Fix thinko in previous change.
11890 2013-04-08 Jakub Jelinek <jakub@redhat.com>
11892 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11893 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11894 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
11895 if possible to compute val.
11896 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11897 For QImode integers don't require anything about precision. Use
11898 const_with_all_bytes_same to find out if the constant doesn't have
11899 repeated bytes in it.
11901 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11903 * config/s390/s390.c (s390_expand_insv): Only accept insertions
11906 2013-04-08 Marek Polacek <polacek@redhat.com>
11908 PR rtl-optimization/48182
11909 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11912 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
11915 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11916 nuses, make sure we have a label.
11918 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11921 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11922 (rs6000_emit_swdiv_low_precision): Remove.
11923 (rs6000_emit_swdiv): Rewrite to handle between one and four
11924 iterations of Newton-Raphson generally; modify required number of
11925 iterations for some cases.
11926 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11928 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
11930 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11931 set-but-unused variable.
11933 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11934 basic blocks of released function bodies garbage-collectable.
11936 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11937 (struct rtl_opt_pass): Add TODO_df_finish.
11939 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11941 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11943 * config/arm/constraints.md (q): New constraint.
11944 * config/arm/ldrdstrd.md: New file.
11945 * config/arm/arm.md (ldrdstrd.md) New include.
11946 (arm_movdi): Use "q" instead of "r" constraint
11947 for double-word memory access.
11948 (movdf_soft_insn): Likewise.
11949 * config/arm/vfp.md (movdi_vfp): Likewise.
11950 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11951 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11952 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11953 (mem_ok_for_ldrd_strd): Likewise.
11954 (output_move_double): Update assertion.
11956 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11958 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11960 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11962 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11963 define_insn_and_split.
11964 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11966 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11968 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11969 define_insn_and_split.
11970 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11971 (shiftsi3_compare): New pattern.
11972 (rrx): New pattern.
11973 * config/arm/unspecs.md (UNSPEC_RRX): New.
11975 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11977 * config/arm/arm.md (negdi_extendsidi): New pattern.
11978 (negdi_zero_extendsidi): Likewise.
11980 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11982 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11983 define_insn_and_split.
11984 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11985 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11987 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11989 * config/arm/arm.md (arm_subdi3): Convert define_insn into
11990 define_insn_and_split.
11991 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11992 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11994 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11996 * config/arm/arm.md (subsi3_carryin): New pattern.
11997 (subsi3_carryin_const): Likewise.
11998 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11999 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12001 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12003 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12005 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12007 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12008 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12010 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12012 * config/arm/arm.c (arm_expand_builtin): Change fcode
12013 type to unsigned int.
12015 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12017 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12019 2013-04-04 Ian Lance Taylor <iant@google.com>
12021 * doc/standards.texi (Standards): The Go frontend supports the Go 1
12024 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
12026 PR middle-end/56729
12027 * df-scan.c (df_insn_delete): Disable failing assert.
12029 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12031 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12032 New function prototype.
12033 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12034 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12035 (arm_builtin_vectorized_function): New function.
12037 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12039 * config/arm/arm_neon_builtins.def: New file.
12040 * config/arm/arm.c (neon_builtin_data): Move contents to
12041 arm_neon_builtins.def.
12042 (enum arm_builtins): Include neon builtin definitions.
12043 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12044 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12046 2013-04-04 Marek Polacek <polacek@redhat.com>
12048 PR tree-optimization/48186
12049 * predict.c (maybe_hot_frequency_p): Return false if
12050 HOT_BB_FREQUENCY_FRACTION is 0.
12051 (cgraph_maybe_hot_edge_p): Likewise.
12053 2013-04-04 Richard Biener <rguenther@suse.de>
12055 PR tree-optimization/56826
12056 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12059 2013-04-04 Richard Biener <rguenther@suse.de>
12061 PR tree-optimization/56213
12062 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12063 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12065 2013-04-04 Richard Biener <rguenther@suse.de>
12067 PR tree-optimization/56837
12068 * tree-loop-distribution.c (classify_partition): For non-zero
12069 values require that the value has the same precision as its
12070 mode to be useful as memset value.
12072 2013-04-03 Nick Clifton <nickc@redhat.com>
12074 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12075 (fmssf4): Use fmsf.s on E3V5 architectures.
12076 (fnmasf4): Use fnmaf.s on E3V5 architectures.
12077 (fnmssf4): Use fnmsf.s on E3V5 architectures.
12079 2013-04-03 Jeff Law <law@redhat.com>
12081 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12082 (lra-eliminations.o): Likewise.
12084 2013-04-03 Teresa Johnson <tejohnson@google.com>
12086 * gcov-io.c (compute_working_sets): Moved most of body of old
12087 compute_working_sets here from profile.c.
12088 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12089 (gcov_working_set_t): Moved typedef here from basic-block.h
12090 (compute_working_set): Declare.
12091 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12092 (get_working_sets): Renamed from compute_working_set,
12093 replace most of body with call to new compute_working_sets.
12094 (get_exec_counts): Replace call to compute_working_sets
12095 to get_working_sets.
12096 * profile.h (get_working_sets): Renamed from compute_working_set.
12097 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12098 to get_working_sets.
12099 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12100 * gcov-dump.c (dump_working_sets): New function.
12102 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
12104 * hwint.c (sext_hwi, zext_hwi): New functions.
12105 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12106 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12107 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12108 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12109 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12110 (sext_hwi, zext_hwi): New functions.
12112 2013-04-03 Jeff Law <law@redhat.com>
12114 PR tree-optimization/56799
12115 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12116 back test for widening conversion erroneously dropped in prior change.
12118 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12121 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12122 instead of next_real_insn.
12124 2013-04-03 Marek Polacek <polacek@redhat.com>
12127 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12129 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12132 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12134 (thumb1_output_casesi): Likewise.
12135 (thumb2_output_casesi): Likewise.
12137 2013-04-03 Richard Biener <rguenther@suse.de>
12139 PR tree-optimization/56817
12140 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12142 (tree_unroll_loops_completely_1): ... new function to manually
12143 walk the loop tree, properly defering outer loops of unrolled
12144 loops to later iterations.
12146 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12148 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12149 (vectorizable_load): Likewise.
12150 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12151 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12153 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12155 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12158 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
12160 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12162 2013-04-03 Bin Cheng <bin.cheng@arm.com>
12164 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12166 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12168 PR tree-optimization/56790
12169 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12172 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12174 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12176 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12177 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12180 2013-04-03 Jakub Jelinek <jakub@redhat.com>
12183 * tree.h (force_folding_builtin_constant_p): New decl.
12184 * builtins.c (force_folding_builtin_constant_p): New variable.
12185 (fold_builtin_constant_p): Fold immediately also if
12186 force_folding_builtin_constant_p.
12188 2013-04-03 Richard Biener <rguenther@suse.de>
12190 PR tree-optimization/56812
12191 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12192 DRs of the same interleaving chain are independent.
12194 2013-04-02 Jason Merrill <jason@redhat.com>
12196 * gdbinit.in (pbb): Use debug fn.
12198 2013-04-02 Lawrence Crowl <crowl@google.com>
12200 * sese.h (struct ivtype_map_elt_s): Remove unused.
12201 (extern debug_ivtype_map): Remove unused.
12202 (extern eq_ivtype_map_elts): Remove unused.
12203 * sese.c (debug_ivtype_map): Removed unused.
12204 (debug_ivtype_map_1): Removed unused.
12205 (debug_ivtype_elt): Remove unused.
12206 (eq_ivtype_map_elts): Remove unused.
12209 2013-04-02 Kai Tietz <ktietz@redhat.com>
12212 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12213 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12214 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12216 (legitimize_pe_coff_symbol): Likewise.
12217 (is_imported_p): New helper-function.
12218 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12219 for Windows x64 targets.
12220 (ix86_expand_prologue): Optimize for pe-coff targets.
12221 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12222 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12223 medium/large code-model.
12224 (legitimize_pic_address): Likewise.
12225 (legitimize_tls_address): Likewise.
12226 (ix86_expand_call): Likewise.
12227 (x86_output_mi_thunk): Likewise.
12228 (get_dllimport_decl): Add new beimport argument.
12229 (construct_plt_address): Don't assert for x64 pe-coff targets.
12230 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12232 (SYMBOL_FLAG_STUBVAR): New macro.
12233 (SYMBOL_REF_STUBVAR_P): Likewise.
12234 * config/i386/winnt.c (stub_list): New structure.
12235 (stub_head): New local variable.
12236 (i386_pe_record_stub): New function.
12237 (i386_pe_file_end): Emit refptr-stubs.
12239 2013-04-02 Jakub Jelinek <jakub@redhat.com>
12241 PR rtl-optimization/56745
12242 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12243 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12246 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12247 and both of them are MEM_REFs, just compare first argument for
12248 equality and attempt to deal even with differing offsets.
12251 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12252 of gimple_clobber_p to be MEM_REF.
12253 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12254 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
12255 after gimplification.
12256 * asan.c (get_mem_ref_of_assignment): Don't instrument
12257 gimple_clobber_p stmts.
12258 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12259 gimple_clobber_p stmt if they have MEM_REF lhs and
12260 are dead because of another gimple_clobber_p stmt.
12261 * tree-ssa-live.c (clear_unused_block_pointer): Treat
12262 gimple_clobber_p stmts like debug stmts.
12263 (remove_unused_locals): Remove clobbers with MEM_REF lhs
12264 that refer to unused VAR_DECLs or uninitialized values.
12265 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12266 gimple_clobber_p stmts if they refer to removed parameters.
12267 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12270 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
12272 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12273 using SWI48 mode attribute.
12275 2013-04-02 Wei Mi <wmi@google.com>
12277 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12278 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12279 *<rotate_insn><mode>3_mask in i386.md.
12281 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
12283 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12285 2013-04-02 Richard Biener <rguenther@suse.de>
12287 PR tree-optimization/56778
12288 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12289 Runtime alias tests are not supported for gather loads.
12290 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12291 stmts referenced from SSA operands before updating SSA form.
12293 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
12294 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12296 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12297 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12298 * config/arm/cortex-a53.md: New file.
12299 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12300 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12301 * config/arm/arm.c (arm_issue_rate): Likewise.
12302 * config/arm/arm-tune.md: Regenerate
12303 * config/arm/arm-tables.opt: Regenerate.
12304 * config/arm/arm-cores.def: Add cortex-a53.
12306 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
12308 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12311 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12313 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12314 scalar load/store operations using B/H registers.
12315 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12317 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12319 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12321 * config/aarch64/aarch64.c
12322 (aarch64_simd_scalar_immediate_valid_for_move): New.
12323 * config/aarch64/aarch64-protos.h
12324 (aarch64_simd_scalar_immediate_valid_for_move): New.
12325 * config/aarch64/constraints.md (Dh, Dq): New.
12326 * config/aarch64/iterators.md (hq): New.
12328 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
12330 * reorg.c (get_branch_condition): Deal with conditional returns.
12331 (fill_simple_delay_slots): Remove dead code dealing with jumps.
12333 2013-04-01 Wei Mi <wmi@google.com>
12335 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12336 Truncate operand 2 using %b asm operand modifier.
12337 (*<shift_insn><mode>3_mask): Ditto.
12338 (*<rotate_insn><mode>3_mask): Ditto.
12340 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
12342 PR middle-end/56798
12343 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12345 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
12347 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12349 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12351 2013-03-30 Lawrence Crowl <crowl@google.com>
12353 * dse.c (clear_alias_sets): Remove never set.
12354 (disqualified_clear_alias_sets): Remove never set.
12355 (clear_alias_mode_pool): Remove never set.
12356 (dse_step0): Remove condition that is never true.
12357 (canon_address): Remove condition that is never true.
12358 (dse_step7): Remove condition that is never true.
12359 (rest_of_handle_dse): Remove condition that is never true.
12360 (rest_of_handle_dse::did_global): Remove never read from above.
12361 (dse_step2_spill): Remove never called from above.
12362 (dse_step5_spill): Remove never called from above.
12364 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
12366 * doc/md.texi (Standard Names) <casesi>: Update documentation for
12367 JUMP_TABLE_DATA changes.
12368 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12369 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12370 (Insns) <jump_table_data>: New entry.
12371 * doc/tm.texi: Regenerate.
12373 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12375 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12376 for table jump at the end of a basic block using tablejump_p.
12377 * targhooks.c (default_invalid_within_doloop): Likewise.
12378 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12379 target hook implementation that is identical to the default hook.
12380 (rs6000_invalid_within_doloop): Remove.
12382 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12383 unused variable from tablejump_p call.
12385 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12386 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12387 (INSN_DELETED_P): Likewise.
12388 (emit_jump_table_data): New prototype.
12389 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12390 after 4th as unused.
12391 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12392 * sched-vis.c (print_insn): Likewise.
12393 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12394 insn for compatibility with back ends that use next_active_insn to
12395 identify jump table data.
12396 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12397 (remove_insn): Likewise.
12398 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12400 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12401 (emit_jump_table_data): New function.
12403 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12404 basic block, a JUMP_TABLE_DATA never is.
12405 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12406 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12407 off from code handling real insns.
12408 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12409 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12410 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12411 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12412 is not a NONDEBUG_INSN_P.
12413 * ira-costs.c (scan_one_insn): Likewise.
12414 * jump.c (mark_all_labels): Likewise.
12415 (mark_jump_label_1): Likewise.
12416 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12417 * lra.c (get_insn_freq): Expect all insns reaching here to be in
12419 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12420 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12421 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12422 JUMP_TABLE_DATA_P insns.
12423 (calculate_elim_costs_all_insns): Likewise.
12424 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12425 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12426 (delete_output_reload): Code style fixups.
12427 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12428 insn flags on this non-insn.
12429 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12430 as scheduling barriers, for pre-change compatibility.
12431 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12432 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12434 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12435 redundant JUMP_TABLE_DATA_P test.
12436 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12437 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12438 (frv_for_each_packet): Likewise.
12439 * config/i386/i386.c (min_insn_size): Likewise.
12440 (ix86_avoid_jump_mispredicts): Likewise.
12441 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12442 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12443 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12444 (mips16_insn_length): Robustify.
12445 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12446 (mips16_split_long_branches): Likewise.
12447 * config/pa/pa.c (pa_combine_instructions): Likewise.
12448 * config/rs6000/rs6000.c (get_next_active_insn): Treat
12449 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12450 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12451 as contributing to pool range lengths.
12452 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12453 Remove redundant JUMP_TABLE_DATA_P test.
12454 (sh_loop_align): Likewise.
12455 (split_branches): Likewise.
12456 (sh_insn_length_adjustment): Likewise.
12457 * config/spu/spu.c (get_branch_target): Likewise.
12459 2013-03-29 Jan Hubicka <jh@suse.cz>
12461 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12462 gcov streaming; stream hot bb threshold to ltrans.
12463 * predict.c (get_hot_bb_threshold): Break out from ....
12464 (maybe_hot_count_p): ... here.
12465 (set_hot_bb_threshold): New function.
12466 * lto-section-in.c (lto_section_name): Add profile.
12467 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12468 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12469 and data-streamer.h
12470 (histogram_entry): New structure.
12471 (histogram, histogram_pool): New global vars.
12472 (histogram_hash): New structure.
12473 (histogram_hash::hash): New method.
12474 (histogram_hash::equal): Likewise.
12475 (account_time_size): New function.
12476 (cmp_counts): New function.
12477 (dump_histogram): New function.
12478 (ipa_profile_generate_summary): New function.
12479 (ipa_profile_write_summary): New function.
12480 (ipa_profile_read_summary): New function.
12481 (ipa_profile): Decide on threshold.
12482 (pass_ipa_profile): Add ipa_profile_write_summary and
12483 ipa_profile_read_summary.
12484 * Makefile.in (ipa.o): Update dependencies.
12485 * lto-streamer.h (LTO_section_ipa_profile): New section.
12487 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
12489 * tree.h (VAR_P): New.
12491 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
12494 * doc/invoke.texi ([-fwhole-program]): Fix typo.
12496 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
12498 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12499 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12500 (control_flow_insn_p): Likewise.
12501 * cfgrtl.c (duplicate_insn_chain): Likewise.
12502 * final.c (get_attr_length_1): Likewise.
12503 (shorten_branches): Likewise.
12504 (final_scan_insn): Likewise.
12505 * function.c (instantiate_virtual_regs): Likewise.
12506 * gcse.c (insert_insn_end_basic_block): Likewise.
12507 * ira-costs.c (scan_one_insn): Likewise.
12508 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12509 * lra.c (check_rtl): Likewise.
12510 * reload1.c (elimination_costs_in_insn): Likewise.
12511 * reorg.c (follow_jumps): Likewise.
12513 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12514 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12515 (thumb_far_jump_used_p): Likewise.
12516 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12517 (workaround_speculation): Likewise.
12518 (add_sched_insns_for_speculation): Likewise.
12519 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12520 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12521 (frv_for_each_packet): Likewise.
12522 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12523 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12524 (final_emit_insn_group_barriers): Likewise.
12525 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12526 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12527 (mips16_insn_length): Likewise.
12528 * config/pa/pa.c (pa_reorg): Likewise.
12529 (pa_combine_instructions): Likewise.
12530 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12531 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12532 (sh_reorg): Likewise.
12533 (split_branches): Likewise.
12534 * config/spu/spu.c (get_branch_target): Likewise.
12536 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12539 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
12541 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12542 Fix declaration name.
12544 2013-03-28 Lawrence Crowl <crowl@google.com>
12546 * graphds.h (struct graph.indicies): Remove unused.
12547 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12548 (SCOP_ORIGINAL_PDDRS): Remove unused.
12549 * sese.h (extern insert_loop_close_phis): Removed unused.
12550 (extern insert_guard_phis): Removed unused.
12551 (extern ivtype_map_elt_info): Removed unused.
12552 (new_ivtype_map_elt): Removed unused.
12553 * sese.c (ivtype_map_elt_info): Removed unused.
12555 2013-03-28 Lawrence Crowl <crowl@google.com>
12557 * Makefile.in: Add several missing include dependences.
12559 (test-dump.o): New. This object is not added to any executable,
12560 but is present for ad-hoc testing.
12562 (debug (const bitmap_head_def &)): New.
12563 (debug (const bitmap_head_def *)): New.
12565 (extern debug (const bitmap_head_def &)): New.
12566 (extern debug (const bitmap_head_def *)): New.
12568 (debug (edge_def &)): New.
12569 (debug (edge_def *)): New.
12571 (debug (basic_block_def &)): New.
12572 (debug (basic_block_def *)): New.
12574 (dump_node (const_tree, int, FILE *)): Correct source file.
12576 (debug (die_struct &)): New.
12577 (debug (die_struct *)): New.
12579 (extern debug (die_struct &)): New.
12580 (extern debug (die_struct *)): New.
12581 * gimple-pretty-print.c
12582 (debug (gimple_statement_d &)): New.
12583 (debug (gimple_statement_d *)): New.
12584 * gimple-pretty-print.h
12585 (extern debug (gimple_statement_d &)): New.
12586 (extern debug (gimple_statement_d *)): New.
12588 (debug (ira_allocno_copy &)): New.
12589 (debug (ira_allocno_copy *)): New.
12590 (debug (ira_allocno &)): New.
12591 (debug (ira_allocno *)): New.
12593 (extern debug (ira_allocno_copy &)): New.
12594 (extern debug (ira_allocno_copy *)): New.
12595 (extern debug (ira_allocno &)): New.
12596 (extern debug (ira_allocno *)): New.
12598 (debug (live_range &)): New.
12599 (debug (live_range *)): New.
12601 (debug (lra_live_range &)): New.
12602 (debug (lra_live_range *)): New.
12604 (debug (lra_live_range &)): New.
12605 (debug (lra_live_range *)): New.
12607 (debug (omega_pb_d &)): New.
12608 (debug (omega_pb_d *)): New.
12610 (extern debug (omega_pb_d &)): New.
12611 (extern debug (omega_pb_d *)): New.
12613 (debug (const rtx_def &)): New.
12614 (debug (const rtx_def *)): New.
12616 (debug_tree (tree): Move within file.
12617 (debug_raw (const tree_node &)): New.
12618 (debug_raw (const tree_node *)): New.
12619 (dump_tree_via_hooks (const tree_node *, int)): New.
12620 (debug (const tree_node &)): New.
12621 (debug (const tree_node *)): New.
12622 (debug_verbose (const tree_node &)): New.
12623 (debug_verbose (const tree_node *)): New.
12624 (debug_head (const tree_node &)): New.
12625 (debug_head (const tree_node *)): New.
12626 (debug_body (const tree_node &)): New.
12627 (debug_body (const tree_node *)): New.
12628 (debug_vec_tree (tree): Move and reimplement in terms of dump.
12629 (debug (vec<tree, va_gc> &)): New.
12630 (debug (vec<tree, va_gc> *)): New.
12632 (extern debug (const rtx_def &)): New.
12633 (extern debug (const rtx_def *)): New.
12635 (debug_raw (simple_bitmap_def &)): New.
12636 (debug_raw (simple_bitmap_def *)): New.
12637 (debug (simple_bitmap_def &)): New.
12638 (debug (simple_bitmap_def *)): New.
12640 (extern debug (simple_bitmap_def &)): New.
12641 (extern debug (simple_bitmap_def *)): New.
12642 (extern debug_raw (simple_bitmap_def &)): New.
12643 (extern debug_raw (simple_bitmap_def *)): New.
12645 (debug (vinsn_def &)): New.
12646 (debug (vinsn_def *)): New.
12647 (debug_verbose (vinsn_def &)): New.
12648 (debug_verbose (vinsn_def *)): New.
12649 (debug (expr_def &)): New.
12650 (debug (expr_def *)): New.
12651 (debug_verbose (expr_def &)): New.
12652 (debug_verbose (expr_def *)): New.
12653 (debug (vec<rtx> &)): New.
12654 (debug (vec<rtx> *)): New.
12656 (extern debug (vinsn_def &)): New.
12657 (extern debug (vinsn_def *)): New.
12658 (extern debug_verbose (vinsn_def &)): New.
12659 (extern debug_verbose (vinsn_def *)): New.
12660 (extern debug (expr_def &)): New.
12661 (extern debug (expr_def *)): New.
12662 (extern debug_verbose (expr_def &)): New.
12663 (extern debug_verbose (expr_def *)): New.
12664 (extern debug (vec<rtx> &)): New.
12665 (extern debug (vec<rtx> *)): New.
12667 (_list_iter_cond_expr): Make inline instead of static.
12669 (debug (sreal &)): New.
12670 (debug (sreal *)): New.
12672 (extern debug (sreal &)): New.
12673 (extern debug (sreal *)): New.
12675 (extern debug_raw (const tree_node &)): New.
12676 (extern debug_raw (const tree_node *)): New.
12677 (extern debug (const tree_node &)): New.
12678 (extern debug (const tree_node *)): New.
12679 (extern debug_verbose (const tree_node &)): New.
12680 (extern debug_verbose (const tree_node *)): New.
12681 (extern debug_head (const tree_node &)): New.
12682 (extern debug_head (const tree_node *)): New.
12683 (extern debug_body (const tree_node &)): New.
12684 (extern debug_body (const tree_node *)): New.
12685 (extern debug (vec<tree, va_gc> &)): New.
12686 (extern debug (vec<tree, va_gc> *)): New.
12688 (debug (struct loop &)): New.
12689 (debug (struct loop *)): New.
12690 (debug_verbose (struct loop &)): New.
12691 (debug_verbose (struct loop *)): New.
12692 * tree-dump.c: Add header dependence.
12694 (extern debug (struct loop &)): New.
12695 (extern debug (struct loop *)): New.
12696 (extern debug_verbose (struct loop &)): New.
12697 (extern debug_verbose (struct loop *)): New.
12699 (debug (data_reference &)): New.
12700 (debug (data_reference *)): New.
12701 (debug (vec<data_reference_p> &)): New.
12702 (debug (vec<data_reference_p> *)): New.
12703 (debug (vec<ddr_p> &)): New.
12704 (debug (vec<ddr_p> *)): New.
12706 (extern debug (data_reference &)): New.
12707 (extern debug (data_reference *)): New.
12708 (extern debug (vec<data_reference_p> &)): New.
12709 (extern debug (vec<data_reference_p> *)): New.
12710 (extern debug (vec<ddr_p> &)): New.
12711 (extern debug (vec<ddr_p> *)): New.
12713 (debug (pt_solution &)): New.
12714 (debug (pt_solution *)): New.
12716 (extern debug (pt_solution &)): New.
12717 (extern debug (pt_solution *)): New.
12719 (debug (_var_map &)): New.
12720 (debug (_var_map *)): New.
12721 (debug (tree_live_info_d &)): New.
12722 (debug (tree_live_info_d *)): New.
12724 (extern debug (_var_map &)): New.
12725 (extern debug (_var_map *)): New.
12726 (extern debug (tree_live_info_d &)): New.
12727 (extern debug (tree_live_info_d *)): New.
12729 2013-03-28 Jan Hubicka <jh@suse.cz>
12731 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12733 2013-03-28 Ian Bolton <ian.bolton@arm.com>
12735 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12736 record only when desired or required.
12738 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
12740 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12741 *vec_extractv2di_1_rex64. Use x64 isa attribute.
12743 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12745 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12746 (*andsi3_compare0_uxtw): New pattern.
12747 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12748 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12750 2013-03-28 Jan Hubicka <jh@suse.cz>
12752 * data-streamer-in.c (streamer_read_gcov_count): New function.
12753 * gimple-streamer-out.c: Include value-prof.h.
12754 (output_gimple_stmt): Output histogram.
12755 (output_bb): Use streamer_write_gcov_count.
12756 * value-prof.c: Include data-streamer.h
12757 (dump_histogram_value): Add HIST_TYPE_MAX.
12758 (stream_out_histogram_value): New function.
12759 (stream_in_histogram_value): New function.
12760 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12761 (stream_out_histogram_value, stream_in_histogram_value): Declare.
12762 * data-streamer-out.c (streamer_write_gcov_count): New function.
12763 (streamer_write_gcov_count_stream): New function.
12764 * lto-cgraph.c (lto_output_edge): Update counter streaming.
12765 (lto_output_node): Likewise.
12766 (input_node, input_edge): Likewise.
12767 * lto-streamer-out.c (output_cfg): Update streaming.
12768 * lto-streamer-in.c (input_cfg): Likewise.
12769 * data-streamer.h (streamer_write_gcov_count,
12770 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12771 * gimple-streamer-in.c: Include value-prof.h
12772 (input_gimple_stmt): Input histograms.
12773 (input_bb): Update profile streaming.
12775 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
12777 * genmodes.c (emit_max_int): New function.
12778 (emit_insn_modes_h): Added call to emit_max_function.
12779 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12781 * machmode.def: Fixed comment.
12783 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
12785 * combine.c (try_combine): Removed useless assert.
12786 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12788 2013-03-28 Marek Polacek <polacek@redhat.com>
12789 Richard Biener <rguenther@suse.de>
12791 PR tree-optimization/56695
12792 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12793 build signed result of a vector comparison.
12794 * tree-cfg.c (verify_gimple_comparison): Check that a result
12795 of a vector comparison has signed type.
12797 2013-03-28 Richard Biener <rguenther@suse.de>
12799 PR tree-optimization/37021
12800 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12801 do not restrict gaps between groups.
12802 * tree-vect-stmts.c (vectorizable_load): Properly account for
12803 a gap between groups.
12805 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
12807 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12808 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12811 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
12813 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12814 * doc/extend.texi (Named Address Spaces): Ditto.
12815 (Variable Attributes): Ditto.
12817 2013-03-27 Kai Tietz <ktietz@redhat.com>
12819 * config.build: Add support for cygwin x64 target.
12820 * config.gcc: Likewise.
12821 * config.host: Likewise.
12822 * configure.ac: Likewise
12823 * configure: Regenerated.
12825 2013-03-27 Kai Tietz <ktietz@redhat.com>
12827 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12828 * config/i386/t-cygwin-w64: New file.
12829 * config/i386/cygwin-w64.h: New file.
12830 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12831 and add support for x64-cygwin target.
12832 (CPP_SPEC): Likewise.
12833 (CXX_WRAP_SPEC_LIST): Undefine before define.
12834 (LIBGCJ_SONAME): Use 15 as version.
12836 2013-03-27 Richard Biener <rguenther@suse.de>
12838 PR tree-optimization/56716
12839 * tree-ssa-structalias.c (perform_var_substitution): Adjust
12840 dumping for ref nodes.
12842 2013-03-27 Martin Jambor <mjambor@suse.cz>
12844 PR tree-optimization/55334
12845 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12846 restricted pointers to arrays.
12848 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
12850 * Makefile.in (.SUFFIXES): Add .cc.
12851 (.c.o): Apply same recipe for implicit rule .cc.o.
12853 2013-03-27 Richard Biener <rguenther@suse.de>
12855 PR tree-optimization/37021
12856 * tree-vect-data-refs.c (vect_check_strided_load): Allow
12857 REALPART/IMAGPART_EXPRs around the supported refs.
12858 * tree-ssa-structalias.c (find_func_aliases): Assume that
12859 floating-point values are not used to transfer pointers.
12861 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
12863 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
12864 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12865 * gcc/doc/tm.texi: Regenerate.
12866 * gcc/targhooks.h (default_has_ifunc_p): New.
12867 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
12868 * gcc/config/linux-protos.h: New file.
12869 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12870 this hook for linux which disables support of indirect functions in
12872 * gcc/config/linux-android.c: New file.
12873 * gcc/config/t-linux-android.c: Ditto.
12874 * gcc/config.gcc: Added new object file linux-android.o.
12875 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
12876 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12877 * gcc/varasm.c (do_assemble_alias): Likewise.
12878 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12879 doesn't support indirect functions.
12880 * configure: Regenerate.
12882 2013-03-27 Bin Cheng <bin.cheng@arm.com>
12885 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12886 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12889 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12891 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12893 2013-03-27 Terry Guo <terry.guo@arm.com>
12895 * config/arm/arm-cores.def: Added core cortex-r7.
12896 * config/arm/arm-tune.md: Regenerated.
12897 * config/arm/arm-tables.opt: Regenerated.
12898 * doc/invoke.texi: Added entry for core cortex-r7.
12900 2013-03-27 Walter Lee <walt@tilera.com>
12902 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12903 double-decrement of next_scratch_regno.
12905 2013-03-27 Walter Lee <walt@tilera.com>
12907 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12909 (insn_v1mulus): Ditto.
12910 (insn_v2muls): Ditto.
12912 2013-03-27 Walter Lee <walt@tilera.com>
12914 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12915 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12917 2013-03-27 Walter Lee <walt@tilera.com>
12919 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12920 (*sibcall_value): Ditto.
12922 2013-03-27 Walter Lee <walt@tilera.com>
12924 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12925 (insn_mnz_v8qi): ... this ...
12926 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
12928 (insn_v<n>mnz): Replaced by ...
12929 (insn_v1mnz): ... this ...
12930 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
12932 (insn_mz_<mode>): Replaced by ...
12933 (insn_mz_v8qi): ... this ...
12934 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
12936 (insn_v<n>mz): Replaced by ...
12937 (insn_v1mz): ... this ...
12938 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
12941 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
12943 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12945 2013-03-26 Roland McGrath <mcgrathr@google.com>
12947 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12948 than fprintf with a non-constant, non-format string.
12950 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
12952 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12953 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
12954 operand 0 predicate.
12955 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12956 attribute. Use general_x64nomem_operand as operand 1 predicate.
12957 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12958 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12959 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12960 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12961 (mov<mode>_insv_1): Remove expander. Merge insn with
12962 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12963 Use general_x64nomem_operand as operand 1 predicate.
12964 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12965 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12966 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12967 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12968 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
12969 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12970 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12971 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12972 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
12973 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12974 (general_x64nomem_operand): Ditto.
12976 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12978 * config/rtems.opt: Add -pthread option.
12980 2013-03-26 Richard Biener <rguenther@suse.de>
12982 * alias.c (find_base_term): Avoid redundant and not used recursion.
12983 (base_alias_check): Get the initial base term from the caller.
12984 (true_dependence_1): Compute and pass base terms to base_alias_check.
12985 (write_dependence_p): Likewise.
12986 (may_alias_p): Likewise.
12988 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
12990 * config/aarch64/aarch64.c (aarch64_classify_address): Support
12991 PC-relative load in SI modes and above only.
12993 2013-03-26 Xinyu Qi <xyqi@marvell.com>
12995 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12996 * config/arm/iwmmxt.md (WCGR0): Update.
12997 (WCGR1, WCGR2, WCGR3): Likewise.
12999 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
13001 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13002 Use x64 and nox64 isa attributes.
13004 2013-03-26 Richard Biener <rguenther@suse.de>
13006 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13007 alignment computations and rely on get_object_alignment_1
13008 for the !TYPE_P case.
13009 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13011 2013-03-26 Walter Lee <walt@tilera.com>
13013 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13014 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13016 2013-03-25 Jeff Law <law@redhat.com>
13018 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13019 check for INTEGRAL_TYPE_P that was missing due to checking in
13020 wrong version of prior patch.
13022 2013-03-25 Walter Lee <walt@tilera.com>
13024 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13025 TILEGX_INSN_SHUFFLEBYTES1.
13026 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13028 (tilegx_builtins): Ditto.
13029 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13031 2013-03-25 Walter Lee <walt@tilera.com>
13033 * config/tilegx/tilegx.md (floatsisf2): New pattern.
13034 (floatunssisf2): New pattern.
13035 (floatsidf2): New pattern.
13036 (floatunssidf2): New pattern.
13038 2013-03-25 Walter Lee <walt@tilera.com>
13040 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13041 tests for constraint J, K, N, P.
13043 2013-03-25 Walter Lee <walt@tilera.com>
13045 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13046 Use indirect/pcrel encoding.
13047 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13050 2013-03-25 Steve Ellcey <sellcey@mips.com>
13052 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13053 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13054 * config/mips/mips.c (mips_option_override): Set IMADD default.
13055 * config/mips/mips.h (PTF_AVOID_IMADD): New.
13056 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13057 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13058 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13059 * doc/invoke.texi (-mimadd/-mno-imadd): New.
13061 2013-03-25 Jeff Law <law@redhat.com>
13063 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13064 slightly to avoid creating and folding useless trees. Simplify
13065 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13067 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
13069 * config/i386/i386.md (*zero_extendsidi2): Merge with
13070 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
13071 * config/i386/predicates.md (x86_64_zext_operand): Rename from
13072 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
13073 targets. Clarify comment.
13075 2013-03-25 Martin Jambor <mjambor@suse.cz>
13077 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13078 pass-through jump functions differently.
13079 (ipa_read_jump_function): Likewise. Also use setter functions to set
13082 2013-03-25 Martin Jambor <mjambor@suse.cz>
13084 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13085 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13087 (ipa_get_indirect_edge_target): New function.
13088 (devirtualization_time_bonus): New parameter known_aggs, pass it to
13089 ipa_get_indirect_edge_target. Update all callers.
13090 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
13091 ipa_get_indirect_edge_target_1 instead of calling
13092 ipa_get_indirect_edge_target.
13093 (create_specialized_node): Pass aggvlas to
13094 ipcp_discover_new_direct_edges.
13096 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13098 * config/arm/arm.md (f_sels, f_seld): New types.
13099 (*cmov<mode>): New pattern.
13100 * config/arm/predicates.md (arm_vsel_comparison_operator): New
13103 2013-03-25 Kai Tietz <ktietz@redhat.com>
13105 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13106 POSIX-printf for mingw-hosted builds.
13108 2013-03-25 Richard Biener <rguenther@suse.de>
13110 PR middle-end/56694
13111 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13112 must-not-throw stmt location.
13114 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13116 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13117 Emit load-acquire versions when acq is true.
13118 (arm_emit_store_exclusive): Add rel parameter.
13119 Emit store-release versions when rel is true.
13120 (arm_split_compare_and_swap): Use acquire-release instructions
13122 of barriers when appropriate.
13123 (arm_split_atomic_op): Likewise.
13124 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13125 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13126 (VUNSPEC_SLX): Likewise.
13127 (VUNSPEC_LDA): Likewise.
13128 (VUNSPEC_STL): Likewise.
13129 * config/arm/sync.md (atomic_load<mode>): New pattern.
13130 (atomic_store<mode>): Likewise.
13131 (arm_load_acquire_exclusive<mode>): Likewise.
13132 (arm_load_acquire_exclusivesi): Likewise.
13133 (arm_load_acquire_exclusivedi): Likewise.
13134 (arm_store_release_exclusive<mode>): Likewise.
13136 2013-03-25 Catherine Moore <clm@codesourcery.com>
13138 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13139 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13140 * config/mip/predicates.md (lwsp_swsp_operand,
13141 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13142 sb16_operand, db4_operand, db7_operand, ib3_operand,
13143 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13144 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13145 andi16_operand): New predicates.
13146 * config/mips/mips.md (compression): New attribute.
13147 (enabled): New attribute.
13148 (length): Consider compression in computing length.
13149 (shift_compression): New code attribute.
13150 (*add<mode>3): New operands. Record compression.
13151 (sub<mode>3): Likewise.
13152 (one_cmpl<mode>2): Likewise.
13153 (*and<mode>3): Likewise.
13154 (*ior<mode>3): Likewise.
13155 (unnamed pattern for xor): Likewise.
13156 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13157 (*<optab><mode>3): Likewise.
13158 (*mov<mode>_internal: Likewise.
13159 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13160 (mips_unsigned_immediate_p): New.
13161 (umips_lwsp_swsp_address_p): New.
13162 (m16_based_address_p): New.
13163 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13164 (mips_unsigned_immediate_p): New prototype.
13165 (lwsp_swsp_address_p): New prototype.
13166 (m16_based_address_p): New prototype.
13167 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13168 (mips_signed_immediate_p): New function.
13169 (m16_based_address_p): New function.
13170 (lwsp_swsp_address_p): New function.
13171 (mips_print_operand_punctuation): Recognize short delay slot insns
13172 for microMIPS.add<mode>3"
13174 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13177 * config/arm/iterators.md (v_cmp_result): New mode attribute.
13178 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13180 2013-03-25 Richard Biener <rguenther@suse.de>
13182 PR tree-optimization/56689
13183 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13186 2013-03-25 Richard Biener <rguenther@suse.de>
13188 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13190 (memory_references): Likewise.
13191 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13192 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13193 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13194 (gather_mem_refs_in_loops): Fold into ...
13195 (analyze_memory_references): ... this. Move initialization
13196 to tree_ssa_lim_initialize.
13197 (fill_always_executed_in): Rename to ...
13198 (fill_always_executed_in_1): ... this.
13199 (fill_always_executed_in): Move contains_call computation to
13200 this new function from ...
13201 (tree_ssa_lim_initialize): ... here.
13202 (tree_ssa_lim): Call fill_always_executed_in.
13204 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
13206 * postreload.c (reload_combine): Fix code detecting returns.
13208 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
13210 * function.c (emit_use_return_register_into_block): On cc0 targets,
13211 do not emit the sequence between cc0 setter and user.
13213 2013-03-25 Kai Tietz <ktietz@redhat.com>
13215 * config/i386/predicates.md (local_symbolic_operand): Interpret
13216 dll-imported symbols as none-local.
13218 2013-03-25 Richard Biener <rguenther@suse.de>
13220 * tree-ssa-loop-im.c (struct depend): Remove.
13221 (struct lim_aux_data): Make depends a vec of gimples.
13222 (free_lim_aux_data): Adjust.
13223 (add_dependency): Likewise.
13224 (set_level): Likewise.
13226 2013-03-25 Richard Biener <rguenther@suse.de>
13228 PR middle-end/56434
13229 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13230 the pointer returned by calls with ECF_MALLOC set.
13232 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
13234 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13236 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
13238 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13239 using MMXMODE mode iterator.
13240 (*move<mode>_internal): Merge with *movv2sf_internal and
13241 *movv2sf_internal_rex64 using MMXMODE mode iterator.
13243 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
13245 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13246 (record_last_mem_set_info): Likewise.
13248 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13249 of XNEWVEC followed by memset.
13250 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13252 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
13254 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13255 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13256 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13257 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13258 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13259 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13260 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13261 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13262 BARRIER_P instead of GET_CODE.
13264 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13266 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13267 inaccuracy in the probing code.
13269 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13270 (ctrapdi4): Likewise.
13272 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13274 * calls.c (expand_call): Add missing guard to code handling return
13275 of non-BLKmode structures in MSB.
13276 * function.c (expand_function_end): Likewise.
13278 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13280 * combine.c (try_combine): Adjust comment. Do not add the set of
13281 insn #0 if the destination indirectly is set or dies in insn #2.
13282 Tidy up code to distribute a new note.
13284 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13286 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13287 also for alternatives 16 and 17.
13289 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13291 * config/i386/sse.md (*mov<mode>_internal): Merge with
13292 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
13293 Emit insn template depending on type attribute. Use
13294 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13295 movd instead of movq mnemonic for interunit moves. Rewrite mode
13296 attribute calculation. Remove unit attribute calculation.
13297 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13298 Set prefix_data16 attribute for DImode ssemov types.
13299 Use Ym instead of y for SSE-MMX conversion alternatives.
13300 Reorder operand constraints.
13302 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
13304 * df.h (df_insn_delete): Adjust prototype.
13305 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13306 and let it decide whether mark the basic block dirty.
13307 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13308 * df-scan.c (df_insn_info_delete): New helper function, split
13309 off from df_insn_delete.
13310 (df_scan_free_bb_info): Use it.
13311 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13313 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
13314 that the insn is actually an insn and it has a non-NULL basic block.
13315 Do not mark basic block dirty if only deleting a DEBUG_INSN.
13317 2013-03-22 Richard Biener <rguenther@suse.de>
13319 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13321 (mem_ref_alloc): Do not allocate them.
13322 (refs_independent_p): Do not query or maintain a cache.
13324 2013-03-22 Richard Biener <rguenther@suse.de>
13326 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13327 (gather_mem_refs_in_loops): Do not compute it.
13328 (analyze_memory_references): Do not allocate it.
13329 (tree_ssa_lim_finalize): Do not free it.
13330 (for_all_locs_in_loop): Do not query all_refs_in_loop.
13332 2013-03-22 Richard Biener <rguenther@suse.de>
13334 * is-a.h (as_a): Use gcc_checking_assert.
13336 2013-03-22 Ian Bolton <ian.bolton@arm.com>
13338 * config/aarch64/aarch64.c (aarch64_print_operand): New
13339 format specifier for printing a constant in hex.
13340 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13341 format specifier for printing second operand.
13343 2013-03-22 Richard Biener <rguenther@suse.de>
13345 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13347 (gather_mem_refs_in_loops): Perform store accumulation here.
13348 (create_vop_ref_mapping_loop): Remove.
13349 (create_vop_ref_mapping): Likewise.
13350 (analyze_memory_references): Initialize refs_stored_in_loop.
13351 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13352 (record_indep_loop): Remove.
13353 (record_dep_loop): New function.
13354 (ref_indep_loop_p_1): Adjust to only walk over references
13355 in the loop, not its subloops.
13356 (ref_indep_loop_p): Rename to ...
13357 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13358 maintaining a more fine-grained cache.
13359 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13360 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13362 2013-03-22 Richard Biener <rguenther@suse.de>
13364 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13365 (struct mem_ref): Make accesses_in_loop a vec of a vec of
13366 aggregate mem_ref_loc.
13367 (free_mem_ref_locs): Inline into ...
13368 (memref_free): ... this and adjust.
13369 (mem_ref_alloc): Adjust.
13370 (mem_ref_locs_alloc): Remove.
13371 (record_mem_ref_loc): Adjust.
13372 (get_all_locs_in_loop): Rewrite into ...
13373 (for_all_locs_in_loop): ... this iterator.
13374 (rewrite_mem_ref_loc): New functor.
13375 (rewrite_mem_refs): Use for_all_locs_in_loop.
13376 (sm_set_flag_if_changed): New functor.
13377 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13378 (ref_always_accessed): New functor.
13379 (ref_always_accessed_p): Use for_all_locs_in_loop.
13381 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13383 * tree-pass.h (PROP_gimple_lvec): New.
13384 * passes.c (dump_properties): Handle PROP_gimple_lvec.
13385 (init_optimization_passes): Move pass_lower_vector.
13386 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13388 (pass_lower_vector): Provide PROP_gimple_lvec.
13389 (pass_lower_vector_ssa): Likewise.
13390 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13392 2013-03-21 Mark Wielaard <mjw@redhat.com>
13394 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13396 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13398 * config/i386/i386.md (*movdi_internal): Disparage slightly
13399 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
13400 conversion alternatives.
13402 2013-03-21 Jakub Jelinek <jakub@redhat.com>
13404 PR middle-end/48087
13405 * diagnostic.def (DK_WERROR): New kind.
13406 * diagnostic.h (werrorcount): Define.
13407 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13408 promoted to DK_ERROR, increment DK_WERROR counter instead of
13410 * toplev.c (toplev_main): Call print_ignored_options even if
13411 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
13412 even if just werrorcount is non-zero.
13415 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13417 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13418 (string_cst_pool_decl): New function.
13419 (optimize_one_addr_into_implicit_ptr): New function.
13420 (resolve_addr_in_expr): Optimize DWARF location expression
13421 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13422 which doesn't live in memory, but has DW_AT_location or
13423 DW_AT_const_value, or refers to a string literal, into
13424 DW_OP_GNU_implicit_pointer.
13425 (optimize_location_into_implicit_ptr): New function.
13426 (resolve_addr): If removing DW_AT_location of a variable because
13427 it was DW_OP_addr of address of the variable, but the variable doesn't
13428 live in memory, try to emit const value attribute for the initializer.
13430 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13432 * tree.h (VECTOR_TYPE_P): New macro.
13433 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13434 TYPE_MODE): Use it.
13435 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13436 VEC_COND_EXPR cannot be lvalues.
13437 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13439 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13441 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13442 Restrict the transformation to equal modes.
13444 2013-03-21 Richard Biener <rguenther@suse.de>
13446 PR tree-optimization/39326
13447 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13448 (MEM_ANALYZABLE): Adjust.
13449 (record_mem_ref_loc): Move bitmap ops ...
13450 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
13451 unanalyzable refs, do not record locations for it.
13452 (analyze_memory_references): Allocate ref zero as shared
13454 (refs_independent_p): Do not test for unanalyzed mems here.
13455 (ref_indep_loop_p_1): Special-case disambiguation against
13456 the unanalyzed ref.
13457 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13459 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
13461 * config/arm/arm-protos.h (tune_params): Add
13462 prefer_neon_for_64bits field.
13463 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13464 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13465 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13466 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13467 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13468 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13469 (arm_option_override): Handle -mneon-for-64bits new option.
13470 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13471 (prefer_neon_for_64bits): Declare new variable.
13472 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13473 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13474 (arch_enabled): Handle new arch types. Remove support for onlya8
13476 (one_cmpldi2): Use new arch names.
13477 (zero_extend<mode>di2, extend<mode>di2): Ditto.
13478 * config/arm/arm.opt (mneon-for-64bits): Add option.
13479 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13480 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13481 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13483 * doc/invoke.texi (-mneon-for-64bits): Document.
13485 2013-03-21 Richard Biener <rguenther@suse.de>
13487 PR tree-optimization/39326
13488 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13489 (sort_bbs_in_loop_postorder_cmp): New function.
13490 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13492 2013-03-21 Richard Biener <rguenther@suse.de>
13494 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13495 (vect_insert_into_interleaving_chain): Likewise.
13496 (vect_drs_dependent_in_basic_block): Inline ...
13497 (vect_slp_analyze_data_ref_dependence): ... here. New function,
13499 (vect_analyze_data_ref_dependence): ... here. Simplify.
13500 (vect_check_interleaving): Simplify.
13501 (vect_analyze_data_ref_dependences): Likewise. Split out ...
13502 (vect_slp_analyze_data_ref_dependences): ... this new function.
13503 (dr_group_sort_cmp): New function.
13504 (vect_analyze_data_ref_accesses): Compute data-reference groups
13505 here instead of in vect_analyze_data_ref_dependence. Use
13506 a more efficient algorithm.
13507 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13508 vect_slp_analyze_data_ref_dependences. Call
13509 vect_analyze_data_ref_accesses earlier.
13510 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13511 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13512 (vect_slp_analyze_data_ref_dependences): New prototype.
13514 2013-03-21 Richard Biener <rguenther@suse.de>
13516 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13517 ref is stored in the loop.
13518 (find_refs_for_sm): Walk only over all stores.
13519 (store_motion_loop): Allocate from lim_bitmap_obstack.
13520 (store_motion): Likewise.
13522 2013-03-21 Richard Biener <rguenther@suse.de>
13524 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13525 Update virtual SSA form.
13527 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13529 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13530 * configure: Regenerate.
13531 * config.in: Regenerate.
13532 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13533 if !HAVE_LD_EH_FRAME_CIEV3.
13535 2013-03-21 Richard Biener <rguenther@suse.de>
13537 * tree-cfg.c (verify_expr_no_block): New function.
13538 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13539 nor DECL_VALUE_EXPR have locations with associated blocks.
13540 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13541 (clear_unused_block_pointer): Remove code dealing with
13542 blocks in DECL_DEBUG_EXPR locations.
13544 2013-03-21 Richard Biener <rguenther@suse.de>
13546 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13547 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
13548 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13549 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13550 instead of DECL_DEBUG_EXPR_IS_FROM.
13551 * gimplify.c (gimplify_modify_expr): Likewise.
13552 * tree-cfg.c (verify_expr_location_1): Likewise.
13553 * tree-complex.c (create_one_component_var): Likewise.
13554 * tree-sra.c (create_access_replacement): Likewise.
13555 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13556 (clear_unused_block_pointer): Likewise.
13557 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13558 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13559 * var-tracking.c (var_debug_decl): Likewise.
13560 (track_expr_p): Likewise.
13561 * tree-inline.c (add_local_variables): Likewise. Set
13562 DECL_HAS_DEBUG_EXPR_P after copying it.
13563 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13564 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
13566 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13569 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13570 * configure: Regenerate.
13571 * config.in: Regenerate.
13572 * config/i386/i386.md (*movdf_internal): Use
13573 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13574 movd instead of movq mnemonic for interunit moves.
13575 (*movdi_internal): Ditto.
13577 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13579 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13580 (abd<mode>_3): New pattern.
13581 (aba<mode>_3): New pattern.
13582 (fabd<mode>_3): New pattern.
13584 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13586 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13587 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13588 occurrence of REGISTER_PREFIX as its empty string.
13590 2013-03-20 Jeff Law <law@redhat.com>
13592 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13593 addititional equivalences for equality comparisons between an SSA_NAME
13594 and a constant where the SSA_NAME was set from a widening conversion.
13596 2013-03-20 Walter Lee <walt@tilera.com>
13598 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13600 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13602 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13603 depending on type attribute.
13604 (*movti_internal): Ditto.
13605 (*movtf_internal): Ditto.
13606 (*movxf_internal): Ditto.
13607 (*movdf_internal): Ditto.
13608 (*movsf_internal): Ditto.
13610 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13612 * config/i386/i386.md (*movti_internal): Set prefix attribute to
13613 maybe_vex for sselog1 and ssemov types.
13614 (*movdi_internal): Reorder operand constraints.
13615 (*movsi_internal): Ditto. Set prefix attribute to
13616 maybe_vex for sselog1 and ssemov types.
13617 (*movtf_internal): Set prefix attribute to maybe_vex
13618 for sselog1 and ssemov types.
13619 (*movdf_internal): Ditto. Set prefix_data16 attribute for
13620 DImode ssemov types. Reorder operand constraints.
13621 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
13622 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
13623 attribute for SImode ssemov types. Reorder operand constraints.
13625 2013-03-20 Martin Jambor <mjambor@suse.cz>
13627 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13628 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13630 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
13632 * config/rs6000/predicates.md (indexed_address, update_address_mem
13633 update_indexed_address_mem): New predicates.
13634 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13635 attribute for load/store instructions.
13636 * config/rs6000/dfp.md (movsd_store): Likewise.
13637 (movsd_load): Likewise.
13638 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13639 (unnamed HI->DI extend define_insn): Likewise.
13640 (unnamed SI->DI extend define_insn): Likewise.
13641 (unnamed QI->SI extend define_insn): Likewise.
13642 (unnamed QI->HI extend define_insn): Likewise.
13643 (unnamed HI->SI extend define_insn): Likewise.
13644 (unnamed HI->SI extend define_insn): Likewise.
13645 (extendsfdf2_fpr): Likewise.
13646 (movsi_internal1): Likewise.
13647 (movsi_internal1_single): Likewise.
13648 (movhi_internal): Likewise.
13649 (movqi_internal): Likewise.
13650 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13651 attribute for load/store instructions.
13652 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13654 (mov<mode>_softfloat): Likewise.
13655 (mov<mode>_hardfloat32): Likewise.
13656 (mov<mode>_hardfloat64): Likewise.
13657 (mov<mode>_softfloat64): Likewise.
13658 (movdi_internal32): Likewise.
13659 (movdi_internal64): Likewise.
13660 (probe_stack_<mode>): Likewise.
13662 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13664 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13665 floating point, and decimal floating point to reload iterator.
13667 * config/rs6000/constraints.md (wl constraint): New constraints to
13668 return FLOAT_REGS if certain options are used to reduce the number
13669 of separate patterns that exist in the file.
13670 (wx constraint): Likewise.
13671 (wz constraint): Likewise.
13673 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13674 -mdebug=reg, print wg, wl, wx, and wz constraints.
13675 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13676 Initialize the reload functions for 64-bit binary/decimal floating
13678 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13679 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13680 create the buffer on the stack to overcome not having a 32-bit
13682 (rs6000_emit_move): Likewise.
13683 (rs6000_secondary_memory_needed_rtx): Likewise.
13684 (rs6000_alloc_sdmode_stack_slot): Likewise.
13685 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13686 via xxlxor, just like DFmode 0.0.
13688 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13689 define as 1 if we are running on a power7 or newer.
13690 (enum r6000_reg_class_enum): Add new constraints.
13692 * config/rs6000/dfp.md (movsd): Delete, combine with binary
13693 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
13694 with other moves by using conditional constraits (wg). Use LFIWZX
13695 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
13696 (movsd splitter): Likewise.
13697 (movsd_hardfloat): Likewise.
13698 (movsd_softfloat): Likewise.
13700 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13701 binary and decimal floating point moves.
13702 (fmove_ok): New attributes to combine binary and decimal floating
13703 point moves, and to combine power6x (mfpgpr) moves along normal
13705 (real_value_to_target): Likewise.
13706 (f32_lr): Likewise.
13707 (f32_lm): Likewise.
13708 (f32_li): Likewise.
13709 (f32_sr): Likewise.
13710 (f32_sm): Likewise.
13711 (f32_si): Likewise.
13712 (movsf): Combine binary and decimal floating point moves. Combine
13713 power6x (mfpgpr) moves with other moves by using conditional
13714 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
13715 (mov<mode> for SFmode/SDmode); Likewise.
13716 (SFmode/SDmode splitters): Likewise.
13717 (movsf_hardfloat): Likewise.
13718 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13719 (movsf_softfloat): Likewise.
13720 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13722 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13723 wx and wz constraints.
13725 * config/rs6000/constraints.md (wg constraint): New constraint to
13726 return FLOAT_REGS if -mmfpgpr (power6x) was used.
13728 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13731 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13732 -mdebug=reg, print wg, wl, wx, and wz constraints.
13733 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13734 Initialize the reload functions for 64-bit binary/decimal floating
13736 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13737 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13738 create the buffer on the stack to overcome not having a 32-bit
13740 (rs6000_emit_move): Likewise.
13741 (rs6000_secondary_memory_needed_rtx): Likewise.
13742 (rs6000_alloc_sdmode_stack_slot): Likewise.
13743 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13744 via xxlxor, just like DFmode 0.0.
13746 * config/rs6000/dfp.md (movdd): Delete, combine with binary
13747 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
13748 with other moves by using conditional constraits (wg). Use LFIWZX
13749 and STFIWX for loading SDmode on power7.
13750 (movdd splitters): Likewise.
13751 (movdd_hardfloat32): Likewise.
13752 (movdd_softfloat32): Likewise.
13753 (movdd_hardfloat64_mfpgpr): Likewise.
13754 (movdd_hardfloat64): Likewise.
13755 (movdd_softfloat64): Likewise.
13757 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13758 64-bit binary and decimal floating point moves.
13759 (FMOVE64X): Likewise.
13760 (movdf): Combine 64-bit binary and decimal floating point moves.
13761 Combine power6x (mfpgpr) moves with other moves by using
13762 conditional constraits (wg).
13763 (mov<mode> for DFmode/DDmode): Likewise.
13764 (DFmode/DDmode splitters): Likewise.
13765 (movdf_hardfloat32): Likewise.
13766 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13767 (movdf_softfloat32): Likewise.
13768 (movdf_hardfloat64_mfpgpr): Likewise.
13769 (movdf_hardfloat64): Likewise.
13770 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13771 (movdf_softfloat64): Likewise.
13772 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13773 (reload_<mode>_load): Move to later in the file so they aren't in
13774 the middle of the floating point move insns.
13775 (reload_<mode>_store): Likewise.
13777 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13780 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13781 constraint if -mdebug=reg.
13782 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13783 Enable using dd reload support if needed.
13785 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13786 binary and decimal floating point moves in rs6000.md.
13787 (movtd_internal): Likewise.
13789 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13790 decimal floating point moves.
13792 (movtf_internal): Likewise.
13793 (mov<mode>_internal, TDmode/TFmode): Likewise.
13794 (movtf_softfloat): Likewise.
13795 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13797 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13798 movdi_internal64, using wg constraint for move direct operations.
13799 (movdi_internal64): Likewise.
13801 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13802 MODES_TIEABLE_P for selected modes. Print the numerical value of
13803 the various virtual registers. Use GPR/FPR first/last values,
13804 instead of hard coding the register numbers. Print which modes
13805 have reload functions registered.
13806 (rs6000_option_override_internal): If -mdebug=reg, trace the options
13807 settings before/after setting cpu, target and subtarget settings.
13808 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13809 and for secondary reload failures in rs6000_secondary_reload_inner.
13810 (rs6000_secondary_reload_fail): Likewise.
13811 (rs6000_secondary_reload_inner): Likewise.
13813 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13814 macros for first/last GPR and FPR registers.
13815 (LAST_GPR_REGNO): Likewise.
13816 (FIRST_FPR_REGNO): Likewise.
13817 (LAST_FPR_REGNO): Likewise.
13819 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13820 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13821 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13822 (vcond<mode><mode>): Likewise.
13823 (vcondu<mode><mode>): Likewise.
13824 (vector_gtu<mode>): Likewise.
13825 (vector_gte<mode>): Likewise.
13826 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13827 to prevent the compiler from converting DImode operations to TImode.
13828 (ior<mode>3): Likewise.
13829 (and<mode>3): Likewise.
13830 (one_cmpl<mode>2): Likewise.
13831 (nor<mode>3): Likewise.
13832 (andc<mode>3): Likewise.
13834 * config/rs6000/constraints.md (wt constraint): New constraint
13835 that returns VSX_REGS if TImode is allowed in VSX registers.
13837 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13838 constant under VSX.
13840 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13841 similar to TImode, but it is restricted to being in the GPRs.
13843 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13844 TImode to occupy a single VSX register.
13846 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13847 -mvsx-timode for power7/power8.
13848 (power7 cpu): Likewise.
13849 (power8 cpu): Likewise.
13851 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13852 sure that TFmode/TDmode take up two registers if they are ever
13853 allowed in the upper VSX registers.
13854 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13856 (rs6000_init_hard_regno_mode_ok): Likewise.
13857 (rs6000_debug_reg_global): Add debugging for PTImode and wt
13858 constraint. Print if LRA is turned on.
13859 (rs6000_option_override_internal): Give an error if -mvsx-timode
13860 and VSX is not enabled.
13861 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
13862 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13863 to reg+offset addressing. Use PTImode when checking offset
13864 addresses for validity.
13865 (reg_offset_addressing_ok_p): Likewise.
13866 (rs6000_legitimate_offset_address_p): Likewise.
13867 (rs6000_legitimize_address): Likewise.
13868 (rs6000_legitimize_reload_address): Likewise.
13869 (rs6000_legitimate_address_p): Likewise.
13870 (rs6000_eliminate_indexed_memrefs): Likewise.
13871 (rs6000_emit_move): Likewise.
13872 (rs6000_secondary_reload): Likewise.
13873 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
13874 reloads to fpr registers to continue to use reg+offset addressing,
13875 but 64-bit reloads to altivec registers need reg+reg addressing.
13876 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13877 it. Treat LO_SUM like a PLUS operation.
13878 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13879 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13880 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13881 registers to share a register with a smaller sized type, since VSX
13882 puts scalars in the upper 64-bits.
13883 (print_operand): Add support for PTImode.
13884 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13885 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13886 registers, but don't have arithmetic support.
13887 (rs6000_memory_move_cost): Add test for VSX.
13888 (rs6000_opt_masks): Add -mvsx-timode.
13890 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13893 (VSr): Use wt constraint for TImode.
13894 (VSv): Drop TImode support.
13895 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13896 (vsx_movti_64bit): Likewise.
13897 (vsx_movti_32bit): Likewise.
13898 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13899 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13900 one '?' on the appropriate output constraint. Do not allow TImode
13901 logical operations on 32-bit systems.
13902 (vsx_ior<mode>3): Likewise.
13903 (vsx_xor<mode>3): Likewise.
13904 (vsx_one_cmpl<mode>2): Likewise.
13905 (vsx_nor<mode>3): Likewise.
13906 (vsx_andc<mode>3): Likewise.
13907 (vsx_concat_<mode>): Likewise.
13908 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13910 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13911 OPTION_MASK_VSX_TIMODE.
13912 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13913 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13915 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13916 (TI2 iterator): New iterator for TImode, PTImode.
13917 (wd mode attribute): Add values for vector types.
13918 (movti_string): Replace TI move operations with operations for TImode
13919 and PTImode. Add support for TImode being allowed in VSX registers.
13920 (mov<mode>_string, TImode/PTImode): Likewise.
13921 (movti_ppc64): Likewise.
13922 (mov<mode>_ppc64, TImode/PTImode): Likewise.
13923 (TI mode splitters): Likewise.
13925 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13928 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
13930 PR tree-optimization/56355
13931 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13932 Also handle integers with undefined overflow.
13934 2013-03-20 Catherine Moore <clm@codesourcery.com>
13935 Maciej W. Rozycki <macro@codesourcery.com>
13936 Tom de Vries <tom@codesourcery.com>
13937 Nathan Sidwell <nathan@codesourcery.com>
13938 Iain Sandoe <iain@codesourcery.com>
13939 Nathan Froyd <froydnj@codesourcery.com>
13940 Chao-ying Fu <fu@mips.com>
13942 * doc/extend.texi: (micromips, nomicromips, nocompression):
13943 Document new function attributes.
13944 * doc/invoke.texi (minterlink-compressed, mmicromips,
13945 m14k, m14ke, m14kec): Document new options.
13946 (minterlink-mips16): Update documentation.
13947 * doc/md.texi (ZC, ZD): Document new constraints.
13948 * configure.ac (gcc_cv_as_micromips): Check if linker
13949 supports the .set micromips directive.
13950 * configure: Regenerate.
13951 * config.in: Regenerate.
13952 * config/mips/mips-tables.opt: Regenerate.
13953 * config/mips/micromips.md: New file.
13954 * constraints.md (ZC, ZD): New constraints.
13955 * config/mips/predicates.md (movep_src_register): New predicate.
13956 (movep_src_operand): New predicate.
13957 (non_volatile_mem_operand): New predicate.
13958 * config/mips/mips.md (multimem): New type.
13959 (length): Differentiate between 17-bit and 18-bit branch offsets.
13960 (MOVEP1, MOVEP2): New mode iterator.
13961 (mov_<load>l): Use ZC constraint.
13962 (mov_<load>r): Likewise.
13963 (mov_<store>l): Likewise.
13964 (mov_<store>r): Likewise.
13965 (*branch_equality<mode>_inverted): Add microMIPS support.
13966 (*branch_equality<mode>): Likewise.
13967 (*jump_absolute): Likewise.
13968 (indirect_jump_<mode>): Likewise.
13969 (tablejump_<mode>): Likewise.
13970 (<optab>_internal): Likewise.
13971 (sibcall_internal): Likewise.
13972 (sibcall_value_internal): Likewise.
13973 (prefetch): Use constraint ZD.
13974 * config/mips/mips.opt (minterlink-compressed): New option.
13975 (minterlink-mips16): Now an alias for minterlink-compressed.
13976 (mmicromips): New option.
13977 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13978 (compare_and_swap_12): Likewise.
13979 (sync_add<mode>): Likewise.
13980 (sync_<optab>_12): Likewise.
13981 (sync_old_<optab>_12): Likewise.
13982 (sync_new_<optab>_12): Likewise.
13983 (sync_nand_12): Likewise.
13984 (sync_old_nand_12): Likewise.
13985 (sync_new_nand_12): Likewise.
13986 (sync_sub<mode>): Likewise.
13987 (sync_old_add<mode>): Likewise.
13988 (sync_old_sub<mode>): Likewise.
13989 (sync_new_add<mode>): Likewise.
13990 (sync_new_sub<mode>): Likewise.
13991 (sync_<optab><mode>): Likewise.
13992 (sync_old_<optab><mode>): Likewise.
13993 (sync_new_<optab><mode>): Likewise.
13994 (sync_nand<mode>): Likewise.
13995 (sync_old_nand<mode>): Likewise.
13996 (sync_new_nand<mode>): Likewise.
13997 (sync_lock_test_and_set<mode>): Likewise.
13998 (test_and_set_12): Likewise.
13999 (atomic_compare_and_swap<mode>): Likewise.
14000 (atomic_exchange<mode>_llsc): Likewise.
14001 (atomic_fetch_add<mode>_llsc): Likewise.
14002 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14003 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14004 (umips_save_restore_pattern_p): Likewise.
14005 (umips_load_store_pair_p): Likewise.
14006 (umips_output_load_store_pair): Likewise.
14007 (umips_movep_target_p): Likewise.
14008 (umips_12bit_offset_address_p): Likewise.
14009 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14010 (mips_base_mips16): Rename this...
14011 (mips_base_compression_flags): ...to this. Update all uses.
14012 (mips_attribute_table): Add micromips, nomicromips and nocompression.
14013 (mips_mips16_decl_p): Delete.
14014 (mips_nomips16_decl_p): Delete.
14015 (mips_get_compress_on_flags): New function.
14016 (mips_get_compress_off_flags): New function.
14017 (mips_get_compress_mode): New function.
14018 (mips_get_compress_on_name): New function.
14019 (mips_get_compress_off_name): New function.
14020 (mips_insert_attributes): Support multiple compression types.
14021 (mips_merge_decl_attributes): Likewise.
14022 (umips_12bit_offset_address_p): New function.
14023 (mips_start_function_definition): Emit .set micromips directive.
14024 (mips_call_may_need_jalx_p): New function.
14025 (mips_function_ok_for_sibcall): Add microMIPS support.
14026 (mips_print_operand_punctuation): Support short delay slots and
14028 (umips_swm_mask, umips_swm_encoding): New.
14029 (umips_build_save_restore): New function.
14030 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14031 (was_mips16_p): Remove.
14032 (old_compression_mode): New.
14033 (mips_set_compression_mode): New function.
14034 (mips_set_current_function): Add microMIPS support.
14035 (mips_option_override): Likewise.
14036 (umips_save_restore_pattern_p): New function.
14037 (umips_output_save_restore): New function.
14038 (umips_load_store_pair_p_1): New function.
14039 (umips_load_store_pair_p): New function.
14040 (umips_output_load_store_pair_1): New function.
14041 (umips_output_load_store_pair): New function.
14042 (umips_movep_target_p) New function.
14043 (mips_prepare_pch_save): Add microMIPS support.
14044 * config/mips/mips.h (TARGET_COMPRESSION): New.
14045 (TARGET_CPU_CPP_BUILTINS): Update macro
14046 to use new compression flags and to support microMIPS.
14047 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14048 (MIPS_ARCH_FLOAT_SPEC): Likewise.
14049 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14050 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14051 (ASM_SPEC): Support mmicromips and mno-micromips.
14052 (M16STORE_REG_P): New macro.
14053 (MIPS_CALL): Support TARGET_MICROMIPS.
14054 (MICROMIPS_J): New macro.
14055 (mips_base_mips16): Rename this...
14056 (mips_base_compression_flags): ...to this.
14057 (UMIPS_12BIT_OFFSET_P): New macro.
14058 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14059 (MULTILIB_DIRNAMES): Likewise.
14060 2013-03-20 Richard Biener <rguenther@suse.de>
14062 PR tree-optimization/56661
14063 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14064 the result does not have to be distinct.
14066 2013-03-20 Richard Biener <rguenther@suse.de>
14068 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14071 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14072 Steven Bosscher <steven@gcc.gnu.org>
14074 PR rtl-optimization/56605
14075 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14077 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14080 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14081 that require movd instead of movq.
14083 2013-03-20 Richard Biener <rguenther@suse.de>
14085 * tree-ssa-structalias.c (struct variable_info): Add pointer
14086 to the first field of an aggregate with sub-vars. Make
14087 this and the pointer to the next subfield its ID.
14088 (vi_next): New function.
14089 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14090 storedanything_id, integer_id): Increment by one.
14091 (new_var_info, get_call_vi, lookup_call_clobber_vi,
14092 get_call_clobber_vi): Adjust.
14093 (solution_set_expand): Simplify and speedup.
14094 (solution_set_add): Inline into ...
14095 (set_union_with_increment): ... this. Adjust accordingly.
14096 (do_sd_constraint): Likewise.
14097 (do_ds_constraint): Likewise.
14098 (do_complex_constraint): Simplify.
14099 (build_pred_graph): Adjust.
14100 (solve_graph): Likewise. Simplify and speedup.
14101 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14102 get_constraint_for_component_ref, get_constraint_for_1,
14103 first_vi_for_offset, first_or_preceding_vi_for_offset,
14104 create_function_info_for, create_variable_info_for_1,
14105 create_variable_info_for, intra_create_variable_infos): Adjust.
14106 (init_base_vars): Push NULL for ID zero.
14107 (compute_points_to_sets): Adjust.
14109 2013-03-20 Richard Biener <rguenther@suse.de>
14111 * cfgloop.c (verify_loop_structure): Streamline and avoid
14112 ICEing on corrupt loop tree.
14113 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14116 2013-03-20 Richard Biener <rguenther@suse.de>
14118 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14119 check whether an SSA update is needed.
14121 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
14123 * config/mips/constraints.md (T): Rename to...
14127 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14128 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14130 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14132 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14133 (*subsi3_carryin_uxtw): Likewise.
14135 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14137 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14138 (*rorsi3_insn_uxtw): Likewise.
14140 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14142 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14143 (*extrsi5_insn_uxtw): Likewise.
14145 2013-03-19 Richard Biener <rguenther@suse.de>
14147 PR tree-optimization/56273
14148 * passes.c (init_optimization_passes): Move second VRP after DOM.
14150 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
14152 * config/i386/i386.md (*movti_internal): Merge from
14153 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
14154 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
14155 nox64 isa attributes.
14157 2013-03-18 Richard Biener <rguenther@suse.de>
14159 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14161 (merge_node_constraints): Likewise.
14162 (build_succ_graph): Likewise.
14163 (valid_graph_edge): Inline into single caller.
14164 (unify_nodes): Likewise. Use bitmap_set_bit return value
14166 (scc_visit): Fix formatting and variable use.
14167 (do_sd_constraint): Use gcc_checking_assert.
14168 (do_ds_constraint): Likewise.
14169 (do_complex_constraint): Likewise.
14170 (condense_visit): Likewise. Cleanup.
14171 (dump_pred_graph): New function.
14172 (perform_var_substitution): Dump the pred-graph before
14173 variable substitution.
14174 (find_equivalent_node): Use gcc_checking_assert.
14175 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14177 2013-03-18 Richard Biener <rguenther@suse.de>
14179 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14180 Remove cond_expr_stmt_list argument and do not gimplify the
14182 (vect_loop_versioning): Adjust.
14183 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14184 Cleanup to use less temporaries.
14185 (vect_create_data_ref_ptr): Cleanup.
14187 2013-03-18 Jakub Jelinek <jakub@redhat.com>
14189 PR tree-optimization/56635
14190 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14191 require types_compatible_p types.
14193 2013-03-18 Nick Clifton <nickc@redhat.com>
14195 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14196 spurious backslash.
14198 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14199 Add missing line to comment describing function.
14201 2013-03-18 Richard Biener <rguenther@suse.de>
14203 PR tree-optimization/56210
14204 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14205 Handle string / character search functions.
14206 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14208 2013-03-18 Richard Biener <rguenther@suse.de>
14210 PR middle-end/56483
14211 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14212 and implement properly.
14213 * gimple.h (gimple_cond_single_var_p): Remove.
14215 2013-03-18 Richard Biener <rguenther@suse.de>
14217 * tree-data-ref.h (find_data_references_in_loop): Declare.
14218 * tree-data-ref.c (get_references_in_stmt): Use a stack
14219 vector pre-allocated in the callers.
14220 (find_data_references_in_stmt): Adjust.
14221 (graphite_find_data_references_in_stmt): Likewise.
14222 (create_rdg_vertices): Likewise.
14223 (find_data_references_in_loop): Export.
14224 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14225 Compute dependences here...
14226 (vect_analyze_data_refs): ...not here. When we encounter
14227 a non-vectorizable data reference in basic-block vectorization
14228 truncate the data reference vector. Do not bother to
14229 fixup data-dependence information for gather loads.
14230 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14231 of data references, as reported.
14233 2013-03-18 Richard Biener <rguenther@suse.de>
14235 PR tree-optimization/3713
14236 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
14237 has_constants and expr.
14238 (stmt_has_constants): Properly valueize SSA names when deciding
14239 whether the stmt has constants.
14241 2013-03-18 Richard Biener <rguenther@suse.de>
14243 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14244 whole function when there is nothing to do.
14245 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14246 * tree-vectorizer.c (vectorize_loops): Update virtual and
14247 loop-closed SSA once.
14248 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14250 2013-03-18 Richard Biener <rguenther@suse.de>
14252 PR middle-end/56113
14253 * domwalk.c (bb_postorder): New global static.
14254 (cmp_bb_postorder): New function.
14255 (walk_dominator_tree): Replace scheme imposing an order for
14256 visiting dominator sons by one sorting them at the time they
14257 are pushed on the stack.
14259 2013-03-18 Richard Biener <rguenther@suse.de>
14261 PR tree-optimization/39326
14262 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14263 (struct mem_ref): Replace mem member with ao_ref typed member.
14264 (MEM_ANALYZABLE): Adjust.
14265 (memref_eq): Likewise.
14266 (mem_ref_alloc): Likewise.
14267 (gather_mem_refs_stmt): Likewise.
14268 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14269 (execute_sm_if_changed_flag_set): Adjust.
14270 (execute_sm): Likewise.
14271 (ref_always_accessed_p): Likewise.
14272 (refs_independent_p): Likewise.
14273 (can_sm_ref_p): Likewise.
14275 2013-03-18 Jakub Jelinek <jakub@redhat.com>
14278 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14279 return 1 even for !unsignedp.
14281 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14283 * config/i386/i386.md (isa): Add x64 and nox64.
14284 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14285 (*pushtf): Enable *roF alternative for x64 isa only.
14286 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
14287 mode attribute of integer alternatives to DImode for TARGET_64BIT.
14288 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
14289 (*movtf_internal): Merge from *movtf_internal_rex64 and
14290 *movtf_internal_sse. Use x64 and nox64 isa attributes.
14291 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
14292 nox64 isa attributes.
14293 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
14294 nox64 isa attributes.
14295 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14297 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14299 * config/alpha/alpha.c (TARGET_LRA_P): New define.
14301 2013-03-17 Jakub Jelinek <jakub@redhat.com>
14304 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14305 class names. Remove trailing comma after "ALL_REGS".
14307 2013-03-16 Jan Hubicka <jh@suse.cz>
14309 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14310 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14311 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14312 of cgraph_get_create_node.
14313 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14315 2013-03-16 Jason Merrill <jason@redhat.com>
14318 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14319 with DW_AT_default_value.
14321 2013-03-16 Jakub Jelinek <jakub@redhat.com>
14323 * BASE-VER: Set to 4.9.0.
14325 2013-03-14 Andi Kleen <ak@linux.intel.com>
14328 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14329 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14330 Document _x* TSX intrinsics.
14332 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
14333 David Holsgrove <david.holsgrove@xilinx.com>
14335 * configure.ac: Add MicroBlaze TLS support detection.
14336 * configure: Regenerate.
14337 * config/microblaze/microblaze-protos.h
14338 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14339 symbol_mentioned_p, label_mentioned_p): Add prototypes.
14340 * config/microblaze/microblaze.c (microblaze_address_type): Add
14341 ADDRESS_TLS and tls_reloc address types.
14342 (microblaze_address_info): Add tls_reloc.
14343 (TARGET_HAVE_TLS): Define.
14344 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14345 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14346 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14347 load_tls_operand, microblaze_call_tls_get_addr,
14348 microblaze_legitimize_tls_address): New functions.
14349 (microblaze_classify_unspec): Handle UNSPEC_TLS.
14350 (get_base_reg): Use microblaze_tls_symbol_p.
14351 (microblaze_classify_address): Handle TLS.
14352 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14353 label_mentioned_p and microblaze_tls_referenced_p.
14354 (microblaze_legitimize_address): Handle TLS.
14355 (microblaze_address_insns): Handle ADDRESS_TLS.
14356 (pic_address_needs_scratch): Handle TLS.
14357 (print_operand_address): Handle TLS.
14358 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14359 (microblaze_expand_move): Handle TLS.
14360 (microblaze_legitimate_constant_p): Check
14361 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14362 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14363 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14364 (PIC_OFFSET_TABLE_REGNUM): Set.
14365 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14366 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14367 (addsi3, movsi_internal2, movdf_internal): Update constraints
14368 * config/microblaze/predicates.md (arith_plus_operand): Define
14369 (move_operand): Redefine as move_src_operand,
14370 check microblaze_tls_referenced_p.
14372 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14374 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14375 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14377 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14379 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14382 2013-03-14 Jakub Jelinek <jakub@redhat.com>
14384 PR tree-optimization/53265
14385 * common.opt (Waggressive-loop-optimizations): New option.
14386 * tree-ssa-loop-niter.c: Include tree-pass.h.
14387 (do_warn_aggressive_loop_optimizations): New function.
14388 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
14389 if number_of_latch_executions returned constant.
14390 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14391 early. If number_of_latch_executions returned constant, set
14392 nb_iterations_upper_bound back to it.
14393 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14395 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14396 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14398 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14399 (MULTILIB_OSDIRNAMES): Set.
14400 * genmultilib: If defaultosdirname doesn't start with :: , set
14401 defaultosdirname2 instead, clear it and emit two . multilib_raw
14402 entries instead of just one.
14404 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
14406 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14407 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14408 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14409 (SUBTARGET_OVERRIDE_OPTIONS): New.
14411 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
14414 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14415 (musermode): Convert to Var(TARGET_USERMODE).
14416 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14417 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14418 * config/sh/sh.c (sh_option_override): Use
14419 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14420 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14422 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14424 (udivsi3_i4_single, divsi3_i4_single): Use
14425 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14427 2013-03-13 Dave Korn <dave.korn.cygwin@....>
14429 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14432 2013-03-13 Richard Biener <rguenther@suse.de>
14434 PR tree-optimization/56608
14435 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14436 calls when vectorizing basic-blocks.
14438 2013-03-13 Jakub Jelinek <jakub@redhat.com>
14441 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14444 2013-03-12 Jakub Jelinek <jakub@redhat.com>
14446 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14448 2013-03-11 Jan Hubicka <jh@suse.cz>
14451 * lto-streamer-out.c (output_symbol_p): Skip references from
14452 constructors of external variables.
14454 2013-03-11 Jan Hubicka <jh@suse.cz>
14456 PR middle-end/56571
14457 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14459 * emit-rtl.c (verify_rtx_sharing): Likewise.
14460 (copy_insn_1): Likewise.
14461 * rtl.c (copy_rtx): Likewise.
14463 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
14466 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14467 output_operand_lossage message.
14469 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
14472 * arm.c (shift_op): Validate RTL pattern on the fly.
14473 (arm_print_operand, case 'S'): Don't use shift_operator to validate
14476 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
14479 * config/pa/pa.md (call_value): Check for calls to powf and direct to
14480 new call patterns that clobber %fr12.
14481 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14482 split and postreload patterns.
14483 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14484 registers %fr12 and %fr12R as call used.
14486 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
14488 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14489 (canon_address, record_store, replace_read, check_mem_read_rtx,
14490 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14491 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14492 rest_of_handle_dse): Likewise.
14494 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
14496 PR middle-end/56524
14497 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14499 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14500 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14501 (save_optabs_if_changed): Replace with...
14502 (init_tree_optimization_optabs): ...this.
14503 * optabs.c (save_optabs_if_changed): Rename to...
14504 (init_tree_optimization_optabs): ...this. Take the optimization node
14505 as argument. Do nothing if the base optabs are already correct.
14506 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14507 to recompute optabs.
14508 * function.h (function): Remove optabs field.
14509 * function.c (invoke_set_current_function_hook): Call
14510 init_tree_optimization_optabs. Use the result to initialize
14513 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
14515 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14516 if GTMA_HAS_NO_INSTRUMENTATION.
14517 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14518 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14519 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14520 * gimple-pretty-print.c (dump_gimple_transaction): Handle
14521 GTMA_HAS_NO_INSTRUMENTATION.
14523 2013-03-08 Jakub Jelinek <jakub@redhat.com>
14525 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14528 2013-03-08 Marek Polacek <polacek@redhat.com>
14529 Jakub Jelinek <jakub@redhat.com>
14531 PR tree-optimization/56478
14532 * predict.c (is_comparison_with_loop_invariant_p): Change the
14533 type of loop_step to tree.
14534 (predict_loops): Adjust.
14535 (predict_iv_comparison): Perform the computations on double_ints.
14537 2013-03-08 Richard Biener <rguenther@suse.de>
14539 PR tree-optimization/56570
14540 * tree-cfg.c (verify_expr_location_1): Verify locations for
14542 * tree-sra.c (create_access_replacement): Strip locations
14543 from DECL_DEBUG_EXPRs.
14545 2013-03-08 Richard Biener <rguenther@suse.de>
14547 * tree-inline.c (expand_call_inline): Do not associate
14548 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14549 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14551 2013-03-08 Richard Biener <rguenther@suse.de>
14553 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14554 or block changes with -Og. Fix for location / block encoding
14555 changes and PHI arguments with locations.
14557 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
14559 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14561 (struct output_info): Likewise.
14562 (register_overhead): Remove bad gcc_assert.
14563 (bitmap_find_bit): If there is only a single bitmap element, do not
14564 count a miss as a search.
14565 (print_statistics): Update for counter type changes.
14566 (dump_bitmap_statistics): Likewise. Print headers such that they
14567 are properly lined up with the printed counters.
14569 2013-03-07 Jakub Jelinek <jakub@redhat.com>
14571 PR tree-optimization/56559
14572 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14573 check that it has only a single use.
14575 2013-03-07 Richard Biener <rguenther@suse.de>
14577 * doc/invoke.texi (fwhole-program): Discourage use in combination
14580 2013-03-06 Jakub Jelinek <jakub@redhat.com>
14582 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14584 PR tree-optimization/56539
14585 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14586 instead of GSI_CONTINUE_LINKING as last argument to
14587 force_gimple_operand_gsi. Adjust function comment.
14589 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14592 PR middle-end/56548
14593 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14594 promoted mode, convert the result back to the original mode.
14596 2013-03-06 Richard Biener <rguenther@suse.de>
14598 PR middle-end/56294
14599 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14600 (insert_updated_phi_nodes_compare_uids): New function.
14601 (update_ssa): Sort symbols_to_rename after UID before
14602 traversing it to insert PHI nodes.
14604 2013-03-06 Richard Biener <rguenther@suse.de>
14606 PR middle-end/50494
14607 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14608 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14611 2013-02-13 Richard Biener <rguenther@suse.de>
14614 * varasm.c (output_constant_def_1): Get the decl representing
14615 the constant as argument.
14616 (output_constant_def): Wrap output_constant_def_1.
14617 (make_decl_rtl): Use output_constant_def_1 with the decl
14618 representing the constant.
14619 (build_constant_desc): Optionally re-use a decl already
14620 representing the constant.
14621 (tree_output_constant_def): Adjust.
14623 2013-03-06 Joey Ye <joey.ye@arm.com>
14626 * gcc.c (convert_white_space): New function.
14627 (main): Handles white space in function name.
14629 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
14632 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14633 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
14634 to SH_DIV_CALL_TABLE for TARGET_SH2.
14635 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14637 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14638 call-table options.
14640 2013-03-05 Sterling Augustine <saugustine@google.com>
14641 Cary Coutant <ccoutant@google.com>
14644 * dwarf2out.c (resolve_addr): Don't call
14645 remove_loc_list_addr_table_entries a second time for the same
14648 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14651 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14652 (avoid_complex_debug_insns): New function.
14653 (expand_debug_locations): Call it.
14655 PR rtl-optimization/56484
14656 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14657 lifetimes of hard registers on small register class machines.
14659 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
14661 * config/microblaze/microblaze-protos.h: Rename
14662 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14663 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14665 (microblaze_fast_interrupt_function_p): New function.
14666 (microblaze_is_interrupt_handler): Rename to
14667 microblaze_is_interrupt_variant and add fast_interrupt check.
14668 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14669 (save_restore_insns): Likewise.
14670 (compute_frame_size): Likewise.
14671 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14672 (microblaze_globalize_label): Likewise.
14673 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14674 * config/microblaze/microblaze.md: Use wrapper
14675 microblaze_is_interrupt_variant.
14677 2013-03-05 Kai Tietz <ktietz@redhat.com>
14679 * sdbout.c (sdbout_one_type): Switch to current function's section
14680 supporting cold/hot.
14682 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
14684 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14687 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14689 PR middle-end/56461
14690 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14691 if VALGRIND_GET_VBITS is defined, temporarily make object
14692 memory all defined, and restore previous valgrind addressability
14693 and definability afterwards. Free this_object at the end.
14695 PR middle-end/56461
14696 * lra.c (lra): Call lra_clear_live_ranges if live_p,
14697 right before calling lra_create_live_ranges, also call it
14698 when clearing live_p. Only call lra_clear_live_ranges
14699 at the end if live_p.
14701 PR middle-end/56461
14702 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14704 2013-03-05 Richard Biener <rguenther@suse.de>
14706 PR tree-optimization/56521
14707 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14710 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
14713 * except.h (remove_unreachable_eh_regions): New prototype.
14714 * except.c (remove_eh_handler_splicer): New function, split out
14715 of remove_eh_handler.
14716 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
14717 warning about running it on many EH regions one at a time.
14718 (remove_unreachable_eh_regions_worker): New function, walk the
14719 EH tree in depth-first order and remove non-marked regions.
14720 (remove_unreachable_eh_regions): New function.
14721 * tree-eh.c (mark_reachable_handlers): New function, split out
14722 from remove_unreachable_handlers.
14723 (remove_unreachable_handlers): Use mark_reachable_handlers and
14724 remove_unreachable_eh_regions.
14725 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14726 and remove_unreachable_eh_regions.
14728 2013-03-05 Richard Biener <rguenther@suse.de>
14730 PR middle-end/56525
14731 * loop-init.c (fix_loop_structure): Remove loops in two stages,
14732 not freeing them until the end.
14734 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14736 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14738 2013-03-05 Richard Biener <rguenther@suse.de>
14740 PR tree-optimization/56270
14741 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14742 of loads after scheduling an SLP instance.
14744 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14746 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14748 (check_gcc_parallelize): Run guality.exp as a separate job from
14749 vect.exp with unsorted.exp and $(dg_target_exps) separately from
14750 struct-layout-1.exp with stackalign.exp.
14752 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14754 PR middle-end/56461
14755 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14756 load_index sbitmap even if some bit in it isn't set.
14758 PR middle-end/56461
14759 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14760 (discover_iteration_bound_by_body_walk): Change queues to
14761 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
14762 spelling in comment. Call safe_push on queues[bound_index] directly.
14763 Release queues[queue_index] in every iteration unconditionally.
14764 Release bounds vector.
14766 PR middle-end/56461
14767 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14768 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14769 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14772 2013-03-05 Richard Biener <rguenther@suse.de>
14775 * tree-inline.c (remap_blocks_to_null): New function.
14776 (expand_call_inline): When expanding a call stmt without
14777 an associated block inline remap all callee blocks to NULL.
14779 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14781 PR rtl-optimization/56494
14782 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14783 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14784 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14786 PR middle-end/56461
14787 * sel-sched-ir.c (free_sched_pools): Release
14788 succs_info_pool.stack[succs_info_pool.max_top] vectors too
14789 if succs_info_pool.max_top isn't -1.
14792 * opts.c (opts_obstack, opts_concat): Moved to...
14793 * opts-common.c (opts_obstack, opts_concat): ... here.
14795 2013-03-04 Jakub Jelinek <jakub@redhat.com>
14797 PR middle-end/56461
14798 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14800 2013-03-04 Martin Jambor <mjambor@suse.cz>
14802 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14803 all appropriate places.
14805 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
14807 PR tree-optimization/56424
14808 * ipa-split.c (split_function): Do not set the RSO flag if result is
14809 not by reference and its type is a register type.
14811 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
14813 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14814 (microblaze_legitimate_pic_operand): Likewise
14815 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14816 new function microblaze_legitimate_pic_operand
14817 * config/microblaze/microblaze-protos.h
14818 (microblaze_legitimate_pic_operand): Declare.
14820 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14822 * config/microblaze/predicates.md (call_insn_simple_operand):
14823 New predicate for supported rtx code types.
14824 * config/microblaze/microblaze.md (call_internal1): Use
14825 call_insn_simple_operand predicate.
14827 2013-03-04 Jakub Jelinek <jakub@redhat.com>
14829 PR middle-end/56461
14830 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14831 partitions.ordered_remove.
14833 PR middle-end/56461
14834 * tree-vect-stmts.c (vectorizable_conversion): Don't call
14835 vec_oprnds0.create (1) for modifier == NONE.
14837 PR middle-end/56461
14838 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14839 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14840 vec_oprnds1 right before pushing anything to it for
14843 PR middle-end/56461
14844 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14845 set nbbs to 0 instead of having separate code path.
14846 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14847 instead of false as last argument if returning NULL.
14849 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
14851 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14852 the attribute is now called "target" instead of "option".
14853 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14854 * doc/tm.texi.in (Target Attributes): Likewise document the correct
14855 attribute/pragma name for TARGET_OPTION_VALID_P and
14856 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
14857 * doc/tm.texi: Regenerated.
14859 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
14861 * config/microblaze/microblaze.c:
14862 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14863 * config/microblaze/microblaze.h: Add -mxl-reorder to
14865 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14866 instructions emitted if TARGET_REORDER.
14867 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14868 or 0 for -m/-mno case, but initialises as 2 to detect default use case
14871 2013-03-01 Xinliang David Li <davidxl@google.com>
14873 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14876 2013-03-01 Jakub Jelinek <jakub@redhat.com>
14878 PR middle-end/56461
14879 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14880 vector even when returning true. Fix up function comment formatting.
14882 PR middle-end/56461
14883 * ira-build.c (ira_loop_nodes_count): New variable.
14884 (create_loop_tree_nodes): Initialize it.
14885 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14887 PR middle-end/56461
14888 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14889 method on dr_chain and result_chain.
14890 * tree-vect-stmts.c (vectorizable_store): Only call
14891 result_chain.create if j == 0.
14893 PR middle-end/56461
14894 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14895 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14896 before overwriting it.
14898 2013-03-01 Tobias Burnus <burnus@net-b.de>
14900 * doc/extended.texi (C Extensions): Change order in @menu
14902 (Other MIPS Built-in Functions): Move last MIPS entry before
14903 "picoChip Built-in Functions".
14904 (SH Built-in Functions): Move after RX Built-in Functions.
14905 * doc/gcc.texi (Introduction): Change order in @menu
14907 * doc/md.texi (Constraints): Ditto.
14908 * gty.texi (Type Information): Ditto.
14909 (User-provided marking routines for template types): Make
14911 * doc/invoke.texi (AArch64 Options): Move before
14912 "Adapteva Epiphany Options".
14914 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
14915 Jakub Jelinek <jakub@redhat.com>
14918 * asan.c (gate_asan): Lookup no_sanitize_address instead of
14919 no_address_safety_analysis attribute.
14920 * doc/extend.texi (no_address_safety_attribute): Rename to
14921 no_sanitize_address attribute, mention no_address_safety_analysis
14922 attribute as deprecated alias.
14924 2013-02-28 Jakub Jelinek <jakub@redhat.com>
14926 PR middle-end/56461
14927 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14928 type to vec<vec<tree> > *.
14929 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
14930 to be vec<tree> instead of vec<tree> *, set vec_defs
14931 to vNULL and call vec_defs.create (number_of_vects), adjust other
14933 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14934 vectorizable_condition): Adjust vect_get_slp_defs callers.
14936 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14938 * config/aarch64/aarch64.c
14939 (aarch64_float_const_representable): Remove unused variable.
14941 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14943 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14945 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14947 * config/aarch64/aarch64-builtins.c
14948 (aarch64_init_simd_builtins): Make static.
14950 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14952 * config/aarch64/aarch64.c
14953 (aarch64_simd_make_constant): Make static.
14955 2013-02-28 Martin Jambor <mjambor@suse.cz>
14957 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14958 with no initialization to the RHS of debug statements.
14960 2013-02-28 Martin Jambor <mjambor@suse.cz>
14962 PR tree-optimization/56294
14963 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14965 (get_access_replacement): Do not call create_access_replacement.
14966 Assert a replacement exists.
14967 (get_repl_default_def_ssa_name): Create the replacement declaration
14970 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14972 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14973 final_end_function.
14975 2013-02-28 Marek Polacek <polacek@redhat.com>
14977 PR rtl-optimization/56466
14978 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14979 if we're changing a loop.
14980 (peel_loops_completely): Likewise.
14982 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
14985 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14987 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
14990 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14991 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14992 INTX_FTYPE_FX, FX_FTYPE_INTX.
14993 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14995 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
14997 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14998 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14999 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15000 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15001 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15002 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15003 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15004 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15005 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15006 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15007 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15008 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15009 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15010 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15011 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15012 (avrxmega6): Increase max flash segments from 5 to 6.
15013 * config/avr/t-multilib: Regenerate.
15014 * config/avr/avr-tables.opt: Regenerate.
15015 * doc/avr-mmcu.texi: Regenerate.
15017 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15019 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15020 (avr_device_to_arch): Rename to avr_device_to_ld.
15021 (avr_device_to_as): New prototype.
15022 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15023 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15024 * config/avr/driver-avr.c (avr_device_to_as): New.
15025 (avr_device_to_arch): Rename to avr_device_to_ld.
15027 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15029 PR middle-end/56461
15030 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15031 method on dr_chain and result_chain.
15033 PR middle-end/56461
15034 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15035 pointer_set_destroy on not_executed_last_iteration.
15037 PR middle-end/56461
15038 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15040 PR middle-end/56461
15041 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15042 FOR_EACH_DEFINED_FUNCTION when freeing state.
15044 PR middle-end/56461
15045 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15047 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15050 PR middle-end/56461
15051 * ipa-cp.c (decide_whether_version_node): Call vec_free on
15052 known_aggs[i].items and release known_aggs vector.
15054 PR middle-end/56461
15055 * ipa-reference.c (propagate): Free node_info even for alias nodes.
15057 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15059 * config/microblaze/microblaze.c (microblaze_emit_compare):
15060 Use xor for EQ/NE comparisions.
15061 * config/microblaze/microblaze.md (cstoresf4): Add constraints
15062 (cbranchsf4): Adjust operator to comparison_operator.
15064 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15066 PR middle-end/56461
15067 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15069 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15070 vec_safe_push, always update *slot.
15071 (redirect_edge_var_map_clear): Use vec_free.
15072 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15073 (free_var_map_entry): Use vec_free.
15074 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15075 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15077 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
15079 PR middle-end/45472
15080 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15081 when the may_trap_p bit of the exprs being merged differs.
15082 Reorder tests for speculativeness in the logical and operator.
15084 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15086 * incpath.c (add_standard_paths): Use reconcat instead of concat
15087 where appropriate and avoid leaking memory.
15089 * opts.h: Include obstack.h.
15090 (opts_concat): New prototype.
15091 (opts_obstack): New declaration.
15092 * opts.c (opts_concat): New function.
15093 (opts_obstack): New variable.
15094 (init_options_struct): Call gcc_init_obstack on opts_obstack.
15095 (finish_options): Use opts_concat instead of concat
15096 and XOBNEWVEC instead of XNEWVEC.
15097 * opts-common.c (generate_canonical_option, decode_cmdline_option,
15098 generate_option): Likewise.
15099 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15100 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15103 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15104 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15106 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15108 PR middle-end/56461
15109 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15111 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15113 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15114 (arm_block_move_unaligned_straight): Likewise.
15115 (arm_adjust_block_mem): Likewise.
15117 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15120 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15121 temp, cond and label.
15122 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15125 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15126 * config/c6x/c6x.h (dbx_register_map): Update declaration.
15129 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15130 of prologue/epilogue functions.
15133 * config/tilegx/tilegx.c (tilegx_expand_prologue):
15134 Remove unused variable cfa_offset.
15135 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15138 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15139 type promotion to unsigned.
15142 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15143 for HOST_WIDE_INT of 32 bit / same size as int.
15144 (arm_block_move_unaligned_straight): Likewise.
15145 (arm_adjust_block_mem): Likewise.
15148 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15151 2013-02-26 Marek Polacek <polacek@redhat.com>
15153 PR tree-optimization/56426
15154 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15156 2013-02-26 Richard Biener <rguenther@suse.de>
15159 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15160 unused variable loops.
15162 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15164 PR tree-optimization/56448
15165 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15166 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15167 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15168 later operands of the references, or even first operand for
15169 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15171 PR tree-optimization/56443
15172 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15173 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15174 to type_for_mode langhook.
15176 2013-02-25 Matt Turner <mattst88@gmail.com>
15178 * doc/invoke.texi: Document r4700.
15180 2013-02-25 Richard Biener <rguenther@suse.de>
15182 PR tree-optimization/56175
15183 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15185 (simplify_bitwise_binary): ... here. Also guard the conversion
15186 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15188 2013-02-25 Catherine Moore <clm@codesourcery.com>
15191 2013-02-24 Catherine Moore <clm@codesourcery.com>
15192 Maciej W. Rozycki <macro@codesourcery.com>
15193 Tom de Vries <tom@codesourcery.com>
15194 Nathan Sidwell <nathan@codesourcery.com>
15195 Iain Sandoe <iain@codesourcery.com>
15196 Nathan Froyd <froydnj@codesourcery.com>
15197 Chao-ying Fu <fu@mips.com>
15199 * doc/extend.texi: (micromips, nomicromips, nocompression):
15200 Document new function attributes.
15201 * doc/invoke.texi (minterlink-compressed, mmicromips,
15202 m14k, m14ke, m14kec): Document new options.
15203 (minterlink-mips16): Update documentation.
15204 * doc/md.texi (ZC, ZD): Document new constraints.
15205 * configure.ac (gcc_cv_as_micromips): Check if linker
15206 supports the .set micromips directive.
15207 * configure: Regenerate.
15208 * config.in: Regenerate.
15209 * config/mips/mips-tables.opt: Regenerate.
15210 * config/mips/micromips.md: New file.
15211 * constraints.md (ZC, AD): New constraints.
15212 * config/mips/predicates.md (movep_src_register): New predicate.
15213 (movep_src_operand): New predicate.
15214 (non_volatile_mem_operand): New predicate.
15215 * config/mips/mips.md (multimem): New type.
15216 (length): Differentiate between 17-bit and 18-bit branch offsets.
15217 (MOVEP1, MOVEP2): New mode iterator.
15218 (mov_<load>l): Use ZC constraint.
15219 (mov_<load>r): Likewise.
15220 (mov_<store>l): Likewise.
15221 (mov_<store>r): Likewise.
15222 (*branch_equality<mode>_inverted): Add microMIPS support.
15223 (*branch_equality<mode>): Likewise.
15224 (*jump_absolute): Likewise.
15225 (indirect_jump_<mode>): Likewise.
15226 (tablejump_<mode>): Likewise.
15227 (<optab>_internal): Likewise.
15228 (sibcall_internal): Likewise.
15229 (sibcall_value_internal): Likewise.
15230 (prefetch): Use constraint ZD.
15231 * config/mips/mips.opt (minterlink-compressed): New option.
15232 (minterlink-mips16): Now an alias for minterlink-compressed.
15233 (mmicromips): New option.
15234 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15235 (compare_and_swap_12): Likewise.
15236 (sync_add<mode>): Likewise.
15237 (sync_<optab>_12): Likewise.
15238 (sync_old_<optab>_12): Likewise.
15239 (sync_new_<optab>_12): Likewise.
15240 (sync_nand_12): Likewise.
15241 (sync_old_nand_12): Likewise.
15242 (sync_new_nand_12): Likewise.
15243 (sync_sub<mode>): Likewise.
15244 (sync_old_add<mode>): Likewise.
15245 (sync_old_sub<mode>): Likewise.
15246 (sync_new_add<mode>): Likewise.
15247 (sync_new_sub<mode>): Likewise.
15248 (sync_<optab><mode>): Likewise.
15249 (sync_old_<optab><mode>): Likewise.
15250 (sync_new_<optab><mode>): Likewise.
15251 (sync_nand<mode>): Likewise.
15252 (sync_old_nand<mode>): Likewise.
15253 (sync_new_nand<mode>): Likewise.
15254 (sync_lock_test_and_set<mode>): Likewise.
15255 (test_and_set_12): Likewise.
15256 (atomic_compare_and_swap<mode>): Likewise.
15257 (atomic_exchange<mode>_llsc): Likewise.
15258 (atomic_fetch_add<mode>_llsc): Likewise.
15259 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15260 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15261 (umips_save_restore_pattern_p): Likewise.
15262 (umips_load_store_pair_p): Likewise.
15263 (umips_output_load_store_pair): Likewise.
15264 (umips_movep_target_p): Likewise.
15265 (umips_12bit_offset_address_p): Likewise.
15266 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15267 (mips_base_mips16): Rename this...
15268 (mips_base_compression_flags): ...to this. Update all uses.
15269 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15270 (mips_mips16_decl_p): Delete.
15271 (mips_nomips16_decl_p): Delete.
15272 (mips_get_compress_on_flags): New function.
15273 (mips_get_compress_off_flags): New function.
15274 (mips_get_compress_mode): New function.
15275 (mips_get_compress_on_name): New function.
15276 (mips_get_compress_off_name): New function.
15277 (mips_insert_attributes): Support multiple compression types.
15278 (mips_merge_decl_attributes): Likewise.
15279 (umips_12bit_offset_address_p): New function.
15280 (mips_start_function_definition): Emit .set micromips directive.
15281 (mips_call_may_need_jalx_p): New function.
15282 (mips_function_ok_for_sibcall): Add microMIPS support.
15283 (mips_print_operand_punctuation): Support short delay slots and
15285 (umips_swm_mask, umips_swm_encoding): New.
15286 (umips_build_save_restore): New function.
15287 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15288 (was_mips16_p): Remove.
15289 (old_compression_mode): New.
15290 (mips_set_compression_mode): New function.
15291 (mips_set_current_function): Add microMIPS support.
15292 (mips_option_override): Likewise.
15293 (umips_save_restore_pattern_p): New function.
15294 (umips_output_save_restore): New function.
15295 (umips_load_store_pair_p_1): New function.
15296 (umips_load_store_pair_p): New function.
15297 (umips_output_load_store_pair_1): New function.
15298 (umips_output_load_store_pair): New function.
15299 (umips_movep_target_p) New function.
15300 (mips_prepare_pch_save): Add microMIPS support.
15301 * config/mips/mips.h (TARGET_COMPRESSION): New.
15302 (TARGET_CPU_CPP_BUILTINS): Update macro
15303 to use new compression flags and to support microMIPS.
15304 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15305 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15306 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15307 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15308 (ASM_SPEC): Support mmicromips and mno-micromips.
15309 (M16STORE_REG_P): New macro.
15310 (MIPS_CALL): Support TARGET_MICROMIPS.
15311 (MICROMIPS_J): New macro.
15312 (mips_base_mips16): Rename this...
15313 (mips_base_compression_flags): ...to this.
15314 (UMIPS_12BIT_OFFSET_P): New macro.
15315 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15316 (MULTILIB_DIRNAMES): Likewise.
15318 2013-02-25 Tom de Vries <tom@codesourcery.com>
15320 PR rtl-optimization/56131
15321 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15322 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15323 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15325 2013-02-25 Tobias Burnus <burnus@net-b.de>
15327 * doc/invoke.texi (-fsanitize=): Move from optimization
15328 to debugging options.
15330 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15332 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15334 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15335 Alexander Monakov <amonakov@ispras.ru>
15337 PR middle-end/56077
15338 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15339 flush pending lists also on non-jumps. Adjust comment.
15341 2013-02-24 Catherine Moore <clm@codesourcery.com>
15342 Maciej W. Rozycki <macro@codesourcery.com>
15343 Tom de Vries <tom@codesourcery.com>
15344 Nathan Sidwell <nathan@codesourcery.com>
15345 Iain Sandoe <iain@codesourcery.com>
15346 Nathan Froyd <froydnj@codesourcery.com>
15347 Chao-ying Fu <fu@mips.com>
15349 * doc/extend.texi: (micromips, nomicromips, nocompression):
15350 Document new function attributes.
15351 * doc/invoke.texi (minterlink-compressed, mmicromips,
15352 m14k, m14ke, m14kec): Document new options.
15353 (minterlink-mips16): Update documentation.
15354 * doc/md.texi (ZC, ZD): Document new constraints.
15355 * configure.ac (gcc_cv_as_micromips): Check if linker
15356 supports the .set micromips directive.
15357 * configure: Regenerate.
15358 * config.in: Regenerate.
15359 * config/mips/mips-tables.opt: Regenerate.
15360 * config/mips/micromips.md: New file.
15361 * constraints.md (ZC, AD): New constraints.
15362 * config/mips/predicates.md (movep_src_register): New predicate.
15363 (movep_src_operand): New predicate.
15364 (non_volatile_mem_operand): New predicate.
15365 * config/mips/mips.md (multimem): New type.
15366 (length): Differentiate between 17-bit and 18-bit branch offsets.
15367 (MOVEP1, MOVEP2): New mode iterator.
15368 (mov_<load>l): Use ZC constraint.
15369 (mov_<load>r): Likewise.
15370 (mov_<store>l): Likewise.
15371 (mov_<store>r): Likewise.
15372 (*branch_equality<mode>_inverted): Add microMIPS support.
15373 (*branch_equality<mode>): Likewise.
15374 (*jump_absolute): Likewise.
15375 (indirect_jump_<mode>): Likewise.
15376 (tablejump_<mode>): Likewise.
15377 (<optab>_internal): Likewise.
15378 (sibcall_internal): Likewise.
15379 (sibcall_value_internal): Likewise.
15380 (prefetch): Use constraint ZD.
15381 * config/mips/mips.opt (minterlink-compressed): New option.
15382 (minterlink-mips16): Now an alias for minterlink-compressed.
15383 (mmicromips): New option.
15384 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15385 (compare_and_swap_12): Likewise.
15386 (sync_add<mode>): Likewise.
15387 (sync_<optab>_12): Likewise.
15388 (sync_old_<optab>_12): Likewise.
15389 (sync_new_<optab>_12): Likewise.
15390 (sync_nand_12): Likewise.
15391 (sync_old_nand_12): Likewise.
15392 (sync_new_nand_12): Likewise.
15393 (sync_sub<mode>): Likewise.
15394 (sync_old_add<mode>): Likewise.
15395 (sync_old_sub<mode>): Likewise.
15396 (sync_new_add<mode>): Likewise.
15397 (sync_new_sub<mode>): Likewise.
15398 (sync_<optab><mode>): Likewise.
15399 (sync_old_<optab><mode>): Likewise.
15400 (sync_new_<optab><mode>): Likewise.
15401 (sync_nand<mode>): Likewise.
15402 (sync_old_nand<mode>): Likewise.
15403 (sync_new_nand<mode>): Likewise.
15404 (sync_lock_test_and_set<mode>): Likewise.
15405 (test_and_set_12): Likewise.
15406 (atomic_compare_and_swap<mode>): Likewise.
15407 (atomic_exchange<mode>_llsc): Likewise.
15408 (atomic_fetch_add<mode>_llsc): Likewise.
15409 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15410 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15411 (umips_save_restore_pattern_p): Likewise.
15412 (umips_load_store_pair_p): Likewise.
15413 (umips_output_load_store_pair): Likewise.
15414 (umips_movep_target_p): Likewise.
15415 (umips_12bit_offset_address_p): Likewise.
15416 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15417 (mips_base_mips16): Rename this...
15418 (mips_base_compression_flags): ...to this. Update all uses.
15419 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15420 (mips_mips16_decl_p): Delete.
15421 (mips_nomips16_decl_p): Delete.
15422 (mips_get_compress_on_flags): New function.
15423 (mips_get_compress_off_flags): New function.
15424 (mips_get_compress_mode): New function.
15425 (mips_get_compress_on_name): New function.
15426 (mips_get_compress_off_name): New function.
15427 (mips_insert_attributes): Support multiple compression types.
15428 (mips_merge_decl_attributes): Likewise.
15429 (umips_12bit_offset_address_p): New function.
15430 (mips_start_function_definition): Emit .set micromips directive.
15431 (mips_call_may_need_jalx_p): New function.
15432 (mips_function_ok_for_sibcall): Add microMIPS support.
15433 (mips_print_operand_punctuation): Support short delay slots and
15435 (umips_swm_mask, umips_swm_encoding): New.
15436 (umips_build_save_restore): New function.
15437 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15438 (was_mips16_p): Remove.
15439 (old_compression_mode): New.
15440 (mips_set_compression_mode): New function.
15441 (mips_set_current_function): Add microMIPS support.
15442 (mips_option_override): Likewise.
15443 (umips_save_restore_pattern_p): New function.
15444 (umips_output_save_restore): New function.
15445 (umips_load_store_pair_p_1): New function.
15446 (umips_load_store_pair_p): New function.
15447 (umips_output_load_store_pair_1): New function.
15448 (umips_output_load_store_pair): New function.
15449 (umips_movep_target_p) New function.
15450 (mips_prepare_pch_save): Add microMIPS support.
15451 * config/mips/mips.h (TARGET_COMPRESSION): New.
15452 (TARGET_CPU_CPP_BUILTINS): Update macro
15453 to use new compression flags and to support microMIPS.
15454 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15455 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15456 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15457 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15458 (ASM_SPEC): Support mmicromips and mno-micromips.
15459 (M16STORE_REG_P): New macro.
15460 (MIPS_CALL): Support TARGET_MICROMIPS.
15461 (MICROMIPS_J): New macro.
15462 (mips_base_mips16): Rename this...
15463 (mips_base_compression_flags): ...to this.
15464 (UMIPS_12BIT_OFFSET_P): New macro.
15465 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15466 (MULTILIB_DIRNAMES): Likewise.
15468 2013-02-24 Jakub Jelinek <jakub@redhat.com>
15471 * target-globals.c (save_target_globals): For init_reg_sets and
15472 target_reinit remporarily set this_fn_optabs to this_target_optabs.
15474 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
15476 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15477 * config/aarch64/t-aarch64
15478 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15480 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
15482 PR inline-asm/56148
15483 * lra-constraints.c (process_alt_operands): Reload operand
15484 conflicting with earlier clobber only if no more other conflicting
15487 2013-02-22 Jakub Jelinek <jakub@redhat.com>
15490 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15491 if not linking a shared library.
15493 2013-02-22 Seth LaForge <sethml@google.com>
15495 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15497 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
15499 * config/arm/arm.md (split for extendsidi): Update condition.
15500 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15501 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15502 (qhs_zextenddi_cstr): Likewise.
15504 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15506 PR middle-end/56420
15507 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15508 avoid signed wrapping.
15509 (expand_mult): Handle properly multiplication by
15510 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
15511 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
15512 in the compiler if coeff is HOST_WIDE_INT_MIN.
15513 (expand_divmod): Don't make ext_op1 static, change it's type to
15514 uhwi. Avoid undefined behavior in -INTVAL (op1).
15516 PR rtl-optimization/50339
15517 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15519 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15520 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15521 into splitting_ashiftrt field.
15522 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15524 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15527 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
15529 PR middle-end/56108
15530 * trans-mem.c (execute_tm_mark): Do not expand transactions that
15531 are sure to go irrevocable.
15533 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
15535 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15536 scalars are valid operands.
15538 2013-02-21 Martin Jambor <mjambor@suse.cz>
15540 PR tree-optimization/56310
15541 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15542 only matching indices and non-negative final offsets.
15543 (intersect_aggregates_with_edge): Pass src_idx to
15544 agg_replacements_to_vector. Pass src_idx insstead of index to
15545 intersect_with_agg_replacements.
15547 2013-02-21 Martin Jambor <mjambor@suse.cz>
15549 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15550 instead of hard-wired defaults.
15552 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
15554 * doc/invoke.texi (MIPS Options): Update documentation of the
15555 floating-point multiply-accumulate instruction restrictions.
15557 2013-02-21 Kostya Serebryany <kcc@google.com>
15559 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15560 asan_shadow_offset on x86_64 linux.
15562 2013-02-21 Richard Biener <rguenther@suse.de>
15564 PR tree-optimization/56415
15566 2013-02-11 Richard Biener <rguenther@suse.de>
15568 PR tree-optimization/56273
15569 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15572 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15575 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15578 PR inline-asm/56405
15579 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15580 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15582 2013-02-20 Jan Hubicka <jh@suse.cz>
15584 PR tree-optimization/56265
15585 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15586 when target is referenced for first time.
15588 2013-02-20 Richard Biener <rguenther@suse.de>
15590 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15591 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15592 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15593 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15594 not return anything.
15595 (rename_ssa_copies): Do not remove unused locals.
15596 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15597 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15598 * passes.c (execute_function_todo): Do not schedule unused locals
15599 removal if cleanup_tree_cfg did something.
15600 * tree-ssa-live.c (remove_unused_locals): Dump statistics
15601 about the number of removed locals.
15603 2013-02-20 Richard Biener <rguenther@suse.de>
15605 PR tree-optimization/56398
15606 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15608 2013-02-20 Martin Jambor <mjambor@suse.cz>
15610 PR tree-optimization/55334
15611 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15612 restricted pointers to arrays.
15614 2013-02-20 Richard Biener <rguenther@suse.de>
15615 Jakub Jelinek <jakub@redhat.com>
15617 PR tree-optimization/56396
15618 * tree-ssa-ccp.c (n_const_val): New static variable.
15619 (get_value): Return NULL for SSA names we don't have a lattice
15621 (ccp_initialize): Initialize n_const_val.
15622 * tree-ssa-copy.c (n_copy_of): New static variable.
15623 (init_copy_prop): Initialize n_copy_of.
15624 (get_value): Return NULL_TREE for SSA names we don't have a
15627 2013-02-20 Martin Jambor <mjambor@suse.cz>
15629 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15631 2013-02-20 Richard Biener <rguenther@suse.de>
15633 * genpreds.c (write_lookup_constraint): Do not compare first
15634 letter of the constraint again.
15636 2013-02-20 Richard Biener <rguenther@suse.de>
15638 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15640 (get_use_iv_cost): Terminate hashtable walk when coming across
15643 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
15645 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15646 reassociation for avx2 targets.
15648 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15650 * config/microblaze/microblaze.c: microblaze_has_clz = 0
15651 Add version check for v8.10.a to enable microblaze_has_clz
15652 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15653 version and TARGET_PATTERN_COMPARE check
15654 * config/microblaze/microblaze.md: New clzsi2 instruction
15656 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15658 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15659 function before branching.
15661 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
15663 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15665 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15667 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
15669 PR middle-end/55889
15670 * sel-sched.c: Include ira.h.
15671 (implicit_clobber_conflict_p): New function.
15672 (moveup_expr): Use it.
15673 * Makefile.in (sel-sched.o): Depend on ira.h.
15675 2013-02-19 Richard Biener <rguenther@suse.de>
15677 PR tree-optimization/56384
15678 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15679 (vn_hash_type): Split out from ...
15680 (vn_hash_constant_with_type): ... here.
15681 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15682 (vn_phi_eq): Compare types from vn_phi_s structure.
15683 (vn_phi_lookup): Populate vn_phi_s type.
15684 (vn_phi_insert): Likewise.
15686 2013-02-19 Jakub Jelinek <jakub@redhat.com>
15688 PR tree-optimization/56350
15689 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15690 if haven't found reduction or nested cycle operand, rather than
15691 asserting we must find it.
15693 PR tree-optimization/56381
15694 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15697 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
15698 Jakub Jelinek <jakub@redhat.com>
15701 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15702 (swap_optab_enable): Same.
15703 (init_all_optabs): Use argument instead of global.
15704 * tree.h (struct tree_optimization_option): New field target_optabs.
15705 * expr.h (init_all_optabs): Add argument to prototype.
15706 (TREE_OPTIMIZATION_OPTABS): New.
15707 (save_optabs_if_changed): Protoize.
15708 * optabs.h: Declare this_fn_optabs.
15709 * optabs.c (save_optabs_if_changed): New.
15710 Declare this_fn_optabs.
15711 (init_optabs): Add argument to init_all_optabs() call.
15712 * function.c (invoke_set_current_function_hook): Handle per
15714 * function.h (struct function): New field optabs.
15715 * config/mips/mips.c (mips_set_mips16_mode): Handle when
15716 optimization_current_node has changed.
15717 * target-globals.h (save_target_globals_default_opts): Protoize.
15718 * target-globals.c (save_target_globals_default_opts): New.
15720 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15723 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15724 registers %fr12 and %fr12R as call used.
15727 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15728 and HImode, require all displacements to be an integer multiple of
15730 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15731 only allow QImode and HImode when reload is in progress and strict is
15732 true. Likewise for symbolic addresses. Use base14_operand to check
15733 displacements in REG+BASE addresses.
15735 2013-02-18 Richard Biener <rguenther@suse.de>
15737 PR tree-optimization/56366
15738 * tree-vect-loop.c (get_initial_def_for_induction): Properly
15739 handle sign-conversion of outer-loop initial induction value.
15741 2013-02-18 Richard Biener <rguenther@suse.de>
15743 PR middle-end/56349
15744 * cfghooks.c (merge_blocks): If we merge a latch into another
15745 block adjust references to it.
15746 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15747 (verify_loop_structure): Verify that a recorded latch is in fact
15750 2013-02-18 Richard Biener <rguenther@suse.de>
15752 PR tree-optimization/56321
15753 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15754 order SSA name release and virtual operand unlinking.
15756 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15758 * config/microblaze/microblaze.md (save_stack_block): Define.
15759 (restore_stack_block): Likewise.
15761 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15763 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15764 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15765 * config/microblaze/microblaze.c (microblaze_option_override):
15766 Bail out early for PIC modes when target does not support PIC.
15768 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15770 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15771 Replace with a microblaze version.
15772 (microblaze_trampoline_init): Adapt for microblaze.
15773 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15776 2013-02-16 Jakub Jelinek <jakub@redhat.com>
15777 Dodji Seketeli <dodji@redhat.com>
15780 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15781 (instrument_mem_region_access): Do not forget to always put
15782 instrumentation of the of 'base' and 'base + len' in a "if (len !=
15783 0) statement, even for cases where either 'base' or 'base + len'
15784 are not instrumented -- because they have been previously
15785 instrumented. Simplify the logic by putting all the statements
15786 instrument 'base + len' inside a sequence, and then insert that
15787 sequence right before the current insertion point. Then, to
15788 instrument 'base + len', just get an iterator on that statement.
15789 And do not forget to update the pointer to iterator the function
15790 received as argument.
15792 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
15794 PR rtl-optimization/56348
15795 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15797 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
15799 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15800 (clean_graph_dump_file): Pass base to start_graph_dump.
15802 2013-02-14 Richard Henderson <rth@redhat.com>
15805 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15807 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
15809 * collect2-aix.h: Define F_LOADONLY.
15811 2013-02-14 Richard Biener <rguenther@suse.de>
15814 * varasm.c (output_constant_def_1): Get the decl representing
15815 the constant as argument.
15816 (output_constant_def): Wrap output_constant_def_1.
15817 (make_decl_rtl): Use output_constant_def_1 with the decl
15818 representing the constant.
15819 (build_constant_desc): Optionally re-use a decl already
15820 representing the constant.
15821 (tree_output_constant_def): Adjust.
15823 2013-02-14 Dodji Seketeli <dodji@redhat.com>
15826 * asan.c (instrument_builtin_call): Really put the length of the
15827 second source argument into src1_len.
15829 2013-02-13 Jakub Jelinek <jakub@redhat.com>
15831 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15832 argument. If it is false, don't create edge from then_bb to
15834 (insert_if_then_before_iter): Pass true to it.
15835 (build_check_stmt): Pass false to it.
15836 (transform_statements): Flush hash table only on extended basic
15837 block boundaries, rather than at the beginning of every bb.
15838 Don't flush hash table on nonfreeing_call_p calls.
15839 * tree-flow.h (nonfreeing_call_p): New prototype.
15840 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15842 2013-02-13 David S. Miller <davem@davemloft.net>
15844 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15846 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
15849 * ira.c (max_regno_before_ira): Move from ...
15851 (fix_reg_equiv_init): Use max_regno_before_ira instead of
15854 2013-02-13 Jakub Jelinek <jakub@redhat.com>
15856 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15858 2013-02-13 Richard Biener <rguenther@suse.de>
15861 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15862 globals in MEM_REFs.
15864 2013-02-13 Richard Biener <rguenther@suse.de>
15866 * loop-init.c (loop_optimizer_init): Clear loop state when
15867 re-initializing preserved loops.
15868 * loop-unswitch.c (unswitch_single_loop): Return whether
15869 we unswitched the loop. Do not verify loop state here.
15870 (unswitch_loops): When we unswitched a loop discover new loops.
15872 2013-02-13 Kostya Serebryany <kcc@google.com>
15874 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15876 * sanitizer.def: Rename __asan_init to __asan_init_v1.
15878 2013-02-12 Dodji Seketeli <dodji@redhat.com>
15880 Avoid instrumenting duplicated memory access in the same basic block
15881 * Makefile.in (asan.o): Add new dependency on hash-table.h
15882 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15883 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15884 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15885 (free_mem_ref_resources, has_mem_ref_been_instrumented)
15886 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15887 (get_mem_ref_of_assignment): New functions.
15888 (get_mem_refs_of_builtin_call): Extract from
15889 instrument_builtin_call and tweak a little bit to make it fit with
15891 (instrument_builtin_call): Use the new
15892 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
15893 of is_gimple_builtin_call.
15894 (instrument_derefs, instrument_mem_region_access): Insert the
15895 instrumented memory reference into the hash table.
15896 (maybe_instrument_assignment): Renamed instrument_assignment into
15897 this, and change it to advance the iterator when instrumentation
15898 actually happened and return true in that case. This makes it
15899 homogeneous with maybe_instrument_assignment, and thus give a
15900 chance to callers to be more 'regular'.
15901 (transform_statements): Clear the memory reference hash table
15902 whenever we enter a new BB, when we cross a function call, or when
15903 we are done transforming statements. Use
15904 maybe_instrument_assignment instead of instrumentation. No more
15905 need to special case maybe_instrument_assignment and advance the
15906 iterator after calling it; it's now handled just like
15907 maybe_instrument_call. Update comment.
15909 2013-02-13 Richard Biener <rguenther@suse.de>
15911 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15912 Fix loop discovery code.
15914 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
15916 PR inline-asm/56148
15917 * lra-constraints.c (process_alt_operands): Match early clobber
15918 operand with itself. Check conflicts with earlyclobber only if
15919 the operand is not reloaded. Prefer to reload conflicting operand
15920 if earlyclobber and matching operands are the same.
15922 2013-02-12 Richard Biener <rguenther@suse.de>
15925 * lto-streamer-out.c (write_symbol): Do not output symbols
15926 for hard register variables.
15928 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
15931 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15932 (umulsidi3_insn, mulsidi3_insn): New insns.
15934 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
15936 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15937 (struct tune_params): Add vec_costs field.
15938 * config/arm/arm.c (arm_builtin_vectorization_cost)
15939 (arm_add_stmt_cost): New functions.
15940 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15941 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15942 (arm_default_vec_cost): New struct of type cpu_vec_costs.
15943 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15944 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15945 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15946 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15948 2013-02-12 Richard Biener <rguenther@suse.de>
15951 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15952 decls again if possible.
15954 2013-02-12 Richard Biener <rguenther@suse.de>
15956 PR middle-end/56288
15957 * tree-ssa.c (verify_ssa_name): Fix check, move
15958 SSA_NAME_IN_FREE_LIST check up.
15960 2013-02-12 Jakub Jelinek <jakub@redhat.com>
15961 Steven Bosscher <steven@gcc.gnu.org>
15963 PR rtl-optimization/56151
15964 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15965 equal to op0 or op1, and last_insn pattern is CODE operation
15966 with MEM dest and one of the operands matches that MEM.
15968 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
15970 * doc/extend.texi: Document Function Multiversioning and "default"
15971 parameter string to target attribute.
15972 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15973 target attribute parameter is "default".
15974 (ix86_compare_version_priority): Remove checks for target attribute.
15975 (ix86_mangle_function_version_assembler_name): Change error to sorry.
15976 Remove check for target attribute equal to NULL. Add assert.
15977 (ix86_generate_version_dispatcher_body): Change error to sorry.
15979 2013-02-11 Iain Sandoe <iain@codesourcery.com>
15980 Jack Howarth <howarth@bromo.med.uc.edu>
15981 Patrick Marlier <patrick.marlier@gmail.com>
15984 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15985 define ENDFILE_SPEC as TM_DESTRUCTOR.
15986 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15988 2013-02-11 Alexander Potapenko <glider@google.com>
15989 Jack Howarth <howarth@bromo.med.uc.edu>
15990 Jakub Jelinek <jakub@redhat.com>
15993 * config/darwin.c (cdtor_record): Rename ctor_record.
15994 (sort_cdtor_records): Rename sort_ctor_records.
15995 (finalize_dtors): New routine to sort destructors by
15996 priority before use in assemble_integer.
15997 (machopic_asm_out_destructor): Use finalize_dtors if needed.
15999 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
16001 PR rtl-optimization/56275
16002 * simplify-rtx.c (avoid_constant_pool_reference): Check that
16003 offset is non-negative and less than cmode size before
16004 calling simplify_subreg.
16006 2013-02-11 Richard Biener <rguenther@suse.de>
16008 PR tree-optimization/56264
16009 * cfgloop.h (fix_loop_structure): Adjust prototype.
16010 * loop-init.c (fix_loop_structure): Return the number of
16011 newly discovered loops.
16012 * tree-cfgcleanup.c (repair_loop_structures): When new loops
16013 are discovered, do a full loop-closed SSA rewrite.
16015 2013-02-11 Richard Biener <rguenther@suse.de>
16017 PR tree-optimization/56273
16018 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16020 (check_array_ref): Fix missing newline in dumps.
16021 (search_for_addr_array): Likewise.
16023 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
16025 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16027 2013-02-09 Jakub Jelinek <jakub@redhat.com>
16030 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16032 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
16034 PR rtl-optimization/56246
16035 * lra-constraints.c (simplify_operand_subreg): Try to reuse
16037 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16038 constraints are satisfied.
16040 2013-02-08 Jeff Law <law@redhat.com>
16043 * emit-rtl.c (reg_is_parm_p): New function.
16044 * regs.h (reg_is_parm_p): New prototype.
16045 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16046 callee-clobbered registers.
16048 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
16051 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16052 If there is no implicit builtin declaration, just return NULL.
16054 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16056 * config/i386/sse.md (FMAMODEM): New mode iterator.
16057 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16058 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16060 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16062 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16063 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16064 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16066 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16068 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16069 (microblaze*-*-elf): Likewise.
16070 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16072 * config/microblaze/microblaze-c.c: Add builtin defines for
16073 _LITTLE_ENDIAN and _BIG_ENDIAN.
16074 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16075 add to TARGET_DEFAULT flags.
16076 Expand ASM_SPEC and LINK_SPEC.
16077 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16078 * config/microblaze/microblaze.md: Update extendsidi2 and
16079 movdi_internal instructions to use low-order / high-order reg
16081 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16082 options and inversemask / mask of LITTLE_ENDIAN.
16083 * config/microblaze/t-microblaze: Expand multilib options to
16084 include mlittle-endian (le) and update exceptions patterns.
16086 2013-02-08 Jakub Jelinek <jakub@redhat.com>
16088 PR rtl-optimization/56195
16089 * lra-constraints.c (get_reload_reg): Don't reuse regs
16090 if they have smaller mode than requested, if they have
16091 wider mode than requested, try to return a SUBREG.
16093 PR tree-optimization/56250
16094 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16095 if type is unsigned and code isn't MULT_EXPR.
16097 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16099 PR tree-optimization/56064
16100 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16101 bits according to mode.
16102 * fixed-value.h (fixed_from_double_int)
16103 (const_fixed_from_double_int): Adjust comments.
16105 2013-02-08 Richard Biener <rguenther@suse.de>
16108 * lto-streamer.h (struct data_in): Remove current_file, current_line
16109 and current_col members.
16110 * lto-streamer-out.c (lto_output_location): Stream changed bits
16111 en-block for efficiency.
16112 * lto-streamer-in.c (clear_line_info): Remove.
16113 (lto_input_location): Cache current file, line and column
16114 globally via local statics. Read changed bits en-block.
16115 (input_function): Do not call clear_line_info.
16116 (lto_read_body): Likewise.
16117 (lto_input_toplevel_asms): Likewise.
16119 2013-02-08 Michael Matz <matz@suse.de>
16121 PR tree-optimization/52448
16122 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16123 (nt_call_phase): New static.
16124 (add_or_mark_expr): Only mark accesses with newer phase than any
16126 (nonfreeing_call_p): New.
16127 (nt_init_block): Update nt_call_phase, mark blocks as visited.
16128 (nt_fini_block): Keep blocks marked as visited.
16129 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16131 2013-02-08 Richard Biener <rguenther@suse.de>
16133 * ira.c (ira): Free broken dominator information.
16135 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16137 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16139 2013-02-08 Marek Polacek <polacek@redhat.com>
16141 * cfgloop.c (verify_loop_structure): Add more checking of headers.
16143 2013-02-08 Richard Biener <rguenther@suse.de>
16145 PR middle-end/56181
16146 * cfgloop.h (flow_loops_find): Adjust.
16147 (bb_loop_header_p): Declare.
16148 * cfgloop.c (bb_loop_header_p): New function split out from ...
16149 (flow_loops_find): ... here. Adjust function signature,
16150 support incremental loop structure update.
16151 (verify_loop_structure): Cleanup. Verify a loop is a loop.
16152 * cfgloopmanip.c (fix_loop_structure): Move ...
16153 * loop-init.c (fix_loop_structure): ... here.
16154 (apply_loop_flags): Split out from ...
16155 (loop_optimizer_init): ... here.
16156 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
16157 in incremental mode, only remove dead loops here.
16159 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16162 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16163 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16164 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16165 (*round<mode>3.libgcc): New insns for fixed-modes.
16166 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16167 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16168 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16169 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16170 implementations. Define to __builtin_avr_absFX,
16171 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16172 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16173 __builtin_avr_countlsFX, respectively.
16174 * config/avr/avr-c.c (target.h): Include it.
16175 (enum avr_builtin_id): New enum.
16176 (avr_resolve_overloaded_builtin): New static function.
16177 (avr_register_target_pragmas): Use it to set
16178 targetm.resolve_overloaded_builtin.
16179 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16180 tree nodes used by DEF_BUILTIN.
16181 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16182 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16183 <AVR_BUILTIN_xxBITS>: Same.
16185 2013-02-08 Richard Biener <rguenther@suse.de>
16187 * cfgloop.c (verify_loop_structure): Properly handle
16188 a loop exiting to another loop header.
16189 * ira-int.h (ira_loops): Remove.
16190 * ira.c (ira_loops): Remove.
16191 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16192 (do_reload): Use loop_optimizer_finalize.
16193 * ira-build.c (create_loop_tree_nodes): Use get_loops and
16194 number_of_loops to access the loop tree.
16195 (more_one_region_p): Likewise.
16196 (finish_loop_tree_nodes): Likewise.
16197 (rebuild_regno_allocno_maps): Likewise.
16198 (mark_loops_for_removal): Likewise.
16199 (mark_all_loops_for_removal): Likewise.
16200 (remove_unnecessary_regions): Likewise.
16201 (ira_build): Likewise.
16202 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16204 2013-02-08 Richard Biener <rguenther@suse.de>
16206 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16207 * ipa-pure-const.c (analyze_function): Avoid calling
16208 mark_irreducible_loops twice.
16209 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16211 2013-02-07 David S. Miller <davem@davemloft.net>
16213 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16215 * var-tracking.c (vt_add_function_parameter): Test the presence of
16216 HAVE_window_save properly and do not remap argument registers when
16217 we have a leaf function.
16219 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
16222 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16224 * config/i386/i386.c (ix86_print_operand): Ditto.
16226 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
16228 * lra-constraints.c (process_alt_operands): Fix recently added comment.
16230 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
16232 PR rtl-optimization/56225
16233 * lra-constraints.c (process_alt_operands): Check that reload hard
16234 reg can hold value for strict_low_part.
16236 2013-02-07 Jakub Jelinek <jakub@redhat.com>
16239 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16240 dwarf2out_end_function.
16241 (in_first_function_p, maybe_at_text_label_p,
16242 first_loclabel_num_not_at_text_label): New variables.
16243 (dwarf2out_var_location): In the first function find out
16244 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16245 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16248 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
16250 PR rtl-optimization/56178
16251 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16252 SUBREG of a register. Tidy up related block of code.
16253 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16254 note if the source is a register or a SUBREG of a register.
16256 2013-02-07 Jakub Jelinek <jakub@redhat.com>
16259 * config/rs6000/rs6000.md (ptrm): New mode attr.
16260 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16261 call_value_indirect_aix<pttrsize>,
16262 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16265 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16267 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16268 if -bnortl. Convert to strcmp and strncmp.
16270 2013-02-07 Alan Modra <amodra@gmail.com>
16273 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16274 addresses won't wrap when offsetting.
16275 (rs6000_secondary_reload): Provide secondary reloads needed for
16276 wrapping LO_SUM addresses.
16278 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
16280 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16281 MACH, just __MACH__.
16283 2013-02-06 Richard Biener <rguenther@suse.de>
16285 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16286 instead of calling fix_loop_structure.
16288 2013-02-06 Jakub Jelinek <jakub@redhat.com>
16290 PR middle-end/56217
16291 * omp-low.c (use_pointer_for_field): Return false if
16292 lower_send_shared_vars doesn't generate any copy-out code.
16294 2013-02-06 Tom de Vries <tom@codesourcery.com>
16296 PR rtl-optimization/56131
16297 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16298 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
16299 of the label is NULL. Add comment.
16301 2013-02-05 Jakub Jelinek <jakub@redhat.com>
16303 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16306 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16307 (STATIC_LIBTSAN_LIBS): Likewise.
16308 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16309 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16310 is defined, don't add anything else beyond that.
16311 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16312 (LINK_COMMAND_SPEC): Use them.
16314 PR tree-optimization/56205
16315 * tree-stdarg.c (check_all_va_list_escapes): Return true if
16316 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16317 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16319 2013-02-05 Richard Biener <rguenther@suse.de>
16321 PR tree-optimization/53342
16322 PR tree-optimization/53185
16323 * tree-vectorizer.h (vect_check_strided_load): Remove.
16324 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16325 not disallow peeling for vectorized strided loads.
16326 (vect_check_strided_load): Make static and simplify.
16327 (vect_analyze_data_refs): Adjust.
16328 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16329 correctly when vectorizing strided loads.
16331 2013-02-05 Richard Biener <rguenther@suse.de>
16333 * doc/install.texi: Refer to ISL, not PPL.
16335 2013-02-05 Jan Hubicka <jh@suse.cz>
16337 PR tree-optimization/55789
16338 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16340 2013-02-05 Jan Hubicka <jh@suse.cz>
16342 PR tree-optimization/55789
16343 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16344 the dead call anyway.
16346 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
16349 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16351 2013-02-04 Alexander Potapenko <glider@google.com>
16352 Jack Howarth <howarth@bromo.med.uc.edu>
16353 Jakub Jelinek <jakub@redhat.com>
16356 * config/darwin.c (sort_ctor_records): Stabilized qsort
16357 on constructor priority by using original position.
16358 (finalize_ctors): New routine to sort constructors by
16359 priority before use in assemble_integer.
16360 (machopic_asm_out_constructor): Use finalize_ctors if needed.
16362 2013-02-04 Jakub Jelinek <jakub@redhat.com>
16365 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16366 about visibility on artificial decls.
16367 * config/sol2.c (solaris_assemble_visibility): Likewise.
16369 2013-02-04 Kai Tietz <ktietz@redhat.com>
16372 * config/i386/i386.c (function_value_ms_64): Add additional valtype
16373 argument and improve checking of return-argument types for 16-byte
16375 (ix86_function_value_1): Add additional valtype argument on call
16376 of function_value_64.
16377 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16378 handling infunction_value_64 function.
16380 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
16382 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16384 2013-02-04 Richard Biener <rguenther@suse.de>
16386 PR tree-optimization/56188
16387 * tree-ssa-structalias.c (label_visit): Consider case with
16388 initially non-empty points-to set.
16389 (perform_var_substitution): Dump node mapping and clean up.
16391 2013-02-04 Richard Guenther <rguenther@suse.de>
16394 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16395 node prevail as last resort.
16396 (lto_symtab_merge_decls): Remove guard on LTRANS here.
16397 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16399 2013-02-04 Richard Biener <rguenther@suse.de>
16401 PR tree-optimization/56113
16402 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16404 (equiv_class_lookup_or_add): ... this.
16405 (label_visit): Adjust and fix error in previous patch.
16406 (perform_var_substitution): Adjust.
16408 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
16410 * config/sh/divtab.c: Fix formatting and comments throughout the file.
16411 * config/sh/sh4-300.md: Likewise.
16412 * config/sh/sh4a.md: Likewise.
16413 * config/sh/constraints.md: Likewise.
16414 * config/sh/sh.md: Likewise.
16415 * config/sh/netbsd-elf.h: Likewise.
16416 * config/sh/predicates.md: Likewise.
16417 * config/sh/sh-protos.h: Likewise.
16418 * config/sh/ushmedia.h: Likewise.
16419 * config/sh/linux.h: Likewise.
16420 * config/sh/sh.c: Likewise.
16421 * config/sh/superh.h: Likewise.
16422 * config/sh/elf.h: Likewise.
16423 * config/sh/sh4.md: Likewise.
16424 * config/sh/sh.h: Likewise.
16426 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16428 * config/pa/constraints.md: Adjust unused letters. Change "T"
16429 constraint to match_test floating_point_store_memory_operand().
16430 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16431 (base14_operand): New.
16432 (floating_point_store_memory_operand): New.
16433 (integer_store_memory_operand): Revise to use base14_operand and
16434 reg_plus_base_memory_operand.
16435 (move_dest_operand): Allow symbolic_memory_operands.
16436 (symbolic_memory_operand): Check for LO_SOM.
16437 (symbolic_operand): Change default case to break.
16438 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16439 CONST_DOUBLE values to be reloaded by putting them into memory when
16440 the destination is a floating point register.
16441 (movdf): Remove code to handle CONST_DOUBLE.
16443 (reload_indf_r1): New.
16444 (reload_insf_r1): New.
16445 Consistently use "Q" and "T" constraints with integer and floating
16446 point move instructions, respectively.
16447 (movdi): Remove FAIL.
16448 Change predicate for source operand unamed DImode move from
16449 general_operand to move_src_operand.
16450 (umulsidi3): Change predicate for destination operand to
16452 Likewise for similar unamed patterns.
16453 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16454 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16455 (hppa_legitimize_address): Simplify mask calculation.
16456 (pa_emit_move_sequence): Revised handling of secondary reloads from
16457 REG+D addresses for floating point loads and stores. Directly handle
16458 loading CONST0_RTX (mode) to a floating point register.
16459 (pa_secondary_reload): Handle reloading DF and SFmode constant values
16460 to floating point registers. Don't restrict secondary reloads to
16461 floating point registers to integer modes. Revise some comments and
16463 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16464 (pa_legitimate_address_p): New.
16465 (pa_legitimize_reload_address): New.
16466 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16467 (STRICT_REG_OK_FOR_BASE_P): New.
16468 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
16469 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16471 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
16472 Andrew Dixie <andrewd@gentrack.com>
16474 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16477 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
16479 * expmed.c (extract_bit_field_1): Pass the full width of the
16480 structure to get_best_reg_extraction_insn.
16482 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16485 * configure.ac (use_cxa_atexit): Add AIX.
16486 * configure: Regenerate.
16488 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16490 2013-02-01 Jakub Jelinek <jakub@redhat.com>
16493 * final.c (need_profile_function): New variable.
16494 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16495 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16496 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16497 notes, targetm.asm_out.function_prologue doesn't emit anything,
16498 HAVE_prologue and profiler should be emitted before prologue,
16499 set need_profile_function instead of emitting it.
16500 (final_scan_insn): If need_profile_function, emit
16501 profile_function on the first NOTE_INSN_BASIC_BLOCK or
16502 NOTE_INSN_FUNCTION_BEG note.
16504 2013-02-01 Richard Henderson <rth@redhat.com>
16506 * config/rs6000/rs6000.md (smulditi3): New.
16509 * config/alpha/alpha.md (umulditi3): New.
16511 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16513 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16514 (ASM_OUTPUT_ALIGNED_LOCAL): New.
16516 2013-02-01 Richard Biener <rguenther@suse.de>
16518 PR tree-optimization/56113
16519 * tree-ssa-structalias.c (label_visit): Reduce work for
16520 single-predecessor nodes.
16522 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
16524 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16525 range isn't testing for zero.
16527 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
16529 PR middle-end/56113
16530 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16532 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
16533 Nick Clifton <nickc@redhat.com>
16535 * config/v850/constraints.md (Q): Define as a memory constraint.
16536 * config/v850/predicates.md (label_ref_operand): New predicate.
16537 (e3v5_shift_operand): New predicate.
16538 (ior_operator): New predicate.
16539 * config/v850/t-v850: Add e3v5 multilib.
16540 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16541 (v850_gen_movdi): Prototype.
16542 * config/v850/v850.c: Add support for e3v5 architecture.
16543 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16545 (construct_save_jarl): Add e3v5 long JARL support.
16546 (v850_adjust_insn_length): New function. Adjust length of call
16547 insns when using e3v5 instructions.
16548 (v850_gen_movdi): New function: Generate instructions to move a
16550 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16551 (CPP_SPEC): Define __v850e3v5__ as appropriate.
16552 (TARGET_USE_FPU): Enable for e3v5.
16553 (CONST_OK_FOR_W): New macro.
16554 (ADJUST_INSN_LENGTH): Define.
16555 * config/v850/v850.md (UNSPEC_LOOP): Define.
16556 (attr cpu): Add v850e3v5.
16557 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16558 (movdi): New pattern.
16559 (movdi_internal): New pattern.
16560 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16561 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16562 (cstoresf4): Likewise.
16563 (cstoredf4): Likewise.
16564 (insv): New pattern.
16565 (rotlso3_a): New pattern.
16566 (rotlsi3_b): New pattern
16567 (rotlsi3_v850e3v5): New pattern.
16568 (doloop_begin): New pattern.
16569 (fix_loop_counter): New pattern.
16570 (doloop_end): New pattern.
16571 (branch_normal): Add e3v5 long branch support.
16572 (branch_invert): Likewise.
16573 (branch_z_normal): Likewise.
16574 (branch_z_invert): Likewise.
16575 (branch_nz_normal): Likewise.
16576 (branch_nz_invert): Likewise.
16577 (call_internal_short): Add e3v5 register-indirect JARL support.
16578 (call_internal_long): Likewise.
16579 (call_value_internal_short): Likewise.
16580 (call_value_internal_long): Likewise.
16581 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16582 (mloop): New option.
16583 * config.gcc: Add support for configuring v840e3v5 target.
16584 * doc/invoke.texi: Document new v850 specific command line options.
16586 2013-01-31 Paul Koning <ni1d@arrl.net>
16590 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16591 children if parent is a class.
16592 (prune_unused_types_prune): Don't add DW_AT_declaration.
16594 2013-01-31 Richard Biener <rguenther@suse.de>
16596 PR tree-optimization/56157
16597 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16598 match up operand with SLP child.
16600 2013-01-31 Jason Merrill <jason@redhat.com>
16603 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16604 parameters the first time.
16605 (gen_scheduled_generic_parms_dies): Check completeness here.
16607 2013-01-31 Richard Biener <rguenther@suse.de>
16609 PR middle-end/53073
16610 * common.opt (faggressive-loop-optimizations): New flag,
16611 enabled by default.
16612 * doc/invoke.texi (faggressive-loop-optimizations): Document.
16613 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16614 infer_loop_bounds_from_undefined by it.
16616 2013-01-31 Richard Biener <rguenther@suse.de>
16618 PR tree-optimization/56150
16619 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16620 visit virtual operands.
16621 (find_uses_to_rename_bb): Likewise.
16623 2013-01-31 Richard Biener <rguenther@suse.de>
16625 PR tree-optimization/56150
16626 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16627 mixed store non-store stmts.
16629 2013-01-30 Jakub Jelinek <jakub@redhat.com>
16632 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16633 LIBASAN_EARLY_SPEC is defined.
16634 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16635 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16637 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16640 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16641 invalid args instead of ICEing on it.
16642 (ix86_valid_target_attribute_tree): Return error_mark_node if
16643 ix86_valid_target_attribute_inner_p failed.
16644 (ix86_valid_target_attribute_p): Return false only if
16645 ix86_valid_target_attribute_tree returned error_mark_node. Allow
16646 target("default") attribute.
16647 (sorted_attr_string): Change argument from const char * to tree,
16648 merge in all target attribute arguments rather than just one.
16649 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
16650 instead of free. Avoid using strcat.
16651 (ix86_mangle_function_version_assembler_name): Mangle
16652 target("default") as if no target attribute is present. Adjust
16653 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
16654 instead of xmalloc and XDELETEVEC instead of free.
16655 (ix86_function_versions): Don't return true if one of the decls
16656 doesn't have target attribute. If they don't and one of the decls
16657 is DECL_FUNCTION_VERSIONED, report an error. Adjust
16658 sorted_attr_string caller. Use XDELETEVEC instead of free.
16659 (ix86_supports_function_versions): Remove.
16660 (make_name): Fix up formatting.
16661 (make_dispatcher_decl): Remove resolver_name and its initialization.
16662 Avoid leaking memory.
16663 (is_function_default_version): Return true if there is
16664 target("default") attribute rather than no target attribute at all.
16665 (make_resolver_func): Avoid leaking memory.
16666 (ix86_generate_version_dispatcher_body): Likewise.
16667 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16668 * target.def (supports_function_versions): Remove.
16669 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16670 * doc/tm.texi: Regenerated.
16672 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
16674 PR rtl-optimization/56144
16675 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16676 for values with side effects.
16678 2013-01-30 Richard Biener <rguenther@suse.de>
16680 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16681 (sparseset_pop): Likewise.
16682 * cfganal.c (compute_idf): Likewise. Increase work-stack size
16683 to be able to use quick_push in the worker loop.
16685 2013-01-30 Marek Polacek <polacek@redhat.com>
16687 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16689 2013-01-30 Richard Biener <rguenther@suse.de>
16692 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16694 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
16696 PR tree-optimization/56064
16697 * fixed-value.c (fixed_from_double_int): New function.
16698 * fixed-value.h (fixed_from_double_int): New prototype.
16699 (const_fixed_from_double_int): New static inline function.
16700 * fold-const.c (native_interpret_fixed): New static function.
16701 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16702 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16703 (native_encode_fixed): New static function.
16704 (native_encode_expr) <FIXED_CST>: Use it.
16705 (native_interpret_int): Move double_int worker code to...
16706 * double-int.c (double_int::from_buffer): ...this new static method.
16707 * double-int.h (double_int::from_buffer): Prototype it.
16709 2013-01-30 Richard Biener <rguenther@suse.de>
16711 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16712 New pointer-map and obstack.
16713 (init_alias_vars): Allocate pointer-map and obstack.
16714 (delete_points_to_sets): Free them.
16715 (find_what_var_points_to): Cache result.
16716 (find_what_p_points_to): Adjust for changed interface of
16717 find_what_var_points_to.
16718 (compute_points_to_sets): Likewise.
16719 (ipa_pta_execute): Likewise.
16721 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16723 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16724 * configure: Regenerate.
16725 * config.in: Regenerate.
16726 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16727 #nobits/#progbits if supported.
16729 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
16732 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16733 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16734 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16736 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16738 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16739 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16741 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16743 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16744 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16746 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16748 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16750 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16751 * config/arm/cortex-a7.md: New bypasses using
16752 arm_mac_accumulator_is_result.
16754 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16756 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
16757 (cortex_a7_neon_mla): Likewise.
16758 (cortex_a7_fpfmad): New reservation.
16759 (cortex_a7_fpmacs): Use ffmas and update required units.
16760 (cortex_a7_fpmuld): Update required units and latency.
16761 (cortex_a7_fpmacd): Likewise.
16762 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16763 (cortex_a7_neon). Likewise.
16764 (bypass) Update participating units.
16766 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16768 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16769 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16771 * config/arm/vfp11.md (vfp_farith): Use ffmas.
16772 (vfp_fmul): Use ffmad.
16773 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16774 (cortex_r4_fmacd): Use ffmad.
16775 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16776 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
16777 (cortex_a9_fmacd): Use ffmad.
16778 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16779 (cortex_a8_vfp_macd): Use ffmad.
16780 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16781 (cortex_a5_fpmacd): Use ffmad.
16782 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16783 (cortex_a15_vfp_macd): Use ffmad.
16784 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16786 2013-01-29 Jason Merrill <jason@redhat.com>
16789 * varasm.c (default_assemble_visibility): Don't warn about
16790 visibility on artificial decls.
16792 2013-01-29 Richard Biener <rguenther@suse.de>
16794 PR tree-optimization/56113
16795 * tree-ssa-structalias.c (equiv_class_lookup): Also return
16797 (label_visit): Free duplicate bitmaps and record the leader instead.
16798 (perform_var_substitution): Adjust.
16800 2013-01-29 Richard Biener <rguenther@suse.de>
16802 PR tree-optimization/55270
16803 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16804 the CFG, schedule loops for fixup.
16806 2013-01-29 Nick Clifton <nickc@redhat.com>
16808 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16811 2013-01-28 Leif Ekblad <leif@rdos.net>
16813 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16814 * config/i386/i386.h (TARGET_RDOS): New macro.
16815 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16816 * config/i386/i386.c (ix86_option_override_internal): For 64bit
16817 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16818 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16819 DEFAULT_LARGE_SECTION_THRESHOLD.
16820 * config/i386/i386.md (R14_REG, R15_REG): New constants.
16821 * config/i386/rdos.h: New file.
16822 * config/i386/rdos64.h: New file.
16824 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
16827 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16830 2013-01-28 Jakub Jelinek <jakub@redhat.com>
16832 PR rtl-optimization/56117
16833 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16834 call cselib_lookup_from_insn on the MEM before calling
16835 add_insn_mem_dependence.
16837 2013-01-28 Richard Biener <rguenther@suse.de>
16839 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16840 to a stmt that didn't have one.
16841 (copy_phis_for_bb): Likewise for PHI arguments.
16842 (copy_debug_stmt): Likewise for debug stmts.
16844 2013-01-28 Richard Biener <rguenther@suse.de>
16846 PR tree-optimization/56034
16847 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16848 (partition_builtin_p): Adjust.
16849 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
16850 it is the last partition.
16851 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16852 up the vertex for the definition.
16853 (classify_partition): Classify whether a partition is a
16854 PKIND_REDUCTION, thus has uses outside of the loop.
16855 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16856 Merge all PKIND_REDUCTION partitions into the last partition.
16857 (tree_loop_distribution): Seed partitions from reductions as well.
16859 2013-01-28 Jakub Jelinek <jakub@redhat.com>
16861 PR tree-optimization/56125
16862 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16863 pow(x,c) into sqrt(x) * powi(x, n/2) or
16864 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16865 optimizing for size.
16866 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16867 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16870 PR tree-optimization/56094
16871 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16872 to UNKNOWN_LOCATION while gimplifying expr.
16874 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
16877 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16878 operand 0 in movabs insn template for -masm=intel asm alternative.
16879 (*movabs<mode>_2): Ditto for operand 1.
16881 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
16884 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16887 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16889 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16892 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16894 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16895 Undef to avoid warning.
16897 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16899 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16900 * configure: Regenerate.
16902 2013-01-25 Jakub Jelinek <jakub@redhat.com>
16904 PR tree-optimization/56098
16905 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16906 for stmts with volatile ops.
16907 (cond_store_replacement): Don't optimize if assign has volatile ops.
16908 (cond_if_else_store_replacement_1): Don't optimize if either
16909 then_assign or else_assign have volatile ops.
16910 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16913 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16915 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16917 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16919 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16920 missing ':' in asm example.
16922 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
16924 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16925 entries into lane and laneq entries.
16926 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16927 Remove AdvSIMD scalar modes.
16928 (aarch64_sq<r>dmulh_laneq<mode>): New.
16929 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16931 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16932 builtin implementations to relfect changes in RTL in aarch64-simd.md.
16933 * config/aarch64/iterators.md (VCOND): New.
16936 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16939 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16940 Add NULL LIBNAME argument to existing definitions.
16941 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16942 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16943 * config/avr/avr.c (DEF_BUILTIN): Same.
16944 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16945 (avr_expand_builtin): Expand to a vanilla call if a libgcc
16946 implementation is available (DECL_ASSEMBLER_NAME is set).
16947 (avr_fold_absfx): New static function.
16948 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16949 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16950 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16951 AVR_BUILTIN_ABSLLK.
16952 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16953 (abshk, absk, abslk, absllk): Provide as static inline functions.
16955 2013-01-25 Marek Polacek <polacek@redhat.com>
16957 PR tree-optimization/56035
16958 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16960 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
16962 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16963 (*movtf_internal_rex64): Add (!o,C) alternative
16964 (*movxf_internal_rex64): Ditto.
16965 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16967 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
16969 * doc/invoke.texi: fix typo.
16970 * doc/objc.texi: fix typo.
16972 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
16974 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16975 for the first two alternatives.
16977 2013-01-24 Diego Novillo <dnovillo@google.com>
16979 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
16980 (ggc-zone.o): Remove.
16981 * configure.ac: Remove option --with-gc.
16982 * configure: Re-generate.
16983 * doc/install.texi: Remove documentation for --with-gc.
16984 * gengtype.c (write_enum_defn): Remove. Update all users.
16985 (write_Types_process_field): Remove generation of gt_e_* argument.
16986 (output_type_enum): Remove. Update all users.
16987 (write_enum_defn): Remove. Update all users.
16988 (enum alloc_zone): Remove. Update all users.
16989 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16990 * ggc-common.c (ggc_splay_alloc): Remove first argument.
16991 Update all callers.
16992 (struct ptr_data): Remove field TYPE. Update all users.
16993 (gt_pch_note_object): Remove argument TYPE. Update all users.
16994 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16996 * ggc-none.c (ggc_alloc_typed_stat): Remove.
16997 (struct alloc_zone): Remove.
16998 (ggc_internal_alloc_zone_stat): Remove.
16999 (ggc_internal_cleared_alloc_zone_stat): Remove.
17000 * ggc-page.c (ggc_alloc_typed_stat): Remove.
17001 (ggc_pch_count_object): Remove last argument. Update all users.
17002 (ggc_pch_alloc_object): Remove last argument. Update all users.
17003 (struct alloc_zone): Remove.
17004 * ggc-zone.c: Remove.
17005 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
17006 (struct alloc_zone): Remove.
17007 (ggc_alloc_typed_stat): Remove.
17008 (ggc_alloc_typed): Remove.
17009 (ggc_splay_alloc): Remove first argument.
17010 (rtl_zone): Remove. Update all users.
17011 (tree_zone): Remove. Update all users.
17012 (tree_id_zone): Remove. Update all users.
17013 (ggc_internal_zone_alloc_stat): Remove. Update all users.
17014 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
17015 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
17016 * tree-ssanames.c: Remove references to zone allocator in comments.
17018 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
17020 * config/avr/avr.c (avr_out_fract): Make register numbers that
17021 might be outside of source operand signed.
17023 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
17025 * config/i386/constraints.md (Yf): New constraint.
17026 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17027 of f constraint to conditionaly disable x87 register preferences.
17028 (*movdf_internal): Ditto.
17029 (*movsf_internal): Ditto.
17031 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
17033 PR inline-asm/55934
17034 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17035 that have operands with impossible constraints.
17036 Add a FIXME for a speed-up opportunity.
17037 * lra-constraints.c (process_alt_operands): Verify that a class
17038 selected from constraints on asms is valid for the operand mode.
17039 (curr_insn_transform): Remove incorrect comment.
17041 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
17043 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17044 TOC operand is a valid symbol ref in the constant pool.
17046 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17048 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17050 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
17053 * config/avr/stdfix.h: New file.
17054 * t-avr (stdfix-gcc.h): New rule to build it.
17055 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17057 2013-01-23 Kostya Serebryany <kcc@google.com>
17059 * config/darwin.h: remove dependency on
17060 CoreFoundation (asan on Mac OS).
17062 2013-01-23 Jakub Jelinek <jakub@redhat.com>
17065 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17066 instead of cmpdi_operand for first comparison operand.
17067 Don't assert that comparison operands aren't both constants.
17069 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17071 * doc/install.texi (Downloading the Source): Update references to
17072 downloading separate components.
17074 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17076 * doc/extend.texi (__int128): Improve grammar.
17078 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
17081 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17082 alternative to (o,r).
17083 (*movdi_internal_rex64): Remove (!o,n) alternative.
17084 (DImode immediate->memory splitter): Remove.
17085 (DImode immediate->memory peephole2): Remove.
17086 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17087 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17088 alternative to (!o,*r).
17089 (*movtf_internal_sse): New pattern.
17090 (*movxf_internal_rex64): New pattern.
17091 (*movxf_internal): Disable for TARGET_64BIT.
17092 (*movdf_internal_rex64): Remove (!o,F) alternative.
17094 2013-01-22 Jakub Jelinek <jakub@redhat.com>
17096 PR middle-end/56074
17097 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17098 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17099 * tree-vect-loop-manip.c (find_loop_location): Also ignore
17100 stmt locations where LOCATION_LOCUS of the stmt location is
17101 UNKNOWN_LOCATION or BUILTINS_LOCATION.
17104 * config/i386/i386.md (UNSPEC_STOS): New.
17105 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17106 *strsetqi_1): Add UNSPEC_STOS.
17108 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
17111 * doc/invoke.texi: Remove left over -Wsynth example.
17113 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17115 PR tree-optimization/56051
17116 * fold-const.c (fold_binary_loc): Don't fold
17117 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17118 a narrowing conversion, or widening conversion from signed
17121 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17123 PR rtl-optimization/56023
17124 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17125 dependent on debug instruction.
17127 2013-01-21 Martin Jambor <mjambor@suse.cz>
17129 PR middle-end/56022
17130 * function.c (allocate_struct_function): Call
17131 invoke_set_current_function_hook earlier.
17133 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17135 * reload1.c (init_reload): Only initialize reload_obstack
17136 during the first call.
17138 2013-01-21 Marek Polacek <polacek@redhat.com>
17140 * cfgloop.c (verify_loop_structure): Fix up grammar.
17142 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
17144 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17145 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17147 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17150 * config/arm/marvell-pj4.md: Update copyright year.
17151 Fix up use of alu to alu_reg and simple_alu_imm.
17153 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17155 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17157 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
17160 * lra-constraints.c (curr_insn_transform): Don't reuse original
17161 insn for secondary memory move when memory mode should be different.
17163 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17165 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17166 atomic_storedi_1): New patterns.
17168 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
17170 btver2 pipeline descriptions.
17171 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17173 * config/i386/i386.md (btver2_decode): New type attributes.
17174 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17176 * config/i386/btver2.md: New file describing btver2 pipelines.
17178 2013-01-19 Andrew Pinski <apinski@cavium.com>
17180 PR tree-optimization/52631
17181 * tree-ssa-sccvn (visit_use): Before looking up the original
17182 statement, try looking up the simplified expression.
17184 2013-01-19 Anthony Green <green@moxielogic.com>
17186 * config/moxie/moxie.c (moxie_expand_prologue): Set
17187 current_function_static_stack_size.
17189 2013-01-18 Jakub Jelinek <jakub@redhat.com>
17191 PR tree-optimization/56029
17192 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17193 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17195 2013-01-18 Sharad Singhai <singhai@google.com>
17197 PR tree-optimization/55995
17198 * dumpfile.c (dump_loc): Print location only if available.
17199 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17201 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
17204 * lra-constraints.c (curr_insn_transform): Reuse original insn for
17205 secondary memory move.
17206 (inherit_reload_reg): Use rclass instead of cl for
17207 check_secondary_memory_needed_p.
17209 2013-01-18 Jakub Jelinek <jakub@redhat.com>
17211 PR middle-end/56015
17212 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17213 the case where writing real complex part of target modifies op1.
17215 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
17217 * config/aarch64/aarch64-simd.md
17218 (aarch64_vcond_internal<mode>): Handle unordered cases.
17219 * config/aarch64/iterators.md (v_cmp_result): New.
17221 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
17222 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17224 * config/arm/marvell-pj4.md: New file.
17225 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17226 * config/arm/arm.md (generic_sched): Add marvell_pj4.
17227 (generic_vfp): Likewise.
17228 * config/arm/arm-cores.def: Add marvell-pj4.
17229 * config/arm/arm-tune.md: Regenerate.
17230 * config/arm/arm-tables.opt: Regenerate.
17231 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17232 * doc/invoke.texi: Document marvell-pj4.
17234 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
17236 * config/aarch64/arm_neon.h: Map scalar types to standard types.
17238 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17243 * var-tracking.c (negative_power_of_two_p): New.
17244 (global_get_addr_cache, local_get_addr_cache): New.
17245 (get_addr_from_global_cache, get_addr_from_local_cache): New.
17246 (vt_canonicalize_addr): Rewrite using the above. Adjust the
17248 (vt_stack_offset_p): Remove.
17249 (vt_canon_true_dep): Always canonicalize loc's address.
17250 (clobber_overlapping_mems): Make sure we have a MEM.
17251 (local_get_addr_clear_given_value): New.
17252 (val_reset): Clear local cached entries.
17253 (compute_bb_dataflow): Create and release the local cache.
17254 Disable duplicate MEMs clobbering.
17255 (emit_notes_in_bb): Clobber MEMs likewise.
17256 (vt_emit_notes): Create and release the local cache.
17257 (vt_initialize, vt_finalize): Create and release the global
17258 cache, respectively.
17259 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17261 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17263 PR libmudflap/53359
17264 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17265 not found in the symtab.
17267 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17270 PR rtl-optimization/55547
17271 PR rtl-optimization/53827
17274 * alias.c (offset_overlap_p): New, factored out of...
17275 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
17276 the conservative special case for symbolic constants. Don't
17277 adjust zero sizes on alignment.
17279 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
17281 PR rtl-optimization/52573
17282 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17283 REG_UNUSED for the same register.
17285 2013-01-17 Richard Biener <rguenther@suse.de>
17286 Marek Polacek <polacek@redhat.com>
17288 PR rtl-optimization/55833
17289 * loop-unswitch.c (unswitch_loops): Move loop verification...
17290 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
17291 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17292 Set it to true when we're removing a loop from hierarchy tree in
17293 an irreducible region.
17294 (fix_bb_placements): Adjust caller.
17295 (fix_loop_placements): Likewise.
17297 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
17299 * config/avr/builtins.def (DEF_BUILTIN): Factor out
17300 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17301 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17302 Remove ID. Adjust comments.
17303 * config/avr/avr-c.c (avr_builtin_name): Remove.
17304 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17305 * config/avr/avr.c (avr_tolower): New static function.
17306 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
17307 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17308 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17311 2013-01-17 Jan Hubicka <jh@suse.cz>
17313 PR tree-optimization/55273
17314 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17316 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
17319 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17320 store through atomic_store<mode>_1.
17321 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17323 2013-01-17 Martin Jambor <mjambor@suse.cz>
17325 PR tree-optimizations/55264
17326 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17327 for virtual methods.
17328 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17329 virtual methods before inlining is over.
17330 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17332 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17335 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
17337 PR rtl-optimization/56005
17338 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17339 pending reads for prefetch.
17341 2013-01-16 Ian Bolton <ian.bolton@arm.com>
17343 * config/aarch64/aarch64.md
17344 (*cstoresi_neg_uxtw): New pattern.
17345 (*cmovsi_insn_uxtw): New pattern.
17346 (*<optab>si3_uxtw): New pattern.
17347 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17348 (*<optab>si3_insn_uxtw): New pattern.
17349 (*bswapsi2_uxtw): New pattern.
17351 2013-01-16 Richard Biener <rguenther@suse.de>
17353 * tree-inline.c (tree_function_versioning): Remove set but
17354 never used variable.
17356 2013-01-16 Richard Biener <rguenther@suse.de>
17358 PR tree-optimization/55964
17359 * tree-flow.h (rename_variables_in_loop): Remove.
17360 (rename_variables_in_bb): Likewise.
17361 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17362 (copy_loop_before): Adjust and delete update-ssa status.
17363 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17364 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
17365 (rename_variables_in_loop): Remove.
17366 (slpeel_update_phis_for_duplicate_loop): Likewise.
17367 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17368 use available cfg machinery instead of duplicating it.
17369 Update PHI nodes and perform poor-mans SSA update here.
17370 (slpeel_tree_peel_loop_to_edge): Adjust.
17372 2013-01-16 Richard Biener <rguenther@suse.de>
17374 PR tree-optimization/54767
17375 PR tree-optimization/53465
17376 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17377 (vrp_visit_phi_node): For PHI arguments coming via backedges
17378 drop all symbolical range information.
17379 (execute_vrp): Compute backedges.
17381 2013-01-16 Richard Biener <rguenther@suse.de>
17383 * doc/install.texi: Update CLooG and ISL requirements to
17386 2013-01-16 Christian Bruel <christian.bruel@st.com>
17389 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17390 (broken_move): Handle UNSPECV_SP_SWITCH_B.
17391 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17393 2013-01-16 DJ Delorie <dj@redhat.com>
17395 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17396 (UNSPECV_SP_SWITCH_E): New.
17397 (sp_switch_1): Change to an unspec.
17398 (sp_switch_2): Change to an unspec. Don't use post-inc when we
17401 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
17403 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17404 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17405 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17406 (expand_mem_thread_fence): Ditto.
17407 (expand_mem_signal_fence): Ditto.
17408 (expand_atomic_load): Ditto.
17409 (expand_atomic_store): Ditto.
17411 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
17413 PR rtl-optimization/55547
17414 PR rtl-optimization/53827
17417 * alias.c (memrefs_conflict_p): Set sizes to negative after
17420 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17423 * function.c (thread_prologue_and_epilogue_insns): Always
17424 add crtl->drap_reg to set_up_by_prologue.set, even if
17425 stack_realign_drap is false.
17427 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
17429 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17430 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17431 *call): Fix indention.
17433 2013-01-15 Tom de Vries <tom@codesourcery.com>
17436 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17439 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
17441 PR rtl-optimization/55153
17442 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17444 2013-01-15 Martin Jambor <mjambor@suse.cz>
17446 PR tree-optimization/55920
17447 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17448 accesses as grp_to_be_debug_replaced.
17450 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17452 PR tree-optimization/55920
17453 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17454 there is non-useless type conversion needed from debug rhs to lhs,
17455 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17457 2013-01-15 Joseph Myers <joseph@codesourcery.com>
17458 Mikael Pettersson <mikpe@it.uu.se>
17461 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17463 (ASM_OUTPUT_CASE_LABEL): Remove.
17464 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17465 * final.c (shorten_branches): Update alignment of labels before
17466 jump tables if CASE_VECTOR_SHORTEN_MODE.
17468 2013-01-15 Richard Biener <rguenther@suse.de>
17471 * system.h: Do not include gmp.h for building host tools.
17473 2013-01-15 Richard Biener <rguenther@suse.de>
17475 PR middle-end/55882
17476 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17477 account for bitpos when computing alignment.
17479 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
17481 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17482 (ix86_target_macros_internal): Likewise.
17484 * config/i386/i386.c (m_CORE2I7): Removed.
17485 (m_CORE_HASWELL): New macro.
17486 (m_CORE_ALL): Likewise.
17487 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17488 (initial_ix86_arch_features): Likewise.
17489 (processor_target_table): Initializations for Core avx2.
17490 (cpu_names): New names "core-avx2".
17491 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17492 PROCESSOR_CORE_HASWELL.
17493 (ix86_issue_rate): New case.
17494 (ia32_multipass_dfa_lookahead): Likewise.
17495 (ix86_sched_init_global): Likewise.
17497 * config/i386/i386.h (TARGET_HASWELL): New macro.
17498 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17499 (processor_type): New PROCESSOR_HASWELL.
17501 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17503 PR tree-optimization/55955
17504 * tree-vect-loop.c (vectorizable_reduction): Give up early on
17505 *SHIFT_EXPR and *ROTATE_EXPR codes.
17507 PR tree-optimization/48766
17508 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17509 -ftrapv disable -fwrapv.
17511 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17514 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17515 etc. to 1 and not to __flash.
17516 Use LL suffix for __INT24_MAX__ with -mint8.
17517 Use ULL suffix for __UINT24_MAX__ with -mint8.
17519 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17521 * config/avr/avr-arch.h
17522 (struct base_arch_s): Use typedef avr_arch_t instead.
17523 (struct arch_info_s): Use typedef avr_arch_info_t instead.
17524 (struct mcu_type_s): Use typedef avr_mcu_t instead.
17525 * config/avr/avr.c: Same.
17526 * config/avr/avr-devices.c: Same.
17527 * config/avr/driver-avr.c: Same.
17528 * config/avr/gen-avr-mmcu-texi.c: Same.
17529 * config/avr/avr-mcus.def: Adjust comment.
17531 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
17533 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17534 * config/aarch64/iterators.md (VALLDI): New.
17536 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
17537 Andi Kleen <ak@linux.intel.com>
17540 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17541 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17544 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17546 * config/avr/avr-stdint.h: Remove trailing blanks.
17547 * config/avr/avr-log.h: Same.
17548 * config/avr/avr-arch.h: Same.
17549 * config/avr/avr-devices.c: Same.
17550 * config/avr/avr-dimode.md: Same.
17551 * config/avr/predicates.md: Same.
17552 * config/avr/avr-c.c: Same. And fix typo.
17554 * config/avr/avr-protos.h: Same. And:
17555 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17556 (init_cumulative_args): Rename to avr_init_cumulative_args.
17557 (expand_prologue): Rename to avr_expand_prologue.
17558 (expand_epilogue): Rename to avr_expand_epilogue.
17559 (adjust_insn_length): Rename to avr_adjust_insn_length.
17560 (notice_update_cc): Rename to avr_notice_update_cc.
17561 (final_prescan_insn): Rename to avr_final_prescan_insn.
17562 * config/avr/avr.c: Same.
17563 * config/avr/avr.h: Same.
17564 * config/avr/avr.md: Remove trailing blanks.
17565 (prologue): Use avr_expand_prologue.
17566 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17568 2013-01-14 Richard Biener <rguenther@suse.de>
17570 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17571 verify_location, collect_subblocks): New functions.
17572 (verify_gimple_in_cfg): Verify that locations only reference
17573 BLOCKs in the functions BLOCK tree.
17575 2013-01-14 Richard Biener <rguenther@suse.de>
17577 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17579 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17581 (insert_out_of_ssa_copy_on_edge): Likewise.
17582 (rewrite_close_phi_out_of_ssa): Likewise.
17583 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17585 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17586 propagated constants.
17587 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17590 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17592 * config/avr/avr-modes.def: Add GPL copyright notice.
17594 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
17596 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17597 MEMMODEL_MASK to determine memory model.
17598 (atomic_store<mode>): Ditto from operands[2].
17599 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17601 2013-01-13 Jakub Jelinek <jakub@redhat.com>
17604 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17605 (fold_gimple_assign): Don't call unshare_expr here.
17606 (fold_ctor_reference): Call unshare_expr.
17608 2013-01-13 Terry Guo <terry.guo@arm.com>
17610 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17611 * doc/fragments.texi: Document MULTILIB_REUSE.
17612 * gcc.c (multilib_reuse): New internal spec.
17613 (set_multilib_dir): Also search multilib from multilib_reuse.
17614 * genmultilib (tmpmultilib3): Refactor code.
17615 (tmpmultilib4): Ditto.
17616 (multilib_reuse): New multilib argument.
17618 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
17620 * Makefile.in: Update copyright.
17622 2013-01-12 Tom de Vries <tom@codesourcery.com>
17624 PR middle-end/55890
17625 * calls.c (expand_call): Check if arg_nr is valid.
17627 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17629 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17630 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17631 documentation. Add missing '__' in front of
17632 __builtin_ia32_packssdw256.
17634 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17637 * config/s390/s390.c (s390_preferred_reload_class): Do not return
17638 NO_REGS for larl operands.
17639 (s390_reload_larl_operand): Use s390_load_address instead of
17642 2013-01-11 Richard Biener <rguenther@suse.de>
17644 * tree-cfg.c (verify_node_sharing_1): Split out from ...
17645 (verify_node_sharing): ... here.
17646 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17648 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
17650 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17651 Substitute TREECHECKING.
17652 * configure: Regenerate.
17653 * Makefile.in (TREECHECKING): New.
17655 2013-01-11 Richard Guenther <rguenther@suse.de>
17657 PR tree-optimization/44061
17658 * tree-vrp.c (extract_range_basic): Compute zero as
17659 value-range for __builtin_constant_p of function parameters.
17661 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
17663 Update copyright years.
17665 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
17667 PR rtl-optimization/55672
17668 * lra-eliminations.c (mark_not_eliminable): Permit addition with
17669 const to be eliminable.
17671 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
17673 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17674 * configure: Regenerate.
17676 2013-01-10 Richard Biener <rguenther@suse.de>
17678 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17680 2013-01-10 Richard Biener <rguenther@suse.de>
17683 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17684 locations for virtual PHI arguments.
17685 (rewrite_update_phi_arguments): Likewise.
17687 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
17689 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17692 2013-01-10 Jakub Jelinek <jakub@redhat.com>
17694 PR tree-optimization/55921
17695 * tree-complex.c (expand_complex_asm): New function.
17696 (expand_complex_operations_1): Call it for GIMPLE_ASM.
17698 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17701 * config/s390/s390.c (s390_symref_operand_p)
17702 (s390_loadrelative_operand_p): Merge the two functions.
17703 (s390_check_qrst_address, print_operand_address): Add parameters
17704 to s390_loadrelative_operand_p invokation.
17705 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17706 (s390_reload_larl_operand, s390_secondary_reload): Use
17707 s390_loadrelative_operand_p instead of s390_symref_operand_p.
17708 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17710 2013-01-09 Mike Stump <mikestump@comcast.net>
17712 * dse.c (record_store): Remove unnecessary assert.
17714 2013-01-09 Jan Hubicka <jh@suse.cz>
17716 PR tree-optimization/55569
17717 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17718 * cfgloop.h (scale_loop_profile): Likewise.
17720 2013-01-09 Jan Hubicka <jh@suse.cz>
17723 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17725 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17727 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
17729 PR middle-end/55114
17730 * expr.h (maybe_emit_group_store): Declare.
17731 * expr.c (maybe_emit_group_store): New function.
17732 * builtins.c (expand_builtin_int_roundingfn): Call it.
17733 (expand_builtin_int_roundingfn_2): Likewise.
17735 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
17737 PR rtl-optimization/55829
17738 * lra-constraints.c (match_reload): Add code for absent output.
17739 (curr_insn_transform): Add code for reloads of matched inputs
17742 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
17744 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17745 attribute of movddup insn to DF.
17746 (*vec_interleave_lowv2df): Ditto.
17747 (vec_dupv2df): Ditto.
17749 2013-01-09 Jan Hubicka <jh@suse.cz>
17751 PR tree-optimiation/55875
17752 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17753 EVERY_ITERATION parameter.
17754 (number_of_iterations_exit): Check if exit is executed every iteration.
17755 (idx_infer_loop_bounds): Similarly here.
17756 (n_of_executions_at_most): Simplify
17757 to only test for cases where statement is dominated by the
17758 particular bound; handle correctly the "postdominance" test.
17759 (scev_probably_wraps_p): Use max loop iterations info
17760 as a global bound first.
17762 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
17763 Nick Clifton <nickc@redhat.com>
17765 * config/v850/v850.md (cbranchsf4): New pattern.
17766 (cstoresf4): New pattern.
17767 (cbranchdf4): New pattern.
17768 (cstoredf4): New pattern.
17769 (movsicc): Disallow floating point comparisons.
17770 (cmpsf_le_insn): Fix order of operators.
17771 (cmpsf_lt_insn): Likewise.
17772 (cmpsf_eq_insn): Likewise.
17773 (cmpdf_le_insn): Likewise.
17774 (cmpdf_lt_insn): Likewise.
17775 (cmpdf_eq_insn): Likewise.
17776 (cmpsf_ge_insn): Use LE comparison.
17777 (cmpdf_ge_insn): Likewise.
17778 (cmpsf_gt_insn): Use LT comparison.
17779 (cmpdf_gt_insn): Likewise.
17780 (cmpsf_ne_insn): Delete pattern.
17781 (cmpdf_ne_insn): Delete pattern.
17782 * config/v850/v850.c (v850_gen_float_compare): Use
17783 gen_cmpdf_eq_insn for NE comparison.
17784 (v850_float_z_comparison_operator)
17785 (v850_float_nz_comparison_operator): Move from here ...
17786 * config/v850/predicates.md: ... to here. Move GT and GE
17787 comparisons into v850_float_z_comparison_operator.
17788 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17790 (v850_float_nz_comparison_operator): Likewise.
17792 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17794 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17795 with calls to gen_insvsi/gen_insvdi.
17797 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
17799 * config/i386/i386.c (initial_ix86_tune_features): Set up
17800 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17802 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
17803 Jakub Jelinek <jakub@redhat.com>
17805 PR tree-optimization/48189
17806 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17807 If nitercst is 0, don't predict the exit edge.
17809 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17811 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17812 in asm_fprintf with reg_names.
17813 (aarch64_print_operand_address): Likewise.
17814 (aarch64_return_addr): Likewise.
17815 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17817 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17819 * config/pa/pa.h (VAL_U6_BITS_P): Define.
17820 (INT_U6_BITS): Likewise.
17821 * config/pa/predicates.md (uint6_operand): New predicate.
17822 (shift5_operand, shift6_operand): Likewise.
17823 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17825 (lshrdi3): Use shift6_operand.
17826 (shrpsi4, shrpdi4): New insn patterns.
17827 (extzv): Delete expander.
17828 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
17829 predicates in unamed zero extract patterns. Tighten common constraint.
17830 (extv): Delete expander.
17831 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
17832 predicates in unamed sign extract patterns. Tighten common constraint.
17833 (insv): Delete expander.
17834 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
17835 predicates in unamed insert patterns. Tighten common constraint.
17836 Change uint32_operand predicate to uint6_operand predicate in unamed
17837 DImode pattern to insert constant values of type 1...1xxxx.
17839 2013-01-04 Jan Hubicka <jh@suse.cz>
17841 PR tree-optimization/55823
17842 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17845 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17846 Uros Bizjak <ubizjak@gmail.com>
17848 PR rtl-optimization/55845
17849 * df-problems.c (can_move_insns_across): Stop scanning at
17850 volatile_insn_p source instruction or give up if
17851 across_from .. across_to range contains any volatile_insn_p
17854 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17856 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17857 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17859 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17860 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17862 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17865 * asan.c (asan_clear_shadow): New function.
17866 (asan_emit_stack_protection): Use it.
17868 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17870 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17871 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17872 with tab instead of space.
17874 2013-01-08 Nick Clifton <nickc@redhat.com>
17876 * config/rl78/rl78.c (rl78_expand_prologue): Always select
17877 register bank 0 at the start of an interrupt handler.
17878 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17881 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17883 * config/aarch64/aarch64-simd.md
17884 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17885 (aarch64_simd_bsl): Likewise.
17886 (aarch64_vcond_internal<mode>): Likewise.
17887 (vcond<mode><mode>): Likewise.
17888 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17889 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17891 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17893 * config/aarch64/aarch64-builtins.c
17894 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17896 2013-01-08 Martin Jambor <mjambor@suse.cz>
17899 * tree-sra.c (analyze_access_subtree): Return true also after
17900 potentially creating a debug-only replacement.
17902 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17904 PR middle-end/55890
17905 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17907 PR tree-optimization/54120
17908 * tree-vrp.c (range_fits_type_p): Don't allow
17909 src_precision < precision from signed vr to unsigned_p
17910 if vr->min or vr->max is negative.
17911 (simplify_float_conversion_using_ranges): Test can_float_p
17912 against CODE_FOR_nothing.
17914 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17915 Richard Biener <rguenther@suse.de>
17917 PR middle-end/55851
17918 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17919 types instead of just INTEGER_TYPE types.
17921 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
17923 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17926 2013-01-07 Steve Ellcey <sellcey@mips.com>
17929 * config/mips/mips.opt: Change mad to mmad to match documentation.
17931 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17934 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17935 .progmemx.data now.
17937 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17940 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17941 (avr_addrspace_t): Add .section_name field.
17942 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17944 (avr_addrspace): Same. Initialize .section_name. Remove last
17945 NULL entry. Put __memx into .progmemx.data.
17946 (progmem_section_prefix): Remove.
17947 (avr_asm_init_sections): No need to initialize progmem_section.
17948 (avr_asm_named_section): Use avr_addrspace[].section_name to get
17949 section name prefix.
17950 (avr_asm_select_section): Ditto. And use get_unnamed_section to
17951 retrieve the progmem section.
17952 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17953 boundary to run over avr_addrspace[].
17954 (avr_register_target_pragmas): Ditto.
17956 2013-01-06 Jakub Jelinek <jakub@redhat.com>
17958 * varasm.c (output_constant_def_contents): For asan_protect_global
17959 protected strings, adjust DECL_ALIGN if needed, before testing for
17961 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17962 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
17964 (output_object_block): For asan protected decls, emit asan padding
17965 after their contents.
17966 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17967 (asan_finish_file): Test it here instead.
17969 2013-01-07 Nick Clifton <nickc@redhat.com>
17970 Matthias Klose <doko@debian.org>
17971 Doug Kwan <dougkwan@google.com>
17972 H.J. Lu <hongjiu.lu@intel.com>
17975 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17977 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17979 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17981 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17983 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17985 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17988 * doc/install.texi (Cross-Compiler-Specific Options): Document
17991 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
17993 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17994 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17995 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17996 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17997 vqmovun_high_s64): Fix source operand number and update copyright.
17999 2013-01-07 Richard Biener <rguenther@suse.de>
18001 PR middle-end/55890
18002 * gimple.h (gimple_call_builtin_p): New overload.
18003 * gimple.c (validate_call): New function.
18004 (gimple_call_builtin_p): Likewise.
18005 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18006 Use gimple_call_builtin_p.
18007 (find_func_clobbers): Likewise.
18008 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18009 (strlen_optimize_stmt): Likewise.
18011 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18013 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18014 (vld1q_dup_*): Likewise.
18015 (vld1_*): Likewise.
18016 (vld1q_*): Likewise.
18017 (vld1_lane_*): Likewise.
18018 (vld1q_lane_*): Likewise.
18020 2013-01-07 Richard Biener <rguenther@suse.de>
18022 * lto-streamer.h (LTO_minor_version): Bump to 2.
18024 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18026 * config/aarch64/aarch64-protos.h
18027 (aarch64_const_double_zero_rtx_p): Rename to...
18028 (aarch64_float_const_zero_rtx_p): ...this.
18029 (aarch64_float_const_representable_p): New.
18030 (aarch64_output_simd_mov_immediate): Likewise.
18031 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18032 move immediate case.
18033 * config/aarch64/aarch64.c
18034 (aarch64_const_double_zero_rtx_p): Rename to...
18035 (aarch64_float_const_zero_rtx_p): ...this.
18036 (aarch64_print_operand): Allow printing of new constants.
18037 (aarch64_valid_floating_const): New.
18038 (aarch64_legitimate_constant_p): Check for valid floating-point
18040 (aarch64_simd_valid_immediate): Likewise.
18041 (aarch64_vect_float_const_representable_p): New.
18042 (aarch64_float_const_representable_p): Likewise.
18043 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18044 (aarch64_output_simd_mov_immediate): New.
18045 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18046 (*movdf_aarch64): Likewise.
18047 * config/aarch64/constraints.md (Ufc): New.
18048 (Y): call aarch64_float_const_zero_rtx.
18049 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18051 2013-01-07 Richard Biener <rguenther@suse.de>
18053 PR tree-optimization/55888
18054 PR tree-optimization/55862
18055 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18056 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18057 not if it is contained therein.
18059 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18061 * config/avr/t-avr: Typo.
18063 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18066 * config/avr/t-avr: Don't automatically rebuild
18067 $(srcdir)/config/avr/t-multilib
18068 $(srcdir)/config/avr/avr-tables.opt
18069 $(srcdir)/doc/avr-mmcu.texi
18070 (avr-mcus): New phony target to build them on request.
18071 (s-avr-mlib, s-avr-mmcu-texi): Remove.
18072 * avr/avr-mcus.def: Adjust comments.
18074 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
18076 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18078 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18080 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18082 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18084 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18086 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
18088 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18089 to generate profiling.
18090 * config/rs6000/aix64.h (LIB_SPEC): Same.
18092 2013-01-04 Andrew Pinski <apinski@cavium.com>
18094 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18096 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18098 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
18100 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18102 (ix86_expand_move): Ditto.
18103 (ix86_zero_extend_to_Pmode): Ditto.
18104 (ix86_expand_call): Ditto.
18105 (ix86_expand_special_args_builtin): Ditto.
18106 (ix86_expand_builtin): Ditto.
18108 2013-01-04 Richard Biener <rguenther@suse.de>
18110 PR tree-optimization/55862
18111 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18112 translating them through PHI nodes.
18114 2013-01-04 Martin Jambor <mjambor@suse.cz>
18116 PR tree-optimization/55755
18117 * tree-sra.c (sra_modify_assign): Do not check that an access has no
18118 children when trying to avoid producing a VIEW_CONVERT_EXPR.
18120 2013-01-04 Marek Polacek <polacek@redhat.com>
18122 PR middle-end/55859
18123 * opts.c (default_options_optimization): Clarify error message.
18125 2013-01-04 Richard Biener <rguenther@suse.de>
18127 PR middle-end/55863
18128 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18131 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18134 * config/pa/pa.md (movsi): Revert previous change.
18135 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18138 2013-01-03 Richard Henderson <rth@redhat.com>
18140 * config/i386/i386.c (ix86_expand_move): Always assign to op1
18141 after eliminating TLS symbols.
18143 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
18146 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18147 * graphite-poly.c (debug_gmp_value): Likewise.
18149 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
18152 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18153 selected code model, define __code_mode_small__, __code_model_medium__,
18154 __code_model_large__, __code_model_32__ or __code_model_kernel__.
18155 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18156 xchg temporary register with %k. Declare temporary register as
18158 [__x86_64__]: For medium and large code models, preserve %rbx register.
18160 2013-01-03 Richard Biener <rguenther@suse.de>
18162 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18163 (dump_subscript): Adjust.
18164 (finalize_ddr_dependent): Do not dump redundant info.
18165 (analyze_siv_subscript): Adjust.
18166 (subscript_dependence_tester): Likewise.
18167 (compute_affine_dependence): Likewise.
18169 2013-01-03 Richard Biener <rguenther@suse.de>
18172 2013-01-03 Richard Biener <rguenther@suse.de>
18174 PR tree-optimization/55857
18175 * tree-vect-stmts.c (vectorizable_load): Do not setup
18176 re-alignment for invariant loads.
18178 2013-01-02 Richard Biener <rguenther@suse.de>
18180 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18181 invariant load do not generate a vector load from the scalar location.
18183 2013-01-03 Richard Biener <rguenther@suse.de>
18185 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18186 for not vectorizing.
18187 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18188 not build INDIRECT_REFs, call get_name once only.
18189 (vect_create_data_ref_ptr): Likewise. Dump base object kind
18190 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18192 2013-01-03 Richard Biener <rguenther@suse.de>
18194 PR tree-optimization/55857
18195 * tree-vect-stmts.c (vectorizable_load): Do not setup
18196 re-alignment for invariant loads.
18198 2013-01-03 Richard Biener <rguenther@suse.de>
18201 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18202 prefer a built-in decl.
18204 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18206 * gcc.c (process_command): Update copyright notice dates.
18207 * gcov.c (print_version): Likewise.
18208 * gcov-dump.c (print_version): Likewise.
18210 PR rtl-optimization/55838
18211 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18212 iv0.step, iv1.step and step.
18214 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18215 Marc Glisse <marc.glisse@inria.fr>
18217 PR tree-optimization/55832
18218 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18219 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18220 integer_{one,zero}_node.
18222 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18225 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18226 * var-tracking.c (reverse_op): Don't add reverse ops to
18227 VALUEs that have already
18228 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18230 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
18232 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18234 2013-01-02 Teresa Johnson <tejohnson@google.com>
18236 * dumpfile.c (dump_loc): Print filename with location.
18237 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18238 new location_t parameter to emit complete unroll message with
18239 new dump framework.
18240 (canonicalize_loop_induction_variables): Compute loops location
18241 and pass to try_unroll_loop_completely.
18242 * loop-unroll.c (report_unroll_peel): New function.
18243 (peel_loops_completely): Use new dump format with location
18244 for main dumpfile message, and invoke report_unroll_peel on success.
18245 (decide_unrolling_and_peeling): Ditto.
18246 (decide_peel_once_rolling): Remove old dumpfile message subsumed
18247 by report_unroll_peel.
18248 (decide_peel_completely): Ditto.
18249 (decide_unroll_constant_iterations): Ditto.
18250 (decide_unroll_runtime_iterations): Ditto.
18251 (decide_peel_simple): Ditto.
18252 (decide_unroll_stupid): Ditto.
18253 * cfgloop.c (get_loop_location): New function.
18254 * cfgloop.h (get_loop_location): Declare.
18256 2013-01-02 Sriraman Tallam <tmsriram@google.com>
18258 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18261 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18263 PR middle-end/55198
18264 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18265 BLKmode objects when EXPAND_MEMORY is specified.
18267 2013-01-02 Sriraman Tallam <tmsriram@google.com>
18269 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18271 (fold_builtin_cpu): Do not share cpu model decls across statements.
18273 2013-01-02 Jason Merrill <jason@redhat.com>
18276 * tree.c (build_array_type_1): Revert earlier change.
18278 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
18280 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18282 * config/aarch64/aarch64-tune.md: Re-generate.
18284 2013-01-02 Richard Biener <rguenther@suse.de>
18286 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18287 invariant load do not generate a vector load from the scalar location.
18289 2013-01-02 Richard Biener <rguenther@suse.de>
18292 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18293 * configure: Regenerate.
18295 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18297 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18298 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18299 (expand_builtin_int_roundingfn_2): Keep the original target around
18300 for the fallback case.
18302 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18304 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18305 to be clear for sign changes.
18307 2013-01-01 Jan Hubicka <jh@suse.cz>
18309 * ipa-inline-analysis.c: Fix formatting.
18311 2013-01-01 Jakub Jelinek <jakub@redhat.com>
18313 PR tree-optimization/55831
18314 * tree-vect-loop.c (get_initial_def_for_induction): Use
18315 gsi_after_labels instead of gsi_start_bb.
18317 Copyright (C) 2013 Free Software Foundation, Inc.
18319 Copying and distribution of this file, with or without modification,
18320 are permitted in any medium without royalty provided the copyright
18321 notice and this notice are preserved.