1 2013-09-25 Tom Tromey <tromey@redhat.com>
3 * Makefile.in (C_TREE_H): Reference c/c-tree.h.
5 2013-09-25 Tom Tromey <tromey@redhat.com>
7 * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
8 to add -DENABLE_SHARED_LIBGCC.
9 (gcc.o): Don't use subshell.
11 2013-09-25 Tom Tromey <tromey@redhat.com>
13 * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
14 * configure.ac: Don't invoke AM_PROG_CC_C_O.
15 (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
16 * configure, config.in: Rebuild.
18 2013-09-25 Tom Tromey <tromey@redhat.com>
20 * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
21 (COMPILE, POSTCOMPILE): New variables.
22 (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
23 (DEPFILES): New variable.
25 * configure.ac: Add checks for dependency checking.
26 * configure, aclocal.m4: Regenerate.
28 2013-09-25 Tom Tromey <tromey@redhat.com>
30 * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
31 ($(ALL_HOST_OBJS)): Move order-only dependency to end
34 2013-09-25 Tom Tromey <tromey@redhat.com>
36 * Makefile.in (generated_files): Add options.h,
37 target-hooks-def.h, insn-opinit.h,
38 common/common-target-hooks-def.h, pass-instances.def,
39 c-family/c-target-hooks-def.h.
41 2013-09-25 Jeff Law <law@redhat.com>
43 * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
44 than foo[foo.length () - 1] to access last member in a vec.
45 * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
47 2013-09-25 Richard Biener <rguenther@suse.de>
50 * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
52 2013-09-25 Jan Hubicka <jh@suse.cz>
54 * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
57 2013-09-25 Marek Polacek <polacek@redhat.com>
60 * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
61 when determining the type name.
63 2013-09-24 Oleg Endo <olegendo@gcc.gnu.org>
65 * config/sh/sh.md: Fix formatting.
67 2013-09-24 Xinliang David Li <davidxl@google.com>
69 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
70 Check max peel iterations parameter.
71 * param.def: New parameter.
72 * doc/invoke.texi: Document New parameter.
74 2013-09-24 Christophe Lyon <christophe.lyon@linaro.org>
76 * gimple-pretty-print.c: Various whitespace tweaks.
77 * tree-core.h: Likewise.
78 * tree-pretty-print.c: Likewise.
79 * tree-ssa-alias.c: Likewise.
80 * tree-ssa-copy.c: Likewise.
81 * tree-ssanames.c: Likewise.
82 * tree-ssanames.h: Likewise.
83 * tree-vrp.c: Likewise.
85 2013-09-24 Alan Modra <amodra@gmail.com>
89 * stmt.c (expand_asm_operands): Call expand_expr with
90 EXPAND_MEMORY for output operands that disallow regs. Don't
91 use EXPAND_WRITE on inout operands.
93 2013-09-24 Richard Biener <rguenther@suse.de>
96 * tree.c (reference_alias_ptr_type): Move ...
97 * alias.c (reference_alias_ptr_type): ... here and implement
98 in terms of the new reference_alias_ptr_type_1.
99 (ref_all_alias_ptr_type_p): New helper.
100 (get_deref_alias_set_1): Drop flag_strict_aliasing here,
101 use ref_all_alias_ptr_type_p.
102 (get_deref_alias_set): Add flag_strict_aliasing check here.
103 (reference_alias_ptr_type_1): New function, split out from ...
104 (get_alias_set): ... here.
105 (alias_ptr_types_compatible_p): New function.
106 * alias.h (reference_alias_ptr_type): Declare.
107 (alias_ptr_types_compatible_p): Likewise.
108 * tree.h (reference_alias_ptr_type): Remove.
109 * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
110 to compare MEM_REF alias types.
112 2013-09-24 Richard Biener <rguenther@suse.de>
114 * tree-vrp.c (vrp_finalize): Check for SSA name presence.
116 2013-09-23 Michael Meissner <meissner@linux.vnet.ibm.com>
118 * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
119 reload helper function arrays into a single array reg_addr.
120 (reload_fpr_gpr): Likewise.
121 (reload_gpr_vsx): Likewise.
122 (reload_vsx_gpr): Likewise.
123 (struct rs6000_reg_addr): Likewise.
124 (reg_addr): Likewise.
125 (rs6000_debug_reg_global): Change rs6000_vector_reload,
126 reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
127 (rs6000_init_hard_regno_mode_ok): Likewise.
128 (rs6000_secondary_reload_direct_move): Likewise.
129 (rs6000_secondary_reload): Likewise.
131 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
132 constraints: wu, ww, and wy. Repurpose wv constraint added during
133 power8 changes. Put wg constraint in alphabetical order.
135 * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
136 for future work to add ISA 2.07 VSX single precision support.
137 (-mvsx-scalar-double): Change default from -1 to 1, update
138 documentation comment.
139 (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
140 (-mupper-regs-df): New debug switch to control whether DF values
141 can go in the traditional Altivec registers.
142 (-mupper-regs-sf): New debug switch to control whether SF values
143 can go in the traditional Altivec registers.
145 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
147 (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
148 loop variables. Rename -mvsx-scalar-memory to -mupper-regs-df.
149 Add new constraints, wu/ww/wy. Repurpose wv constraint.
150 (rs6000_debug_legitimate_address_p): Print if we are running
151 before, during, or after reload.
152 (rs6000_secondary_reload): Add a comment.
153 (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
155 * config/rs6000/constraints.md (wa constraint): Sort w<x>
156 constraints. Update documentation string.
157 (wd constraint): Likewise.
158 (wf constraint): Likewise.
159 (wg constraint): Likewise.
160 (wn constraint): Likewise.
161 (ws constraint): Likewise.
162 (wt constraint): Likewise.
163 (wx constraint): Likewise.
164 (wz constraint): Likewise.
165 (wu constraint): New constraint for ISA 2.07 SFmode scalar
167 (ww constraint): Likewise.
168 (wy constraint): Likewise.
169 (wv constraint): Repurpose ISA 2.07 constraint that we not used in
170 the previous submissions.
171 * doc/md.texi (PowerPC and IBM RS6000): Likewise.
173 2013-09-23 Richard Sandiford <rdsandiford@googlemail.com>
175 * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
176 (REG_BR_PROB): Say that the probability is stored in an int_list.
177 * reg-notes.def: Update commentary to mention INT_LIST.
178 * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
180 * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
181 * rtlanal.c (int_reg_note_p): New function.
182 (alloc_reg_note): Assert that the note does not have an int argument.
183 (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
184 * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
185 * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
186 rather than an INSN_LIST. Handle INT_LIST.
187 * ifcvt.c (cond_exec_process_insns): Take the probability as an int
188 rather than an rtx. Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
189 (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
190 Manipulate them as ints rather than rtxes.
191 * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
192 * regmove.c (copy_src_to_dest): Likewise.
193 * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
195 * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
196 into the cases that need it.
197 * config/arm/arm.c (arm_unwind_emit): Likewise.
199 * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
200 * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
201 * loop-doloop.c (add_test, doloop_modify): Likewise.
202 * loop-unswitch.c (compare_and_jump_seq): Likewise.
203 * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
204 * predict.c (combine_predictions_for_insn): Likewise.
205 * print-rtl.c (print_rtx): Handle INT_LIST.
206 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
207 * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
208 * config/arm/arm.c (emit_unlikely_jump): Likewise.
209 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
210 (ix86_split_fp_branch, predict_jump): Likewise.
211 * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
212 * config/sh/sh.c (expand_cbranchsi4): Likewise.
213 * config/spu/spu.c (ea_load_store_inline): Likewise.
215 * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
216 value of a REG_BR_PROB note.
217 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
218 (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
219 * emit-rtl.c (try_split): Likewise.
220 * predict.c (br_prob_note_reliable_p): Likewise.
221 (invert_br_probabilities, combine_predictions_for_insn): Likewise.
222 * reorg.c (mostly_true_jump): Likewise.
223 * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
224 * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
225 * config/i386/i386.c (ix86_print_operand): Likewise.
226 * config/ia64/ia64.c (ia64_print_operand): Likewise.
227 * config/mmix/mmix.c (mmix_print_operand): Likewise.
228 * config/rs6000/rs6000.c (output_cbranch): Likewise.
229 * config/s390/s390.c (s390_expand_tbegin): Likewise.
230 * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
231 * config/sparc/sparc.c (output_cbranch): Likewise.
232 * config/spu/spu.c (get_branch_target): Likewise.
233 * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
234 * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
236 2013-09-23 Jan Hubicka <jh@suse.cz>
238 * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
240 * ipa-utils.h (possible_polymorphic_call_target_p): New function.
241 * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
243 * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
244 (gimple_fold_call): Dump inconsistent devirtualizations; add
245 sanity check for type based devirtualizations.
246 * ipa-prop.c: Include ipa-utils.h
247 (ipa_intraprocedural_devirtualization): Add sanity check.
248 (try_make_edge_direct_virtual_call): Likewise.
250 2013-09-23 Eric Botcazou <ebotcazou@adacore.com>
252 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
253 assignment statements.
255 2013-09-23 Kugan Vivekanandarajah <kuganv@linaro.org>
257 * gimple-pretty-print.c (dump_ssaname_info): New function.
258 (dump_gimple_phi): Call it.
259 (pp_gimple_stmt_1): Likewise.
260 * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
261 (range_info_def): Declare.
262 * tree-pretty-print.c (pp_double_int): New function.
263 (dump_generic_node): Call it.
264 * tree-pretty-print.h (pp_double_int): Declare.
265 * tree-ssa-alias.c (dump_alias_info): Check pointer type.
266 * tree-ssanames.h (range_info_def): New structure.
267 (value_range_type): Move definition here.
268 (set_range_info, value_range_type, duplicate_ssa_name_range_info):
270 * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
272 (set_range_info): New function.
273 (get_range_info): Likewise.
274 (duplicate_ssa_name_range_info): Likewise.
275 (duplicate_ssa_name_fn): Check pointer type and call
276 duplicate_ssa_name_range_info.
277 * tree-ssa-copy.c (fini_copy_prop): Likewise.
278 * tree-vrp.c (value_range_type): Remove definition, now in
280 (vrp_finalize): Call set_range_info to update value range of
282 * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
283 (SSA_NAME_RANGE_INFO): New macro.
285 2013-09-23 Richard Biener <rguenther@suse.de>
287 PR tree-optimization/58464
288 * tree-ssa-pre.c (phi_trans_lookup): Remove.
289 (phi_trans_add): Change to add conditionally on being not
291 (phi_translate_1): Remove recursion detection here.
292 (phi_translate): Pre-seed the cache with NULL to catch
293 recursion here in a more generic way.
294 (bitmap_find_leader): Adjust comment.
295 (get_representative_for): Dump value-numbers.
296 (create_expression_by_pieces): Likewise.
297 (insert_into_preds_of_block): Likewise.
299 2013-09-23 Christian Bruel <christian.bruel@st.com>
302 * config/sh/sh.md (movsf_ie): Allow fpul_operand.
303 * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
305 2013-09-23 James Greenhalgh <james.greenhalgh@arm.com>
308 2013-09-20 Renlin Li <renlin.li@arm.com>
310 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
311 (aarch64_expand_epilogue): Likewise.
312 (aarch64_legitimize_reload_address): Likewise.
314 2013-09-22 Eric Botcazou <ebotcazou@adacore.com>
316 * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
317 NULL_TREE before pushing them onto the vector. Likewise for labels.
319 2013-09-21 Eric Botcazou <ebotcazou@adacore.com>
321 * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
322 comparison operators when -fno-trapping-math is in effect.
323 * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
324 comparison operators in TFmode and assert that unsupported operators
326 (ia64_print_operand): Likewise.
328 2013-09-21 Jan Hubicka <jh@suse.cz>
330 * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
331 (sse_typeless_stores): Enable for core
332 (sse_load0_by_pxor): Likewise.
333 (four_jump_limit): Disable for core.
334 (pad_returns): Likewise.
335 (avoid_vector_decode): Likewise.
336 (fuse_cmp_and_branch): Enable for cores.
337 * i386.c (x86_accumulate_outgoing_args): Disable for cores.
339 2013-09-20 John David Anglin <danglin@gcc.gnu.org>
342 * config/pa/pa.c (pa_option_override): Disable auto increment and
343 decrement instructions until reload is completed.
345 * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
346 __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
347 and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
349 2013-09-20 DJ Delorie <dj@redhat.com>
350 Nick Clifton <nickc@redhat.com>
352 * config/rl78/rl78.c: Various whitespace and comment tweaks.
353 (need_to_save): Save bank 0 on interrupts.
354 (characterize_address): Strip far address wrappers.
355 (rl78_as_legitimate_address): Likewise.
356 (transcode_memory_rtx): Likewise.
357 (rl78_peep_movhi_p): Disable this peephole after devirt.
358 (rl78_propogate_register_origins): Forget all origins when a
360 * config/rl78/rl78-virt.md: Various whitespace tweaks.
361 * config/rl78/rl78-real.md: Various whitespace tweaks. Additional
363 * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
364 * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
365 * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
367 * config/rl78/constraints.md: Various whitespace and paren tweaks.
369 2013-09-20 John David Anglin <danglin@gcc.gnu.org>
371 * config/pa/pa.md: In "scc" insn patterns, change output template to
372 handle const0_rtx in reg_or_0_operand operands.
374 2013-09-20 Martin Husemann <martin@NetBSD.org>
377 * config/vax/vax.c (vax_output_int_move): Use D format specifier.
378 * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
380 2013-09-20 Richard Biener <rguenther@suse.de>
383 * tree-scalar-evolution.c (struct scev_info_str): Shrink by
384 remembering SSA name version and block index.
385 (new_scev_info_str): Adjust.
386 (hash_scev_info): Likewise. Also hash the block index.
387 (eq_scev_info): Adjust.
388 (find_var_scev_info): Likewise.
389 (struct instantiate_cache_entry): Remove.
390 (struct instantiate_cache_type): Use a htab to map name, block
392 (instantiate_cache_type::~instantiate_cache_type): Adjust.
393 (get_instantiated_value_entry): Likewise.
394 (hash_idx_scev_info, eq_idx_scev_info): New functions.
395 (instantiate_scev_name): Adjust.
397 2013-09-20 Jeff Law <law@redhat.com>
399 * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
401 2013-09-20 Yufeng Zhang <yufeng.zhang@arm.com>
403 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
404 Call aarch64_simd_expand_args to update op[argc].
406 2013-09-20 Basile Starynkevitch <basile@starynkevitch.net>
408 * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
411 2013-09-20 Basile Starynkevitch <basile@starynkevitch.net>
413 * gengtype.c (file_rules): Added rule for *.cc files.
414 (get_output_file_with_visibility): Give fatal message when no
417 2013-09-20 Renlin Li <renlin.li@arm.com>
419 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
420 (aarch64_expand_epilogue): Likewise.
421 (aarch64_legitimize_reload_address): Likewise.
423 2013-09-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
426 * expr.c (expand_assignment): Remove misalignp code path.
428 2013-09-20 Marek Polacek <polacek@redhat.com>
431 * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
432 TYPE_PRECISION. Add asserts.
434 2013-09-20 Richard Biener <rguenther@suse.de>
436 PR tree-optimization/58453
437 * tree-loop-distribution.c (distribute_loop): Apply the cost
438 model for -ftree-loop-distribute-patterns, too.
440 2013-09-20 Richard Biener <rguenther@suse.de>
443 * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
444 make type comparison less strict.
446 2013-09-20 Alan Modra <amodra@gmail.com>
448 * configure: Regenerate.
449 * aclocal.m4: Regenerate.
451 2013-09-20 Marek Polacek <polacek@redhat.com>
454 * doc/extend.texi: Document that attribute used is meant to be used
455 on variables with static storage duration.
457 2013-09-19 Jakub Jelinek <jakub@redhat.com>
459 PR tree-optimization/58472
460 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
461 simd_lane_access set inv_p = false.
462 * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
463 the simduid magic VAR_DECL.
465 2013-09-19 Jan Hubicka <jh@suse.cz>
467 * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
469 2013-09-17 Jeff Law <law@redhat.com>
471 * tree-ssa-dom.c (record_temporary_equivalences): New function
472 split out of dom_opt_dom_walker::after_dom_children.
473 (dom_opt_dom_walker::thread_across_edge): Move common code
474 in here from dom_opt_dom_walker::after_dom_children.
475 (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
477 2013-09-19 Jan Hubicka <jh@suse.cz>
479 * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
480 (TARGET_GENERIC): Use PROCESOR_GENERIC
481 (enum processor_type): Unify generic32 and 64.
482 * i386.md (cpu): Likewise.
483 * x86-tune.def (use_leave): Enable for generic32.
484 (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
485 * athlon.md: Change generic64 to generic in all occurences.
486 * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
487 (ix86_target_macros_internal): Likewise.
488 * driver-i386.c (host_detect_local_cpu): Likewise.
489 * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename to ..
490 (generic_memcpy, generic_memset, generic_cost): This one.
491 (generic32_memcpy, generic32_memset, generic32_cost): Remove.
492 (m_GENERIC32, m_GENERIC64): Remove.
493 (m_GENERIC): Turn into one flag.
494 (processor_target): Unify generic tunnings.
495 (ix86_option_override_internal): Replace generic32/64 by generic.
496 (ix86_issue_rate): Likewise.
497 (ix86_adjust_cost): Likewise.
499 2013-09-19 Jan Hubicka <jh@suse.cz>
501 * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
504 2013-09-19 Jakub Jelinek <jakub@redhat.com>
506 * omp-low.c (expand_omp_sections): Always pass len - 1 to
507 GOMP_sections_start, even if !exit_reachable.
509 2013-09-18 Vladimir Makarov <vmakarov@redhat.com>
511 * lra-constraints.c (need_for_all_save_p): Use macro
512 HARD_REGNO_CALL_PART_CLOBBERED.
513 * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
514 set up pseudo conflict hard regs.
516 2013-09-18 Michael Meissner <meissner@linux.vnet.ibm.com>
519 * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
522 2013-09-18 Vladimir Makarov <vmakarov@redhat.com>
524 PR rtl-optimization/58438
525 * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
526 * lra-constraints.c (undo_optional_reloads): Keep optional reloads
527 from previous subpasses.
529 2013-09-18 Richard Earnshaw <rearnsha@arm.com>
531 * arm.c (arm_get_frame_offsets): Validate architecture supports
532 LDRD/STRD before accepting the tuning preference.
533 (arm_expand_prologue): Likewise.
534 (arm_expand_epilogue): Likewise.
536 2013-09-18 Richard Biener <rguenther@suse.de>
538 PR tree-optimization/58417
539 * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
540 have chrecs with symbols defined in the loop as operands.
541 (chrec_fold_multiply): Likewise.
542 * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
543 parameters before folding binary operations.
544 (struct instantiate_cache_entry_hasher): Remove.
545 (struct instantiate_cache_type): Use a pointer-map.
546 (instantiate_cache_type::instantiate_cache_type): New function.
547 (instantiate_cache_type::get): Likewise.
548 (instantiate_cache_type::set): Likewise.
549 (instantiate_cache_type::~instantiate_cache_type): Adjust.
550 (get_instantiated_value_entry): Likewise.
551 (global_cache): New global.
552 (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
553 instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
554 instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
555 (instantiate_scev_name): Adjust.
556 (instantiate_scev): Construct global instead of local cache.
557 (resolve_mixers): Likewise.
559 2013-09-18 Daniel Morris <danielm@ecoscentric.com>
560 Paolo Carlini <paolo.carlini@oracle.com>
563 * doc/implement-cxx.texi: Fix references to the C++ standards.
565 2013-09-18 Jakub Jelinek <jakub@redhat.com>
567 * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
568 * tree-vect-data-refs.c (vect_analyze_data_refs): For
569 simd_lane_access drs, update also DR_ALIGNED_TO.
571 2013-09-18 Marek Polacek <polacek@redhat.com>
574 * doc/extend.texi: Document no_sanitize_undefined attribute.
575 * builtins.c (fold_builtin_0): Don't sanitize function if it has the
576 no_sanitize_undefined attribute.
578 2013-09-18 Nick Clifton <nickc@redhat.com>
580 * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
581 (ASM_DECLARE_FUNCTION_NAME): Define.
583 2013-09-17 Trevor Saunders <tsaunders@mozilla.com>
585 * compare-elim.c (find_comparison_dom_walker): New class
586 (find_comparisons_in_bb): Rename to
587 find_comparison_dom_walker::before_dom_children
588 (find_comparisons): Adjust
589 * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
591 (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
592 * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
593 (init_walk_dominator_tree): Remove declaration.
594 (fini_walk_dominator_tree): Remove declaration.
595 * fwprop.c (single_def_use_dom_walker): New class
596 (single_def_use_enter_block): Convert to
597 single_def_use_dom_walker::before_dom_children.
598 (single_def_use_leave_block): Convert to
599 single_def_use_dom_walker::after_dom_children.
600 (build_single_def_use_links): Adjust.
601 * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
603 (find_candidates_in_block): Convert to
604 find_candidates_dom_walker::before_dom_children.
605 (execute_strength_reduction): Adjust.
606 * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
607 (sese_dom_walker): New class.
608 (sese_dom_walker::sese_dom_walker): New constructor.
609 (sese_dom_walker::~sese_dom_walker): New destructor.
610 (build_sese_conditions_before): Convert to
611 sese_dom_walker::before_dom_children.
612 (build_sese_conditions_after): Convert to
613 sese_dom_walker::after_dom_children.
614 (build_poly_scop): Adjust
615 * tree-into-ssa.c (rewrite_dom_walker): New class
616 (rewrite_enter_block): Convert to
617 rewrite_dom_walker::before_dom_children.
618 (rewrite_leave_block): Convert to
619 rewrite_dom_walker::after_dom_children.
620 (rewrite_update_dom_walker): New class.
621 (rewrite_update_enter_block): Convert to
622 rewrite_update_dom_walker::before_dom_children.
623 (rewrite_update_leave_block): Convert to
624 rewrite_update_dom_walker::after_dom_children.
625 (rewrite_blocks, rewrite_into_ssa): Adjust.
626 (mark_def_dom_walker): New class.
627 (mark_def_dom_walker::mark_def_dom_walker): New constructor.
628 (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
629 (mark_def_sites_blocks): Convert to
630 mark_def_dom_walker::before_dom_children.
631 (mark_def_site_blocks): Remove.
632 * tree-ssa-dom.c (dom_opt_dom_walker): New class.
633 (tree_ssa_dominator_optimize): Adjust.
634 (dom_thread_across_edge): Convert to method
635 dom_opt_dom_walker::thread_across_edge.
636 (dom_opt_enter_block): Convert to member function
637 dom_opt_dom_walker::before_dom_children.
638 (dom_opt_leave_block): Convert to member function
639 dom_opt_dom_walker::after_dom_children.
640 * tree-ssa-dse.c (dse_dom_walker): New class.
641 (dse_enter_block): Convert to member function
642 dse_dom_walker::before_dom_children.
643 (tree_ssa_dse): Adjust.
644 * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
645 (determine_invariantness_stmt): Convert to method
646 invariantness_dom_walker::before_dom_children.
647 (determine_invariantness): Remove
648 (move_computations_dom_walker): New class.
649 (move_computations_stmt): Convert to method
650 move_computations_dom_walker::before_dom_children.
651 (move_computations, tree_ssa_lim): Adjust.
652 * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
653 (nt_init_block): Make method
654 notrappping_dom_walker::before_dom_children.
655 (nt_fini_block): Make
656 method nontrapping_dom_walker::after_dom_children.
657 (get_non_trapping): Adjust.
658 * tree-ssa-pre.c (eliminate_dom_walker): New class.
659 (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
660 (eliminate_leave_block): Make method.
661 eliminate_dom_walker::after_dom_children.
663 * tree-ssa-strlen.c (strlen_dom_walker): New class.
664 (strlen_enter_block): Make method
665 strlen_dom_walker::before_dom_children.
666 (strlen_leave_block): Make
667 method strlen_dom_walker::after_dom_children.
668 (tree_ssa_strlen): Adjust.
669 * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
670 (tree_ssa_uncprop): Adjust.
671 (uncprop_leave_block): Make method
672 uncprop_dom_walker::after_dom_children.
673 (uncprop_leave_block): Make method
674 uncprop_dom_walker::before_dom_children.
676 2013-09-18 Bin Cheng <bin.cheng@arm.com>
678 * config/arm/arm.c (thumb1_reorg): Search for flag setting insn
679 before branch in same basic block. Check both src and dest of
682 2013-09-17 Nick Clifton <nickc@redhat.com>
684 * config/rl78/rl78-real.md (bf): New pattern.
686 * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
687 (rl78_print_operand): Do not put a # before a %B.
688 * config/rl78/rl78.opt: Tweak doc strings.
690 2013-09-17 DJ Delorie <dj@redhat.com>
692 * config/rl78/constraints.md (Wcv): Allow up to $r31.
693 * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
694 (rl78_option_override): Likewise, if -mallregs.
695 (is_virtual_register): Likewise.
696 * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
697 (REGNO_OK_FOR_BASE_P): Likewise.
698 * config/rl78/rl78.opt (-mallregs): New.
700 2013-09-17 Nick Clifton <nickc@redhat.com>
702 * config/rl78/rl78.c (need_to_save): Change return type to bool.
703 For interrupt functions: save all call clobbered registers if the
704 interrupt handler is not a leaf function.
705 (rl78_expand_prologue): Always recompute the frame information.
706 For interrupt functions: only select bank 0 if one of the bank 0
707 registers is going to be psuhed.
709 2013-09-17 DJ Delorie <dj@redhat.com>
711 * config/rl78/constraints.md: For each W* constraint, rename to C*
712 and create a W* constraint that checks for an optional ES: prefix
714 * config/rl78/rl78.md (UNS_ES_ADDR): New.
715 (es_addr): New. Used to wrap far addresses.
716 * config/rl78/rl78-protos.h (rl78_es_addr): New.
718 * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
719 wrapped far addresses.
720 (rl78_print_operand_1): Unwrap far addresses before processing.
721 (rl78_lo16): Wrap far addresses in unspecs.
724 (insn_ok_now): Check for not-yet-wrapped far addresses.
725 (transcode_memory_rtx): Properly re-wrap far addresses.
727 2013-09-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
729 * config/sparc/t-rtems: Add leon3 multilibs.
731 2013-09-17 Cong Hou <congh@google.com>
733 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
734 when checking the dot production pattern. The type of rhs operand
735 of multiply is now checked correctly.
737 2013-09-17 Jeff Law <law@redhat.com>
739 * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
740 edge implied equivalences into successor phis.
741 * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
742 here from tree-ssa-threadedge.c.
743 (mark_threaded_blocks): When threading through a joiner, if both
744 successors of the joiner's clone reach the same block, verify the
745 PHI arguments are equal. If not, cancel the jump threading request.
746 * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
747 tree-ssa-threadupdate.c
748 (thread_across_edge): Don't check PHI argument equality when
749 threading through joiner block here.
751 2013-09-17 Andrew MacLeod <amacleod@redhat.com>
753 * tree-flow.h (ssa_undefined_value_p): Remove prototype.
754 * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
755 (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
756 make_pass_early_warn_uninitialized): Move to tree-ssa-uninit.c.
757 * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c
758 (has_undefined_value_p): New. Pass dependant parts of
759 ssa_undefined_value_p.
760 (uninit_undefined_value_p): Use has_undefined_value_p.
761 (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
762 make_pass_early_warn_uninitialized): Move from tree-ssa.c
763 * tree-ssa.h: Adjust prototypes
765 2013-09-17 Jan Hubicka <jh@suse.cz>
768 * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
769 * ipa-inline.c (can_inline_edge_p): Do not downgrade
770 FUNCTION_NOT_OPTIMIZED.
771 * ipa-inline-analysis.c (compute_inline_parameters): Function
772 not optimized is not inlinable unless it is alwaysinline.
773 (inline_analyze_function): Force calls in not optimized
774 function not inlinable.
776 2013-09-17 Jan Hubicka <jh@suse.cz>
779 * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
781 * ipa.c (function_and_variable_visibility): Likewise.
782 * ipa-profile.c (ipa_profile): Likewise.
784 2013-09-17 Bernd Edlinger <bernd.edlinger@hotmail.de>
787 * cgraph.c (cgraph_function_body_availability): Check for ifunc
788 attribute, and don't inline the resolver in this case.
790 2013-09-17 Teresa Johnson <tejohnson@google.com>
792 * coverage.c (get_coverage_counts): Add missing newline.
794 2013-09-17 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
796 PR tree-optimization/58088
797 * fold-const.c (mask_with_trailing_zeros): New function.
798 (fold_binary_loc): Make sure we don't recurse infinitely
799 when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
800 Use mask_with_trailing_zeros where appropriate.
802 2013-09-17 Yuri Rumyantsev <ysrumyan@gmail.com>
804 * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
805 of 'prev' var to get better distance estimation.
807 2013-09-17 Eric Botcazou <ebotcazou@adacore.com>
809 * tree-inline.h (struct copy_body_data): Add transform_parameter.
810 * tree-inline.c (is_parameter_of): New predicate.
811 (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
812 a parameter has been remapped.
813 (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
814 (optimize_inline_calls): Initialize transform_parameter.
815 (copy_gimple_seq_and_replace_locals): Likewise.
816 (tree_function_versioning): Likewise.
817 (maybe_inline_call_in_expr): Likewise.
819 2013-09-17 Nick Clifton <nickc@redhat.com>
821 * config/msp430/msp430-protos.h: Add prototypes for new functions.
822 * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
824 (is_attr_func): New function.
825 (msp430_is_interrupt_func): New function.
826 (is_naked_func): New function.
827 (is_reentrant_func): New function.
828 (is_critical_func): New function.
829 (msp430_start_function): Add annotations for function attributes.
830 (msp430_attr): New function.
831 (msp430_attribute_table): New.
832 (msp430_function_section): New function.
833 (TARGET_ASM_FUNCTION_SECTION): Define.
834 (msp430_builtin): New enum.
835 (msp430_init_builtins): New function.
836 (msp430_builtin_devl): New function.
837 (msp430_expand_builtin): New function.
838 (TARGET_INIT_BUILTINS): Define.
839 (TARGET_EXPAND_BUILTINS): Define.
840 (TARGET_BUILTIN_DECL): Define.
841 (msp430_expand_prologue): Add support for naked, interrupt,
842 critical and reentrant functions.
843 (msp430_expand_epilogue): Likewise.
844 (msp430_print_operand): Handle 'O' character.
845 * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
847 * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
848 UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
849 (pushm): Use a 'n' rather than an 'i' constraint.
850 (msp_return): Add generation of the interrupt return instruction.
851 (disable_interrupts): New pattern.
852 (enable_interrupts): New pattern.
853 (push_intr_state): New pattern.
854 (pop_intr_state): New pattern.
855 (bic_SR): New pattern.
856 (bis_SR): New pattern.
857 * doc/extend.texi: Document MSP430 function attributes and builtin
860 2013-09-17 Richard Biener <rguenther@suse.de>
862 PR tree-optimization/58432
863 * tree-loop-distribution.c (tree_loop_distribution): Also
864 scan PHIs for outside loop uses and seed a partition from them.
866 2013-09-17 Bin Cheng <bin.cheng@arm.com>
868 * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
869 (restructure_reference): Call backtrace_base_for_ref.
871 2013-09-17 Alan Modra <amodra@gmail.com>
874 * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
876 2013-09-16 DJ Delorie <dj@redhat.com>
878 * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
879 vregs location for RL78/G10.
880 (rl78_expand_prologue): Avoid SEL on G10.
881 (rl78_expand_epilogue): Likewise.
882 (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
883 * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
884 __RL78_G10__ when appropriate.
885 (ASM_SPEC): Pass -mg10 along to the assembler.
886 * config/rl78/rl78.md (sel_rb): Disable for G10.
887 * config/rl78/rl78.opt: Add -mg10 option.
888 * config/rl78/t-rl78: Add -mg10 multilib.
890 2013-09-16 Xinliang David Li <davidxl@google.com>
892 * tree-if-conv.c (main_tree_if_conversion): Check new flag.
893 * omp-low.c (omp_max_vf): Ditto.
894 (expand_omp_simd): Ditto.
895 * tree-vectorizer.c (vectorize_loops): Ditto.
896 (gate_vect_slp): Ditto.
897 (gate_increase_alignment): Ditto.
898 * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
899 * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
900 (gate_tree_vectorize): Name change.
901 (tree_vectorize): Ditto.
902 (pass_vectorize::gate): Call new function.
903 (pass_vectorize::execute): Ditto.
904 opts.c: O3 default setting change.
905 (finish_options): Check new flag.
906 * doc/invoke.texi: Document new flags.
907 * common.opt: New flags.
909 2013-09-16 Andreas Schwab <schwab@linux-m68k.org>
911 * doc/tm.texi.in (Cond Exec Macros): Remove node.
912 (Condition Code): Don't reference it.
913 * doc/tm.texi: Regenerate.
915 2013-09-16 Vladimir Makarov <vmakarov@redhat.com>
918 * lra-constraints.c (undo_optional_reloads): Consider all optional
919 reload even if it did not get a hard reg.
921 2013-09-16 Teresa Johnson <tejohnson@google.com>
923 * dumpfile.c (dump_loc): Remove newline emission.
924 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
925 emission to dump_printf_loc calls where missing.
926 (vect_mark_for_runtime_alias_test): Ditto.
927 (vect_analyze_data_ref_dependence): Ditto.
928 (vect_analyze_data_ref_dependences): Ditto.
929 (vect_slp_analyze_data_ref_dependence): Ditto.
930 (vect_slp_analyze_data_ref_dependences): Ditto.
931 (vect_compute_data_ref_alignment): Ditto.
932 (vect_update_misalignment_for_peel): Ditto.
933 (vect_verify_datarefs_alignment): Ditto.
934 (vector_alignment_reachable_p): Ditto.
935 (vect_get_data_access_cost): Ditto.
936 (vect_enhance_data_refs_alignment): Ditto.
937 (vect_find_same_alignment_drs): Ditto.
938 (vect_analyze_data_refs_alignment): Ditto.
939 (vect_analyze_group_access): Ditto.
940 (vect_analyze_data_ref_access): Ditto.
941 (vect_analyze_data_ref_accesses): Ditto.
942 (vect_prune_runtime_alias_test_list): Ditto.
943 (vect_analyze_data_refs): Ditto.
944 (vect_create_addr_base_for_vector_ref): Ditto.
945 (vect_create_data_ref_ptr): Ditto.
946 (vect_grouped_store_supported): Ditto.
947 (vect_grouped_load_supported): Ditto.
948 * value-prof.c (check_counter): Ditto.
949 (check_ic_target): Ditto.
950 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
951 (vect_recog_widen_mult_pattern): Ditto.
952 (vect_recog_widen_sum_pattern): Ditto.
953 (vect_recog_over_widening_pattern): Ditto.
954 (vect_recog_widen_shift_pattern): Ditto.
955 (vect_recog_rotate_pattern): Ditto.
956 (vect_recog_vector_vector_shift_pattern): Ditto.
957 (vect_recog_divmod_pattern): Ditto.
958 (vect_recog_mixed_size_cond_pattern): Ditto.
959 (vect_recog_bool_pattern): Ditto.
960 (vect_pattern_recog_1): Ditto.
961 (vect_pattern_recog): Ditto.
962 * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
963 (vect_is_simple_iv_evolution): Ditto.
964 (vect_analyze_scalar_cycles_1): Ditto.
965 (vect_get_loop_niters): Ditto.
966 (vect_analyze_loop_1): Ditto.
967 (vect_analyze_loop_form): Ditto.
968 (vect_analyze_loop_operations): Ditto.
969 (vect_analyze_loop_2): Ditto.
970 (vect_analyze_loop): Ditto.
971 (report_vect_op): Ditto.
972 (vect_is_slp_reduction): Ditto.
973 (vect_is_simple_reduction_1): Ditto.
974 (vect_get_known_peeling_cost): Ditto.
975 (vect_estimate_min_profitable_iters): Ditto.
976 (vect_model_reduction_cost): Ditto.
977 (vect_model_induction_cost): Ditto.
978 (get_initial_def_for_induction): Ditto.
979 (vect_create_epilog_for_reduction): Ditto.
980 (vectorizable_reduction): Ditto.
981 (vectorizable_induction): Ditto.
982 (vectorizable_live_operation): Ditto.
983 (vect_loop_kill_debug_uses): Ditto.
984 (vect_transform_loop): Ditto.
985 * tree-vect-stmts.c (vect_mark_relevant): Ditto.
986 (vect_stmt_relevant_p): Ditto.
987 (process_use): Ditto.
988 (vect_mark_stmts_to_be_vectorized): Ditto.
989 (vect_model_simple_cost): Ditto.
990 (vect_model_promotion_demotion_cost): Ditto.
991 (vect_model_store_cost): Ditto.
992 (vect_get_store_cost): Ditto.
993 (vect_model_load_cost): Ditto.
994 (vect_get_load_cost): Ditto.
995 (vect_init_vector_1): Ditto.
996 (vect_get_vec_def_for_operand): Ditto.
997 (vect_finish_stmt_generation): Ditto.
998 (vectorizable_call): Ditto.
999 (vectorizable_conversion): Ditto.
1000 (vectorizable_assignment): Ditto.
1001 (vectorizable_shift): Ditto.
1002 (vectorizable_operation): Ditto.
1003 (vectorizable_store): Ditto.
1004 (vectorizable_load): Ditto.
1005 (vectorizable_condition): Ditto.
1006 (vect_analyze_stmt): Ditto.
1007 (vect_transform_stmt): Ditto.
1008 (vect_is_simple_use): Ditto.
1009 * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
1010 (vect_can_advance_ivs_p): Ditto.
1011 (vect_update_ivs_after_vectorizer): Ditto.
1012 (vect_do_peeling_for_loop_bound): Ditto.
1013 (vect_gen_niters_for_prolog_loop): Ditto.
1014 (vect_update_inits_of_drs): Ditto.
1015 (vect_create_cond_for_alias_checks): Ditto.
1016 * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
1017 (vect_build_slp_tree_1): Ditto.
1018 (vect_supported_load_permutation_p): Ditto.
1019 (vect_analyze_slp_instance): Ditto.
1020 (vect_analyze_slp): Ditto.
1021 (vect_make_slp_decision): Ditto.
1022 (vect_detect_hybrid_slp): Ditto.
1023 (vect_bb_vectorization_profitable_p): Ditto.
1024 (vect_slp_analyze_bb_1): Ditto.
1025 (vect_update_slp_costs_according_to_vf): Ditto.
1026 (vect_get_mask_element): Ditto.
1027 (vect_transform_slp_perm_load): Ditto.
1028 (vect_schedule_slp_instance): Ditto.
1029 (vect_schedule_slp): Ditto.
1030 (vect_slp_transform_bb): Ditto.
1031 * profile.c (read_profile_edge_counts): Ditto.
1032 (compute_branch_probabilities): Ditto.
1033 * coverage.c (get_coverage_counts): Ditto.
1035 2013-09-16 Diego Novillo <dnovillo@google.com>
1037 * tree-core.h: Add missing comment lines from refactoring
1040 2013-09-16 Jan Hubicka <jh@suse.cz>
1042 * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
1043 abstract functions; for static functions check the presence
1046 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
1048 * config/aarch64/aarch64-simd-builtins.def (fma): New.
1049 * config/aarch64/aarch64-simd.md
1050 (aarch64_mla_elt<mode>): New.
1051 (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1052 (aarch64_mls_elt<mode>): Likewise.
1053 (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1054 (aarch64_fma4_elt<mode>): Likewise.
1055 (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1056 (aarch64_fma4_elt_to_128v2df): Likewise.
1057 (aarch64_fma4_elt_to_64df): Likewise.
1058 (fnma<mode>4): Likewise.
1059 (aarch64_fnma4_elt<mode>): Likewise.
1060 (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1061 (aarch64_fnma4_elt_to_128v2df): Likewise.
1062 (aarch64_fnma4_elt_to_64df): Likewise.
1063 * config/aarch64/iterators.md (VDQSF): New.
1064 * config/aarch64/arm_neon.h
1065 (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
1066 (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
1068 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
1070 * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
1071 (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1072 (aarch64_mul3_elt_to_128df): Likewise.
1073 (aarch64_mul3_elt_to_64v2df): Likewise.
1074 * config/aarch64/iterators.md (VEL): Also handle DFmode.
1076 (VMUL_CHANGE_NLANES) Likewise.
1079 * config/aarch64/arm_neon.h
1080 (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
1082 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
1084 * config/aarch64/arm_neon.h
1085 (vcvtx_high_f32_f64): Fix parameters.
1087 2013-09-16 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1088 Uros Bizjak <ubizjak@gmail.com>
1090 * config/alpha.c: Include tree-ssanames.h.
1092 2013-09-16 Richard Biener <rguenther@suse.de>
1094 * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
1095 (dot_rdg_1): Handle control_dd.
1096 (create_edge_for_control_dependence): New function.
1097 (create_rdg_edges): Add control dependences if asked for.
1098 (build_rdg): Likewise.
1099 (generate_loops_for_partition): If there are not necessary
1100 control stmts remove all their dependencies.
1101 (collect_condition_stmts, rdg_flag_loop_exits): Remove.
1102 (distribute_loop): Pass on control dependences.
1103 (tree_loop_distribution): Compute control dependences and remove
1104 restriction on number of loop nodes.
1106 2013-09-16 Jakub Jelinek <jakub@redhat.com>
1108 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
1111 2013-09-16 Richard Sandiford <rdsandiford@googlemail.com>
1113 * cse.c (try_const_anchors): Punt on CC modes.
1115 2013-09-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1117 * config/vax/constraints.md (T): Add missing CONSTANT_P check.
1119 2013-09-14 John David Anglin <danglin@gcc.gnu.org>
1122 * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
1125 2013-09-14 Iain Sandoe <iain@codesourcery.com>
1128 * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
1130 (darwin_objc1_section): Likewise.
1131 (darwin_file_end): Emit Image Info section when required.
1133 2013-09-14 Jan Hubicka <jh@suse.cz>
1135 * tree-into-ssa.c (gate_into_ssa): New.
1136 (pass_data_build_ssa): Use it.
1137 * cgraph.h (expand_thunk): Update prototype.
1138 * cgraphunit.c (analyze_function): Expand thunks early.
1139 (expand_thunk): Fix DECL_CONTEXT of reust_decl;
1140 build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
1141 set lowered flag; do not add new function.
1142 (assemble_thunks_and_aliases): Update.
1143 * tree-ssa.c (gate_init_datastructures): New gate.
1144 (pass_data_init_datastructures): Use it.
1146 2013-09-14 Iain Sandoe <iain@codesourcery.com>
1149 * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
1150 xmm register set described in the psABI.
1152 2013-09-13 Evgeny Gavrin <e.gavrin@samsung.com>
1154 * dwarf2out.c (should_emit_struct_debug): Add check
1155 for type_decl variable is not NULL.
1157 2013-09-13 Jacek Caban <jacek@codeweavers.com>
1159 * config.gcc: Use new winnt-c.c target hooks
1160 * config/t-winnt: New file
1161 * config/winnt-c.c: New file
1162 * doc/tm.texi.in: Document new hook
1163 * doc/tm.texi: Regenerated
1165 2013-09-13 Jan Hubicka <jh@suse.cz>
1168 * ipa-inline.c (check_callers): New function.
1169 (check_caller_edge): Remove.
1170 (want_inline_function_to_all_callers_p): Also permit alises that are
1172 (inline_to_all_callers): Terminate the walk when devirtualization
1173 introduce new calls.
1175 2013-09-13 Jan Hubicka <jh@suse.cz>
1177 * ipa-inline-analysis.c (struct growth_data): Add node.
1178 (do_estimate_growth_1): Fix detection of recursion.
1180 2013-09-13 Jakub Jelinek <jakub@redhat.com>
1182 PR tree-optimization/58392
1183 * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
1184 to avoid shadowing of outer loop variable. If
1185 saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
1186 replace_by_duplicate_decl simduid of loops that have it set and
1187 set dest_cfun->has_simduid_loops and/or
1188 dest_cfun->has_force_vect_loops.
1189 * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
1190 instead of maybe_lookup_decl.
1191 * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
1192 Use id->blocks_to_copy instead of blocks_to_copy. Adjust recursive
1193 call. Copy over force_vect and copy and remap simduid. Set
1194 cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
1195 (copy_cfg_body): Remove blocks_to_copy argument. Use
1196 id->blocks_to_copy instead of blocks_to_copy. Adjust copy_loops
1197 caller. Don't set cfun->has_simduid_loops and/or
1198 cfun->has_force_vect_loops here.
1199 (copy_body): Remove blocks_to_copy argument. Adjust copy_cfg_body
1201 (expand_call_inline, tree_function_versioning): Adjust copy_body
1204 2013-09-13 Martin Jambor <mjambor@suse.cz>
1207 * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
1208 the assert if the edge was a speculative one.
1210 2013-09-13 Richard Biener <rguenther@suse.de>
1212 * tree-data-ref.h (known_dependences_p): Move here ...
1213 * tree-loop-distribution.c (known_dependences_p): ... from here.
1214 (dump_rdg_component, debug_rdg_component): Remove.
1216 (generate_loops_for_partition): Use gimple_uid instead of
1217 relying on matching stmt visit order.
1218 (rdg_build_partitions): Take starting stmt vector.
1219 (ldist_gen): Merge into ...
1220 (distribute_loop): ... this function. Do not compute starting
1222 * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
1224 2013-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1226 * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
1227 Set type attribute correctly. Set predicable_short_it attribute.
1228 (cmpsi_shiftsi): Remove %? from output template.
1230 2013-09-13 Richard Biener <rguenther@suse.de>
1232 * tree-loop-distribution.c (struct rdg_component,
1233 rdg_defs_used_in_other_loops_p, free_rdg_components,
1234 rdg_build_components): Remove.
1235 (stmts_from_loop): Do not record virtual PHIs.
1236 (generate_loops_for_partition): Skip virtual PHIs.
1237 (build_rdg_partition_for_component): Rename to ...
1238 (build_rdg_partition_for_vertex): ... this and adjust.
1239 (rdg_build_partitions): Take a vector of starting vertices
1240 instead of components. Remove unnecessary leftover handling.
1241 (ldist_gen): Do not build components or record other stores.
1242 (distribute_loop): Do not distribute loops containing stmts
1245 2013-09-13 Christian Bruel <christian.bruel@st.com>
1248 * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
1250 2013-09-13 Kai Tietz <ktietz@redhat.com>
1252 * config.gcc: Separate cases for mingw and cygwin targets,
1253 and add 64-bit cygwin target case.
1255 * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
1256 dll-export inline-functions.
1257 * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
1260 2013-09-13 Jeff Law <law@redhat.com>
1264 2013-09-06 Jeff Law <law@redhat.com>
1266 * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1267 edge implied equivalences into successor phis.
1269 2013-09-12 DJ Delorie <dj@redhat.com>
1271 * config/rl78/rl78-virt.md: Change from | to \; for asm line
1274 2013-09-12 Brooks Moses <bmoses@google.com>
1277 * Makefile.in: Do not install driver binaries in $(target)/bin.
1279 2013-09-12 DJ Delorie <dj@redhat.com>
1281 * config/rl78/rl78.opt (mrelax): New.
1282 * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
1283 * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
1285 * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
1287 (rl78_expand_epilogue): Likewise.
1289 2013-09-12 Vladimir Makarov <vmakarov@redhat.com>
1292 * lra-eliminations.c (remove_reg_equal_offset_note): New.
1293 (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
1294 pointer elimination with using remove_reg_equal_offset_note.
1296 2013-09-12 DJ Delorie <dj@redhat.com>
1298 * config/msp430/: New port.
1299 * config.gcc (msp430): Added.
1300 * doc/invoke.texi: Document MSP430 options.
1301 * doc/install.texi: Document msp430-elf
1302 * doc/md.texi: Document msp430-elf
1303 * doc/contrib.texi: Document msp430-elf
1305 * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
1308 2013-09-12 Martin Jambor <mjambor@suse.cz>
1311 * ipa-prop.c (remove_described_reference): Give up if the edge in the
1312 reference descriptor is NULL.
1313 (ipa_edge_removal_hook): If owning a reference descriptor, set its
1316 2013-09-12 Andrew MacLeod <amacleod@redhat.com>
1318 * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
1319 (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
1320 num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
1321 * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
1322 make_temp_ssa_name): move to tree-ssanames.h
1323 * tree-ssa-alias.h: Move prototype.
1324 * tree-ssa.h: Include tree-ssanames.h.
1325 * tree-ssanames.c (FREE_SSANAMES): Move to here.
1326 * tree-ssanames.h: New. Move items from tree-flow*.h
1327 * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
1329 2013-09-12 Richard Biener <rguenther@suse.de>
1331 PR tree-optimization/58404
1332 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
1333 propagate non-invariant addresses into dereferences wrapped
1334 in component references.
1336 2013-09-12 Richard Biener <rguenther@suse.de>
1338 PR tree-optimization/58402
1339 * passes.def: Move pass_late_warn_uninitialized later.
1341 2013-09-12 Andrew MacLeod <amacleod@redhat.com>
1343 * tree-ssa.h: New. Move content from tree-flow.h and
1345 * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
1346 Move prototypes belonging to tree-ssa.c.
1347 * tree-flow-inline.h (redirect_edge_var_map_def,
1348 redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
1350 * gimple.h: Adjust prototypes.
1351 * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
1353 * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
1354 * tree.h: Move prototype to tree-ssa.h.
1355 * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
1356 * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
1357 * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
1358 cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
1359 cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
1360 ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
1361 gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
1362 gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
1363 graphite-blocking.c, graphite-clast-to-gimple.c,
1364 graphite-dependences.c, graphite-interchange.c,
1365 graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
1366 graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
1367 ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
1368 ipa-reference.c, ipa-split.c, ipa-utils.c,
1369 loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
1370 lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
1371 passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
1372 tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1373 tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
1374 tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
1375 tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
1376 tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
1377 tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
1378 tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
1379 tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
1380 tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
1381 tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
1382 tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
1383 value-prof.c, var-tracking.c,
1384 varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
1386 2013-09-12 Richard Biener <rguenther@suse.de>
1388 PR tree-optimization/58396
1389 * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
1390 (build_rdg): Take a loop-nest parameter, fix memleaks.
1391 (distribute_loop): Compute loop-nest here and pass it to build_rdg.
1393 2013-09-12 Yuri Rumyantsev <ysrumyan@gmail.com>
1395 * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
1398 2013-09-12 Cameron McInally <cameron.mcinally@nyu.edu>
1400 * doc/extend.texi: Fix errors in x86 FMA builtin naming.
1401 The FMA instruction names should have a 'v' prefix.
1403 2013-09-12 Richard Biener <rguenther@suse.de>
1405 * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
1406 (dot_rdg): Use popen instead of system in optional code.
1407 (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
1408 (already_processed_vertex_p): Adjust.
1409 (has_anti_or_output_dependence, predecessor_has_mem_write,
1410 mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
1411 rdg_flag_uses): Remove.
1412 (rdg_flag_vertex): Simplify.
1413 (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
1415 (build_rdg_partition_for_component): Process the first vertex
1416 of a component only.
1417 (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
1419 2013-09-12 Alan Modra <amodra@gmail.com>
1421 * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
1423 2013-09-11 DJ Delorie <dj@redhat.com>
1424 Nick Clifton <nickc@redhat.com>
1426 * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
1427 (rl78_stack_based_mem): New.
1428 * config/rl78/constraints.md (Iv08): New.
1437 * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
1439 (movsi): Change from expand to insn-and-split.
1440 (ashrsi3): Clobber AX.
1444 * config/rl78/rl78.md (CC_REG): Fix.
1445 (addsi3): Allow memory and immediate operands.
1446 (addsi3_internal): Split into...
1447 (addsi3_internal_virt): ...new, and ...
1448 (addsi3_internal_real): ...new.
1450 (subsi3_internal_virt): New.
1451 (subsi3_internal_real): New.
1452 (mulsi3): Add memory operand.
1453 (mulsi3_rl78): Likewise.
1454 (mulsi3_g13): Likewise.
1455 * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
1456 (cbranchqi4_real): Add more constraint options.
1457 (cbranchhi4_real): Expand pattern.
1458 (cbranchhi4_real_signed): New.
1459 (cbranchhi4_real_inverted): New.
1460 (cbranchsi4_real_lt): New.
1461 (cbranchsi4_real_ge): New.
1462 (cbranchsi4_real_signed): New.
1463 (cbranchsi4_real): New.
1465 * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
1467 (lshrsi3_virt): Likewise.
1468 (ashlsi3_virt): Likewise.
1469 (cbranchqi4_virt_signed): New.
1470 (cbranchhi4_virt_signed): New.
1471 (cbranchsi4_virt): New.
1472 * config/rl78/rl78.c: Whitespace fixes throughout.
1473 (move_elim_pass): New.
1474 (pass_data_rl78_move_elim): New.
1475 (pass_rl78_move_elim): New.
1476 (make_pass_rl78_move_elim): New.
1477 (rl78_devirt_info): Run devirt earlier.
1478 (rl78_move_elim_info): New.
1479 (rl78_asm_file_start): Register it.
1480 (rl78_split_movsi): New.
1481 (rl78_as_legitimate_address): Allow virtual base registers when
1483 (rl78_addr_space_convert): Remove spurious debug stuff.
1484 (rl78_print_operand_1): Add z,s,S,r,E modifiers.
1485 (rl78_print_operand): More cases for not printing '#'.
1486 (rl78_expand_compare): Remove most of the logic.
1487 (content_memory): New.
1488 (clear_content_memory): New.
1489 (get_content_index): New.
1490 (get_content_name): New.
1491 (display_content_memory): New.
1492 (update_content): New.
1493 (record_content): New.
1494 (already_contains): New.
1495 (insn_ok_now): Re-recog insns with virtual registers.
1496 (add_postponed_content_update): New.
1497 (process_postponed_content_update): New.
1498 (gen_and_emit_move): New.
1499 (transcode_memory_rtx): Record new location content.
1500 Use gen_and_emit_move.
1501 (force_into_acc): New.
1502 (move_to_acc): Use gen_and_emit_move.
1503 (move_from_acc): Likewise.
1504 (move_acc_to_reg): Likewise.
1505 (move_to_x): Likewise.
1506 (move_to_hl): Likewise.
1507 (move_to_de): Likewise.
1508 (rl78_alloc_physical_registers_op1): Record location content.
1509 (has_constraint): New.
1510 (rl78_alloc_physical_registers_op2): Record location content.
1512 (rl78_alloc_physical_registers_ro1): Likewise.
1513 (rl78_alloc_physical_registers_cmp): Likewise.
1514 (rl78_alloc_physical_registers_umul): Likewise.
1515 (rl78_alloc_address_registers_macax): New.
1516 (rl78_alloc_physical_registers): Initialize and set location
1517 content memory as needed.
1518 (rl78_reorg): Make sure split2 is called.
1519 (rl78_rtx_costs): New.
1521 2013-09-11 Richard Sandiford <rdsandiford@googlemail.com>
1523 * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
1524 for (not (neg ...)) and (neg (not ...)) cases.
1526 2013-09-11 Richard Biener <rguenther@suse.de>
1529 * passes.def: Split critical edges before late uninit warning passes.
1530 * tree-cfg.c (pass_split_crit_edges): Implement clone method.
1532 2013-09-11 Jakub Jelinek <jakub@redhat.com>
1534 PR tree-optimization/58385
1535 * fold-const.c (build_range_check): If both low and high are NULL,
1536 use omit_one_operand_loc to preserve exp side-effects.
1538 2013-09-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1540 * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
1542 2013-09-11 Richard Biener <rguenther@suse.de>
1544 * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
1545 dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
1546 dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
1547 create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
1548 stmts_from_loop, known_dependences_p, build_empty_rdg,
1549 build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
1550 * tree-loop-distribution.c: ... here.
1551 * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
1552 RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
1553 RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
1554 struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
1555 * tree-loop-distribution.c: ... here.
1556 * tree-loop-distribution.c: Include gimple-pretty-print.h.
1557 (struct partition_s): Add loops member.
1558 (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
1559 rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
1560 build_rdg_partition_for_component, rdg_build_partitions): Adjust.
1562 2013-09-11 Alexander Ivchenko <alexander.ivchenko@intel.com>
1563 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
1564 Sergey Lega <sergey.s.lega@intel.com>
1565 Anna Tikhonova <anna.tikhonova@intel.com>
1566 Ilya Tocar <ilya.tocar@intel.com>
1567 Andrey Turetskiy <andrey.turetskiy@intel.com>
1568 Ilya Verbin <ilya.verbin@intel.com>
1569 Kirill Yukhin <kirill.yukhin@intel.com>
1570 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
1572 * config/i386/constraints.md (k): New.
1574 * config/i386/i386.c (const regclass_map): Add new mask registers.
1575 (dbx_register_map): Ditto.
1576 (dbx64_register_map): Ditto.
1577 (svr4_dbx_register_map): Ditto.
1578 (ix86_conditional_register_usage): Squash mask registers if AVX512F is
1580 (ix86_preferred_reload_class): Disable constants for mask registers.
1581 (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
1582 (ix86_hard_regno_mode_ok): Support new mask registers.
1583 (x86_order_regs_for_local_alloc): Ditto.
1584 * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
1585 (FIXED_REGISTERS): Add new mask registers.
1586 (CALL_USED_REGISTERS): Ditto.
1587 (REG_ALLOC_ORDER): Ditto.
1588 (VALID_MASK_REG_MODE): New.
1589 (FIRST_MASK_REG): Ditto.
1590 (LAST_MASK_REG): Ditto.
1591 (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
1592 (MAYBE_MASK_CLASS_P): New.
1593 (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
1594 (REG_CLASS_CONTENTS): Ditto.
1595 (MASK_REGNO_P): New.
1596 (ANY_MASK_REG_P): Ditto.
1597 (HI_REGISTER_NAMES): Add new mask registers.
1598 * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
1599 MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
1601 (attribute "type"): Add mskmov, msklog.
1602 (attribute "length_immediate"): Support them.
1603 (attribute "memory"): Ditto.
1604 (attribute "prefix_0f"): Ditto.
1605 (*movhi_internal): Support new mask registers.
1606 (*movqi_internal): Ditto.
1607 (define_split): Split out clobber pattern is a logic
1608 insn on mask registers.
1609 (*k<logic><mode>): New.
1610 (*andhi_1): Extend to support mask regs.
1611 (*andqi_1): Extend to support mask regs.
1613 (define_split): Split and-not to and and not if operands
1615 (*<code><mode>_1): Separate HI mode to new pattern...
1616 (*<code>hi_1): This.
1617 (*<code>qi_1): Extend to support mask regs.
1619 (kortestzhi): Ditto.
1620 (kortestchi): Ditto.
1622 (*one_cmpl<mode>2_1): Remove HImode and handle it...
1623 (*one_cmplhi2_1): ...Here, now with mask registers support.
1624 (*one_cmplqi2_1): Support new mask registers.
1625 (HI/QImode arithmetics splitter): Don't split if mask registers
1627 (HI/QImode not splitter): Ditto.
1628 * config/i386/predicated.md (mask_reg_operand): New.
1629 (general_reg_operand): Ditto.
1631 2013-09-11 Alexander Ivchenko <alexander.ivchenko@intel.com>
1633 * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
1634 * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
1636 2013-09-10 Jeff Law <law@redhat.com>
1638 PR tree-optimization/58380
1639 * tree-ssa-threadupdate.c (thread_block): Recognize another case
1640 of threading through a buried loop header.
1642 * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
1643 return value for single successor case.
1645 2013-09-10 Jan Hubicka <jh@suse.cz>
1647 * ipa-devirt.c (ipa_devirt): Enable with LTO.
1649 2013-09-10 Richard Earnshaw <rearnsha@arm.com>
1652 * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
1653 support conditional execution.
1654 (combine_vcvt_f64_<FCVTI32typename>): Likewise.
1656 2013-09-10 Vladimir Makarov <vmakarov@redhat.com>
1658 * lra.c (lra): Clear lra_optional_reload_pseudos before every
1660 * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
1661 Check destination too to check move insn.
1662 (undo_optional_reloads): Add check that the original peudo did not
1663 changed its allocation and the optional reload was inherited on last
1664 inheritance pass. Break loop after deciding to keep optional reload.
1665 (lra_undo_inheritance): Add check that inherited pseudo still in
1668 2013-09-10 James Greenhalgh <james.greenhalgh@arm.com>
1670 * config/aarch64/aarch64.md (generic_sched): New.
1671 * config/aarch64/aarch64-generic.md (load): Make conditional
1672 on generic_sched attribute.
1673 (nonload): Likewise.
1675 2013-09-10 Jan Hubicka <jh@suse.cz>
1677 * lto-cgraph.c: Include ipa-utils.h.
1678 (compute_ltrans_boundary): Also add possible targets into the boundary.
1680 2013-09-10 Jan Hubicka <jh@suse.cz>
1682 * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
1683 VAR_DECL of vtable rather than full expression.
1685 2013-09-10 Jan Hubicka <jh@suse.cz>
1686 Paolo Carlini <paolo.carlini@oracle.com>
1688 * cgraphunit.c (analyze_functions): Save input_location, set it
1689 to UNKNOWN_LOCATION and restore it at the end.
1691 2013-09-10 Martin Jambor <mjambor@suse.cz>
1693 * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
1694 represented by a thunk.
1696 2013-09-10 Jeff Law <law@redhat.com>
1698 PR tree-optimization/58343
1699 * tree-ssa-threadupdate.c (thread_block): Identify and disable
1700 jump threading requests through loop headers buried in the middle
1701 of a jump threading path.
1703 * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
1704 in return value/type.
1706 2013-09-10 Jakub Jelinek <jakub@redhat.com>
1708 PR rtl-optimization/58365
1709 * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
1710 resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
1713 2013-09-10 Richard Biener <rguenther@suse.de>
1715 * tree-data-ref.h (build_rdg): Drop all parameters but loop.
1716 * tree-data-ref.c (create_rdg_vertices): Collect all data
1717 references, signal failure to the caller, use data-ref API.
1718 (build_rdg): Compute data references only once. Maintain lifetime
1719 of data references and data dependences from within RDG.
1720 (free_rdg): Free dependence relations.
1721 * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
1722 inventing extra dependences.
1723 (distribute_loop): Update for RDG API changes.
1725 2013-09-10 Kai Tietz <ktietz@redhat.com>
1727 * doc/invoke.texi (fms-extensions): Document changed
1728 behavior for ms-abi targets.
1729 * config/i386/i386.c (ix86_option_override_internal):
1730 Set default value of option -fms-extension for ms-abi targets.
1732 2013-09-10 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
1734 * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
1736 2013-09-10 Alan Modra <amodra@gmail.com>
1739 * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
1741 2013-09-10 Alan Modra <amodra@gmail.com>
1743 * config/rs6000/predicates.md (add_cint_operand): New.
1744 (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
1745 * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
1746 using add_cint_operand.
1747 (largetoc_high_plus_aix): Likewise.
1749 2013-09-09 Jakub Jelinek <jakub@redhat.com>
1751 PR tree-optimization/58364
1752 * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
1753 BOOLEAN_TYPE, only invert in_p and continue with arg0 if
1754 the current range can't be an unconditional true or false.
1756 2013-09-09 James Greenhalgh <james.greenhalgh@arm.com>
1758 * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
1760 2013-09-09 Uros Bizjak <ubizjak@gmail.com>
1762 * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
1764 2013-09-09 Paolo Carlini <paolo.carlini@oracle.com>
1767 * doc/invoke.texi (-Wdelete-incomplete): Document it.
1769 2013-09-09 Ian Bolton <ian.bolton@arm.com>
1771 * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
1772 NO_REGS for immediate that can't be moved directly into FP_REGS.
1774 2013-09-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1776 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
1777 comparison with negated operand.
1778 * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
1781 2013-09-09 Richard Biener <rguenther@suse.de>
1784 * cfgloopmanip.c (fix_bb_placements): When fixing the placement
1785 of a subloop record all its block as affecting loop-closed SSA form.
1787 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1789 * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
1790 of an rtx/bitpos pair.
1791 (store_fixed_bit_field): Update accordingly.
1793 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1795 * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
1797 * builtins.c (expand_errno_check): Likewise.
1798 * dwarf2cfi.c (init_return_column_size): Likewise.
1799 * except.c (sjlj_mark_call_sites): Likewise.
1800 * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
1801 * lra-constraints.c (emit_inc): Likewise.
1802 * ree.c (combine_set_extension): Likewise.
1803 * regmove.c (fixup_match_2): Likewise.
1804 * reload1.c (inc_for_reload): Likewise.
1806 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1808 * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
1809 (force_to_mode, simplify_shift_const_1, simplify_comparison):
1810 Use gen_int_mode with the mode of the associated simplify_* call.
1811 * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
1812 * expmed.c (expand_shift_1): Likewise.
1813 * function.c (instantiate_virtual_regs_in_insn): Likewise.
1814 * loop-iv.c (iv_number_of_iterations): Likewise.
1815 * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1816 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
1818 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1820 * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
1821 of the associated expand_* call.
1822 (asan_emit_stack_protection): Likewise.
1823 * builtins.c (round_trampoline_addr): Likewise.
1824 * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
1825 * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
1826 (emit_store_flag): Likewise.
1827 * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
1828 (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
1830 * function.c (instantiate_virtual_regs_in_insn): Likewise.
1831 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1832 * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1833 * modulo-sched.c (generate_prolog_epilog): Likewise.
1834 * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
1835 (expand_ctz, expand_ffs, expand_unop): Likewise.
1837 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1839 * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
1840 of the associated gen_rtx_* call.
1841 * caller-save.c (init_caller_save): Likewise.
1842 * combine.c (find_split_point, make_extraction): Likewise.
1843 (make_compound_operation): Likewise.
1844 * dwarf2out.c (mem_loc_descriptor): Likewise.
1845 * explow.c (plus_constant, probe_stack_range): Likewise.
1846 * expmed.c (expand_mult_const): Likewise.
1847 * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
1848 * reload1.c (init_reload): Likewise.
1849 * valtrack.c (cleanup_auto_inc_dec): Likewise.
1850 * var-tracking.c (adjust_mems): Likewise.
1851 * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
1852 rather than gen_rtx_fmt_ee.
1854 2013-09-09 Jan Hubicka <jh@suse.cz>
1857 * value-prof.c (gimple_ic): Copy also abnormal edges.
1859 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
1861 * asan.c (asan_shadow_cst): Use gen_int_mode.
1863 2013-09-08 Jan Hubicka <jh@suse.cz>
1865 * ipa-profile.c: Add toplevel comment.
1866 (ipa_propagate_frequency_1): Be more conservative when profile is read.
1867 (contains_hot_call_p): New function.
1868 (ipa_propagate_frequency): Set frequencies based on counts when
1870 * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
1871 profile; do not tamper with profile after inlining if it is read.
1873 2013-09-08 Jan Hubicka <jh@suse.cz>
1875 * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
1878 2013-09-08 Jan Hubicka <jh@suse.cz>
1880 * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
1883 2013-09-08 Jeff Law <law@redhat.com>
1886 * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
1889 2013-09-08 Andi Kleen <ak@linux.intel.com>
1891 * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
1893 2013-09-08 Jan Hubicka <jh@suse.cz>
1895 * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
1897 2013-09-08 Richard Sandiford <rdsandiford@googlemail.com>
1899 * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
1900 for non-debug insns.
1901 * lra.c (new_insn_reg): Take the containing insn as a parameter.
1902 Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
1903 (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
1906 2013-09-08 Jan Hubicka <jh@suse.cz>
1908 * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
1909 targets and devirtualize to BUILT_IN_UNREACHABLE.
1910 * timevar.def (TV_IPA_UNREACHABLE): New timevar.
1911 * ipa.c (walk_polymorphic_call_targets): New function.
1912 (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
1913 functions; use the new timevar.
1914 * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
1915 was removed from the program.
1916 (record_binfo): If BINFO corresponds to an anonymous namespace, we may
1917 not consider it in the walk when its vtable is dead.
1918 (possible_polymorphic_call_targets_1): Pass anonymous flag to
1920 (devirt_variable_node_removal_hook): New function.
1921 (possible_polymorphic_call_targets): Also register
1922 devirt_variable_node_removal_hook.
1923 (ipa_devirt): Do not do non-speculative devirtualization.
1924 (gate_ipa_devirt): One execute if devirtualizing speculatively.
1926 2013-09-08 Jan Hubicka <jh@suse.cz>
1928 * cgraph.h (varpool_node_hook, varpool_node_hook_list,
1929 varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
1930 varpool_remove_variable_insertion_hook): Declare.
1931 * varpool.c (varpool_node_hook_list): New structure.
1932 (first_varpool_node_removal_hook,
1933 first_varpool_variable_insertion_hook): New variables.
1934 (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
1935 varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
1936 varpool_remove_variable_insertion_hook,
1937 varpool_call_variable_insertion_hooks): New functions.
1938 (varpool_remove_node): Use it.
1940 2013-09-08 Paolo Carlini <paolo.carlini@oracle.com>
1943 * diagnostic.c (diagnostic_build_prefix): When s.file is
1944 "<built-in>" don't output line and column numbers.
1946 2013-09-06 Jan Hubicka <jh@suse.cz>
1948 * cgraphunit.c (expand_thunk): Get body before touching arguments.
1949 * lto-streamer-out.c: Stream thunks, too.
1950 * lto-streamer-in.c (input_function): Pop cfun here
1951 (lto_read_body): Instead of here.
1953 2013-09-06 Caroline Tice <cmtice@google.com>
1955 * doc/install.texi: Add documentation for the --enable-vtable-verify
1956 and the --disable-libvtv configure options.
1958 2013-09-06 Jeff Law <law@redhat.com>
1960 * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1961 edge implied equivalences into successor phis.
1963 2013-09-06 Joern Rennecke <joern.rennecke@embecosm.com>
1965 * resource.c (mark_referenced_resources): Handle COND_EXEC.
1967 2013-09-06 Claudiu Zissulescu <claziss@synopsys.com>
1969 * resource.c (mark_target_live_regs): Compute resources taking
1970 into account if a call is predicated or not.
1972 2013-09-06 Eric Botcazou <ebotcazou@adacore.com>
1974 * toplev.c (output_stack_usage): Be prepared for suffixes created by
1975 the compiler in the function names.
1977 2013-09-06 Jan Hubicka <jh@suse.cz>
1980 * ipa-inline.c (has_caller_p): New function.
1981 (want_inline_function_to_all_callers_p): Use it.
1982 (sum_callers, inline_to_all_callers): Break out from ...
1983 (ipa_inline): ... here.
1985 2013-09-06 Jan Hubicka <jh@suse.cz>
1987 * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
1988 only when AVX is enabled.
1990 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1992 * config/aarch64/aarch64.md
1993 (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
1994 is fpsimd_<load/store>2.
1995 (load_pair<mode>): Likewise.
1996 (store_pair<mode>): Likewise.
1998 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2000 * config/arm/types.md (type): Add "mrs" type.
2001 * config/aarch64/aarch64.md
2002 (aarch64_load_tp_hard): Make type "mrs".
2004 (load_tp_hard): Make type "mrs".
2005 * config/arm/cortex-a15.md: Update with new attributes.
2006 * config/arm/cortex-a5.md: Update with new attributes.
2007 * config/arm/cortex-a53.md: Update with new attributes.
2008 * config/arm/cortex-a7.md: Update with new attributes.
2009 * config/arm/cortex-a8.md: Update with new attributes.
2010 * config/arm/cortex-a9.md: Update with new attributes.
2011 * config/arm/cortex-m4.md: Update with new attributes.
2012 * config/arm/cortex-r4.md: Update with new attributes.
2013 * config/arm/fa526.md: Update with new attributes.
2014 * config/arm/fa606te.md: Update with new attributes.
2015 * config/arm/fa626te.md: Update with new attributes.
2016 * config/arm/fa726te.md: Update with new attributes.
2018 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2020 * config/aarch64/aarch64.md
2021 (*movti_aarch64): Use "multiple" for type where v8type is "move2".
2022 (*movtf_aarch64): Likewise.
2024 (thumb1_movdi_insn): Use "multiple" for type where more than one
2025 instruction is used for a move.
2026 (*arm32_movhf): Likewise.
2027 (*thumb_movdf_insn): Likewise.
2029 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2031 * config/arm/types.md (type): Rename fcpys to fmov.
2033 (*arm_movsi_vfp): Rename type fcpys as fmov.
2034 (*thumb2_movsi_vfp): Likewise
2035 (*movhf_vfp_neon): Likewise
2036 (*movhf_vfp): Likewise
2037 (*movsf_vfp): Likewise
2038 (*thumb2_movsf_vfp): Likewise
2039 (*movsfcc_vfp): Likewise
2040 (*thumb2_movsfcc_vfp): Likewise
2041 * config/aarch64/aarch64-simd.md
2042 (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
2043 * config/aarch64/aarch64.md
2044 (*movsi_aarch64): Replace type mov_reg with fmovs.
2045 (*movdi_aarch64): Likewise
2046 (*movsf_aarch64): Likewise
2047 (*movdf_aarch64): Likewise
2049 (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
2050 * config/arm/iwmmxt.md
2051 (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
2052 * config/arm/arm1020e.md: Update with new attributes.
2053 * config/arm/cortex-a15-neon.md: Update with new attributes.
2054 * config/arm/cortex-a5.md: Update with new attributes.
2055 * config/arm/cortex-a53.md: Update with new attributes.
2056 * config/arm/cortex-a7.md: Update with new attributes.
2057 * config/arm/cortex-a8-neon.md: Update with new attributes.
2058 * config/arm/cortex-a9.md: Update with new attributes.
2059 * config/arm/cortex-m4-fpu.md: Update with new attributes.
2060 * config/arm/cortex-r4f.md: Update with new attributes.
2061 * config/arm/marvell-pj4.md: Update with new attributes.
2062 * config/arm/vfp11.md: Update with new attributes.
2064 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2066 * config/aarch64/aarch64.md
2067 (*madd<mode>): Fix type attribute.
2068 (*maddsi_uxtw): Likewise.
2069 (*msub<mode>): Likewise.
2070 (*msubsi_uxtw): Likewise.
2071 (<su_optab>maddsidi4): Likewise.
2072 (<su_optab>msubsidi4): Likewise.
2074 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2076 * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
2077 * config/arm/arm.md (core_cycles): Remove fdiv.
2078 * config/arm/vfp.md:
2079 (*sqrtsf2_vfp): Update for attribute changes.
2080 (*sqrtdf2_vfp): Likewise.
2081 * config/aarch64/aarch64.md:
2082 (sqrt<mode>2): Update for attribute changes.
2083 * config/arm/arm1020e.md: Update with new attributes.
2084 * config/arm/cortex-a15-neon.md: Update with new attributes.
2085 * config/arm/cortex-a5.md: Update with new attributes.
2086 * config/arm/cortex-a53.md: Update with new attributes.
2087 * config/arm/cortex-a7.md: Update with new attributes.
2088 * config/arm/cortex-a8-neon.md: Update with new attributes.
2089 * config/arm/cortex-a9.md: Update with new attributes.
2090 * config/arm/cortex-m4-fpu.md: Update with new attributes.
2091 * config/arm/cortex-r4f.md: Update with new attributes.
2092 * config/arm/marvell-pj4.md: Update with new attributes.
2093 * config/arm/vfp11.md: Update with new attributes.
2095 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2097 * config/arm/types.md
2098 (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
2099 * config/aarch64/aarch64.md
2100 (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
2102 (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
2103 (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
2104 (float<GPI:mode><GPF:mode>2): Likewise.
2106 (*truncsisf2_vfp): Update with new attributes.
2107 (*truncsidf2_vfp): Likewise.
2108 (fixuns_truncsfsi2): Likewise.
2109 (fixuns_truncdfsi2): Likewise.
2110 (*floatsisf2_vfp): Likewise.
2111 (*floatsidf2_vfp): Likewise.
2112 (floatunssisf2): Likewise.
2113 (floatunssidf2): Likewise.
2114 (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
2115 (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
2116 * config/arm/arm1020e.md: Update with new attributes.
2117 * config/arm/cortex-a15-neon.md: Update with new attributes.
2118 * config/arm/cortex-a5.md: Update with new attributes.
2119 * config/arm/cortex-a53.md: Update with new attributes.
2120 * config/arm/cortex-a7.md: Update with new attributes.
2121 * config/arm/cortex-a8-neon.md: Update with new attributes.
2122 * config/arm/cortex-a9.md: Update with new attributes.
2123 * config/arm/cortex-m4-fpu.md: Update with new attributes.
2124 * config/arm/cortex-r4f.md: Update with new attributes.
2125 * config/arm/marvell-pj4.md: Update with new attributes.
2126 * config/arm/vfp11.md: Update with new attributes.
2128 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2130 * config/aarch64/arm_neon.h
2131 (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
2132 (vqtbx<1,2,3,4><q>_s8): Likewise.
2134 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2136 * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
2137 * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
2138 (add<mode>3): Add type attribute.
2139 (add<mode>3): Likewise.
2140 (usadd<mode>3): Likewise.
2141 (ssadd<mode>3): Likewise.
2142 (sub<mode>3): Likewise.
2143 (sub<mode>3): Likewise.
2144 (ussub<mode>3): Likewise.
2145 (sssub<mode>3): Likewise.
2146 (ssmulsa3): Likewise.
2147 (usmulusa3): Likewise.
2148 (arm_usatsihi): Likewise.
2150 (*movdi_vfp): Add types for all instructions.
2151 (*movdi_vfp_cortexa8): Likewise.
2152 (*movhf_vfp_neon): Likewise.
2153 (*movhf_vfp): Likewise.
2154 (*movdf_vfp): Likewise.
2155 (*thumb2_movdf_vfp): Likewise.
2156 (*thumb2_movdfcc_vfp): Likewise.
2157 * config/arm/arm.md: Add type attribute to all insn patterns.
2158 (*thumb1_adddi3): Add type attribute.
2159 (*arm_adddi3): Likewise.
2160 (*adddi_sesidi_di): Likewise.
2161 (*adddi_zesidi_di): Likewise.
2162 (*thumb1_addsi3): Likewise.
2163 (addsi3_compare0): Likewise.
2164 (*addsi3_compare0_scratch): Likewise.
2165 (*compare_negsi_si): Likewise.
2166 (cmpsi2_addneg): Likewise.
2167 (*addsi3_carryin_<optab>): Likewise.
2168 (*addsi3_carryin_alt2_<optab>): Likewise.
2169 (*addsi3_carryin_clobercc_<optab>): Likewise.
2170 (*subsi3_carryin): Likewise.
2171 (*subsi3_carryin_const): Likewise.
2172 (*subsi3_carryin_compare): Likewise.
2173 (*subsi3_carryin_compare_const): Likewise.
2174 (*arm_subdi3): Likewise.
2175 (*thumb_subdi3): Likewise.
2176 (*subdi_di_zesidi): Likewise.
2177 (*subdi_di_sesidi): Likewise.
2178 (*subdi_zesidi_di): Likewise.
2179 (*subdi_sesidi_di): Likewise.
2180 (*subdi_zesidi_ze): Likewise.
2181 (thumb1_subsi3_insn): Likewise.
2182 (*arm_subsi3_insn): Likewise.
2183 (*anddi3_insn): Likewise.
2184 (*anddi_zesidi_di): Likewise.
2185 (*anddi_sesdi_di): Likewise.
2186 (*ne_zeroextracts): Likewise.
2187 (*ne_zeroextracts): Likewise.
2188 (*ite_ne_zeroextr): Likewise.
2189 (*ite_ne_zeroextr): Likewise.
2190 (*anddi_notdi_di): Likewise.
2191 (*anddi_notzesidi): Likewise.
2192 (*anddi_notsesidi): Likewise.
2193 (andsi_notsi_si): Likewise.
2194 (thumb1_bicsi3): Likewise.
2195 (*iordi3_insn): Likewise.
2196 (*iordi_zesidi_di): Likewise.
2197 (*iordi_sesidi_di): Likewise.
2198 (*thumb1_iorsi3_insn): Likewise.
2199 (*xordi3_insn): Likewise.
2200 (*xordi_zesidi_di): Likewise.
2201 (*xordi_sesidi_di): Likewise.
2202 (*arm_xorsi3): Likewise.
2203 (*andsi_iorsi3_no): Likewise.
2204 (*smax_0): Likewise.
2205 (*smax_m1): Likewise.
2206 (*arm_smax_insn): Likewise.
2207 (*smin_0): Likewise.
2208 (*arm_smin_insn): Likewise.
2209 (*arm_umaxsi3): Likewise.
2210 (*arm_uminsi3): Likewise.
2211 (*minmax_arithsi): Likewise.
2212 (*minmax_arithsi_): Likewise.
2213 (*satsi_<SAT:code>): Likewise.
2214 (arm_ashldi3_1bit): Likewise.
2215 (arm_ashrdi3_1bit): Likewise.
2216 (arm_lshrdi3_1bit): Likewise.
2217 (*arm_negdi2): Likewise.
2218 (*thumb1_negdi2): Likewise.
2219 (*arm_negsi2): Likewise.
2220 (*thumb1_negsi2): Likewise.
2221 (*negdi_extendsid): Likewise.
2222 (*negdi_zero_extend): Likewise.
2223 (*arm_abssi2): Likewise.
2224 (*thumb1_abssi2): Likewise.
2225 (*arm_neg_abssi2): Likewise.
2226 (*thumb1_neg_abss): Likewise.
2227 (one_cmpldi2): Likewise.
2228 (extend<mode>di2): Likewise.
2229 (*compareqi_eq0): Likewise.
2230 (*arm_extendhisi2addsi): Likewise.
2231 (*arm_movdi): Likewise.
2232 (*thumb1_movdi_insn): Likewise.
2233 (*arm_movt): Likewise.
2234 (*thumb1_movsi_insn): Likewise.
2235 (pic_add_dot_plus_four): Likewise.
2236 (pic_add_dot_plus_eight): Likewise.
2237 (tls_load_dot_plus_eight): Likewise.
2238 (*thumb1_movhi_insn): Likewise.
2239 (*thumb1_movsf_insn): Likewise.
2240 (*movdf_soft_insn): Likewise.
2241 (*thumb_movdf_insn): Likewise.
2242 (cbranchsi4_insn): Likewise.
2243 (cbranchsi4_scratch): Likewise.
2244 (*negated_cbranchsi4): Likewise.
2245 (*tbit_cbranch): Likewise.
2246 (*tlobits_cbranch): Likewise.
2247 (*tstsi3_cbranch): Likewise.
2248 (*cbranchne_decr1): Likewise.
2249 (*addsi3_cbranch): Likewise.
2250 (*addsi3_cbranch_scratch): Likewise.
2251 (*arm_cmpdi_insn): Likewise.
2252 (*arm_cmpdi_unsig): Likewise.
2253 (*arm_cmpdi_zero): Likewise.
2254 (*thumb_cmpdi_zero): Likewise.
2255 (*deleted_compare): Likewise.
2256 (*mov_scc): Likewise.
2257 (*mov_negscc): Likewise.
2258 (*mov_notscc): Likewise.
2259 (*cstoresi_eq0_thumb1_insn): Likewise.
2260 (cstoresi_nltu_thumb1): Likewise.
2261 (cstoresi_ltu_thu): Likewise.
2262 (thumb1_addsi3_addgeu): Likewise.
2263 (*arm_jump): Likewise.
2264 (*thumb_jump): Likewise.
2265 (*check_arch2): Likewise.
2266 (arm_casesi_internal): Likewise.
2267 (thumb1_casesi_dispatch): Likewise.
2268 (*arm_indirect_jump): Likewise.
2269 (*thumb1_indirect_jump): Likewise.
2271 (*and_scc): Likewise.
2272 (*ior_scc): Likewise.
2273 (*compare_scc): Likewise.
2274 (*cond_move): Likewise.
2275 (*cond_arith): Likewise.
2276 (*cond_sub): Likewise.
2277 (*cmp_ite0): Likewise.
2278 (*cmp_ite1): Likewise.
2279 (*cmp_and): Likewise.
2280 (*cmp_ior): Likewise.
2281 (*ior_scc_scc): Likewise.
2282 (*ior_scc_scc_cmp): Likewise.
2283 (*and_scc_scc): Likewise.
2284 (*and_scc_scc_cmp): Likewise.
2285 (*and_scc_scc_nod): Likewise.
2286 (*negscc): Likewise.
2287 (movcond_addsi): Likewise.
2288 (movcond): Likewise.
2289 (*ifcompare_plus_move): Likewise.
2290 (*if_plus_move): Likewise.
2291 (*ifcompare_move_plus): Likewise.
2292 (*if_move_plus): Likewise.
2293 (*ifcompare_arith_arith): Likewise.
2294 (*if_arith_arith): Likewise.
2295 (*ifcompare_arith_move): Likewise.
2296 (*if_arith_move): Likewise.
2297 (*ifcompare_move_arith): Likewise.
2298 (*if_move_arith): Likewise.
2299 (*ifcompare_move_not): Likewise.
2300 (*if_move_not): Likewise.
2301 (*ifcompare_not_move): Likewise.
2302 (*if_not_move): Likewise.
2303 (*ifcompare_shift_move): Likewise.
2304 (*if_shift_move): Likewise.
2305 (*ifcompare_move_shift): Likewise.
2306 (*if_move_shift): Likewise.
2307 (*ifcompare_shift_shift): Likewise.
2308 (*ifcompare_not_arith): Likewise.
2309 (*ifcompare_arith_not): Likewise.
2310 (*if_arith_not): Likewise.
2311 (*ifcompare_neg_move): Likewise.
2312 (*if_neg_move): Likewise.
2313 (*ifcompare_move_neg): Likewise.
2314 (*if_move_neg): Likewise.
2315 (prologue_thumb1_interwork): Likewise.
2316 (*cond_move_not): Likewise.
2317 (*sign_extract_onebit): Likewise.
2318 (*not_signextract_onebit): Likewise.
2319 (stack_tie): Likewise.
2320 (align_4): Likewise.
2321 (align_8): Likewise.
2322 (consttable_end): Likewise.
2323 (consttable_1): Likewise.
2324 (consttable_2): Likewise.
2325 (consttable_4): Likewise.
2326 (consttable_8): Likewise.
2327 (consttable_16): Likewise.
2328 (*thumb1_tablejump): Likewise.
2329 (prefetch): Likewise.
2330 (force_register_use): Likewise.
2331 (thumb_eh_return): Likewise.
2332 (load_tp_hard): Likewise.
2333 (load_tp_soft): Likewise.
2334 (tlscall): Likewise.
2335 (*arm_movtas_ze): Likewise.
2336 (*arm_rev): Likewise.
2337 (*arm_revsh): Likewise.
2338 (*arm_rev16): Likewise.
2339 * config/arm/thumb2.md
2340 (*thumb2_smaxsi3): Likewise.
2341 (*thumb2_sminsi3): Likewise.
2342 (*thumb32_umaxsi3): Likewise.
2343 (*thumb2_uminsi3): Likewise.
2344 (*thumb2_negdi2): Likewise.
2345 (*thumb2_abssi2): Likewise.
2346 (*thumb2_neg_abss): Likewise.
2347 (*thumb2_movsi_insn): Likewise.
2348 (tls_load_dot_plus_four): Likewise.
2349 (*thumb2_movhi_insn): Likewise.
2350 (*thumb2_mov_scc): Likewise.
2351 (*thumb2_mov_negs): Likewise.
2352 (*thumb2_mov_negs): Likewise.
2353 (*thumb2_mov_nots): Likewise.
2354 (*thumb2_mov_nots): Likewise.
2355 (*thumb2_movsicc_): Likewise.
2356 (*thumb2_movsfcc_soft_insn): Likewise.
2357 (*thumb2_indirect_jump): Likewise.
2358 (*thumb2_and_scc): Likewise.
2359 (*thumb2_ior_scc): Likewise.
2360 (*thumb2_ior_scc_strict_it): Likewise.
2361 (*thumb2_cond_move): Likewise.
2362 (*thumb2_cond_arith): Likewise.
2363 (*thumb2_cond_ari): Likewise.
2364 (*thumb2_cond_sub): Likewise.
2365 (*thumb2_negscc): Likewise.
2366 (*thumb2_movcond): Likewise.
2367 (thumb2_casesi_internal): Likewise.
2368 (thumb2_casesi_internal_pic): Likewise.
2369 (*thumb2_alusi3_short): Likewise.
2370 (*thumb2_mov<mode>_shortim): Likewise.
2371 (*thumb2_addsi_short): Likewise.
2372 (*thumb2_subsi_short): Likewise.
2373 (thumb2_addsi3_compare0): Likewise.
2374 (*thumb2_cbz): Likewise.
2375 (*thumb2_cbnz): Likewise.
2376 (*thumb2_one_cmplsi2_short): Likewise.
2377 (*thumb2_negsi2_short): Likewise.
2378 (*orsi_notsi_si): Likewise.
2379 * config/arm/arm1020e.md: Update with new attributes.
2380 * config/arm/arm1026ejs.md: Update with new attributes.
2381 * config/arm/arm1136jfs.md: Update with new attributes.
2382 * config/arm/arm926ejs.md: Update with new attributes.
2383 * config/arm/cortex-a15.md: Update with new attributes.
2384 * config/arm/cortex-a5.md: Update with new attributes.
2385 * config/arm/cortex-a53.md: Update with new attributes.
2386 * config/arm/cortex-a7.md: Update with new attributes.
2387 * config/arm/cortex-a8.md: Update with new attributes.
2388 * config/arm/cortex-a9.md: Update with new attributes.
2389 * config/arm/cortex-m4.md: Update with new attributes.
2390 * config/arm/cortex-r4.md: Update with new attributes.
2391 * config/arm/fa526.md: Update with new attributes.
2392 * config/arm/fa606te.md: Update with new attributes.
2393 * config/arm/fa626te.md: Update with new attributes.
2394 * config/arm/fa726te.md: Update with new attributes.
2396 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
2398 * config/aarch64/aarch64-simd.md
2399 (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
2400 <vwx> iterator to ensure correct register choice.
2401 (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
2402 (aarch64_sqdmull_n<mode>): Likewise.
2403 (aarch64_sqdmull2_n<mode>_internal): Likewise.
2404 * config/aarch64/arm_neon.h
2405 (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
2406 (vml<as><q>_n_<su>16): Likewise.
2407 (vml<as>l_high_lane<q>_<su>16): Likewise.
2408 (vml<as>l_high_n_<su>16): Likewise.
2409 (vml<as>l_lane<q>_<su>16): Likewise.
2410 (vml<as>l_n_<su>16): Likewise.
2411 (vmul<q>_lane<q>_<su>16): Likewise.
2412 (vmul<q>_n_<su>16): Likewise.
2413 (vmull_lane<q>_<su>16): Likewise.
2414 (vmull_n_<su>16): Likewise.
2415 (vmull_high_lane<q>_<su>16): Likewise.
2416 (vmull_high_n_<su>16): Likewise.
2417 (vqrdmulh<q>_n_s16): Likewise.
2419 2013-09-06 Tejas Belagod <tejas.belagod@arm.com>
2421 * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
2422 have the correct lane parameter.
2424 2013-09-06 Richard Biener <rguenther@suse.de>
2426 * cfganal.c (control_dependences::~control_dependences):
2427 Properly free all of the vector.
2429 2013-09-06 Kirill Yukhin <kirill.yukhin@intel.com>
2432 * config/i386/i386.c (ix86_conditional_register_usage):
2433 Proper initialize extended SSE registers.
2435 2013-09-06 Jan Hubicka <jh@suse.cz>
2437 PR tree-optimization/58311
2438 * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
2440 2013-09-06 Jan Hubicka <jh@suse.cz>
2442 * Makefile.in (tree-sra.o): Update dependencies.
2443 * tree-sra.c: Include ipa-utils.h
2444 (scan_function): Use recursive_call_p.
2445 (has_caller_p): New function.
2446 (cgraph_for_node_and_aliases): Count also callers of aliases.
2448 2013-09-06 Jan Hubicka <jh@suse.cz>
2451 * cgraph.h (symtab_semantically_equivalent_p): Declare.
2452 * tree-tailcall.c: Include ipa-utils.h.
2453 (find_tail_calls): Use it.
2454 * ipa-pure-const.c (check_call): Likewise.
2455 * ipa-utils.c (recursive_call_p): New function.
2456 * ipa-utils.h (recursive_call_p): Dclare.
2457 * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
2458 (symtab_semantically_equivalent_p): New function.
2459 * Makefile.in (tree-tailcall.o): Update dependencies.
2461 2013-09-06 Eric Botcazou <ebotcazou@adacore.com>
2463 * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
2466 2013-09-06 Richard Biener <rguenther@suse.de>
2468 * lto-streamer.h (lto_global_var_decls): Remove.
2469 * Makefile.in (OBJS): Remove lto-symtab.o.
2470 (lto-symtab.o): Remove.
2471 (GTFILES): Remove lto-symtab.c
2472 * lto-symtab.c: Move to lto/
2474 2013-09-06 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2476 * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
2477 (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
2478 (UNSPEC_FPINT_RINT): New constant definitions.
2479 (FPINT, fpint_name, fpint_roundingmode): New integer iterator
2480 definition with 2 attributes.
2481 ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
2482 ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
2485 2013-09-06 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2487 * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
2488 ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
2489 Set the mnemonic attribute to "bcr_flush". Set the "z196prop"
2490 attribute to "z196_alone".
2491 * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
2494 2013-09-06 Richard Biener <rguenther@suse.de>
2496 * basic-block.h (class control_dependences): New.
2497 * tree-ssa-dce.c (control_dependence_map): Remove.
2499 (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
2500 (set_control_dependence_map_bit, clear_control_dependence_bitmap,
2501 find_pdom, find_control_dependence, find_all_control_dependences):
2503 (mark_control_dependent_edges_necessary,
2504 find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
2505 tree_dce_done, perform_tree_ssa_dce): Adjust.
2506 * cfganal.c (set_control_dependence_map_bit,
2507 clear_control_dependence_bitmap, find_pdom, find_control_dependence,
2508 find_all_control_dependences): Move from tree-ssa-dce.c and
2509 implement as methods of control_dependences class.
2510 (control_dependences::control_dependences): New.
2511 (control_dependences::~control_dependences): Likewise.
2512 (control_dependences::get_edges_dependent_on): Likewise.
2513 (control_dependences::get_edge): Likewise.
2515 2013-09-04 Jan Hubicka <jh@suse.cz>
2517 * tree.c (types_same_for_odr): Drop overactive check.
2518 * ipa-devirt.c (hash_type_name): Likewise.
2520 2013-09-04 Jan Hubicka <jh@suse.cz>
2522 * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
2523 (analyze_functions): ... here.
2525 2013-09-04 Jan Hubicka <jh@suse.cz>
2528 * cgraphunit.c (analyze_functions): Clear AUX fields
2529 after processing; initialize assembler name has.
2531 2013-09-05 Jeff Law <law@redhat.com>
2533 * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
2534 from thread_around_empty_block. Record threading path into PATH.
2535 Recurse if threading through the initial block is successful.
2536 (thread_across_edge): Corresponding changes to slightly simplify.
2538 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
2540 * config/aarch64/aarch64.md
2541 (type): Remove frecpe, frecps, frecpx.
2542 (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
2543 fix to be a TARGET_SIMD instruction.
2544 (aarch64_frecps): Remove.
2545 * config/aarch64/aarch64-simd.md
2546 (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
2547 (aarch64_frecps<mode>): Handle all float/vector of float modes.
2549 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
2550 Sofiane Naci <sofiane.naci@arm.com>
2552 * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
2553 into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
2554 Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
2555 "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
2556 "logic_reg", "logics_reg", "rev". Expand "arlo_shift" into
2557 "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
2558 "logics_shift_imm". Expand "arlo_shift_reg" into "alu_shift_reg",
2559 "alus_shift_reg", "logic_shift_reg", "logics_shift_reg". Expand "clz"
2560 into "clz, "rbit". Rename "shift" to "shift_imm".
2561 * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
2562 changes. Update for attribute changes all occurrences of arlo_* and
2564 * config/arm/arm-fixed.md: Update for attribute changes
2565 all occurrences of arlo_* types.
2566 * config/arm/thumb2.md: Update for attribute changes all occurrences
2568 * config/arm/arm.c (xscale_sched_adjust_cost): (rtx insn, rtx
2569 (cortexa7_older_only): Likewise.
2570 (cortexa7_younger): Likewise.
2571 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2572 (1020alu_shift_op): Likewise.
2573 (1020alu_shift_reg_op): Likewise.
2574 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2575 (alu_shift_op): Likewise.
2576 (alu_shift_reg_op): Likewise.
2577 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2578 (11_alu_shift_op): Likewise.
2579 (11_alu_shift_reg_op): Likewise.
2580 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2581 (9_alu_shift_reg_op): Likewise.
2582 * config/arm/cortex-a15.md (cortex_a15_alu): Update for
2584 (cortex_a15_alu_shift): Likewise.
2585 (cortex_a15_alu_shift_reg): Likewise.
2586 * config/arm/cortex-a5.md (cortex_a5_alu): Update for
2588 (cortex_a5_alu_shift): Likewise.
2589 * config/arm/cortex-a53.md (cortex_a53_alu): Update for
2591 (cortex_a53_alu_shift): Likewise.
2592 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
2594 (cortex_a7_alu_reg): Likewise.
2595 (cortex_a7_alu_shift): Likewise.
2596 * config/arm/cortex-a8.md (cortex_a8_alu): Update for
2598 (cortex_a8_alu_shift): Likewise.
2599 (cortex_a8_alu_shift_reg): Likewise.
2600 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2601 (cortex_a9_dp_shift): Likewise.
2602 * config/arm/cortex-m4.md (cortex_m4_alu): Update for
2604 * config/arm/cortex-r4.md
2605 (cortex_r4_alu): Update for attribute changes.
2606 (cortex_r4_mov): Likewise.
2607 (cortex_r4_alu_shift_reg): Likewise.
2608 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2609 (526_alu_shift_op): Likewise.
2610 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2611 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2612 (626te_alu_shift_op): Likewise.
2613 * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
2614 (726te_alu_shift_op): Likewise.
2615 (726te_alu_shift_reg_op): Likewise.
2616 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2617 (mp626_alu_shift_op): Likewise.
2618 * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
2619 (pj4_alu_conds): Likewise.
2620 (pj4_shift): Likewise.
2621 (pj4_shift_conds): Likewise.
2622 (pj4_alu_shift): Likewise.
2623 (pj4_alu_shift_conds): Likewise.
2624 * config/aarch64/aarch64.md: Update for attribute change
2625 all occurrences of arlo_* and shift* types.
2627 2013-09-05 Mike Stump <mikestump@comcast.net>
2629 * tree.h: Move documentation for tree_function_decl to tree-core.h
2630 with the declaration.
2632 2013-09-05 Peter Bergner <bergner@vnet.ibm.com>
2635 * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
2636 looking for widest mode.
2638 2013-09-05 Eric Botcazou <ebotcazou@adacore.com>
2640 * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
2642 2013-09-05 Richard Biener <rguenther@suse.de>
2644 PR tree-optimization/58137
2645 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
2646 Do not create vectors of pointers.
2647 * tree-vect-loop.c (get_initial_def_for_induction): Use proper
2648 types for the components of the vector initializer.
2649 * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
2650 allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
2652 2013-09-05 Martin Jambor <mjambor@suse.cz>
2654 * ipa-prop.c (remove_described_reference): Accept missing references,
2655 return false if that hppens, otherwise return true.
2656 (cgraph_node_for_jfunc): New function.
2657 (try_decrement_rdesc_refcount): Likewise.
2658 (try_make_edge_direct_simple_call): Use them.
2659 (ipa_edge_removal_hook): Remove references from rdescs.
2660 (ipa_edge_duplication_hook): Clone rdescs and their references
2661 when the new edge has the same caller as the old one.
2662 * cgraph.c (cgraph_resolve_speculation): Remove speculative
2663 reference before removing any edges.
2665 2013-09-05 Richard Earnshaw <rearnsha@arm.com>
2667 * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
2669 * thumb2.md (thumb2_storewb_parisi): New pattern.
2671 2013-09-05 Yufeng Zhang <yufeng.zhang@arm.com>
2673 * config/aarch64/aarch64-option-extensions.def: Add
2674 AARCH64_OPT_EXTENSION of 'crc'.
2675 * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
2676 (AARCH64_ISA_CRC): Ditto.
2677 * doc/invoke.texi (-march and -mcpu feature modifiers): Add
2678 description of the CRC extension.
2680 2013-09-05 Alexander Ivchenko <alexander.ivchenko@intel.com>
2682 * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
2683 * config/rs6000/linux.h: Ditto.
2684 * alpha/linux.h: Ditto.
2685 * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
2686 no_c99_libc_has_function.
2687 * config/c6x/uclinux-elf.h: Ditto.
2688 * config/lm32/uclinux-elf.h: Ditto.
2689 * config/m68k/uclinux.h: Ditto.
2690 * config/moxie/uclinux.h: Ditto.
2691 * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
2692 (crisv32-*-linux*, cris-*-linux*): Ditto.
2693 * config/bfin/bfin.c: Include "tm_p.h".
2695 2013-09-05 Richard Biener <rguenther@suse.de>
2697 * tree-vect-loop.c (vect_analyze_loop_operations): Properly
2698 check for a definition without a basic-block.
2700 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
2701 Sofiane Naci <sofiane.naci@arm.com>
2703 * config/aarch64/aarch64.md
2704 (*movti_aarch64): Rename r_2_f and f_2_r.
2705 (*movsf_aarch64): Likewise.
2706 (*movdf_aarch64): Likewise.
2707 (*movtf_aarch64): Likewise.
2708 (aarch64_movdi_<mode>low): Likewise.
2709 (aarch64_movdi_<mode>high): Likewise.
2710 (aarch64_mov<mode>high_di): Likewise.
2711 (aarch64_mov<mode>low_di): Likewise.
2712 (aarch64_movtilow_tilow): Likewise.
2713 * config/arm/arm.md (attribute "neon_type"): Delete. Move attribute
2714 values to config/arm/types.md
2715 (attribute "conds"): Update for attribute change.
2716 (anddi3_insn): Likewise.
2717 (iordi3_insn): Likewise.
2718 (xordi3_insn): Likewise.
2719 (one_cmpldi2): Likewise.
2720 * config/arm/types.md (type): Add Neon types.
2721 * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
2722 use "type" attribute.
2723 (movmisalign<mode>_neon_store): Likewise.
2724 (movmisalign<mode>_neon_load): Likewise.
2725 (vec_set<mode>_internal): Likewise.
2726 (vec_setv2di_internal): Likewise.
2727 (vec_extract<mode>): Likewise.
2728 (vec_extractv2di): Likewise.
2729 (add<mode>3_neon): Likewise.
2730 (adddi3_neon): Likewise.
2731 (sub<mode>3_neon): Likewise.
2732 (subdi3_neon): Likewise.
2733 (mul<mode>3_neon): Likewise.
2734 (mul<mode>3add<mode>_neon): Likewise.
2735 (mul<mode>3neg<mode>add<mode>_neon): Likewise.
2736 (fma<VCVTF:mode>4)): Likewise.
2737 (fma<VCVTF:mode>4_intrinsic): Likewise.
2738 (fmsub<VCVTF:mode>4)): Likewise.
2739 (fmsub<VCVTF:mode>4_intrinsic): Likewise.
2740 (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
2741 (ior<mode>3): Likewise.
2742 (and<mode>3): Likewise.
2743 (anddi3_neon): Likewise.
2744 (orn<mode>3_neon): Likewise.
2745 (orndi3_neon): Likewise.
2746 (bic<mode>3_neon): Likewise.
2747 (bicdi3_neon): Likewise.
2748 (xor<mode>3): Likewise.
2749 (one_cmpl<mode>2): Likewise.
2750 (abs<mode>2): Likewise.
2751 (neg<mode>2): Likewise.
2752 (umin<mode>3_neon): Likewise.
2753 (umax<mode>3_neon): Likewise.
2754 (smin<mode>3_neon): Likewise.
2755 (smax<mode>3_neon): Likewise.
2756 (vashl<mode>3): Likewise.
2757 (vashr<mode>3_imm): Likewise.
2758 (vlshr<mode>3_imm): Likewise.
2759 (ashl<mode>3_signed): Likewise.
2760 (ashl<mode>3_unsigned): Likewise.
2761 (neon_load_count): Likewise.
2762 (ashldi3_neon_noclobber): Likewise.
2763 (signed_shift_di3_neon): Likewise.
2764 (unsigned_shift_di3_neon): Likewise.
2765 (ashrdi3_neon_imm_noclobber): Likewise.
2766 (lshrdi3_neon_imm_noclobber): Likewise.
2767 (widen_ssum<mode>3): Likewise.
2768 (widen_usum<mode>3): Likewise.
2769 (quad_halves_<code>v4si): Likewise.
2770 (quad_halves_<code>v4sf): Likewise.
2771 (quad_halves_<code>v8hi): Likewise.
2772 (quad_halves_<code>v16qi): Likewise.
2773 (reduc_splus_v2di): Likewise.
2774 (neon_vpadd_internal<mode>): Likewise.
2775 (neon_vpsmin<mode>): Likewise.
2776 (neon_vpsmax<mode>): Likewise.
2777 (neon_vpumin<mode>): Likewise.
2778 (neon_vpumax<mode>): Likewise.
2779 (ss_add<mode>_neon): Likewise.
2780 (us_add<mode>_neon): Likewise.
2781 (ss_sub<mode>_neon): Likewise.
2782 (us_sub<mode>_neon): Likewise.
2783 (neon_vadd<mode>_unspec): Likewise.
2784 (neon_vaddl<mode>): Likewise.
2785 (neon_vaddw<mode>): Likewise.
2786 (neon_vhadd<mode>): Likewise.
2787 (neon_vqadd<mode>): Likewise.
2788 (neon_vaddhn<mode>): Likewise.
2789 (neon_vmul<mode>): Likewise.
2790 (neon_vmla<mode>): Likewise.
2791 (neon_vmlal<mode>): Likewise.
2792 (neon_vmls<mode>): Likewise.
2793 (neon_vmlsl<mode>): Likewise.
2794 (neon_vqdmulh<mode>): Likewise.
2795 (neon_vqdmlal<mode>): Likewise.
2796 (neon_vqdmlsl<mode>): Likewise.
2797 (neon_vmull<mode>): Likewise.
2798 (neon_vqdmull<mode>): Likewise.
2799 (neon_vsub<mode>_unspec): Likewise.
2800 (neon_vsubl<mode>): Likewise.
2801 (neon_vsubw<mode>): Likewise.
2802 (neon_vqsub<mode>): Likewise.
2803 (neon_vhsub<mode>): Likewise.
2804 (neon_vsubhn<mode>): Likewise.
2805 (neon_vceq<mode>): Likewise.
2806 (neon_vcge<mode>): Likewise.
2807 (neon_vcgeu<mode>): Likewise.
2808 (neon_vcgt<mode>): Likewise.
2809 (neon_vcgtu<mode>): Likewise.
2810 (neon_vcle<mode>): Likewise.
2811 (neon_vclt<mode>): Likewise.
2812 (neon_vcage<mode>): Likewise.
2813 (neon_vcagt<mode>): Likewise.
2814 (neon_vtst<mode>): Likewise.
2815 (neon_vabd<mode>): Likewise.
2816 (neon_vabdl<mode>): Likewise.
2817 (neon_vaba<mode>): Likewise.
2818 (neon_vabal<mode>): Likewise.
2819 (neon_vmax<mode>): Likewise.
2820 (neon_vmin<mode>): Likewise.
2821 (neon_vpaddl<mode>): Likewise.
2822 (neon_vpadal<mode>): Likewise.
2823 (neon_vpmax<mode>): Likewise.
2824 (neon_vpmin<mode>): Likewise.
2825 (neon_vrecps<mode>): Likewise.
2826 (neon_vrsqrts<mode>): Likewise.
2827 (neon_vqabs<mode>): Likewise.
2828 (neon_vqneg<mode>): Likewise.
2829 (neon_vcls<mode>): Likewise.
2830 (clz<mode>2): Likewise.
2831 (popcount<mode>2): Likewise.
2832 (neon_vrecpe): Likewise.
2833 (neon_vrsqrte): Likewise.
2834 (neon_vget_lane<mode>_sext_internal): Likewise.
2835 (neon_vget_lane<mode>_zext_internal): Likewise.
2836 (neon_vdup_n<mode>): Likewise.
2837 (neon_vdup_nv2di): Likewise.
2838 (neon_vdpu_lane<mode>_internal): Likewise.
2839 (neon_vswp<mode>): Likewise.
2840 (float<mode><V_cvtto>2): Likewise.
2841 (floatuns<mode><V_cvtto>2): Likewise.
2842 (fix_trunc<mode><V_cvtto>)2): Likewise
2843 (fixuns_trunc<mode><V_cvtto)2): Likewise.
2844 (neon_vcvt<mode>): Likewise.
2845 (neon_vcvtv4sfv4hf): Likewise.
2846 (neon_vcvtv4hfv4sf): Likewise.
2847 (neon_vcvt_n<mode>): Likewise.
2848 (neon_vmovn<mode>): Likewise.
2849 (neon_vqmovn<mode>): Likewise.
2850 (neon_vqmovun<mode>): Likewise.
2851 (neon_vmovl<mode>): Likewise.
2852 (neon_vmul_lane<mode>): Likewise.
2853 (neon_vmull_lane<mode>): Likewise.
2854 (neon_vqdmull_lane<mode>): Likewise.
2855 (neon_vqdmulh_lane<mode>): Likewise.
2856 (neon_vmla_lane<mode>): Likewise.
2857 (neon_vmlal_lane<mode>): Likewise.
2858 (neon_vqdmlal_lane<mode>): Likewise.
2859 (neon_vmls_lane<mode>): Likewise.
2860 (neon_vmlsl_lane<mode>): Likewise.
2861 (neon_vqdmlsl_lane<mode>): Likewise.
2862 (neon_vext<mode>): Likewise.
2863 (neon_vrev64<mode>): Likewise.
2864 (neon_vrev32<mode>): Likewise.
2865 (neon_vrev16<mode>): Likewise.
2866 (neon_vbsl<mode>_internal): Likewise.
2867 (neon_vshl<mode>): Likewise.
2868 (neon_vqshl<mode>): Likewise.
2869 (neon_vshr_n<mode>): Likewise.
2870 (neon_vshrn_n<mode>): Likewise.
2871 (neon_vqshrn_n<mode>): Likewise.
2872 (neon_vqshrun_n<mode>): Likewise.
2873 (neon_vshl_n<mode>): Likewise.
2874 (neon_vqshl_n<mode>): Likewise.
2875 (neon_vqshlu_n<mode>): Likewise.
2876 (neon_vshll_n<mode>): Likewise.
2877 (neon_vsra_n<mode>): Likewise.
2878 (neon_vsri_n<mode>): Likewise.
2879 (neon_vsli_n<mode>): Likewise.
2880 (neon_vtbl1v8qi): Likewise.
2881 (neon_vtbl2v8qi): Likewise.
2882 (neon_vtbl3v8qi): Likewise.
2883 (neon_vtbl4v8qi): Likewise.
2884 (neon_vtbx1v8qi): Likewise.
2885 (neon_vtbx2v8qi): Likewise.
2886 (neon_vtbx3v8qi): Likewise.
2887 (neon_vtbx4v8qi): Likewise.
2888 (neon_vtrn<mode>_internal): Likewise.
2889 (neon_vzip<mode>_internal): Likewise.
2890 (neon_vuzp<mode>_internal): Likewise.
2891 (neon_vld1<mode>): Likewise.
2892 (neon_vld1_lane<mode>): Likewise.
2893 (neon_vld1_dup<mode>): Likewise.
2894 (neon_vld1_dupv2di): Likewise.
2895 (neon_vst1<mode>): Likewise.
2896 (neon_vst1_lane<mode>): Likewise.
2897 (neon_vld2<mode>): Likewise.
2898 (neon_vld2_lane<mode>): Likewise.
2899 (neon_vld2_dup<mode>): Likewise.
2900 (neon_vst2<mode>): Likewise.
2901 (neon_vst2_lane<mode>): Likewise.
2902 (neon_vld3<mode>): Likewise.
2903 (neon_vld3qa<mode>): Likewise.
2904 (neon_vld3qb<mode>): Likewise.
2905 (neon_vld3_lane<mode>): Likewise.
2906 (neon_vld3_dup<mode>): Likewise.
2907 (neon_vst3<mode>): Likewise.
2908 (neon_vst3qa<mode>): Likewise.
2909 (neon_vst3qb<mode>): Likewise.
2910 (neon_vst3_lane<mode>): Likewise.
2911 (neon_vld4<mode>): Likewise.
2912 (neon_vld4qa<mode>): Likewise.
2913 (neon_vld4qb<mode>): Likewise.
2914 (neon_vld4_lane<mode>): Likewise.
2915 (neon_vld4_dup<mode>): Likewise.
2916 (neon_vst4<mode>): Likewise.
2917 (neon_vst4qa<mode>): Likewise.
2918 (neon_vst4qb<mode>): Likewise.
2919 (neon_vst4_lane<mode>): Likewise.
2920 (neon_vec_unpack<US>_lo_<mode>): Likewise.
2921 (neon_vec_unpack<US>_hi_<mode>): Likewise.
2922 (neon_vec_<US>mult_lo_<mode>): Likewise.
2923 (neon_vec_<US>mult_hi_<mode>): Likewise.
2924 (neon_vec_<US>shiftl_<mode>): Likewise.
2925 (neon_unpack<US>_<mode>): Likewise.
2926 (neon_vec_<US>mult_<mode>): Likewise.
2927 (vec_pack_trunc_<mode>): Likewise.
2928 (neon_vec_pack_trunk_<mode>): Likewise.
2929 (neon_vabd<mode>_2): Likewise.
2930 (neon_vabd<mode>_3): Likewise.
2931 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2932 (thumb2_movsi_vfp): Likewise.
2933 (movdi_vfp): Likewise.
2934 (movdi_vfp_cortexa8): Likewise.
2935 (movhf_vfp_neon): Likewise.
2936 (movhf_vfp): Likewiwse.
2937 (movsf_vfp): Likewiwse.
2938 (thumb2_movsf_vfp): Likewiwse.
2939 (movdf_vfp): Likewise.
2940 (thumb2_movdf_vfp): Likewise.
2941 (movsfcc_vfp): Likewise.
2942 (thumb2_movsfcc_vfp): Likewise.
2943 (movdfcc_vfp): Likewise.
2944 (thumb2_movdfcc_vfp): Likewise.
2945 * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
2946 * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
2947 (v10_v2c): Likewise.
2948 * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
2950 (cortex_a15_neon_int_2): Likewise.
2951 (cortex_a15_neon_int_3): Likewise.
2952 (cortex_a15_neon_int_4): Likewise.
2953 (cortex_a15_neon_int_5): Likewise.
2954 (cortex_a15_neon_vqneg_vqabs): Likewise.
2955 (cortex_a15_neon_vmov): Likewise.
2956 (cortex_a15_neon_vaba): Likewise.
2957 (cortex_a15_neon_vaba_qqq): Likewise.
2958 (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2959 (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2960 (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
2961 scalar_64_32_long_scalar): Likewise.
2962 (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2963 (cortex_a15_neon_mla_qqq_8_16): Likewise.
2964 (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2965 lotype_qdd_64_32_long): Likewise.
2966 (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2967 (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2968 (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2969 (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2970 (cortex_a15_neon_shift_1): Likewise.
2971 (cortex_a15_neon_shift_2): Likewise.
2972 (cortex_a15_neon_shift_3): Likewise.
2973 (cortex_a15_neon_vshl_ddd): Likewise.
2974 (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2975 (cortex_a15_neon_vsra_vrsra): Likewise.
2976 (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2977 (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2978 (cortex_a15_neon_fp_vmul_ddd): Likewise.
2979 (cortex_a15_neon_fp_vmul_qqd): Likewise.
2980 (cortex_a15_neon_fp_vmla_ddd): Likewise.
2981 (cortex_a15_neon_fp_vmla_qqq): Likewise.
2982 (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2983 (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2984 (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2985 (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2986 (cortex_a15_neon_bp_simple): Likewise.
2987 (cortex_a15_neon_bp_2cycle): Likewise.
2988 (cortex_a15_neon_bp_3cycle): Likewise.
2989 (cortex_a15_neon_vld1_1_2_regs): Likewise.
2990 (cortex_a15_neon_vld1_3_4_regs): Likewise.
2991 (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2992 (cortex_a15_neon_vld2_4_regs): Likewise.
2993 (cortex_a15_neon_vld3_vld4): Likewise.
2994 (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2995 (cortex_a15_neon_vst1_3_4_regs): Likewise.
2996 (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
2997 (cortex_a15_neon_vst3_vst4): Likewise.
2998 (cortex_a15_neon_vld1_vld2_lane): Likewise.
2999 (cortex_a15_neon_vld3_vld4_lane" 10
3000 (cortex_a15_neon_vst1_vst2_lane): Likewise.
3001 (cortex_a15_neon_vst3_vst4_lane): Likewise.
3002 (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3003 (cortex_a15_neon_ldm_2): Likewise.0
3004 (cortex_a15_neon_stm_2): Likewise.
3005 (cortex_a15_neon_mcr): Likewise.
3006 (cortex_a15_neon_mcr_2_mcrr): Likewise.
3007 (cortex_a15_neon_mrc): Likewise.
3008 (cortex_a15_neon_mrrc): Likewise.
3009 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3011 (cortex_a15_alu_shift): Likewise.
3012 (cortex_a15_alu_shift_reg): Likewise.
3013 (cortex_a15_mult32): Likewise.
3014 (cortex_a15_mult64): Likewise.
3015 (cortex_a15_block): Likewise.
3016 (cortex_a15_branch): Likewise.
3017 (cortex_a15_load1): Likewise.
3018 (cortex_a15_load3): Likewise.
3019 (cortex_a15_store1): Likewise.
3020 (cortex_a15_store3): Likewise.
3021 (cortex_a15_call): Likewise.
3022 * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
3023 (cortex_a5_f2r): Likewise.
3024 * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
3026 (cortex_a53_f2r): Likewise.
3027 * config/arm/cortex-a7.md
3028 (cortex_a7_branch): Update for attribute change.
3029 (cortex_a7_call): Likewise.
3030 (cortex_a7_alu_imm): Likewise.
3031 (cortex_a7_alu_reg): Likewise.
3032 (cortex_a7_alu_shift): Likewise.
3033 (cortex_a7_mul): Likewise.
3034 (cortex_a7_load1): Likewise.
3035 (cortex_a7_store1): Likewise.
3036 (cortex_a7_load2): Likewise.
3037 (cortex_a7_store2): Likewise.
3038 (cortex_a7_load3): Likewise.
3039 (cortex_a7_store3): Likewise.
3040 (cortex_a7_load4): Likewise.
3041 (cortex_a7_store4): Likewise.
3042 (cortex_a7_fpalu): Likewise.
3043 (cortex_a7_fconst): Likewise.
3044 (cortex_a7_fpmuls): Likewise.
3045 (cortex_a7_neon_mul): Likewise.
3046 (cortex_a7_fpmacs): Likewise.
3047 (cortex_a7_neon_mla: Likewise.
3048 (cortex_a7_fpmuld: Likewise.
3049 (cortex_a7_fpmacd: Likewise.
3050 (cortex_a7_fpfmad: Likewise.
3051 (cortex_a7_fdivs: Likewise.
3052 (cortex_a7_fdivd: Likewise.
3053 (cortex_a7_r2f: Likewise.
3054 (cortex_a7_f2r: Likewise.
3055 (cortex_a7_f_flags: Likewise.
3056 (cortex_a7_f_loads: Likewise.
3057 (cortex_a7_f_loadd: Likewise.
3058 (cortex_a7_f_stores: Likewise.
3059 (cortex_a7_f_stored: Likewise.
3060 (cortex_a7_neon): Likewise.
3061 * config/arm/cortex-a8-neon.md
3062 (cortex_a8_neon_mrc): Update for attribute change.
3063 (cortex_a8_neon_mrrc): Likewise.
3064 (cortex_a8_neon_int_1): Likewise.
3065 (cortex_a8_neon_int_2): Likewise.
3066 (cortex_a8_neon_int_3): Likewise.
3067 (cortex_a8_neon_int_4): Likewise.
3068 (cortex_a8_neon_int_5): Likewise.
3069 (cortex_a8_neon_vqneg_vqabs): Likewise.
3070 (cortex_a8_neon_vmov): Likewise.
3071 (cortex_a8_neon_vaba): Likewise.
3072 (cortex_a8_neon_vaba_qqq): Likewise.
3073 (cortex_a8_neon_vsma): Likewise.
3074 (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3075 (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3076 (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3077 long_scalar): Likewise.
3078 (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3079 (cortex_a8_neon_mla_qqq_8_16): Likewise.
3080 (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3081 long_scalar_qdd_64_32_long): Likewise.
3082 (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3083 (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3084 (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3085 (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3086 (cortex_a8_neon_shift_1): Likewise.
3087 (cortex_a8_neon_shift_2): Likewise.
3088 (cortex_a8_neon_shift_3): Likewise.
3089 (cortex_a8_neon_vshl_ddd): Likewise.
3090 (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3091 (cortex_a8_neon_vsra_vrsra): Likewise.
3092 (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3093 (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3094 (cortex_a8_neon_fp_vsum): Likewise.
3095 (cortex_a8_neon_fp_vmul_ddd): Likewise.
3096 (cortex_a8_neon_fp_vmul_qqd): Likewise.
3097 (cortex_a8_neon_fp_vmla_ddd): Likewise.
3098 (cortex_a8_neon_fp_vmla_qqq): Likewise.
3099 (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3100 (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3101 (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3102 (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3103 (cortex_a8_neon_bp_simple): Likewise.
3104 (cortex_a8_neon_bp_2cycle): Likewise.
3105 (cortex_a8_neon_bp_3cycle): Likewise.
3106 (cortex_a8_neon_ldr): Likewise.
3107 (cortex_a8_neon_str): Likewise.
3108 (cortex_a8_neon_vld1_1_2_regs): Likewise.
3109 (cortex_a8_neon_vld1_3_4_regs): Likewise.
3110 (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3111 (cortex_a8_neon_vld2_4_regs): Likewise.
3112 (cortex_a8_neon_vld3_vld4): Likewise.
3113 (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3114 (cortex_a8_neon_vst1_3_4_regs): Likewise.
3115 (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3116 (cortex_a8_neon_vst3_vst4): Likewise.
3117 (cortex_a8_neon_vld1_vld2_lane): Likewise.
3118 (cortex_a8_neon_vld3_vld4_lane): Likewise.
3119 (cortex_a8_neon_vst1_vst2_lane): Likewise.
3120 (cortex_a8_neon_vst3_vst4_lane): Likewise.
3121 (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3122 (cortex_a8_neon_mcr): Likewise.
3123 (cortex_a8_neon_mcr_2_mcrr): Likewise.
3124 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3125 * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
3127 (ca9_neon_mrrc): Likewise.
3128 (cortex_a9_neon_int_1): Likewise.
3129 (cortex_a9_neon_int_2): Likewise.
3130 (cortex_a9_neon_int_3): Likewise.
3131 (cortex_a9_neon_int_4): Likewise.
3132 (cortex_a9_neon_int_5): Likewise.
3133 (cortex_a9_neon_vqneg_vqabs): Likewise.
3134 (cortex_a9_neon_vmov): Likewise.
3135 (cortex_a9_neon_vaba): Likewise.
3136 (cortex_a9_neon_vaba_qqq): Likewise.
3137 (cortex_a9_neon_vsma): Likewise.
3138 (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3139 (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3140 (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3141 long_scalar): Likewise.
3142 (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3143 (cortex_a9_neon_mla_qqq_8_16): Likewise.
3144 (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3145 long_scalar_qdd_64_32_long): Likewise.
3146 (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3147 (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3148 (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3149 (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3150 (cortex_a9_neon_shift_1): Likewise.
3151 (cortex_a9_neon_shift_2): Likewise.
3152 (cortex_a9_neon_shift_3): Likewise.
3153 (cortex_a9_neon_vshl_ddd): Likewise.
3154 (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3155 (cortex_a9_neon_vsra_vrsra): Likewise.
3156 (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3157 (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3158 (cortex_a9_neon_fp_vsum): Likewise.
3159 (cortex_a9_neon_fp_vmul_ddd): Likewise.
3160 (cortex_a9_neon_fp_vmul_qqd): Likewise.
3161 (cortex_a9_neon_fp_vmla_ddd): Likewise.
3162 (cortex_a9_neon_fp_vmla_qqq): Likewise.
3163 (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3164 (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3165 (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3166 (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3167 (cortex_a9_neon_bp_simple): Likewise.
3168 (cortex_a9_neon_bp_2cycle): Likewise.
3169 (cortex_a9_neon_bp_3cycle): Likewise.
3170 (cortex_a9_neon_ldr): Likewise.
3171 (cortex_a9_neon_str): Likewise.
3172 (cortex_a9_neon_vld1_1_2_regs): Likewise.
3173 (cortex_a9_neon_vld1_3_4_regs): Likewise.
3174 (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3175 (cortex_a9_neon_vld2_4_regs): Likewise.
3176 (cortex_a9_neon_vld3_vld4): Likewise.
3177 (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3178 (cortex_a9_neon_vst1_3_4_regs): Likewise.
3179 (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3180 (cortex_a9_neon_vst3_vst4): Likewise.
3181 (cortex_a9_neon_vld1_vld2_lane): Likewise.
3182 (cortex_a9_neon_vld3_vld4_lane): Likewise.
3183 (cortex_a9_neon_vst1_vst2_lane): Likewise.
3184 (cortex_a9_neon_vst3_vst4_lane): Likewise.
3185 (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3186 (cortex_a9_neon_mcr): Likewise.
3187 (cortex_a9_neon_mcr_2_mcrr): Likewise.
3188 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3189 (cortex_a9_fps): Likewise.
3190 * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
3192 (cortex_m4_fmuls): Likewise.
3193 * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
3195 (cortex_r4_mrc): Likewise.
3196 * config/arm/iterators.md: Update comment referring to neon_type.
3197 * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
3198 (iwmmxt_movsi_insn): Likewise.
3199 * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
3201 (pj4_core_to_vfp): Likewise.
3202 * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
3204 * config/arm/vfp11.md (vfp_fload): Update for attribute change.
3205 (vfp_fstore): Likewise.
3206 * doc/md.texi: Change references to neon_type to refer to type.
3208 2013-09-04 Dodji Seketeli <dodji@redhat.com>
3210 * tree.h (DECL_BUILT_IN): Fix typo in comment.
3212 2013-09-04 David Edelsohn <dje.gcc@gmail.com>
3214 * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
3217 2013-09-04 Easwaran Raman <eraman@google.com>
3220 * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
3221 (build_and_add_sum): Use it.
3222 (appears_later_in_bb): Simplify code.
3224 2013-09-04 Teresa Johnson <tejohnson@google.com>
3226 * dumpfile.c (dump_finish): Don't close stderr/stdout.
3228 2013-09-04 James Greenhalgh <james.greenhalgh@arm.com>
3230 * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
3232 2013-09-04 Jan Hubicka <jh@suse.cz>
3234 * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
3235 * ipa-devirt.c: Include diganostic.h
3236 (odr_type_d): Add types and types_set.
3237 (hash_type_name): Work for types with vtables during LTO.
3238 (odr_hasher::remove): Fix comment; destroy types_set.
3239 (add_type_duplicate): New function,
3240 (get_odr_type): Use it.
3241 (dump_type_inheritance_graph): Dump type duplicates.
3242 * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
3244 * tree.c (types_same_for_odr): Give exact answers on types with
3247 2013-09-04 Dodji Seketeli <dodji@redhat.com>
3249 * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
3250 explaining their differences.
3252 2013-09-04 Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
3254 * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
3256 2013-09-03 Jeff Law <law@redhat.com>
3258 * tree-ssa-threadedge.c (thread_across_edge): Record entire path
3259 when not threading through a joiner block. Pass joiner/no joiner
3260 state to register_jump_thread.
3261 * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
3262 state from argument rather than implying on path length.
3263 Dump the entire jump thread path into debugging dump.
3264 * tree-flow.h (register_jump_thread): Update prototype.
3266 2013-08-29 Xinliang David Li <davidxl@google.com>
3268 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3269 Remove a trivial gcc_assert.
3271 2013-08-29 Xinliang David Li <davidxl@google.com>
3273 * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
3274 * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
3275 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3276 Delay base decl alignment adjustment.
3277 * tree-vectorizer.c (vect_destroy_datarefs): New function.
3278 * tree-vectorizer.h: New data structure.
3279 (set_dr_misalignment): New function.
3280 (dr_misalignment): Ditto.
3281 * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
3282 (vectorizable_load): Ditto.
3283 (ensure_base_align): New function.
3284 (vectorize_loops): Add dbg_cnt support.
3285 (execute_vect_slp): Ditto.
3286 * dbgcnt.def: New debug counter.
3287 * Makefile: New dependency.
3289 2013-09-03 Meador Inge <meadori@codesourcery.com>
3293 2013-08-30 Meador Inge <meadori@codesourcery.com>
3295 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3297 2013-09-03 David Edelsohn <dje.gcc@gmail.com>
3299 * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
3300 function descriptor.
3302 2013-09-03 Richard Biener <rguenther@suse.de>
3304 * tree-affine.c (add_elt_to_tree): Fix association issue,
3305 avoid useless converts and make sure to always return a
3306 properly typed result.
3308 2013-09-03 Richard Biener <rguenther@suse.de>
3311 * fold-const.c (negate_expr_p): Fix division case.
3312 (negate_expr): Likewise.
3314 2013-09-03 Richard Biener <rguenther@suse.de>
3317 * tree-streamer-out.c: Include tm.h.
3318 * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
3320 2013-09-03 Jan Hubicka <jh@suse.cz>
3322 * tree-profile.c (tree_profiling): Cleanup CFG when done.
3324 2013-09-03 Alan Modra <amodra@gmail.com>
3326 * config.gcc (powerpc*-*-linux*): Add support for little-endian
3327 multilibs to big-endian target and vice versa.
3328 * config/rs6000/t-linux64: Use := assignment on all vars.
3329 (MULTILIB_EXTRA_OPTS): Remove fPIC.
3330 (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
3331 * config/rs6000/t-linux64le: New file.
3332 * config/rs6000/t-linux64bele: New file.
3333 * config/rs6000/t-linux64lebe: New file.
3335 2013-09-02 Jan Hubicka <jh@suse.cz>
3337 * ipa-inline-transform.c (inline_transform): Do not
3338 optimize_inline_calls when not optimizing.
3340 2013-09-02 Jan Hubicka <jh@suse.cz>
3342 * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
3343 duplicated nodes for assembler names.
3344 * symtab.c (symtab_unregister_node): Do not attempt to unlink
3345 hard registers from assembler name hash.
3347 2013-09-02 Jan Hubicka <jh@suse.cz>
3349 * ipa-split.c (execute_split_functions): Split externally visible
3350 functions called once.
3352 2013-09-02 Martin Jambor <mjambor@suse.cz>
3355 * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
3356 linked list. When finding the correct duplicate, also consider also
3357 the caller in additon to its inlined_to node.
3359 2013-09-02 James Greenhalgh <james.greenhalgh@arm.com>
3361 * config/aarch64/aarch64-simd-builtins.def
3362 (dup_lane_scalar): Remove.
3363 * config/aarch64/aarch64-simd.md
3364 (aarch64_simd_dup): Add 'w->w' alternative.
3365 (aarch64_dup_lane<mode>): Allow for VALL.
3366 (aarch64_dup_lane_scalar<mode>): Remove.
3367 (aarch64_dup_lane_<vswap_width_name><mode>): New.
3368 (aarch64_get_lane_signed<mode>): Add w->w altenative.
3369 (aarch64_get_lane_unsigned<mode>): Likewise.
3370 (aarch64_get_lane<mode>): Likewise.
3371 * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
3372 (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
3373 * config/aarch64/iterators.md (VSWAP_WIDTH): New.
3374 (VCON): Change container of V2SF.
3375 (vswap_width_name): Likewise.
3376 * config/aarch64/arm_neon.h
3377 (__aarch64_vdup_lane_any): New.
3378 (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3379 (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
3380 (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3382 2013-09-02 Eric Botcazou <ebotcazou@adacore.com>
3385 * expr.c (emit_move_complex): Do not move complex FP values as parts if
3386 the source or the destination is a single hard register.
3388 2013-09-02 Richard Biener <rguenther@suse.de>
3391 * tree-scalar-evolution.c (instantiate_scev_name): Allow
3394 2013-09-02 Richard Biener <rguenther@suse.de>
3396 * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
3397 arithmetic to sizetype.
3399 2013-09-02 Bin Cheng <bin.cheng@arm.com>
3401 * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
3402 Find auto-increment use both before and after candidate.
3404 2013-09-02 Marek Polacek <polacek@redhat.com>
3406 * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
3408 2013-09-01 Jan Hubicka <jh@suse.cz>
3410 * Makefile.in: Add ipa-profile.o
3411 (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
3412 * cgraph.c (struct cgraph_propagate_frequency_data,
3413 cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
3414 ipa-profile.c; replace cgraph_ by ipa_ prefix.
3415 * cgraph.h (cgraph_propagate_frequency): Remove.
3416 * ipa-inline-analysis.c: Include ipa-utils.h;
3417 drop duplicated cfgloop.h.
3418 (inline_update_callee_summaries): Update.
3419 * ipa-profile.c: New file.
3420 * ipa-utils.h (ipa_propagate_frequency): Declare.
3421 * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
3422 data-streamer.h, value-prof.h.
3423 (symtab_remove_unreachable_nodes): Update profile.
3424 (struct histogram_entry, histogram, histogram_pool, histogram_hash,
3425 account_time_size, cmp_counts, dump_histogram,
3426 ipa_profile_generate_summary, ipa_profile_write_summary,
3427 ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
3428 pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
3429 Move to ipa-profile.c.
3431 2013-09-01 John David Anglin <danglin@gcc.gnu.org>
3433 * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
3435 2013-09-01 Jan Hubicka <jh@suse.cz>
3437 * common.opt (fdevirtualize-speculatively): New function.
3438 * invoke.texi (fdevirtualize-speculatively): Document.
3439 * ipa-devirt.c: Include ipa-inline.h
3440 (likely_target_p): New function.
3441 (ipa_devirt): New function.
3442 (gate_ipa_devirt): New function.
3443 (pass_data_ipa_devirt): New static var.
3444 (pass_ipa_devirt): Likewise.
3445 (make_pass_ipa_devirt): New function.
3446 * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
3447 (common_handle_option): Disable devirtualization when
3448 value range profiling is available.
3449 * passes.def (pass_ipa_devirt): Add.
3450 * timever.def (TV_IPA_DEVIRT): New timevar.
3451 * tree-pass.h (make_pass_ipa_devirt):
3453 2013-09-01 Iain Sandoe <iain@codesourcery.com>
3455 * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
3456 include sanitize(undefined).
3458 2013-08-31 Diego Novillo <dnovillo@google.com>
3460 * Makefile.in (TREE_CORE_H): Define.
3462 (GTFILES): Add tree-core.h.
3463 * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
3465 * tree-core.h: New file.
3466 Move all data structures, enum, typedefs, global
3467 declarations and constants from ...
3470 2013-08-31 Jan Hubicka <jh@suse.cz>
3472 * bulitins.c (expand_builtin): Do not early exit for gcov
3473 instrumented functions.
3475 2013-08-31 Marek Polacek <polacek@redhat.com>
3477 * ubsan.c: Include tm_p.h.
3479 2013-08-31 Jan Hubicka <jh@suse.cz>
3481 * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
3484 * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
3485 * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
3486 * tree-cfg.c (verify_gimple_label): ... here.
3487 * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
3488 (ipa_merge_profiles): New function.
3489 * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
3490 (lto_input_function_body): Likewise.
3491 * ipa-utils.h (ipa_merge_profiles): Declare.
3492 * lto-streamer.h (lto_input_function_body): Update prototype.
3493 (emit_label_in_global_context_p): Remove.
3494 * lto-symtab.c: Include ipa-utils.h
3495 (lto_cgraph_replace_node): Use ipa_merge_profiles.
3497 2013-08-31 Jan Hubicka <jh@suse.cz>
3499 * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
3501 2013-08-31 Jan Hubicka <jh@suse.cz>
3503 * basic-block.h (apply_scale): Make scale parmeter gcov_type.
3505 2013-08-31 Uros Bizjak <ubizjak@gmail.com>
3507 * config/alpha/alpha.c (alpha_emit_conditional_move): Update
3508 "cmp" RTX before signed_comparison_operator check to account
3511 2013-08-30 Jan Hubicka <jh@suse.cz>
3513 * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
3514 (detect_type_change_1): Rename to ...
3515 (detect_type_change): ... this one; early return on non-polymorphic
3517 (detect_type_change_ssa): Add comp_type parameter; update
3518 use of detect_type_change.
3519 (compute_complex_assign_jump_func): Add param_type parameter;
3520 update use of detect_type_change_ssa.
3521 (compute_complex_ancestor_jump_func): Likewise.
3522 (ipa_get_callee_param_type): New function.
3523 (ipa_compute_jump_functions_for_edge): Compute parameter type;
3524 update calls to the jump function computation functions.
3526 2013-08-30 Teresa Johnson <tejohnson@google.com>
3527 Steven Bosscher <steven@gcc.gnu.org>
3529 * cfgrtl.c (fixup_new_cold_bb): New routine.
3530 (commit_edge_insertions): Invoke fixup_partitions.
3531 (find_partition_fixes): New routine.
3532 (fixup_partitions): Ditto.
3533 (verify_hot_cold_block_grouping): Update comments.
3534 (rtl_verify_edges): Invoke find_partition_fixes.
3535 (rtl_verify_bb_pointers): Update comments.
3536 (rtl_verify_bb_layout): Ditto.
3537 * basic-block.h (probably_never_executed_edge_p): Declare.
3538 (fixup_partitions): Ditto.
3539 * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
3540 * bb-reorder.c (sanitize_hot_paths): New function.
3541 (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
3543 * predict.c (probably_never_executed_edge_p): New routine.
3544 * cfg.c (check_bb_profile): Add partition insanity warnings.
3546 2013-08-30 Meador Inge <meadori@codesourcery.com>
3548 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3550 2013-08-30 Marek Polacek <polacek@redhat.com>
3552 * Makefile.in (ubsan.o): Add.
3553 (c-family/c-ubsan.o): Add.
3554 (builtins.o): Add ubsan.h dependency.
3555 * ubsan.h: New file.
3556 * ubsan.c: New file.
3557 * common.opt: Add -fsanitize=undefined option.
3558 (flag_sanitize): Add variable.
3559 (fsanitize=): Add option. Add Driver.
3560 (fsanitize=thread): Remove option.
3561 (fsanitize=address): Likewise.
3562 (static-libubsan): New option.
3563 * doc/invoke.texi: Document the new flag and -static-libubsan.
3564 * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
3565 (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
3566 * builtin-attrs.def (ATTR_COLD): Define.
3567 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3568 * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
3569 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
3570 * flag-types.h (sanitize_code): New enum.
3571 * opts.c (common_handle_option): Parse command line arguments
3572 of -fsanitize=. Add -fsanitize=unreachable option.
3573 * varasm.c (get_variable_section): Adjust.
3574 (assemble_noswitch_variable): Likewise.
3575 (assemble_variable): Likewise.
3576 (output_constant_def_contents): Likewise.
3577 (categorize_decl_for_section): Likewise.
3578 (place_block_symbol): Likewise.
3579 (output_object_block): Likewise.
3580 * builtins.def: Likewise.
3581 * toplev.c (compile_file): Likewise.
3582 (process_options): Likewise.
3583 * cppbuiltin.c: Likewise.
3584 * tsan.c (tsan_pass): Likewise.
3585 (tsan_gate): Likewise.
3586 (tsan_gate_O0): Likewise.
3587 * cfgexpand.c (partition_stack_vars): Likewise.
3588 (expand_stack_vars): Likewise.
3589 (defer_stack_allocation): Likewise.
3590 (expand_used_vars): Likewise.
3591 * cfgcleanup.c (old_insns_match_p): Likewise.
3592 * asan.c (asan_finish_file): Likewise.
3593 (asan_instrument): Likewise.
3594 (gate_asan): Likewise.
3595 (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
3596 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3597 (asan_global_struct): Use pointer_sized_int_node instead
3598 calling build_nonstandard_integer_type.
3599 (initialize_sanitizer_builtins): Likewise.
3600 (asan_finish_file): Likewise.
3601 * gcc.c: Document %{%:function(args):X}.
3602 (static_spec_functions): Add sanitize.
3603 (handle_spec_function): Add retval_nonnull argument and if non-NULL,
3604 store funcval != NULL there.
3605 (do_spec_1): Adjust handle_spec_function caller.
3606 (handle_braces): Allow %:function(args) as condition.
3607 (sanitize_spec_function): New function.
3608 (ADD_STATIC_LIBUBSAN_LIBS): Define.
3609 (LIBUBSAN_SPEC): Likewise.
3610 (LIBUBSAN_EARLY_SPEC): Likewise.
3611 (SANITIZER_SPEC): Handle libubsan.
3612 (SANITIZER_EARLY_SPEC): Likewise.
3613 * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
3614 instead of fsanitize=address.
3615 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
3616 instead of fsanitize=address*.
3617 * builtins.c: Include ubsan.h.
3618 (fold_builtin_0): Instrument __builtin_unreachable.
3619 * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
3620 instead of flag_asan.
3621 * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
3622 (pointer_sized_int_node): Define.
3623 * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
3625 2013-08-30 Mike Stump <mikestump@comcast.net>
3627 * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
3630 2013-08-29 Jan Hubicka <jh@suse.cz>
3632 * cgraph.c (cgraph_function_body_availability): Handle weakref
3634 * passes.def: Remove pass_fixup_cfg.
3635 * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
3636 track when we need to remove functions.
3637 (gate_ipa_inline): Execute inlining always; add comment why.
3638 (pass_data_ipa_inline): Remove TODO_remove_functions.
3639 * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
3640 do not produce summaries.
3641 * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
3642 (symtab_nonoverwritable_alias): Assert we are not called on weakref.
3643 * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
3644 constant pool and vtable.
3646 2013-08-30 Tejas Belagod <tejas.belagod@arm.com>
3648 * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
3649 New arm_neon.h's internal macros to specify 64-bit constants.
3650 Avoid using stdint.h's macros.
3652 2013-08-30 Joern Rennecke <joern.rennecke@embecosm.com>
3654 * recog.c (verify_changes): Verify that changes[i].old is non-zero
3655 before applying REG_P.
3657 2013-08-30 Jakub Jelinek <jakub@redhat.com>
3659 PR tree-optimization/58277
3660 * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
3661 after seeing too many stmts with vdef in between dombb and current
3662 bb, invalidate everything.
3664 2013-08-30 Richard Biener <rguenther@suse.de>
3666 * fold-const.c (fold_single_bit_test): Fix overflow test.
3668 2013-08-30 Eric Botcazou <ebotcazou@adacore.com>
3670 * function.c (assign_parm_setup_reg): For a parameter passed by pointer
3671 and which can live in a register, always retrieve the value on entry.
3672 * var-tracking.c (add_stores): Treat the copy on entry for a parameter
3673 passed by invisible reference specially.
3674 (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
3675 (vt_add_function_parameter): Correctly deal with a parameter passed by
3676 invisible reference.
3678 2013-08-30 Jan Hubicka <jh@suse.cz>
3680 * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
3682 2013-08-30 Richard Biener <rguenther@suse.de>
3684 PR tree-optimization/58228
3685 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
3686 allow invariant loads in nested loop vectorization.
3688 2013-08-30 Richard Biener <rguenther@suse.de>
3690 PR tree-optimization/58223
3691 * tree-loop-distribution.c (has_anti_dependence): Rename to ...
3692 (has_anti_or_output_dependence): ... this and adjust to also
3693 look for output dependences.
3694 (mark_nodes_having_upstream_mem_writes): Adjust.
3695 (rdg_flag_uses): Likewise.
3697 2013-08-30 Richard Biener <rguenther@suse.de>
3699 PR tree-optimization/58010
3700 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
3701 assert that we have a loop-closed PHI.
3703 2013-08-29 Jan Hubicka <jh@suse.cz>
3705 * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
3706 * cgraph.c (cgraph_release_function_body): Free decl_in_state.
3707 * lto-section-in.c (lto_free_function_in_decl_state): New function.
3708 (lto_free_function_in_decl_state_for_node): New function.
3710 2013-08-29 Xinliang David Li <davidxl@google.com>
3712 * loop-unroll.c (report_unroll_peel): Minor message change.
3713 * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
3714 Emit alignment peeling message with default -fopt-info.
3715 (vect_loop_versioning): Emit loop version info message.
3716 * tree-vectorizer.c (vectorize_loops): Minor message change.
3717 (execute_vect_slp): Ditto.
3719 2013-08-29 Eric Botcazou <ebotcazou@adacore.com>
3721 * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
3722 of the clone from the DECL_NAME of the original function.
3724 2013-08-29 Oleg Endo <olegendo@gcc.gnu.org>
3726 * passes.c (register_pass): Add overload.
3727 * tree-pass.h (register_pass): Forward declare it. Add comment.
3729 2013-08-29 Jan Hubicka <jh@suse.cz>
3731 * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
3732 DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
3733 * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
3734 DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3735 (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
3736 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
3737 Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3738 (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
3740 2013-08-29 Teresa Johnson <tejohnson@google.com>
3742 * dumpfile.c (dump_loc): Output column number.
3743 * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
3744 * doc/invoke.texi: Document optall -fopt-info flag.
3745 * profile.c (read_profile_edge_counts): Use new dump framework.
3746 (compute_branch_probabilities): Ditto.
3747 * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
3748 when pass not in any opt group.
3749 * pass_manager.h (pass_manager::get_pass_profile): New method.
3750 * value-prof.c (check_counter): Use new dump framework.
3751 (check_ic_target): Ditto.
3752 * coverage.c (get_coverage_counts): Ditto.
3753 (coverage_init): Setup new dump framework.
3755 2013-08-29 Richard Biener <rguenther@suse.de>
3757 PR tree-optimization/58246
3758 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
3759 handle the dominance check inside a basic-block.
3761 2013-08-29 Richard Biener <rguenther@suse.de>
3764 * tree-ssa-copy.c (may_propagate_copy): Allow propagating
3765 of default defs that appear in abnormal PHI nodes.
3767 2013-08-29 Richard Biener <rguenther@suse.de>
3769 PR tree-optimization/57685
3770 * tree-vrp.c (register_edge_assert_for_1): Recurse only for
3771 single-use operands to avoid exponential complexity.
3773 2013-08-28 Dehao Chen <dehao@google.com>
3775 * ipa-inline.c (edge_badness): Fix integer underflow.
3777 2013-08-28 Uros Bizjak <ubizjak@gmail.com>
3779 * gtm-builtins.def (_ITM_free): Declare leaf.
3781 2013-08-28 Jakub Jelinek <jakub@redhat.com>
3784 * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
3785 (*tls_local_dynamic_base_64_largepic): Likewise.
3786 (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
3787 Remove predicate from call operand.
3788 * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
3789 return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
3791 2013-08-28 Jeff Law <law@redhat.com>
3793 * tree-ssa-threadedge.c (thread_around_empty_block): Remove
3794 checks for the number of predecessors and successors allowed.
3795 * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
3796 which require copying a joiner block if there is a request which
3797 is a subpath that requires no joiner block copying.
3799 2013-08-28 Jan Hubicka <jh@suse.cz>
3801 * lto-streamer-out.c (DFS_write_tree_body): Drop
3802 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
3803 (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
3804 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
3805 * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
3806 Do not read DECL_ERROR_ISSUED.
3807 (unpack_ts_decl_with_vis_value_fields): Do not read
3809 (lto_input_ts_binfo_tree_pointers): Do not read
3810 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
3811 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
3812 write DECL_ERROR_ISSUED..
3813 (pack_ts_decl_with_vis_value_fields): Do not write
3815 (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
3816 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
3817 * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
3818 * tree.h (tree_decl_common): Update comment.
3819 (DECL_ERROR_ISSUED): Remove.
3821 2013-08-28 Jakub Jelinek <jakub@redhat.com>
3824 * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
3826 2013-08-28 Jan Hubicka <jh@suse.cz>
3828 * builtins.def (free): Declare leaf.
3830 2013-08-27 David Malcolm <dmalcolm@redhat.com>
3833 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
3834 * configure: Regenerate.
3836 2013-08-27 Martin Jambor <mjambor@suse.cz>
3838 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
3839 (ipa_ancestor_jf_data): Likewise.
3840 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
3841 (ipa_get_jf_pass_through_type_preserved): New function.
3842 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
3843 (ipa_get_jf_ancestor_type_preserved): New function.
3844 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
3845 (ipa_get_jf_ancestor_result): Likewise.
3846 (propagate_vals_accross_pass_through): Use
3847 ipa_get_jf_pass_through_result to do all the value mappings.
3848 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
3849 type_preserved flag.
3850 (ipa_set_jf_cst_copy): New function.
3851 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
3852 (ipa_set_jf_arith_pass_through): Likewise.
3853 (ipa_set_ancestor_jf): Likewise.
3854 (compute_complex_assign_jump_func): Set type_preserved instead of
3856 (ipa_compute_jump_functions_for_edge): Likewise.
3857 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
3858 (update_jump_functions_after_inlining): Update type_preserved.
3859 Explicitely create jump functions when combining one with pass_through.
3860 (ipa_write_jump_function): Stream the type_preserved flags.
3861 (ipa_read_jump_function): Likewise.
3863 2013-08-27 Jakub Jelinek <jakub@redhat.com>
3864 Aldy Hernandez <aldyh@redhat.com>
3866 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
3867 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
3868 * function.h (struct function): Add has_force_vect_loops and
3870 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
3871 * gimple.c (gimple_build_omp_critical): Add KIND argument and
3873 * gimple.def: Update CLAUSES comments.
3874 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
3875 (gimple_build_omp_for): Add argument to prototype.
3876 (gimple_omp_for_kind): New.
3877 (gimple_omp_for_set_kind): New.
3878 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
3879 GOVD_DATA_SHARE_CLASS.
3880 (enum omp_region_type): Add ORT_SIMD.
3881 (gimple_add_tmp_var): Handle ORT_SIMD.
3882 (gimplify_var_or_parm_decl): Same.
3883 (is_gimple_stmt): Same.
3884 (omp_firstprivatize_variable): Same.
3885 (omp_add_variable): Only use splay_tree_insert if lookup failed.
3886 (omp_notice_variable): Handle ORT_SIMD.
3887 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
3888 (omp_check_private): Handle ORT_SIMD.
3889 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3891 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
3892 Handle OMP_CLAUSE_LASTPRIVATE.
3893 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3895 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
3896 (gimplify_expr): Handle OMP_SIMD.
3897 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
3898 (expand_GOMP_SIMD_VF): New.
3899 (expand_GOMP_SIMD_LAST_LANE): New.
3900 * internal-fn.def (GOMP_SIMD_LANE): New.
3901 (GOMP_SIMD_VF): New.
3902 (GOMP_SIMD_LAST_LANE): New.
3903 * omp-low.c: Include target.h.
3904 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
3906 (check_omp_nesting_restrictions): Same.
3908 (lower_rec_simd_input_clauses): New.
3909 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
3911 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
3912 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
3913 (expand_omp_build_assign): New.
3914 (expand_omp_for_init_counts): New.
3915 (expand_omp_for_init_vars): New.
3916 (extract_omp_for_update_vars): New.
3917 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
3918 and rewrite accordingly.
3919 (expand_omp_simd): New.
3920 (expand_omp_for): Use expand_omp_simd.
3921 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
3922 (lower_omp_for): Do not lower the body.
3923 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
3925 * tree-flow.h (find_omp_clause): Remove prototype.
3926 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
3927 forcing vectorization of the loop, or if flag_tree_vectorize.
3928 (gate_tree_if_conversion): Similarly.
3929 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
3930 gimple_build_omp_for.
3931 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
3932 * tree-parloops (create_parallel_loop): Pass kind argument to
3933 gimple_build_omp_for.
3934 * tree-pretty-print.c (dump_omp_clause): Add cases for
3935 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
3936 OMP_CLAUSE__SIMDUID_.
3937 (dump_generic_node): Handle OMP_SIMD.
3938 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
3939 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
3940 unroll OMP_SIMD loops here.
3941 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
3942 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
3944 (vect_analyze_data_refs): Handle simd loops.
3945 * tree-vect-loop.c (vectorizable_live_operation): Handle
3947 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
3948 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
3949 (vectorizable_load): Same.
3950 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
3951 (struct simduid_to_vf): New.
3952 (simduid_to_vf::hash): New.
3953 (simduid_to-vf::equal): New.
3954 (struct simd_array_to_simduid): New.
3955 (simd_array_to_simduid::hash): New.
3956 (simd_array_to_simduid::equal): New.
3957 (adjust_simduid_builtins): New.
3958 (struct note_simd_array_uses_struct): New.
3959 (note_simd_array_uses_cb): New.
3960 (note_simd_array_uses): New.
3961 (vectorize_loops): Handle simd hints and adjust simd builtins
3963 * tree-vectorizer.h (struct _stmt_vec_info): Add
3964 simd_lane_access_p field.
3965 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
3966 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
3967 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
3968 (omp_clause_code_name): Same.
3969 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
3970 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
3971 * tree.def (OMP_SIMD): New entry.
3972 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
3973 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
3974 (OMP_CLAUSE_DECL): Adjust range for new clauses.
3975 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
3976 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
3977 (OMP_CLAUSE_LINEAR_STEP): New.
3978 (OMP_CLAUSE_SAFELEN_EXPR): New.
3979 (OMP_CLAUSE__SIMDUID__DECL): New.
3980 (find_omp_clause): New prototype.
3982 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
3984 * config/i386/driver-i386.c (host_detect_local_cpu): Update
3985 Haswell processor detection.
3987 2013-08-27 Christian Widmer <shadow@umbrox.de>
3990 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
3991 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
3992 AVX2 capable processors.
3994 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
3996 * config/aarch64/arm_neon.h: Replace all inline asm implementations
3997 of vget_low_* with implementations in terms of other intrinsics.
3999 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
4002 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
4003 values to -1, 0 and 1.
4005 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
4007 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
4008 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
4009 (<optab><mode>3_insn): Remove.
4010 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
4011 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
4012 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
4013 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
4014 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
4015 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
4016 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
4017 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
4018 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
4019 (ror<mode>3_insn): Likewise.
4020 * config/aarch64/predicates.md (aarch64_simd_register): New.
4022 2013-08-27 Richard Biener <rguenther@suse.de>
4024 PR tree-optimization/57521
4025 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
4026 one edge is non-critical.
4027 (find_phi_replacement_condition): Make sure to use a non-critical
4028 edge. Cleanup and remove old bug workarounds.
4029 (bb_postdominates_preds): Remove.
4030 (if_convertible_loop_p_1): Do not compute post-dominators.
4031 (combine_blocks): Do not free post-dominators.
4032 (main_tree_if_conversion): Likewise.
4033 (pass_data_if_conversion): Add TODO_verify_ssa.
4035 2013-08-27 DJ Delorie <dj@redhat.com>
4037 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
4039 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
4041 * function.c (assign_parm_find_data_types): Set passed_mode and
4042 nominal_mode to the TYPE_MODE of nominal_type for the built
4043 pointer type in case of the struct-pass-by-reference.
4045 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
4047 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
4048 (UINT16_TYPE): Change default to "unsigned int".
4050 * config/avr/avr.opt (mfract-convert-truncate): New option.
4051 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
4052 is set, round negative fractional integers according to n1169
4053 when converting to integer types.
4055 2013-08-26 Jan Hubicka <jh@suse.cz>
4057 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
4058 methods can not be called indirectly when their address is not taken.
4060 2013-08-26 Jan Hubicka <jh@suse.cz>
4062 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
4065 2013-08-26 Jan Hubicka <jh@suse.cz>
4067 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
4070 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
4072 * reload.c (find_valid_class): Allow classes that do not include
4073 FIRST_PSEUDO_REGISTER - 1.
4075 2013-08-26 Jan Hubicka <jh@suse.cz>
4077 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
4078 fix edge count/frequency when speculation failed; fix type check
4079 for the direct call.
4081 2013-08-26 Jan Hubicka <jh@suse.cz>
4083 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
4085 2013-08-26 Jan Hubicka <jh@suse.cz>
4087 * ipa-inline-transform.c (inline_transform): Be ready for basic block
4088 to be changed by edge redirection.
4090 2013-08-26 Jan Hubicka <jh@suse.cz>
4092 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
4093 formating; add sanity check.
4094 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
4095 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
4096 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
4097 (resolve_noninline_speculation): Update callee keys, too.
4099 2013-08-26 Jan Hubicka <jh@suse.cz>
4101 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
4102 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
4104 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
4106 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
4109 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
4111 * config/i386/i386.c (ix86_debug_options): Remove prototype.
4112 (x86_64_elf_select_section): Ditto.
4113 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
4115 (ix86_pass_by_reference): Ditto.
4116 (output_set_got): Ditto.
4117 (ix86_unary_operator_ok): Ditto.
4118 (ix86_expand_builtin): Ditto.
4120 2013-08-23 Jan Hubicka <jh@suse.cz>
4122 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
4124 2013-08-23 Jan Hubicka <jh@suse.cz>
4126 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
4127 (tree_decl_with_vis): Add FINAL field.
4129 2013-08-23 Jeff Law <law@redhat.com>
4131 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
4132 the debugging dump when the expression is fully redundant.
4134 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
4136 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
4137 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
4138 * pretty-print.c (pp_formatted_text_data): Likewise.
4139 (pp_write_text_to_stream): Likewise.
4140 (pp_write_text_as_dot_label_to_stream): Likewise.
4141 (pp_append_r): Likewise.
4142 (pp_format): Likewise.
4143 (pp_flush): Likewise.
4144 (pp_clear_output_area): Likewise.
4145 (pp_append_text): Likewise.
4146 (pp_formatted_text): Likewise.
4147 (pp_remaining_character_count_for_line): Likewise.
4148 (pp_newline): Likewise.
4149 (pp_character): Likewise.
4150 (output_buffer::~output_buffer): Define.
4151 (pretty_printer::~pretty_printer): Destruct output buffer.
4152 * pretty-print.h (output_buffer::~output_buffer): Declare.
4153 (pretty_printer::~pretty_printer): Declare virtual.
4155 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
4158 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
4159 HOST_WIDE_INT_M1U): New macros.
4160 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
4161 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
4162 unsigned -1 for lshift.
4163 * cse.c (cse_insn): Likewise.
4164 * double-int.c (rshift_double, lshift_double): Likewise.
4165 * builtins.c (fold_builtin_bitop): Likewise.
4166 * combine.c (force_to_mode): Likewise.
4167 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
4168 * simplify-rtx.c (simplify_const_unary_operation,
4169 simplify_const_binary_operation): Likewise.
4170 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
4171 check_va_list_escapes): Likewise.
4172 * rtlanal.c (nonzero_bits1): Likewise.
4173 * expmed.c (expand_smod_pow2): Likewise.
4174 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
4176 2013-08-23 Jan Hubicka <jh@suse.cz>
4178 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
4179 as having address taken.
4181 2013-08-23 Jan Hubicka <jh@suse.cz>
4183 * ipa-utils.h (method_class_type): Declare.
4184 * ipa-devirt.c (method_class_type): Export.
4186 * cgraphunit.c (analyze_functions): Do basic devirtualization;
4187 do not walk base classes of anonymous types.
4189 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
4191 PR rtl-optimization/58220
4193 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
4194 handle SEQUENCE insns properly.
4196 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
4198 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
4200 (pp_newline_and_indent): Likewise.
4201 (pp_separate_with): Likewise.
4202 * pretty-print.c (pp_newline_and_flush): Define.
4203 (pp_newline_and_indent): Likewise.
4204 (pp_separate_with): Likewise.
4206 2013-08-23 Jakub Jelinek <jakub@redhat.com>
4209 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
4210 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
4212 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
4214 * config/i386/predicates.md (ext_sse_reg_operand): New.
4215 * config/i386/i386.md (*movti_internal): Use
4216 predicate to determine if EVEX is needed.
4217 (*movsi_internal): Ditto.
4218 (*movdf_internal): Ditto.
4219 (*movsf_internal): Ditto.
4220 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
4222 2013-08-23 Jakub Jelinek <jakub@redhat.com>
4224 PR tree-optimization/58209
4225 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
4226 (find_tail_calls): Give up for pointer result types if m is non-NULL.
4227 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
4228 emit POINTER_PLUS_EXPR.
4229 (create_tailcall_accumulator): For pointer result type accumulate in
4232 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
4234 * configure.ac: Add backslashes missing from the last change.
4235 * configure: Regenerate.
4237 2013-08-22 Jan Hubicka <jh@susue.cz>
4239 * ipa.c (function_and_variable_visibility): First remember function
4240 was global and then make it local.
4242 2013-08-22 Julian Brown <julian@codesourcery.com>
4244 * configure.ac: Add aarch64 to list of arches which use "nop" in
4246 * configure: Regenerate.
4248 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4250 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
4251 gnu_libc_has_function.
4252 * config/s390/tpf.h: Likewise.
4254 2013-08-22 Jan Hubicka <jh@susue.cz>
4256 * timevar.c (validate_phases): Add cast.
4258 2013-08-22 Jan Hubicka <jh@susue.cz>
4260 * timevar.c (validate_phases): Use size_t for memory.
4261 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
4263 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
4265 * pretty-print.h (output_buffer::output_buffer): Declare.
4266 (pretty_printer::pretty_printer): Likewise.
4267 (pp_construct): Remove.
4268 * pretty-print.c (output_buffer::output_buffer): Define.
4269 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
4270 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
4271 (print_gimple_expr): Likewise.
4272 (print_gimple_seq): Likewise.
4273 (gimple_dump_bb): Likewise.
4274 * sched-vis.c (dump_value_slim): Likewise.
4275 (dump_insn_slim): Likewise.
4276 (dump_rtl_slim): Likewise.
4277 (str_pattern_slim): Likewise.
4278 * tree-mudflap.c (mf_varname_tree): Likewise.
4279 * graph.c (print_graph_cfg): Likewise.
4280 (start_graph_dump): Likewise.
4281 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
4283 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
4284 pretty printer initialization.
4285 * coretypes.h (diagnostic_context): Remove superflous type alias
4287 (pretty_printer): Likewise. Declare directly as a class.
4288 (pretty_print_info): Remove declaration as class.
4289 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
4290 and pp_clear_output_area.
4291 (asan_add_global): Likewise.
4293 2013-08-22 Jan Hubicka <jh@suse.cz>
4295 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
4296 * ipa-utils.h (update_type_inheritance_graph): Declare.
4297 (possible_polymorphic_call_target_p): Declare.
4298 (possible_polymorphic_call_target_p): New.
4299 * ipa-devirt.c: Update toplevel comments.
4300 (cached_polymorphic_call_targets): Move up.
4301 (odr_type_d): Move ID down.
4302 (polymorphic_type_binfo_p): Update comment.
4303 (odr_hasher::remove): Likewise;
4304 (get_odr_type): Set anonymous_namespace.
4305 (dump_odr_type): Dump it.
4306 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
4307 (maybe_record_node): Record node in cached_polymorphic_call_targets.
4308 (record_binfo): Add comment.
4309 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
4311 (devirt_node_removal_hook): Do not iCE when cache is freed.
4312 (possible_polymorphic_call_target_p): New predicate.
4313 (update_type_inheritance_graph): New function.
4315 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
4316 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
4317 Sergey Lega <sergey.s.lega@intel.com>
4318 Anna Tikhonova <anna.tikhonova@intel.com>
4319 Ilya Tocar <ilya.tocar@intel.com>
4320 Andrey Turetskiy <andrey.turetskiy@intel.com>
4321 Ilya Verbin <ilya.verbin@intel.com>
4322 Kirill Yukhin <kirill.yukhin@intel.com>
4323 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
4325 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
4326 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
4327 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
4328 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
4329 (OPTION_MASK_ISA_AVX2_UNSET): Update.
4330 (OPTION_MASK_ISA_AVX512F_UNSET): New.
4331 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
4332 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
4333 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
4334 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
4335 OPT_mavx512pf, OPT_mavx512er cases.
4336 * config/i386/constraints.md (v): New constraint.
4337 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
4338 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
4339 (bit_AVX512CD): New.
4340 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4341 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
4342 * config/i386/i386-c.c (ix86_target_macros_internal):
4343 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
4345 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
4346 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
4347 * config/i386/i386.c (regclass_map, dbx_register_map)
4348 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
4349 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
4350 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
4352 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
4353 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
4354 -mavx512pf options. Fix formatting.
4355 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
4356 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
4357 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
4358 -mavx512cd, -mavx512pf options.
4359 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
4360 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
4361 (ix86_preferred_output_reload_class): Replace SSE_REGS with
4363 (ix86_hard_regno_mode_ok): Support 512-bit registers.
4364 (ix86_set_reg_reg_cost): Ditto.
4365 (x86_order_regs_for_local_alloc): Ditto.
4366 (MAX_VECT_LEN): Extend to 64-byte.
4367 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
4368 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
4369 (TARGET_AVX512ER, TARGET_AVX512CD): New.
4370 (BIGGEST_ALIGNMENT): Extend to 512-bits.
4371 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
4372 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
4373 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
4374 (SSE_REG_MODE_P): Support new modes.
4375 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
4376 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
4377 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
4378 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
4379 (REG_CLASS_CONTENTS): Add new registers.
4380 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
4381 (EXT_REX_SSE_REGNO_P): New.
4382 (HI_REGISTER_NAMES): Add new registers.
4383 * config/i386/i386.md: Define constants for new registers.
4384 (mode): Add new 512-bit modes.
4385 (prefix): Support evex prefix.
4386 (isa): Support avx512f, noavx512f, fma_avx512f.
4387 (ssemodesuffix): Add new 512-bit modes.
4389 (*movxi_internal_avx512f): Ditto.
4390 (*movdi_internal): Replace constraint "x" with the new constraint "v".
4392 (*movsi_internal): Likewise.
4393 (*movdf_internal): Likewise.
4394 (*movsf_internal): Likewise.
4395 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
4396 (<code><mode>3): Likewise.
4397 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
4399 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
4400 with the new constraint "v".
4401 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
4403 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
4404 with the new constraint "v".
4405 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
4406 (<sse2>_storedqu<avxsizesuffix>): Likewise.
4407 (*<plusminus_insn><mode>3): Likewise.
4408 (<sse>_vm<plusminus_insn><mode>3): Likewise.
4409 (*mul<mode>3): Likewise.
4410 (<sse>_vmmul<mode>3): Likewise.
4411 (<sse>_div<mode>3): Likewise.
4412 (<sse>_vmdiv<mode>3): Likewise.
4413 (<sse>_sqrt<mode>2): Likewise.
4414 (<sse>_vmsqrt<mode>2): Likewise.
4415 (*<code><mode>3_finite): Likewise.
4416 (*<code><mode>3) <smaxmin>: Likewise.
4417 (<sse>_vm<code><mode>3): Likewise.
4418 (*<code><mode>3) <any_logic>: Likewise.
4419 (*fma_fmadd_<mode>): Likewise.
4420 (*fma_fmsub_<mode>): Likewise.
4421 (*fma_fnmadd_<mode>): Likewise.
4422 (*fma_fnmsub_<mode>): Likewise.
4423 (*fma_fmaddsub_<mode>): Likewise.
4424 (*fma_fmsubadd_<mode>): Likewise.
4425 (*fmai_fmadd_<mode>): Likewise.
4426 (*fmai_fmsub_<mode>): Likewise.
4427 (*fmai_fnmadd_<mode>): Likewise.
4428 (*fmai_fnmsub_<mode>): Likewise.
4429 (sse_cvtsi2ss): Likewise.
4430 (sse_cvtsi2ssq): Likewise.
4431 (sse_cvtss2si): Likewise.
4432 (sse_cvtss2si_2): Likewise.
4433 (sse_cvtss2siq): Likewise.
4434 (sse_cvtss2siq_2): Likewise.
4435 (sse_cvttss2si): Likewise.
4436 (sse_cvtss2siq_2): Likewise.
4437 (float<sseintvecmodelower><mode>2): Likewise.
4438 (sse2_cvtsd2si_2): Likewise.
4439 (sse2_cvtsd2siq_2): Likewise.
4440 (*<plusminus_insn><mode>3): Likewise.
4441 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
4442 (*<sse4_1_avx2>_mul<mode>3): Likewise.
4443 (ashr<mode>3): Likewise.
4444 (<shift_insn><mode>3): Likewise.
4445 (avx2_<code><mode>3): Likewise.
4446 (*avx2_<code><mode>3): Likewise.
4447 (*andnot<mode>3): Likewise.
4448 (*<code><mode>3) <any_logic>: Likewise.
4449 (abs<mode>2): Likewise.
4450 (avx2_permvar<mode>): Likewise.
4451 (avx2_perm<mode>_1): Likewise.
4452 (*avx_vpermilp<mode>): Likewise.
4453 (avx_vpermilvar<mode>3): Likewise.
4454 (avx2_ashrv<mode>): Likewise.
4455 (avx2_<shift_insn>v<mode>): Likewise.
4456 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
4458 * doc/rtl.texi: Document XImode.
4460 2013-08-21 Jeff Law <law@redhat.com>
4462 * tree-flow.h (register_jump_thread): Pass vector of edges
4463 instead of each important edge.
4464 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
4465 thread path into a vector and pass that to register_jump_thread.
4466 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
4467 passed in edge vector to the current 3-edge form.
4470 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
4472 * dce.c (fini_dce): Call df_analyze again just in case
4473 delete_unmarked_insns removed anything.
4475 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
4477 * reload.h (struct reg_equivs): Rename to ..
4478 (struct reg_equivs_s): .. this.
4480 2013-08-20 Martin Liska <marxin.liska@gmail.com>
4482 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
4484 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4486 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
4488 2013-08-21 Jeff Law <law@redhat.com>
4490 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
4491 simplify assignments too. If the RHS collapses to a singleton
4492 range, then return the value for the range.
4494 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
4496 * config/i386/sse.md (V16): Rename to...
4498 (mov<mode>): Update iterator name.
4499 (*mov<mode>_internal): Ditto.
4500 (push<mode>1): Ditto.
4501 (movmisalign<mode>): Ditto.
4503 2013-08-20 Jan Hubicka <jh@suse.cz>
4506 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
4507 entry for direct edges.
4508 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
4510 2013-08-20 David Malcolm <dmalcolm@redhat.com>
4512 Revert my last two changes, r201865 and r201864:
4515 2013-08-20 David Malcolm <dmalcolm@redhat.com>
4517 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4518 instances can own GC refs.
4520 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4521 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4522 (gcc::context::gt_pch_nx): Likewise.
4523 (gcc::context::gt_pch_nx): Likewise.
4524 * ggc.h (gt_ggc_mx <T>): New.
4525 (gt_pch_nx_with_op <T>): New.
4526 (gt_pch_nx <T>): New.
4527 * passes.c (opt_pass::gt_ggc_mx): New.
4528 (opt_pass::gt_pch_nx): New.
4529 (opt_pass::gt_pch_nx_with_op): New.
4530 (pass_manager::gt_ggc_mx): New.
4531 (pass_manager::gt_pch_nx): New.
4532 (pass_manager::gt_pch_nx_with_op): New.
4533 (pass_manager::operator new): Use
4534 ggc_internal_cleared_alloc_stat rather than xcalloc.
4535 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4536 (pass_manager::gt_ggc_mx): New.
4537 (pass_manager::gt_pch_nx): New.
4538 (pass_manager::gt_pch_nx_with_op): New.
4539 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4540 (opt_pass::operator new): New.
4541 (opt_pass::gt_ggc_mx): New.
4542 (opt_pass::gt_pch_nx): New.
4543 (opt_pass::gt_pch_nx_with_op): New.
4546 2013-08-20 David Malcolm <dmalcolm@redhat.com>
4548 * Makefile.in (GTFILES): Add context.h.
4549 * context.c (gcc::context::operator new): New.
4550 (gcc::context::gt_ggc_mx): New.
4551 (gcc::context::gt_pch_nx): New.
4552 (gcc::context::gt_pch_nx): New.
4553 * context.h (gcc::context): Add GTY((user)) marking.
4554 (gcc::context::operator new): New.
4555 (gcc::context::gt_ggc_mx): New.
4556 (gcc::context::gt_pch_nx): New.
4557 (gcc::context::gt_pch_nx): New.
4558 (g): Add GTY marking.
4559 (gt_ggc_mx (gcc::context *)): New.
4560 (gt_pch_nx (gcc::context *)): New.
4561 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4562 void *cookie)): New.
4563 * gengtype.c (open_base_files) <ifiles>: Add context.h.
4565 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
4567 * dce.c (fini_dce): Call df_analyze again just in case
4568 delete_unmarked_insns removed anything.
4570 2013-08-20 Teresa Johnson <tejohnson@google.com>
4572 PR rtl-optimizations/57451
4573 * final.c (reemit_insn_block_notes): Prevent lexical blocks
4574 from crossing split section boundaries.
4576 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4578 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
4579 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
4580 with MULTLIB_DEFAULTS.
4582 2013-08-20 Nick Clifton <nickc@redhat.com>
4584 * target.def (narrow_volatile_bitfield): Note that the default
4585 value is false, not !TARGET_STRICT_ALIGN.
4586 * doc/tm.texi: Regenerate.
4588 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
4590 Fix LIB_SPEC for systems without libpthread.
4592 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
4593 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
4595 * config/i386/linux-common.h: Likewise.
4596 * config/mips/linux-common.h: Likewise.
4598 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
4600 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
4603 2013-08-20 David Malcolm <dmalcolm@redhat.com>
4605 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4606 instances can own GC refs.
4608 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4609 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4610 (gcc::context::gt_pch_nx): Likewise.
4611 (gcc::context::gt_pch_nx): Likewise.
4612 * ggc.h (gt_ggc_mx <T>): New.
4613 (gt_pch_nx_with_op <T>): New.
4614 (gt_pch_nx <T>): New.
4615 * passes.c (opt_pass::gt_ggc_mx): New.
4616 (opt_pass::gt_pch_nx): New.
4617 (opt_pass::gt_pch_nx_with_op): New.
4618 (pass_manager::gt_ggc_mx): New.
4619 (pass_manager::gt_pch_nx): New.
4620 (pass_manager::gt_pch_nx_with_op): New.
4621 (pass_manager::operator new): Use
4622 ggc_internal_cleared_alloc_stat rather than xcalloc.
4623 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4624 (pass_manager::gt_ggc_mx): New.
4625 (pass_manager::gt_pch_nx): New.
4626 (pass_manager::gt_pch_nx_with_op): New.
4627 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4628 (opt_pass::operator new): New.
4629 (opt_pass::gt_ggc_mx): New.
4630 (opt_pass::gt_pch_nx): New.
4631 (opt_pass::gt_pch_nx_with_op): New.
4633 2013-08-20 David Malcolm <dmalcolm@redhat.com>
4635 * Makefile.in (GTFILES): Add context.h.
4636 * context.c (gcc::context::operator new): New.
4637 (gcc::context::gt_ggc_mx): New.
4638 (gcc::context::gt_pch_nx): New.
4639 (gcc::context::gt_pch_nx): New.
4640 * context.h (gcc::context): Add GTY((user)) marking.
4641 (gcc::context::operator new): New.
4642 (gcc::context::gt_ggc_mx): New.
4643 (gcc::context::gt_pch_nx): New.
4644 (gcc::context::gt_pch_nx): New.
4645 (g): Add GTY marking.
4646 (gt_ggc_mx (gcc::context *)): New.
4647 (gt_pch_nx (gcc::context *)): New.
4648 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4649 void *cookie)): New.
4650 * gengtype.c (open_base_files) <ifiles>: Add context.h.
4652 2013-08-20 Alan Modra <amodra@gmail.com>
4655 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
4656 (rs6000_emit_epilogue): Likewise.
4658 2013-08-19 Dehao Chen <dehao@google.com>
4660 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
4662 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
4663 Jakub Jelinek <jakub@redhat.com>
4665 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
4666 (BUILT_IN_FABSD64): Likewise.
4667 (BUILT_IN_FABSD128): Likewise.
4668 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
4669 (fold_builtin_1): Likewise.
4670 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
4671 destination and source operands.
4672 (*abstd2_fpr): Likewise.
4673 (*nabstd2_fpr): Likewise.
4675 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
4677 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
4680 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
4682 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
4685 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
4687 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
4688 * builtins.c (default_libc_has_function): New.
4689 (gnu_libc_has_function): Ditto.
4690 (no_c99_libc_has_function): Ditto.
4691 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
4692 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
4693 (fold_builtin_sincos): Likewise.
4694 (fold_builtin_cexp): Likewise.
4695 * builtins.def (DEF_C94_BUILTIN): Likewise.
4696 (DEF_C99_BUILTIN): Likewise.
4697 (DEF_C99_C90RES_BUILTIN): Likewise.
4698 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
4699 definitions to using this define.
4700 * config/darwin-protos.h (darwin_libc_has_function): New.
4701 * config/darwin.c (darwin_libc_has_function): Ditto.
4702 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
4703 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
4704 * config/darwin.h: Ditto.
4705 * config/elfos.h: Ditto.
4706 * config/freebsd.h: Ditto.
4707 * config/i386/cygming.h: Ditto.
4708 * config/i386/djgpp.h: Ditto.
4709 * config/i386/i386-interix.h: Ditto.
4710 * config/microblaze/microblaze.h: Ditto.
4711 * config/mmix/mmix.h: Ditto.
4712 * config/gnu-user.h: Ditto.
4713 * config/ia64/hpux.h: Ditto.
4714 * config/pa/pa-hpux.h: Ditto.
4715 * config/pdp11/pdp11.h: Ditto.
4716 * config/picochip/picochip.h: Ditto.
4717 * config/linux.h: Ditto.
4718 * config/netbsd.h: Ditto.
4719 * config/openbsd.h: Ditto.
4720 * config/rs6000/aix43.h: Ditto.
4721 * config/rs6000/aix51.h: Ditto.
4722 * config/rs6000/aix52.h: Ditto.
4723 * config/rs6000/aix53.h: Ditto.
4724 * config/rs6000/aix61.h: Ditto.
4725 * config/rs6000/darwin.h: Ditto.
4726 * config/rs6000/linux.h: Ditto.
4727 * config/rs6000/linux64.h: Ditto.
4728 * config/s390/tpf.h: Ditto.
4729 * config/sol2-10.h: Ditto.
4730 * config/sol2.h: Ditto.
4731 * config/vms/vms.h: Ditto.
4732 * config/vxworks.h: Ditto.
4733 * config/linux-android.c (linux_android_libc_has_function):
4734 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
4735 * config/linux-protos.h (linux_android_libc_has_function):
4737 * config/i386/i386.c (ix86_libc_has_function): New.
4738 * config/i386/i386-protos.h
4739 (ix86_libc_has_function): New declaration.
4740 * config/i386/i386.md
4741 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
4742 ("isinf<mode>2): Likewise.
4743 * convert.c (convert_to_integer): Using new target hook
4744 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
4745 TARGET_C99_FUNCTIONS.
4746 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
4747 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
4748 * coretypes.h (function_class): New enum for different
4749 classes of functions.
4750 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
4751 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
4752 (TARGET_HAS_SINCOS): Likewise.
4753 (TARGET_LIBC_HAS_FUNCTION): New.
4754 * doc/tm.texi: Regenerated.
4755 * targhooks.h (default_libc_has_function): New declaration.
4756 (no_c99_libc_has_function): Ditto.
4757 (gnu_libc_has_function): Ditto.
4758 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
4759 and TARGET_HAS_SINCOS.
4761 2013-08-18 Jan Hubicka <jh@suse.cz>
4763 * Makeifle-in (ipa-devirt.o): New.
4764 (GTFILES): Add ipa-utils.h and ipa-devirt.c
4765 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
4766 (analyze_functions): Look into possible targets of polymorphic call.
4767 * dumpfile.c (dump_files): Add type-inheritance dump.
4768 * dumpfile.h (TDI_inheritance): New.
4769 * ipa-devirt.c: New file.
4770 * ipa-utils.h (odr_type_d): Forward declare.
4771 (odr_type): New type.
4772 (build_type_inheritance_graph): Declare.
4773 (possible_polymorphic_call_targets): Declare and introduce inline
4774 variant when only edge is pased.
4775 (dump_possible_polymorphic_call_targets): Likewise.
4776 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
4777 * tree.c (type_in_anonymous_namespace_p): Break out from ...
4778 (types_same_for_odr): ... here.
4779 * tree.h (type_in_anonymous_namespace_p): Declare.
4781 2013-08-18 Jakub Jelinek <jakub@redhat.com>
4783 PR tree-optimization/58006
4784 * tree-parloops.c (take_address_of): Don't ICE if get_name
4786 (eliminate_local_variables_stmt): Remove clobber stmts.
4788 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
4790 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
4791 error message is issued for an alias to undefined symbol.
4793 2013-08-18 Jan Hubicka <jh@suse.cz>
4795 * cgraph.c (cgraph_create_indirect_edge): Discover
4796 polymorphic calls and record basic info into indirect_info.
4797 * gimple-fold.c (gimple_fold_call): When doing BINFO based
4798 devirtualization, ignore objc function calls.
4799 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
4800 call with no parm index info.
4801 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
4802 * tree.c (virtual_method_call_p): New function.
4803 * tree.h (virtual_method_call_p): Declare.
4805 2013-08-16 Jan Hubicka <jh@suse.cz>
4808 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
4810 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
4812 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
4815 2013-08-16 David Malcolm <dmalcolm@redhat.com>
4817 * gengtype.c (type_for_name): Add special-case support for
4818 locating types within the "gcc::" namespace.
4819 (open_base_files): Emit a "using namespace gcc" directive.
4821 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4824 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
4825 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
4827 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
4828 array instead of each individual operand as a separate argument.
4829 (emit_fusion_gpr_load): Likewise.
4830 (expand_fusion_gpr_load): Add new function declaration.
4832 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
4833 signature to have the operands passed as an array, instead of as
4834 separate arguments. Allow ZERO_EXTEND to be in the memory
4835 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
4836 depend on the register live/dead flags when peepholes are run.
4837 (expand_fusion_gpr_load): New function to be called from the
4838 peephole2 pass, to change the register that addis sets to be the
4840 (emit_fusion_gpr_load): Change the calling signature to have the
4841 operands passed as an array, instead of as separate arguments.
4842 Allow ZERO_EXTEND to be in the memory address, and also
4843 SIGN_EXTEND if -mpower8-fusion-sign.
4845 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
4847 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
4848 adjust the register addis loads up in the peephole2 pass. Do not
4849 depend on the register live/dead state when the peephole pass is done.
4851 2013-08-16 David Malcolm <dmalcolm@redhat.com>
4853 * gengtype.c (create_user_defined_type): Ensure that the kind
4854 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
4855 declaration is seen before the GTY((user)) marking.
4857 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
4860 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
4862 2013-08-16 Jan Hubicka <jh@suse.cz>
4864 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
4865 arugment expected_type.
4866 (gimple_fold_call): Use it.
4867 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
4868 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
4869 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
4870 (try_make_edge_direct_virtual_call): Likewise.
4871 * tree.c (obj_type_ref_class): New.
4872 * tree.h (obj_type_ref_class): Use it.
4874 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
4876 * sched-vis.c (rtl_slim_pp_initialized): Remove.
4877 (rtl_slim_pp): Likewise.
4878 (init_rtl_slim_pretty_print): Likewise.
4879 (dump_value_slim): Don't call it. Use local pretty printer.
4880 (dump_insn_slim): Likewise.
4881 (dump_rtl_slim): Likewise.
4882 (str_pattern_slim): Likewise.
4883 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
4886 2013-08-16 Jakub Jelinek <jakub@redhat.com>
4888 PR tree-optimization/58164
4889 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
4890 walk gimple_goto_dest of GIMPLE_GOTO.
4892 PR tree-optimization/58165
4893 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
4894 bi_call must be the last stmt in a bb, don't split_block, instead
4895 use fallthru edge from it and give up if there is none.
4896 Release conds vector when returning early.
4898 2013-08-14 Xinliang David Li <davidxl@google.com>
4900 * config/i386/i386.c (ix86_option_override_internal):
4901 Remove unused variable and field.
4903 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4906 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
4907 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
4908 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
4909 and Linux, correct BLKmode alignment when 128-bit alignment is
4910 required and compatibility flag is not set.
4911 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
4912 for zero-size arguments when compatibility flag is not set.
4914 2013-08-14 Jakub Jelinek <jakub@redhat.com>
4916 PR tree-optimization/58145
4917 * tree-sra.c (build_ref_for_offset): If prev_base has
4918 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
4920 2013-08-14 Xinliang David Li <davidxl@google.com>
4922 * config/i386/i386.c (ix86_option_override_internal):
4923 Fix uninitialized variable error.
4925 2013-08-14 Xinliang David Li <davidxl@google.com>
4927 * config/i386/i386.opt: Define two new options.
4928 * config/i386/x86-tune.def: Add arch selector field in macros.
4929 * config/i386/i386.h: Adjust macro definition.
4930 * config/i386/i386.c (ix86_option_override_internal):
4932 (parse_mtune_ctrl_str): New function.
4933 (set_ix86_tune_features): New function.
4934 (ix86_function_specific_restore): Call the new helper function.
4936 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
4938 PR rtl-optimization/57662
4939 * sel-sched.c (code_motion_process_successors): When the current insn
4940 is removed after the recursive traversal, break from the loop.
4941 Add comments and debug printouts.
4943 2013-08-14 Jakub Jelinek <jakub@redhat.com>
4944 Alexandre Oliva <aoliva@redhat.com>
4947 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
4948 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
4949 there also UNSPEC_PLTOFF.
4951 2013-08-14 Marek Polacek <polacek@redhat.com>
4953 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
4954 NUM_CONDITIONS bit positions.
4956 2013-08-13 Cary Coutant <ccoutant@google.com>
4958 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
4959 (attr_checksum): Hash vector contents instead of pointer.
4960 (attr_checksum_ordered): Likewise.
4962 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
4964 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
4965 when Pmode != word_mode. Add length_address attribute.
4966 (sse3_monitor_<mode>): Merge from sse3_monitor and
4967 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
4968 Pmode != word_mode. Update insn length attribute.
4969 * config/i386/i386.c (ix86_option_override_internal): Update
4970 ix86_gen_monitor selection for merged sse3_monitor insn.
4972 2013-08-13 Julian Brown <julian@codesourcery.com>
4974 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
4975 perform invalid legitimization on greater-than-word-size modes for
4978 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
4980 * ira.c (setup_class_translate_array): Use aclass instead of cl
4981 for classes not fully covered by allocno classes.
4983 2013-08-13 Jakub Jelinek <jakub@redhat.com>
4985 PR tree-optimization/57661
4986 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
4987 * tree-inline.c (tree_function_versioning): Initialize it.
4988 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
4989 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
4990 that is not being copied.
4993 * asan.c (instrument_strlen_call): Fix typo in comment.
4994 Use char * type even for the lhs of POINTER_PLUS_EXPR.
4996 2013-08-13 Steve Ellcey <sellcey@mips.com>
4998 * config/mips/mips.md (prefetch): Use lw instead of ld on
4999 loongson in 32bit mode.
5001 2013-08-13 Nick Clifton <nickc@redhat.com>
5003 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
5005 2013-08-13 Jan Hubicka <jh@suse.cz>
5007 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
5008 introduced edge; fix typo in sanity check.
5009 (cgraph_resolve_speculation): Export; improve diagnostic.
5010 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
5011 speculation at type mismatch.
5012 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
5013 (cgraph_resolve_speculation): Declare.
5014 (symtab_can_be_discarded): New function.
5015 * value-prof.c (gimple_ic_transform): Remove actual transform code.
5016 * ipa-inline-transform.c (speculation_removed): New global var.
5017 (clone_inlined_nodes): See if speculation can be removed.
5018 (inline_call): If speculations was removed, we growths may not match.
5019 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
5020 (speculation_useful_p): New function.
5021 (resolve_noninline_speculation): New function.
5022 (inline_small_functions): Resolve useless speculations.
5023 * ipa-inline.h (speculation_useful_p): Declare
5024 * ipa.c (can_replace_by_local_alias): Simplify.
5025 (ipa_profile): Produce speculative calls in non-lto, too;
5026 add simple cost model; produce local aliases.
5028 2013-08-13 David Malcolm <dmalcolm@redhat.com>
5030 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
5033 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
5035 * config/i386/i386.c (ix86_function_versions): Use error + inform.
5037 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
5039 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
5040 iterator instead of X87MODEF.
5042 2013-08-12 Perez Read <netfirewall@gmail.com>
5045 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
5046 operand 0 for intel asm alternative.
5047 (*movabs<mode>_2): Ditto for operand 1.
5049 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
5051 * config/aarch64/arm_none.h
5052 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
5054 2013-08-12 Nick Clifton <nickc@redhat.com>
5056 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
5058 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
5060 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
5061 expand for QI/HImode operand to produce more effictive code for
5062 unsigned char(short) --> float(double) conversion.
5064 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
5066 * doc/invoke.texi: Mention that -ftls-model does not force the final
5069 2013-08-12 Marek Polacek <polacek@redhat.com>
5070 Marc Glisse <marc.glisse@inria.fr>
5072 PR tree-optimization/57980
5073 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
5074 when creating -1 constant.
5076 2013-08-10 Jan Hubicka <jh@suse.cz>
5078 Workaround binutils PR14342.
5079 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
5080 (gimple_init_edge_profiler): Likewise.
5081 (gimple_gen_ic_func_profiler): Likewise.
5083 2013-08-09 Jan Hubicka <jh@suse.cz>
5085 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
5087 2013-08-09 Xinliang David Li <davidxl@google.com>
5089 * config/i386/stringop.def: New file.
5090 * config/i386/stringop.opt: New file.
5091 * config/i386/i386-opts.h: Include stringopt.def.
5092 * config/i386/i386.opt: Include stringopt.opt.
5093 * config/i386/i386.c (ix86_option_override_internal):
5094 Override default size based stringop inline strategies with options.
5095 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
5098 2013-08-09 Jan Hubicka <jh@suse.cz>
5100 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
5102 2013-08-09 Jan Hubicka <jh@suse.cz>
5104 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
5106 (dump_cgraph_node): Dump profile-id.
5107 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
5108 and common_target_probability.
5109 * lto-cgraph.c (lto_output_edge): Stream common targets.
5110 (lto_output_node): Stream profile ids.
5111 (input_node): Stream profile ids.
5112 (input_edge): Stream common targets.
5113 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
5114 * ipa.c: Include value-prof.h
5115 (ipa_profile_generate_summary): Turn indirect call statement histograms
5116 into common targets.
5117 (ipa_profile): Turn common targets into speculative edges.
5119 2013-08-09 Jan Hubicka <jh@suse.cz>
5121 * cgraph.h (cgraph_node): Add profile_id.
5122 * value-prof.c (cgraph_node_map): Turn into pointer_map.
5123 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
5124 (del_node_map): Update.
5125 (find_func_by_funcdef_no): Replace by ...
5126 (find_func_by_profile_id): ... this one.
5127 (gimple_ic_transform): Do not remove useful histograms when
5128 speculation is not done; dump info when indirect call removal
5130 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
5131 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
5132 (__gcov_indirect_call_profiler_v2): .. this one.
5133 * profile.h (init_node_map): Update.
5134 * coverage.c (coverage_compute_profile_id): New function.
5135 * coverage.h (coverage_compute_profile_id): Declare.
5136 * tree-profile.c (init_ic_make_global_vars): Make
5137 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
5138 (gimple_init_edge_profiler): Update prototype of
5139 __gcov_indirect_call_profiler.
5140 (gimple_gen_ic_func_profiler): Simplify.
5141 (tree_profiling): Use init_node_map
5143 2013-08-09 Jan Hubicka <jh@suse.cz>
5145 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
5146 non-speculative refs.
5147 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
5148 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
5149 (cgraph_set_call_stmt): Likewise.
5150 (cgraph_create_edge_1): Fix release checking compilatoin;
5152 (cgraph_free_edge): Free indirect info.
5153 (cgraph_turn_edge_to_speculative): New function.
5154 (cgraph_speculative_call_info): New function.
5155 (cgraph_make_edge_direct): Return direct edge; handle speculation.
5156 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
5157 (dump_cgraph_node): Dump speculation.
5158 (verify_edge_count_and_frequency): Accept speculative edges.
5159 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
5160 (verify_cgraph_node): Handle speculation.
5161 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
5162 (cgraph_set_call_stmt): Update prototype.
5163 (cgraph_make_edge_direct): Update prototype.
5164 (cgraph_speculative_call_info): Declare.
5165 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
5166 to change; update call of ipa_find_references.
5167 * ipa-ref.c (ipa_record_reference): Fix return value; clear
5168 lto_stmt_uid and speculative flags.
5169 (ipa_dump_references): Dump speculation.
5170 (ipa_clone_references): Clone speculative flag.
5171 (ipa_clone_referring): Likewise.
5172 (ipa_clone_ref): New function.
5173 (ipa_find_reference): Look into lto_stmt_uids
5174 (ipa_clear_stmts_in_references): Do not clear speculative calls.
5175 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
5176 (ipa_find_reference): Update declaration.
5177 (ipa_clone_ref): Declare.
5178 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
5179 stream speculative flag.
5180 (lto_output_ref): Stream statements uids and speculation.
5181 (input_ref): Likewise.
5182 (input_edge): Stream speuclation.
5183 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
5184 (cgraph_set_call_stmt_including_clones): Handle speculation.
5185 * ipa-inline.c (heap_edge_removal_hook): New function.
5186 (inline_small_functions): Register it.
5187 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
5188 also initialize refs.
5189 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
5191 (try_make_edge_direct_simple_call): Likewise.
5192 (try_make_edge_direct_simple_call): Likewise.
5193 (update_indirect_edges_after_inlining): Likewise.
5194 (remove_described_reference): Look proper lto_stmt_uid.
5195 (propagate_controlled_uses): Likewise.
5196 (propagate_controlled_uses): Liekwise.
5197 * tree-inline.c (copy_bb): Copy speculative edges.
5198 (redirect_all_calls): New function.
5199 (copy_cfg_body): Do redirection after loop info is updated.
5200 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
5202 2013-08-09 Jan Hubicka <jh@suse.cz>
5204 * lto-streamer-out.c (output_function): Renumber PHIs.
5205 * lto-streamer-in.c (input_function): Likewise.
5207 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
5209 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
5210 (get_lane_unsigned): Likewise.
5211 (dup_lane_scalar): Likewise.
5212 (get_lane): enable for VALL.
5213 * config/aarch64/aarch64-simd.md
5214 (aarch64_dup_lane_scalar<mode>): Remove.
5215 (aarch64_get_lane_signed<mode>): Likewise.
5216 (aarch64_get_lane_unsigned<mode>): Likewise.
5217 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
5218 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
5219 (aarch64_get_lane<mode>): Enable for all vector modes.
5220 (aarch64_get_lanedi): Remove misleading constraints.
5221 * config/aarch64/arm_neon.h
5222 (__aarch64_vget_lane_any): Define.
5223 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
5224 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
5225 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
5226 * config/aarch64/iterators.md (VDQQH): New.
5230 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
5232 * configure.ac: Add GAS check for LEON instructions on SPARC.
5233 * configure: Regenerate.
5234 * config.in: Likewise.
5235 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
5237 * config/sparc/sparc.opt (LEON, LEON3): New masks.
5238 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
5240 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
5241 (AS_LEON_FLAG): New macro.
5242 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
5243 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
5244 Deal with LEON and LEON3 for the memory model.
5245 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
5246 (atomic_compare_and_swap<mode>_1): Likewise.
5247 (*atomic_compare_and_swap<mode>_1): Likewise.
5249 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
5251 * config/arm/neon.md (vcond): Fix floating-point vector
5252 comparisons against 0.
5254 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
5256 * lra-constraints.c (emit_spill_move): Remove assert.
5257 (process_alt_operands): Add more debugging
5258 output. Increase reject for spilling into memory. Decrease
5259 reject for reloading scratch.
5260 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
5262 2013-08-08 Steve Ellcey <sellcey@mips.com>
5264 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
5265 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
5266 micromips incompatible. Add nan2008.
5267 (MULTILIB_DIRNAMES): Add nan2008.
5268 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5269 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
5270 and micromips incompatible. Add nan2008.
5271 (MULTILIB_DIRNAMES): Add nan2008.
5272 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5274 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
5276 PR rtl-optimization/58079
5277 * combine.c (combine_simplify_rtx): Avoid using SUBST if
5278 simplify_comparison has widened a comparison with an integer.
5280 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5282 * config/arm/neon.md (movmisalign<mode>): Disable when we
5283 don't allow unaligned accesses.
5284 (*movmisalign<mode>_neon_store): Likewise.
5285 (*movmisalign<mode>_neon_load): Likewise.
5286 (*movmisalign<mode>_neon_store): Likewise.
5287 (*movmisalign<mode>_neon_load): Likewise.
5289 2013-08-08 Jan Hubicka <jh@suse.cz>
5291 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
5292 (make_pass_rebuild_cgraph_edges): Also clear references.
5293 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
5294 * ipa-inline-transform.c (inline_transform): Remove all references
5296 * cgraphunit.c (expand_function): Remove all references after
5298 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
5299 (ipa_find_reference): Rewrite to iterator.
5300 (remove_stmt_references): Likewise.
5301 (ipa_clear_stmts_in_references): New function.
5302 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
5303 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
5305 * ipa-split.c (split_function): Remove references in split function.
5307 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
5310 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
5311 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
5313 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
5316 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
5317 suggested mode for the assignment isn't compatible with the
5320 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5323 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
5325 2013-08-07 Xinliang David Li <davidxl@google.com>
5327 * config/i386/i386.opt: New option -mtune-ctrl=.
5328 * config/i386/x86-tune.def: New file.
5329 * config/i386/i386.h: include x86-tune.def.
5330 * config/i386/i386.c (ix86_option_override_internal):
5331 Parsing -mtune-ctrl= option and set tune features.
5333 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
5336 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
5337 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
5340 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
5342 * rtl.h (update_alignments): Declare.
5343 * final.c (grow_label_align): New function extracted from...
5344 (shorten_branches): ...here. Call it.
5345 (update_alignments): New function.
5346 * reorg.c (sibling_labels): New variable.
5347 (get_label_before): Add SIBLING parameter. If it is non-zero, push
5348 the new label along with it onto the sibling_labels vector.
5349 (fill_simple_delay_slots): Adjust call to get_label_before.
5350 (fill_slots_from_thread): Likewise.
5351 (relax_delay_slots): Likewise.
5352 (make_return_insns): Likewise.
5353 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
5355 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
5357 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
5358 document its semantics.
5359 (diagnostic_report_diagnostic): Adjust accordingly.
5361 2013-08-07 David Malcolm <dmalcolm@redhat.com>
5363 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
5364 (sparc_option_override): ...and port to new C++ pass API.
5365 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
5367 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
5369 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
5371 2013-08-06 Caroline Tice <cmtice@google.com>
5373 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
5374 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
5375 * tree-pass.h: Add pass_vtable_verify.
5376 * varasm.c (assemble_variable): Add code to properly set the comdat
5377 section and name for the .vtable_map_vars section.
5378 (assemble_vtyv_preinit_initializer): New function.
5379 (default_sectin_type_flags): Make sure .vtable_map_vars section has
5381 * output.h: Add function decl for assemble_vtv_preinit_initializer.
5382 * vtable-verify.c: New file.
5383 * vtable-verify.h: New file.
5384 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
5385 initialiation levels.
5386 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
5387 * passes.def: Insert pass_vtable_verify.
5388 * aclocal.m4: Reorder includes.
5389 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
5390 -fvtv-counts options.
5391 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
5392 as appropriate, if -fvtable-verify=... is used.
5393 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
5394 -fvtable-verify=... is used.
5395 * Makefile.in (OBJS): Add vtable-verify.o to list.
5396 (vtable-verify.o): Add new build rule.
5397 (GTFILES): Add vtable-verify.c to list.
5398 * common.opt (fvtable-verify=): New flag.
5399 (vtv_priority): Values for fvtable-verify= flag.
5400 (fvtv-counts): New flag.
5401 (fvtv-debug): New flag.
5402 * tree.h (save_vtable_map_decl): New extern function decl.
5404 2013-08-07 David Malcolm <dmalcolm@redhat.com>
5406 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
5407 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
5408 (pass_data_rl78_devirt): ...new pass_data instance and...
5409 (make_pass_rl78_devirt): ...new function.
5410 (rl78_asm_file_start): Port pass registration to new C++ API.
5412 2013-08-07 David Malcolm <dmalcolm@redhat.com>
5414 * coretypes.h (rtl_opt_pass): Add.
5415 (gcc::context): Add.
5416 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
5417 (epiphany_init): Port to new C++ pass API.
5418 (epiphany_optimize_mode_switching): Likewise.
5419 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
5420 (pass_manager::get_pass_mode_switching): New.
5421 (pass_manager::get_pass_peephole2): New.
5422 * mode-switching.c (pass_mode_switching): Add clone method.
5423 * recog.c (pass_peephole2): Add clone method.
5424 (pass_split_all_insns): Add clone method.
5426 2013-08-06 David Malcolm <dmalcolm@redhat.com>
5428 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
5429 (mips_option_override): ...here, porting to new C++ API for passes.
5431 2013-08-06 Jan Hubicka <jh@suse.cz>
5433 * cgraph.c (cgraph_get_body): New function based on lto.c
5435 * cgraph.h (cgraph_get_body): Declare.
5436 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
5438 * cgraphunit.c (expand_function): Get body prior expanding.
5439 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
5440 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
5442 * passes.c (do_per_function_toporder): Get body.
5443 * tree-inline.c (expand_call_inline): Get body prior inlining it.
5444 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
5446 2013-08-06 Martin Jambor <mjambor@suse.cz>
5449 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
5450 is not re-finalized. Rename second parameter to no_collect.
5452 2013-08-06 Martin Jambor <mjambor@suse.cz>
5455 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
5456 MEM_REF has proper alignment information.
5458 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
5461 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
5463 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
5464 rtx (*) (rtx, ...) with insn_gen_fn.
5465 * genoutput.c (output_insn_data): Cast gen_? function pointers to
5466 insn_gen_fn::stored_funcptr. Add initializer braces.
5468 2013-08-05 David Malcolm <dmalcolm@redhat.com>
5470 Rewrite how instances of passes are cloned to remove assumptions
5471 about their sizes (thus allowing pass subclasses to have
5472 additional data fields, albeit non-GC-managed ones at this point).
5474 * passes.c (make_pass_instance): Now that passes have clone
5475 methods, rewrite this function to eliminate XNEW and memcpy
5476 calls that used hardcoded sizes. Since this function no longer
5477 creates pass instances, rename it to...
5478 (add_pass_instance): ...this. Document the old way that passes were
5479 numbered and flagged, and rework this function to continue using it.
5480 (next_pass_1): Add an initial_pass argument for use by
5482 (position_pass): When adding multiple instances of a pass, use
5483 the pass's clone method, rather than relying on the XNEW/memcpy
5484 within the former make_pass_instance (now add_pass_instance).
5485 (pass_manager::pass_manager): When invoking next_pass_1, also supply
5486 the initial instance of the current pass within the pass manager.
5488 2013-08-05 David Malcolm <dmalcolm@redhat.com>
5490 This is the automated part of the conversion of passes from C
5491 structs to C++ classes.
5493 Patch autogenerated by refactor_passes.py from
5494 https://github.com/davidmalcolm/gcc-refactoring-scripts
5495 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
5497 * asan.c (pass_asan): Convert from a global struct to a subclass of
5498 gimple_opt_pass along with...
5499 (pass_data_asan): ...new pass_data instance and...
5500 (make_pass_asan): ...new function.
5501 (pass_asan_O0): Convert from a global struct to a subclass of
5502 gimple_opt_pass along with...
5503 (pass_data_asan_O0): ...new pass_data instance and...
5504 (make_pass_asan_O0): ...new function.
5505 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
5506 subclass of rtl_opt_pass along with...
5507 (pass_data_inc_dec): ...new pass_data instance and...
5508 (make_pass_inc_dec): ...new function.
5509 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
5510 a subclass of rtl_opt_pass along with...
5511 (pass_data_reorder_blocks): ...new pass_data instance and...
5512 (make_pass_reorder_blocks): ...new function.
5513 (pass_duplicate_computed_gotos): Convert from a global struct to a
5514 subclass of rtl_opt_pass along with...
5515 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
5516 (make_pass_duplicate_computed_gotos): ...new function.
5517 (pass_partition_blocks): Convert from a global struct to a subclass of
5518 rtl_opt_pass along with...
5519 (pass_data_partition_blocks): ...new pass_data instance and...
5520 (make_pass_partition_blocks): ...new function.
5521 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
5522 struct to a subclass of rtl_opt_pass along with...
5523 (pass_data_branch_target_load_optimize1): ...new pass_data instance
5525 (make_pass_branch_target_load_optimize1): ...new function.
5526 (pass_branch_target_load_optimize2): Convert from a global struct to a
5527 subclass of rtl_opt_pass along with...
5528 (pass_data_branch_target_load_optimize2): ...new pass_data instance
5530 (make_pass_branch_target_load_optimize2): ...new function.
5531 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
5532 of rtl_opt_pass along with...
5533 (pass_data_jump): ...new pass_data instance and...
5534 (make_pass_jump): ...new function.
5535 (pass_jump2): Convert from a global struct to a subclass of
5536 rtl_opt_pass along with...
5537 (pass_data_jump2): ...new pass_data instance and...
5538 (make_pass_jump2): ...new function.
5539 * cfgexpand.c (pass_expand): Convert from a global struct to a
5540 subclass of rtl_opt_pass along with...
5541 (pass_data_expand): ...new pass_data instance and...
5542 (make_pass_expand): ...new function.
5543 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
5544 of rtl_opt_pass along with...
5545 (pass_data_free_cfg): ...new pass_data instance and...
5546 (make_pass_free_cfg): ...new function.
5547 (pass_into_cfg_layout_mode): Convert from a global struct to a
5548 subclass of rtl_opt_pass along with...
5549 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
5550 (make_pass_into_cfg_layout_mode): ...new function.
5551 (pass_outof_cfg_layout_mode): Convert from a global struct to a
5552 subclass of rtl_opt_pass along with...
5553 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
5554 (make_pass_outof_cfg_layout_mode): ...new function.
5555 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
5556 struct to a subclass of gimple_opt_pass along with...
5557 (pass_data_build_cgraph_edges): ...new pass_data instance and...
5558 (make_pass_build_cgraph_edges): ...new function.
5559 (pass_rebuild_cgraph_edges): Convert from a global struct to a
5560 subclass of gimple_opt_pass along with...
5561 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
5562 (make_pass_rebuild_cgraph_edges): ...new function.
5563 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
5564 subclass of gimple_opt_pass along with...
5565 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
5567 (make_pass_remove_cgraph_callee_edges): ...new function.
5568 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
5569 struct to a subclass of rtl_opt_pass along with...
5570 (pass_data_stack_adjustments): ...new pass_data instance and...
5571 (make_pass_stack_adjustments): ...new function.
5572 * combine.c (pass_combine): Convert from a global struct to a subclass
5573 of rtl_opt_pass along with...
5574 (pass_data_combine): ...new pass_data instance and...
5575 (make_pass_combine): ...new function.
5576 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
5577 global struct to a subclass of rtl_opt_pass along with...
5578 (pass_data_compare_elim_after_reload): ...new pass_data instance
5580 (make_pass_compare_elim_after_reload): ...new function.
5581 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
5582 of rtl_opt_pass along with...
5583 (pass_data_rtl_cprop): ...new pass_data instance and...
5584 (make_pass_rtl_cprop): ...new function.
5585 * cse.c (pass_cse): Convert from a global struct to a subclass of
5586 rtl_opt_pass along with...
5587 (pass_data_cse): ...new pass_data instance and...
5588 (make_pass_cse): ...new function.
5589 (pass_cse2): Convert from a global struct to a subclass of
5590 rtl_opt_pass along with...
5591 (pass_data_cse2): ...new pass_data instance and...
5592 (make_pass_cse2): ...new function.
5593 (pass_cse_after_global_opts): Convert from a global struct to a
5594 subclass of rtl_opt_pass along with...
5595 (pass_data_cse_after_global_opts): ...new pass_data instance and...
5596 (make_pass_cse_after_global_opts): ...new function.
5597 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
5598 of rtl_opt_pass along with...
5599 (pass_data_ud_rtl_dce): ...new pass_data instance and...
5600 (make_pass_ud_rtl_dce): ...new function.
5601 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
5602 rtl_opt_pass along with...
5603 (pass_data_fast_rtl_dce): ...new pass_data instance and...
5604 (make_pass_fast_rtl_dce): ...new function.
5605 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
5606 a subclass of rtl_opt_pass along with...
5607 (pass_data_df_initialize_opt): ...new pass_data instance and...
5608 (make_pass_df_initialize_opt): ...new function.
5609 (pass_df_initialize_no_opt): Convert from a global struct to a
5610 subclass of rtl_opt_pass along with...
5611 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
5612 (make_pass_df_initialize_no_opt): ...new function.
5613 (pass_df_finish): Convert from a global struct to a subclass of
5614 rtl_opt_pass along with...
5615 (pass_data_df_finish): ...new pass_data instance and...
5616 (make_pass_df_finish): ...new function.
5617 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
5618 rtl_opt_pass along with...
5619 (pass_data_rtl_dse1): ...new pass_data instance and...
5620 (make_pass_rtl_dse1): ...new function.
5621 (pass_rtl_dse2): Convert from a global struct to a subclass of
5622 rtl_opt_pass along with...
5623 (pass_data_rtl_dse2): ...new pass_data instance and...
5624 (make_pass_rtl_dse2): ...new function.
5625 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
5626 subclass of rtl_opt_pass along with...
5627 (pass_data_dwarf2_frame): ...new pass_data instance and...
5628 (make_pass_dwarf2_frame): ...new function.
5629 * except.c (pass_set_nothrow_function_flags): Convert from a global
5630 struct to a subclass of rtl_opt_pass along with...
5631 (pass_data_set_nothrow_function_flags): ...new pass_data instance
5633 (make_pass_set_nothrow_function_flags): ...new function.
5634 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
5635 subclass of rtl_opt_pass along with...
5636 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
5638 (make_pass_convert_to_eh_region_ranges): ...new function.
5639 * final.c (pass_compute_alignments): Convert from a global struct to a
5640 subclass of rtl_opt_pass along with...
5641 (pass_data_compute_alignments): ...new pass_data instance and...
5642 (make_pass_compute_alignments): ...new function.
5643 (pass_final): Convert from a global struct to a subclass of
5644 rtl_opt_pass along with...
5645 (pass_data_final): ...new pass_data instance and...
5646 (make_pass_final): ...new function.
5647 (pass_shorten_branches): Convert from a global struct to a subclass of
5648 rtl_opt_pass along with...
5649 (pass_data_shorten_branches): ...new pass_data instance and...
5650 (make_pass_shorten_branches): ...new function.
5651 (pass_clean_state): Convert from a global struct to a subclass of
5652 rtl_opt_pass along with...
5653 (pass_data_clean_state): ...new pass_data instance and...
5654 (make_pass_clean_state): ...new function.
5655 * function.c (pass_instantiate_virtual_regs): Convert from a global
5656 struct to a subclass of rtl_opt_pass along with...
5657 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
5658 (make_pass_instantiate_virtual_regs): ...new function.
5659 (pass_leaf_regs): Convert from a global struct to a subclass of
5660 rtl_opt_pass along with...
5661 (pass_data_leaf_regs): ...new pass_data instance and...
5662 (make_pass_leaf_regs): ...new function.
5663 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
5664 subclass of rtl_opt_pass along with...
5665 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
5667 (make_pass_thread_prologue_and_epilogue): ...new function.
5668 (pass_match_asm_constraints): Convert from a global struct to a
5669 subclass of rtl_opt_pass along with...
5670 (pass_data_match_asm_constraints): ...new pass_data instance and...
5671 (make_pass_match_asm_constraints): ...new function.
5672 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
5673 subclass of rtl_opt_pass along with...
5674 (pass_data_rtl_fwprop): ...new pass_data instance and...
5675 (make_pass_rtl_fwprop): ...new function.
5676 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
5677 rtl_opt_pass along with...
5678 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
5679 (make_pass_rtl_fwprop_addr): ...new function.
5680 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
5681 rtl_opt_pass along with...
5682 (pass_data_rtl_pre): ...new pass_data instance and...
5683 (make_pass_rtl_pre): ...new function.
5684 (pass_rtl_hoist): Convert from a global struct to a subclass of
5685 rtl_opt_pass along with...
5686 (pass_data_rtl_hoist): ...new pass_data instance and...
5687 (make_pass_rtl_hoist): ...new function.
5688 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
5689 subclass of gimple_opt_pass along with...
5690 (pass_data_lower_cf): ...new pass_data instance and...
5691 (make_pass_lower_cf): ...new function.
5692 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
5693 from a global struct to a subclass of gimple_opt_pass along with...
5694 (pass_data_strength_reduction): ...new pass_data instance and...
5695 (make_pass_strength_reduction): ...new function.
5696 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
5697 of rtl_opt_pass along with...
5698 (pass_data_rtl_ifcvt): ...new pass_data instance and...
5699 (make_pass_rtl_ifcvt): ...new function.
5700 (pass_if_after_combine): Convert from a global struct to a subclass of
5701 rtl_opt_pass along with...
5702 (pass_data_if_after_combine): ...new pass_data instance and...
5703 (make_pass_if_after_combine): ...new function.
5704 (pass_if_after_reload): Convert from a global struct to a subclass of
5705 rtl_opt_pass along with...
5706 (pass_data_if_after_reload): ...new pass_data instance and...
5707 (make_pass_if_after_reload): ...new function.
5708 * init-regs.c (pass_initialize_regs): Convert from a global struct to
5709 a subclass of rtl_opt_pass along with...
5710 (pass_data_initialize_regs): ...new pass_data instance and...
5711 (make_pass_initialize_regs): ...new function.
5712 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
5713 of ipa_opt_pass_d along with...
5714 (pass_data_ipa_cp): ...new pass_data instance and...
5715 (make_pass_ipa_cp): ...new function.
5716 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
5717 global struct to a subclass of gimple_opt_pass along with...
5718 (pass_data_inline_parameters): ...new pass_data instance and...
5719 (make_pass_inline_parameters): ...new function.
5720 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
5721 subclass of gimple_opt_pass along with...
5722 (pass_data_early_inline): ...new pass_data instance and...
5723 (make_pass_early_inline): ...new function.
5724 (pass_ipa_inline): Convert from a global struct to a subclass of
5725 ipa_opt_pass_d along with...
5726 (pass_data_ipa_inline): ...new pass_data instance and...
5727 (make_pass_ipa_inline): ...new function.
5728 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
5729 struct to a subclass of gimple_opt_pass along with...
5730 (pass_data_local_pure_const): ...new pass_data instance and...
5731 (make_pass_local_pure_const): ...new function.
5732 (pass_ipa_pure_const): Convert from a global struct to a subclass of
5733 ipa_opt_pass_d along with...
5734 (pass_data_ipa_pure_const): ...new pass_data instance and...
5735 (make_pass_ipa_pure_const): ...new function.
5736 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
5737 to a subclass of ipa_opt_pass_d along with...
5738 (pass_data_ipa_reference): ...new pass_data instance and...
5739 (make_pass_ipa_reference): ...new function.
5740 * ipa-split.c (pass_split_functions): Convert from a global struct to
5741 a subclass of gimple_opt_pass along with...
5742 (pass_data_split_functions): ...new pass_data instance and...
5743 (make_pass_split_functions): ...new function.
5744 (pass_feedback_split_functions): Convert from a global struct to a
5745 subclass of gimple_opt_pass along with...
5746 (pass_data_feedback_split_functions): ...new pass_data instance and...
5747 (make_pass_feedback_split_functions): ...new function.
5748 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
5749 global struct to a subclass of simple_ipa_opt_pass along with...
5750 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
5752 (make_pass_ipa_function_and_variable_visibility): ...new function.
5753 (pass_ipa_free_inline_summary): Convert from a global struct to a
5754 subclass of simple_ipa_opt_pass along with...
5755 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
5756 (make_pass_ipa_free_inline_summary): ...new function.
5757 (pass_ipa_whole_program_visibility): Convert from a global struct to a
5758 subclass of ipa_opt_pass_d along with...
5759 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
5761 (make_pass_ipa_whole_program_visibility): ...new function.
5762 (pass_ipa_profile): Convert from a global struct to a subclass of
5763 ipa_opt_pass_d along with...
5764 (pass_data_ipa_profile): ...new pass_data instance and...
5765 (make_pass_ipa_profile): ...new function.
5766 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
5767 ipa_opt_pass_d along with...
5768 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
5769 (make_pass_ipa_cdtor_merge): ...new function.
5770 * ira.c (pass_ira): Convert from a global struct to a subclass of
5771 rtl_opt_pass along with...
5772 (pass_data_ira): ...new pass_data instance and...
5773 (make_pass_ira): ...new function.
5774 (pass_reload): Convert from a global struct to a subclass of
5775 rtl_opt_pass along with...
5776 (pass_data_reload): ...new pass_data instance and...
5777 (make_pass_reload): ...new function.
5778 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
5779 subclass of rtl_opt_pass along with...
5780 (pass_data_cleanup_barriers): ...new pass_data instance and...
5781 (make_pass_cleanup_barriers): ...new function.
5782 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
5783 of rtl_opt_pass along with...
5784 (pass_data_loop2): ...new pass_data instance and...
5785 (make_pass_loop2): ...new function.
5786 (pass_rtl_loop_init): Convert from a global struct to a subclass of
5787 rtl_opt_pass along with...
5788 (pass_data_rtl_loop_init): ...new pass_data instance and...
5789 (make_pass_rtl_loop_init): ...new function.
5790 (pass_rtl_loop_done): Convert from a global struct to a subclass of
5791 rtl_opt_pass along with...
5792 (pass_data_rtl_loop_done): ...new pass_data instance and...
5793 (make_pass_rtl_loop_done): ...new function.
5794 (pass_rtl_move_loop_invariants): Convert from a global struct to a
5795 subclass of rtl_opt_pass along with...
5796 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
5797 (make_pass_rtl_move_loop_invariants): ...new function.
5798 (pass_rtl_unswitch): Convert from a global struct to a subclass of
5799 rtl_opt_pass along with...
5800 (pass_data_rtl_unswitch): ...new pass_data instance and...
5801 (make_pass_rtl_unswitch): ...new function.
5802 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
5803 subclass of rtl_opt_pass along with...
5804 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
5806 (make_pass_rtl_unroll_and_peel_loops): ...new function.
5807 (pass_rtl_doloop): Convert from a global struct to a subclass of
5808 rtl_opt_pass along with...
5809 (pass_data_rtl_doloop): ...new pass_data instance and...
5810 (make_pass_rtl_doloop): ...new function.
5811 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
5812 a subclass of rtl_opt_pass along with...
5813 (pass_data_lower_subreg): ...new pass_data instance and...
5814 (make_pass_lower_subreg): ...new function.
5815 (pass_lower_subreg2): Convert from a global struct to a subclass of
5816 rtl_opt_pass along with...
5817 (pass_data_lower_subreg2): ...new pass_data instance and...
5818 (make_pass_lower_subreg2): ...new function.
5819 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
5820 struct to a subclass of ipa_opt_pass_d along with...
5821 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
5822 (make_pass_ipa_lto_gimple_out): ...new function.
5823 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
5824 of ipa_opt_pass_d along with...
5825 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
5826 (make_pass_ipa_lto_finish_out): ...new function.
5827 * mode-switching.c (pass_mode_switching): Convert from a global struct
5828 to a subclass of rtl_opt_pass along with...
5829 (pass_data_mode_switching): ...new pass_data instance and...
5830 (make_pass_mode_switching): ...new function.
5831 * modulo-sched.c (pass_sms): Convert from a global struct to a
5832 subclass of rtl_opt_pass along with...
5833 (pass_data_sms): ...new pass_data instance and...
5834 (make_pass_sms): ...new function.
5835 * omp-low.c (pass_expand_omp): Convert from a global struct to a
5836 subclass of gimple_opt_pass along with...
5837 (pass_data_expand_omp): ...new pass_data instance and...
5838 (make_pass_expand_omp): ...new function.
5839 (pass_lower_omp): Convert from a global struct to a subclass of
5840 gimple_opt_pass along with...
5841 (pass_data_lower_omp): ...new pass_data instance and...
5842 (make_pass_lower_omp): ...new function.
5843 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
5844 of gimple_opt_pass along with...
5845 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
5846 (make_pass_diagnose_omp_blocks): ...new function.
5847 * passes.c (pass_early_local_passes): Convert from a global struct to
5848 a subclass of simple_ipa_opt_pass along with...
5849 (pass_data_early_local_passes): ...new pass_data instance and...
5850 (make_pass_early_local_passes): ...new function.
5851 (pass_all_early_optimizations): Convert from a global struct to a
5852 subclass of gimple_opt_pass along with...
5853 (pass_data_all_early_optimizations): ...new pass_data instance and...
5854 (make_pass_all_early_optimizations): ...new function.
5855 (pass_all_optimizations): Convert from a global struct to a subclass
5856 of gimple_opt_pass along with...
5857 (pass_data_all_optimizations): ...new pass_data instance and...
5858 (make_pass_all_optimizations): ...new function.
5859 (pass_all_optimizations_g): Convert from a global struct to a subclass
5860 of gimple_opt_pass along with...
5861 (pass_data_all_optimizations_g): ...new pass_data instance and...
5862 (make_pass_all_optimizations_g): ...new function.
5863 (pass_rest_of_compilation): Convert from a global struct to a subclass
5864 of rtl_opt_pass along with...
5865 (pass_data_rest_of_compilation): ...new pass_data instance and...
5866 (make_pass_rest_of_compilation): ...new function.
5867 (pass_postreload): Convert from a global struct to a subclass of
5868 rtl_opt_pass along with...
5869 (pass_data_postreload): ...new pass_data instance and...
5870 (make_pass_postreload): ...new function.
5871 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
5872 subclass of rtl_opt_pass along with...
5873 (pass_data_gcse2): ...new pass_data instance and...
5874 (make_pass_gcse2): ...new function.
5875 * postreload.c (pass_postreload_cse): Convert from a global struct to
5876 a subclass of rtl_opt_pass along with...
5877 (pass_data_postreload_cse): ...new pass_data instance and...
5878 (make_pass_postreload_cse): ...new function.
5879 * predict.c (pass_profile): Convert from a global struct to a subclass
5880 of gimple_opt_pass along with...
5881 (pass_data_profile): ...new pass_data instance and...
5882 (make_pass_profile): ...new function.
5883 (pass_strip_predict_hints): Convert from a global struct to a subclass
5884 of gimple_opt_pass along with...
5885 (pass_data_strip_predict_hints): ...new pass_data instance and...
5886 (make_pass_strip_predict_hints): ...new function.
5887 * recog.c (pass_peephole2): Convert from a global struct to a subclass
5888 of rtl_opt_pass along with...
5889 (pass_data_peephole2): ...new pass_data instance and...
5890 (make_pass_peephole2): ...new function.
5891 (pass_split_all_insns): Convert from a global struct to a subclass of
5892 rtl_opt_pass along with...
5893 (pass_data_split_all_insns): ...new pass_data instance and...
5894 (make_pass_split_all_insns): ...new function.
5895 (pass_split_after_reload): Convert from a global struct to a subclass
5896 of rtl_opt_pass along with...
5897 (pass_data_split_after_reload): ...new pass_data instance and...
5898 (make_pass_split_after_reload): ...new function.
5899 (pass_split_before_regstack): Convert from a global struct to a
5900 subclass of rtl_opt_pass along with...
5901 (pass_data_split_before_regstack): ...new pass_data instance and...
5902 (make_pass_split_before_regstack): ...new function.
5903 (pass_split_before_sched2): Convert from a global struct to a subclass
5904 of rtl_opt_pass along with...
5905 (pass_data_split_before_sched2): ...new pass_data instance and...
5906 (make_pass_split_before_sched2): ...new function.
5907 (pass_split_for_shorten_branches): Convert from a global struct to a
5908 subclass of rtl_opt_pass along with...
5909 (pass_data_split_for_shorten_branches): ...new pass_data instance
5911 (make_pass_split_for_shorten_branches): ...new function.
5912 * ree.c (pass_ree): Convert from a global struct to a subclass of
5913 rtl_opt_pass along with...
5914 (pass_data_ree): ...new pass_data instance and...
5915 (make_pass_ree): ...new function.
5916 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
5917 subclass of rtl_opt_pass along with...
5918 (pass_data_stack_regs): ...new pass_data instance and...
5919 (make_pass_stack_regs): ...new function.
5920 (pass_stack_regs_run): Convert from a global struct to a subclass of
5921 rtl_opt_pass along with...
5922 (pass_data_stack_regs_run): ...new pass_data instance and...
5923 (make_pass_stack_regs_run): ...new function.
5924 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
5925 subclass of rtl_opt_pass along with...
5926 (pass_data_cprop_hardreg): ...new pass_data instance and...
5927 (make_pass_cprop_hardreg): ...new function.
5928 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
5929 subclass of rtl_opt_pass along with...
5930 (pass_data_reginfo_init): ...new pass_data instance and...
5931 (make_pass_reginfo_init): ...new function.
5932 * regmove.c (pass_regmove): Convert from a global struct to a subclass
5933 of rtl_opt_pass along with...
5934 (pass_data_regmove): ...new pass_data instance and...
5935 (make_pass_regmove): ...new function.
5936 * regrename.c (pass_regrename): Convert from a global struct to a
5937 subclass of rtl_opt_pass along with...
5938 (pass_data_regrename): ...new pass_data instance and...
5939 (make_pass_regrename): ...new function.
5940 * reorg.c (pass_delay_slots): Convert from a global struct to a
5941 subclass of rtl_opt_pass along with...
5942 (pass_data_delay_slots): ...new pass_data instance and...
5943 (make_pass_delay_slots): ...new function.
5944 (pass_machine_reorg): Convert from a global struct to a subclass of
5945 rtl_opt_pass along with...
5946 (pass_data_machine_reorg): ...new pass_data instance and...
5947 (make_pass_machine_reorg): ...new function.
5948 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
5949 of rtl_opt_pass along with...
5950 (pass_data_sched): ...new pass_data instance and...
5951 (make_pass_sched): ...new function.
5952 (pass_sched2): Convert from a global struct to a subclass of
5953 rtl_opt_pass along with...
5954 (pass_data_sched2): ...new pass_data instance and...
5955 (make_pass_sched2): ...new function.
5956 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
5957 to a subclass of rtl_opt_pass along with...
5958 (pass_data_stack_ptr_mod): ...new pass_data instance and...
5959 (make_pass_stack_ptr_mod): ...new function.
5960 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
5961 to a subclass of rtl_opt_pass along with...
5962 (pass_data_rtl_store_motion): ...new pass_data instance and...
5963 (make_pass_rtl_store_motion): ...new function.
5964 * tracer.c (pass_tracer): Convert from a global struct to a subclass
5965 of gimple_opt_pass along with...
5966 (pass_data_tracer): ...new pass_data instance and...
5967 (make_pass_tracer): ...new function.
5968 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
5969 to a subclass of gimple_opt_pass along with...
5970 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
5971 (make_pass_diagnose_tm_blocks): ...new function.
5972 (pass_lower_tm): Convert from a global struct to a subclass of
5973 gimple_opt_pass along with...
5974 (pass_data_lower_tm): ...new pass_data instance and...
5975 (make_pass_lower_tm): ...new function.
5976 (pass_tm_init): Convert from a global struct to a subclass of
5977 gimple_opt_pass along with...
5978 (pass_data_tm_init): ...new pass_data instance and...
5979 (make_pass_tm_init): ...new function.
5980 (pass_tm_mark): Convert from a global struct to a subclass of
5981 gimple_opt_pass along with...
5982 (pass_data_tm_mark): ...new pass_data instance and...
5983 (make_pass_tm_mark): ...new function.
5984 (pass_tm_edges): Convert from a global struct to a subclass of
5985 gimple_opt_pass along with...
5986 (pass_data_tm_edges): ...new pass_data instance and...
5987 (make_pass_tm_edges): ...new function.
5988 (pass_tm_memopt): Convert from a global struct to a subclass of
5989 gimple_opt_pass along with...
5990 (pass_data_tm_memopt): ...new pass_data instance and...
5991 (make_pass_tm_memopt): ...new function.
5992 (pass_ipa_tm): Convert from a global struct to a subclass of
5993 simple_ipa_opt_pass along with...
5994 (pass_data_ipa_tm): ...new pass_data instance and...
5995 (make_pass_ipa_tm): ...new function.
5996 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
5997 subclass of gimple_opt_pass along with...
5998 (pass_data_call_cdce): ...new pass_data instance and...
5999 (make_pass_call_cdce): ...new function.
6000 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
6001 subclass of gimple_opt_pass along with...
6002 (pass_data_build_cfg): ...new pass_data instance and...
6003 (make_pass_build_cfg): ...new function.
6004 (pass_split_crit_edges): Convert from a global struct to a subclass of
6005 gimple_opt_pass along with...
6006 (pass_data_split_crit_edges): ...new pass_data instance and...
6007 (make_pass_split_crit_edges): ...new function.
6008 (pass_warn_function_return): Convert from a global struct to a
6009 subclass of gimple_opt_pass along with...
6010 (pass_data_warn_function_return): ...new pass_data instance and...
6011 (make_pass_warn_function_return): ...new function.
6012 (pass_warn_function_noreturn): Convert from a global struct to a
6013 subclass of gimple_opt_pass along with...
6014 (pass_data_warn_function_noreturn): ...new pass_data instance and...
6015 (make_pass_warn_function_noreturn): ...new function.
6016 (pass_warn_unused_result): Convert from a global struct to a subclass
6017 of gimple_opt_pass along with...
6018 (pass_data_warn_unused_result): ...new pass_data instance and...
6019 (make_pass_warn_unused_result): ...new function.
6020 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
6021 a subclass of gimple_opt_pass along with...
6022 (pass_data_merge_phi): ...new pass_data instance and...
6023 (make_pass_merge_phi): ...new function.
6024 * tree-complex.c (pass_lower_complex): Convert from a global struct to
6025 a subclass of gimple_opt_pass along with...
6026 (pass_data_lower_complex): ...new pass_data instance and...
6027 (make_pass_lower_complex): ...new function.
6028 (pass_lower_complex_O0): Convert from a global struct to a subclass of
6029 gimple_opt_pass along with...
6030 (pass_data_lower_complex_O0): ...new pass_data instance and...
6031 (make_pass_lower_complex_O0): ...new function.
6032 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
6033 subclass of gimple_opt_pass along with...
6034 (pass_data_lower_eh): ...new pass_data instance and...
6035 (make_pass_lower_eh): ...new function.
6036 (pass_refactor_eh): Convert from a global struct to a subclass of
6037 gimple_opt_pass along with...
6038 (pass_data_refactor_eh): ...new pass_data instance and...
6039 (make_pass_refactor_eh): ...new function.
6040 (pass_lower_resx): Convert from a global struct to a subclass of
6041 gimple_opt_pass along with...
6042 (pass_data_lower_resx): ...new pass_data instance and...
6043 (make_pass_lower_resx): ...new function.
6044 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
6045 of gimple_opt_pass along with...
6046 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
6047 (make_pass_lower_eh_dispatch): ...new function.
6048 (pass_cleanup_eh): Convert from a global struct to a subclass of
6049 gimple_opt_pass along with...
6050 (pass_data_cleanup_eh): ...new pass_data instance and...
6051 (make_pass_cleanup_eh): ...new function.
6052 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
6053 to a subclass of simple_ipa_opt_pass along with...
6054 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
6055 (make_pass_ipa_lower_emutls): ...new function.
6056 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
6057 a subclass of gimple_opt_pass along with...
6058 (pass_data_if_conversion): ...new pass_data instance and...
6059 (make_pass_if_conversion): ...new function.
6060 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
6061 subclass of gimple_opt_pass along with...
6062 (pass_data_build_ssa): ...new pass_data instance and...
6063 (make_pass_build_ssa): ...new function.
6064 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
6065 global struct to a subclass of gimple_opt_pass along with...
6066 (pass_data_loop_distribution): ...new pass_data instance and...
6067 (make_pass_loop_distribution): ...new function.
6068 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
6069 subclass of gimple_opt_pass along with...
6070 (pass_data_mudflap_1): ...new pass_data instance and...
6071 (make_pass_mudflap_1): ...new function.
6072 (pass_mudflap_2): Convert from a global struct to a subclass of
6073 gimple_opt_pass along with...
6074 (pass_data_mudflap_2): ...new pass_data instance and...
6075 (make_pass_mudflap_2): ...new function.
6076 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
6077 subclass of gimple_opt_pass along with...
6078 (pass_data_mudflap_1): ...new pass_data instance and...
6079 (make_pass_mudflap_1): ...new function.
6080 (pass_mudflap_2): Convert from a global struct to a subclass of
6081 gimple_opt_pass along with...
6082 (pass_data_mudflap_2): ...new pass_data instance and...
6083 (make_pass_mudflap_2): ...new function.
6084 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
6085 gimple_opt_pass along with...
6086 (pass_data_nrv): ...new pass_data instance and...
6087 (make_pass_nrv): ...new function.
6088 (pass_return_slot): Convert from a global struct to a subclass of
6089 gimple_opt_pass along with...
6090 (pass_data_return_slot): ...new pass_data instance and...
6091 (make_pass_return_slot): ...new function.
6092 * tree-object-size.c (pass_object_sizes): Convert from a global struct
6093 to a subclass of gimple_opt_pass along with...
6094 (pass_data_object_sizes): ...new pass_data instance and...
6095 (make_pass_object_sizes): ...new function.
6096 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
6097 global struct to a subclass of gimple_opt_pass along with...
6098 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
6100 (make_pass_cleanup_cfg_post_optimizing): ...new function.
6101 (pass_fixup_cfg): Convert from a global struct to a subclass of
6102 gimple_opt_pass along with...
6103 (pass_data_fixup_cfg): ...new pass_data instance and...
6104 (make_pass_fixup_cfg): ...new function.
6105 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
6106 (make_pass_mudflap_1): ...new function.
6107 (pass_mudflap_2): Replace declaration with that of...
6108 (make_pass_mudflap_2): ...new function.
6109 (pass_asan): Replace declaration with that of...
6110 (make_pass_asan): ...new function.
6111 (pass_asan_O0): Replace declaration with that of...
6112 (make_pass_asan_O0): ...new function.
6113 (pass_tsan): Replace declaration with that of...
6114 (make_pass_tsan): ...new function.
6115 (pass_tsan_O0): Replace declaration with that of...
6116 (make_pass_tsan_O0): ...new function.
6117 (pass_lower_cf): Replace declaration with that of...
6118 (make_pass_lower_cf): ...new function.
6119 (pass_refactor_eh): Replace declaration with that of...
6120 (make_pass_refactor_eh): ...new function.
6121 (pass_lower_eh): Replace declaration with that of...
6122 (make_pass_lower_eh): ...new function.
6123 (pass_lower_eh_dispatch): Replace declaration with that of...
6124 (make_pass_lower_eh_dispatch): ...new function.
6125 (pass_lower_resx): Replace declaration with that of...
6126 (make_pass_lower_resx): ...new function.
6127 (pass_build_cfg): Replace declaration with that of...
6128 (make_pass_build_cfg): ...new function.
6129 (pass_early_tree_profile): Replace declaration with that of...
6130 (make_pass_early_tree_profile): ...new function.
6131 (pass_cleanup_eh): Replace declaration with that of...
6132 (make_pass_cleanup_eh): ...new function.
6133 (pass_sra): Replace declaration with that of...
6134 (make_pass_sra): ...new function.
6135 (pass_sra_early): Replace declaration with that of...
6136 (make_pass_sra_early): ...new function.
6137 (pass_early_ipa_sra): Replace declaration with that of...
6138 (make_pass_early_ipa_sra): ...new function.
6139 (pass_tail_recursion): Replace declaration with that of...
6140 (make_pass_tail_recursion): ...new function.
6141 (pass_tail_calls): Replace declaration with that of...
6142 (make_pass_tail_calls): ...new function.
6143 (pass_tree_loop): Replace declaration with that of...
6144 (make_pass_tree_loop): ...new function.
6145 (pass_tree_loop_init): Replace declaration with that of...
6146 (make_pass_tree_loop_init): ...new function.
6147 (pass_lim): Replace declaration with that of...
6148 (make_pass_lim): ...new function.
6149 (pass_tree_unswitch): Replace declaration with that of...
6150 (make_pass_tree_unswitch): ...new function.
6151 (pass_predcom): Replace declaration with that of...
6152 (make_pass_predcom): ...new function.
6153 (pass_iv_canon): Replace declaration with that of...
6154 (make_pass_iv_canon): ...new function.
6155 (pass_scev_cprop): Replace declaration with that of...
6156 (make_pass_scev_cprop): ...new function.
6157 (pass_empty_loop): Replace declaration with that of...
6158 (make_pass_empty_loop): ...new function.
6159 (pass_record_bounds): Replace declaration with that of...
6160 (make_pass_record_bounds): ...new function.
6161 (pass_graphite): Replace declaration with that of...
6162 (make_pass_graphite): ...new function.
6163 (pass_graphite_transforms): Replace declaration with that of...
6164 (make_pass_graphite_transforms): ...new function.
6165 (pass_if_conversion): Replace declaration with that of...
6166 (make_pass_if_conversion): ...new function.
6167 (pass_loop_distribution): Replace declaration with that of...
6168 (make_pass_loop_distribution): ...new function.
6169 (pass_vectorize): Replace declaration with that of...
6170 (make_pass_vectorize): ...new function.
6171 (pass_slp_vectorize): Replace declaration with that of...
6172 (make_pass_slp_vectorize): ...new function.
6173 (pass_complete_unroll): Replace declaration with that of...
6174 (make_pass_complete_unroll): ...new function.
6175 (pass_complete_unrolli): Replace declaration with that of...
6176 (make_pass_complete_unrolli): ...new function.
6177 (pass_parallelize_loops): Replace declaration with that of...
6178 (make_pass_parallelize_loops): ...new function.
6179 (pass_loop_prefetch): Replace declaration with that of...
6180 (make_pass_loop_prefetch): ...new function.
6181 (pass_iv_optimize): Replace declaration with that of...
6182 (make_pass_iv_optimize): ...new function.
6183 (pass_tree_loop_done): Replace declaration with that of...
6184 (make_pass_tree_loop_done): ...new function.
6185 (pass_ch): Replace declaration with that of...
6186 (make_pass_ch): ...new function.
6187 (pass_ccp): Replace declaration with that of...
6188 (make_pass_ccp): ...new function.
6189 (pass_phi_only_cprop): Replace declaration with that of...
6190 (make_pass_phi_only_cprop): ...new function.
6191 (pass_build_ssa): Replace declaration with that of...
6192 (make_pass_build_ssa): ...new function.
6193 (pass_build_alias): Replace declaration with that of...
6194 (make_pass_build_alias): ...new function.
6195 (pass_build_ealias): Replace declaration with that of...
6196 (make_pass_build_ealias): ...new function.
6197 (pass_dominator): Replace declaration with that of...
6198 (make_pass_dominator): ...new function.
6199 (pass_dce): Replace declaration with that of...
6200 (make_pass_dce): ...new function.
6201 (pass_dce_loop): Replace declaration with that of...
6202 (make_pass_dce_loop): ...new function.
6203 (pass_cd_dce): Replace declaration with that of...
6204 (make_pass_cd_dce): ...new function.
6205 (pass_call_cdce): Replace declaration with that of...
6206 (make_pass_call_cdce): ...new function.
6207 (pass_merge_phi): Replace declaration with that of...
6208 (make_pass_merge_phi): ...new function.
6209 (pass_split_crit_edges): Replace declaration with that of...
6210 (make_pass_split_crit_edges): ...new function.
6211 (pass_pre): Replace declaration with that of...
6212 (make_pass_pre): ...new function.
6213 (pass_profile): Replace declaration with that of...
6214 (make_pass_profile): ...new function.
6215 (pass_strip_predict_hints): Replace declaration with that of...
6216 (make_pass_strip_predict_hints): ...new function.
6217 (pass_lower_complex_O0): Replace declaration with that of...
6218 (make_pass_lower_complex_O0): ...new function.
6219 (pass_lower_complex): Replace declaration with that of...
6220 (make_pass_lower_complex): ...new function.
6221 (pass_lower_vector): Replace declaration with that of...
6222 (make_pass_lower_vector): ...new function.
6223 (pass_lower_vector_ssa): Replace declaration with that of...
6224 (make_pass_lower_vector_ssa): ...new function.
6225 (pass_lower_omp): Replace declaration with that of...
6226 (make_pass_lower_omp): ...new function.
6227 (pass_diagnose_omp_blocks): Replace declaration with that of...
6228 (make_pass_diagnose_omp_blocks): ...new function.
6229 (pass_expand_omp): Replace declaration with that of...
6230 (make_pass_expand_omp): ...new function.
6231 (pass_expand_omp_ssa): Replace declaration with that of...
6232 (make_pass_expand_omp_ssa): ...new function.
6233 (pass_object_sizes): Replace declaration with that of...
6234 (make_pass_object_sizes): ...new function.
6235 (pass_strlen): Replace declaration with that of...
6236 (make_pass_strlen): ...new function.
6237 (pass_fold_builtins): Replace declaration with that of...
6238 (make_pass_fold_builtins): ...new function.
6239 (pass_stdarg): Replace declaration with that of...
6240 (make_pass_stdarg): ...new function.
6241 (pass_early_warn_uninitialized): Replace declaration with that of...
6242 (make_pass_early_warn_uninitialized): ...new function.
6243 (pass_late_warn_uninitialized): Replace declaration with that of...
6244 (make_pass_late_warn_uninitialized): ...new function.
6245 (pass_cse_reciprocals): Replace declaration with that of...
6246 (make_pass_cse_reciprocals): ...new function.
6247 (pass_cse_sincos): Replace declaration with that of...
6248 (make_pass_cse_sincos): ...new function.
6249 (pass_optimize_bswap): Replace declaration with that of...
6250 (make_pass_optimize_bswap): ...new function.
6251 (pass_optimize_widening_mul): Replace declaration with that of...
6252 (make_pass_optimize_widening_mul): ...new function.
6253 (pass_warn_function_return): Replace declaration with that of...
6254 (make_pass_warn_function_return): ...new function.
6255 (pass_warn_function_noreturn): Replace declaration with that of...
6256 (make_pass_warn_function_noreturn): ...new function.
6257 (pass_cselim): Replace declaration with that of...
6258 (make_pass_cselim): ...new function.
6259 (pass_phiopt): Replace declaration with that of...
6260 (make_pass_phiopt): ...new function.
6261 (pass_forwprop): Replace declaration with that of...
6262 (make_pass_forwprop): ...new function.
6263 (pass_phiprop): Replace declaration with that of...
6264 (make_pass_phiprop): ...new function.
6265 (pass_tree_ifcombine): Replace declaration with that of...
6266 (make_pass_tree_ifcombine): ...new function.
6267 (pass_dse): Replace declaration with that of...
6268 (make_pass_dse): ...new function.
6269 (pass_nrv): Replace declaration with that of...
6270 (make_pass_nrv): ...new function.
6271 (pass_rename_ssa_copies): Replace declaration with that of...
6272 (make_pass_rename_ssa_copies): ...new function.
6273 (pass_sink_code): Replace declaration with that of...
6274 (make_pass_sink_code): ...new function.
6275 (pass_fre): Replace declaration with that of...
6276 (make_pass_fre): ...new function.
6277 (pass_check_data_deps): Replace declaration with that of...
6278 (make_pass_check_data_deps): ...new function.
6279 (pass_copy_prop): Replace declaration with that of...
6280 (make_pass_copy_prop): ...new function.
6281 (pass_vrp): Replace declaration with that of...
6282 (make_pass_vrp): ...new function.
6283 (pass_uncprop): Replace declaration with that of...
6284 (make_pass_uncprop): ...new function.
6285 (pass_return_slot): Replace declaration with that of...
6286 (make_pass_return_slot): ...new function.
6287 (pass_reassoc): Replace declaration with that of...
6288 (make_pass_reassoc): ...new function.
6289 (pass_rebuild_cgraph_edges): Replace declaration with that of...
6290 (make_pass_rebuild_cgraph_edges): ...new function.
6291 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
6292 (make_pass_remove_cgraph_callee_edges): ...new function.
6293 (pass_build_cgraph_edges): Replace declaration with that of...
6294 (make_pass_build_cgraph_edges): ...new function.
6295 (pass_local_pure_const): Replace declaration with that of...
6296 (make_pass_local_pure_const): ...new function.
6297 (pass_tracer): Replace declaration with that of...
6298 (make_pass_tracer): ...new function.
6299 (pass_warn_unused_result): Replace declaration with that of...
6300 (make_pass_warn_unused_result): ...new function.
6301 (pass_diagnose_tm_blocks): Replace declaration with that of...
6302 (make_pass_diagnose_tm_blocks): ...new function.
6303 (pass_lower_tm): Replace declaration with that of...
6304 (make_pass_lower_tm): ...new function.
6305 (pass_tm_init): Replace declaration with that of...
6306 (make_pass_tm_init): ...new function.
6307 (pass_tm_mark): Replace declaration with that of...
6308 (make_pass_tm_mark): ...new function.
6309 (pass_tm_memopt): Replace declaration with that of...
6310 (make_pass_tm_memopt): ...new function.
6311 (pass_tm_edges): Replace declaration with that of...
6312 (make_pass_tm_edges): ...new function.
6313 (pass_split_functions): Replace declaration with that of...
6314 (make_pass_split_functions): ...new function.
6315 (pass_feedback_split_functions): Replace declaration with that of...
6316 (make_pass_feedback_split_functions): ...new function.
6317 (pass_strength_reduction): Replace declaration with that of...
6318 (make_pass_strength_reduction): ...new function.
6319 (pass_ipa_lower_emutls): Replace declaration with that of...
6320 (make_pass_ipa_lower_emutls): ...new function.
6321 (pass_ipa_function_and_variable_visibility): Replace declaration with
6323 (make_pass_ipa_function_and_variable_visibility): ...new function.
6324 (pass_ipa_tree_profile): Replace declaration with that of...
6325 (make_pass_ipa_tree_profile): ...new function.
6326 (pass_early_local_passes): Replace declaration with that of...
6327 (make_pass_early_local_passes): ...new function.
6328 (pass_ipa_whole_program_visibility): Replace declaration with that
6330 (make_pass_ipa_whole_program_visibility): ...new function.
6331 (pass_ipa_lto_gimple_out): Replace declaration with that of...
6332 (make_pass_ipa_lto_gimple_out): ...new function.
6333 (pass_ipa_increase_alignment): Replace declaration with that of...
6334 (make_pass_ipa_increase_alignment): ...new function.
6335 (pass_ipa_inline): Replace declaration with that of...
6336 (make_pass_ipa_inline): ...new function.
6337 (pass_ipa_free_lang_data): Replace declaration with that of...
6338 (make_pass_ipa_free_lang_data): ...new function.
6339 (pass_ipa_free_inline_summary): Replace declaration with that of...
6340 (make_pass_ipa_free_inline_summary): ...new function.
6341 (pass_ipa_cp): Replace declaration with that of...
6342 (make_pass_ipa_cp): ...new function.
6343 (pass_ipa_reference): Replace declaration with that of...
6344 (make_pass_ipa_reference): ...new function.
6345 (pass_ipa_pure_const): Replace declaration with that of...
6346 (make_pass_ipa_pure_const): ...new function.
6347 (pass_ipa_pta): Replace declaration with that of...
6348 (make_pass_ipa_pta): ...new function.
6349 (pass_ipa_lto_finish_out): Replace declaration with that of...
6350 (make_pass_ipa_lto_finish_out): ...new function.
6351 (pass_ipa_tm): Replace declaration with that of...
6352 (make_pass_ipa_tm): ...new function.
6353 (pass_ipa_profile): Replace declaration with that of...
6354 (make_pass_ipa_profile): ...new function.
6355 (pass_ipa_cdtor_merge): Replace declaration with that of...
6356 (make_pass_ipa_cdtor_merge): ...new function.
6357 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
6359 (make_pass_cleanup_cfg_post_optimizing): ...new function.
6360 (pass_init_datastructures): Replace declaration with that of...
6361 (make_pass_init_datastructures): ...new function.
6362 (pass_fixup_cfg): Replace declaration with that of...
6363 (make_pass_fixup_cfg): ...new function.
6364 (pass_expand): Replace declaration with that of...
6365 (make_pass_expand): ...new function.
6366 (pass_instantiate_virtual_regs): Replace declaration with that of...
6367 (make_pass_instantiate_virtual_regs): ...new function.
6368 (pass_rtl_fwprop): Replace declaration with that of...
6369 (make_pass_rtl_fwprop): ...new function.
6370 (pass_rtl_fwprop_addr): Replace declaration with that of...
6371 (make_pass_rtl_fwprop_addr): ...new function.
6372 (pass_jump): Replace declaration with that of...
6373 (make_pass_jump): ...new function.
6374 (pass_jump2): Replace declaration with that of...
6375 (make_pass_jump2): ...new function.
6376 (pass_lower_subreg): Replace declaration with that of...
6377 (make_pass_lower_subreg): ...new function.
6378 (pass_cse): Replace declaration with that of...
6379 (make_pass_cse): ...new function.
6380 (pass_fast_rtl_dce): Replace declaration with that of...
6381 (make_pass_fast_rtl_dce): ...new function.
6382 (pass_ud_rtl_dce): Replace declaration with that of...
6383 (make_pass_ud_rtl_dce): ...new function.
6384 (pass_rtl_dce): Replace declaration with that of...
6385 (make_pass_rtl_dce): ...new function.
6386 (pass_rtl_dse1): Replace declaration with that of...
6387 (make_pass_rtl_dse1): ...new function.
6388 (pass_rtl_dse2): Replace declaration with that of...
6389 (make_pass_rtl_dse2): ...new function.
6390 (pass_rtl_dse3): Replace declaration with that of...
6391 (make_pass_rtl_dse3): ...new function.
6392 (pass_rtl_cprop): Replace declaration with that of...
6393 (make_pass_rtl_cprop): ...new function.
6394 (pass_rtl_pre): Replace declaration with that of...
6395 (make_pass_rtl_pre): ...new function.
6396 (pass_rtl_hoist): Replace declaration with that of...
6397 (make_pass_rtl_hoist): ...new function.
6398 (pass_rtl_store_motion): Replace declaration with that of...
6399 (make_pass_rtl_store_motion): ...new function.
6400 (pass_cse_after_global_opts): Replace declaration with that of...
6401 (make_pass_cse_after_global_opts): ...new function.
6402 (pass_rtl_ifcvt): Replace declaration with that of...
6403 (make_pass_rtl_ifcvt): ...new function.
6404 (pass_into_cfg_layout_mode): Replace declaration with that of...
6405 (make_pass_into_cfg_layout_mode): ...new function.
6406 (pass_outof_cfg_layout_mode): Replace declaration with that of...
6407 (make_pass_outof_cfg_layout_mode): ...new function.
6408 (pass_loop2): Replace declaration with that of...
6409 (make_pass_loop2): ...new function.
6410 (pass_rtl_loop_init): Replace declaration with that of...
6411 (make_pass_rtl_loop_init): ...new function.
6412 (pass_rtl_move_loop_invariants): Replace declaration with that of...
6413 (make_pass_rtl_move_loop_invariants): ...new function.
6414 (pass_rtl_unswitch): Replace declaration with that of...
6415 (make_pass_rtl_unswitch): ...new function.
6416 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
6417 (make_pass_rtl_unroll_and_peel_loops): ...new function.
6418 (pass_rtl_doloop): Replace declaration with that of...
6419 (make_pass_rtl_doloop): ...new function.
6420 (pass_rtl_loop_done): Replace declaration with that of...
6421 (make_pass_rtl_loop_done): ...new function.
6422 (pass_web): Replace declaration with that of...
6423 (make_pass_web): ...new function.
6424 (pass_cse2): Replace declaration with that of...
6425 (make_pass_cse2): ...new function.
6426 (pass_df_initialize_opt): Replace declaration with that of...
6427 (make_pass_df_initialize_opt): ...new function.
6428 (pass_df_initialize_no_opt): Replace declaration with that of...
6429 (make_pass_df_initialize_no_opt): ...new function.
6430 (pass_reginfo_init): Replace declaration with that of...
6431 (make_pass_reginfo_init): ...new function.
6432 (pass_inc_dec): Replace declaration with that of...
6433 (make_pass_inc_dec): ...new function.
6434 (pass_stack_ptr_mod): Replace declaration with that of...
6435 (make_pass_stack_ptr_mod): ...new function.
6436 (pass_initialize_regs): Replace declaration with that of...
6437 (make_pass_initialize_regs): ...new function.
6438 (pass_combine): Replace declaration with that of...
6439 (make_pass_combine): ...new function.
6440 (pass_if_after_combine): Replace declaration with that of...
6441 (make_pass_if_after_combine): ...new function.
6442 (pass_ree): Replace declaration with that of...
6443 (make_pass_ree): ...new function.
6444 (pass_partition_blocks): Replace declaration with that of...
6445 (make_pass_partition_blocks): ...new function.
6446 (pass_match_asm_constraints): Replace declaration with that of...
6447 (make_pass_match_asm_constraints): ...new function.
6448 (pass_regmove): Replace declaration with that of...
6449 (make_pass_regmove): ...new function.
6450 (pass_split_all_insns): Replace declaration with that of...
6451 (make_pass_split_all_insns): ...new function.
6452 (pass_fast_rtl_byte_dce): Replace declaration with that of...
6453 (make_pass_fast_rtl_byte_dce): ...new function.
6454 (pass_lower_subreg2): Replace declaration with that of...
6455 (make_pass_lower_subreg2): ...new function.
6456 (pass_mode_switching): Replace declaration with that of...
6457 (make_pass_mode_switching): ...new function.
6458 (pass_sms): Replace declaration with that of...
6459 (make_pass_sms): ...new function.
6460 (pass_sched): Replace declaration with that of...
6461 (make_pass_sched): ...new function.
6462 (pass_ira): Replace declaration with that of...
6463 (make_pass_ira): ...new function.
6464 (pass_reload): Replace declaration with that of...
6465 (make_pass_reload): ...new function.
6466 (pass_clean_state): Replace declaration with that of...
6467 (make_pass_clean_state): ...new function.
6468 (pass_branch_prob): Replace declaration with that of...
6469 (make_pass_branch_prob): ...new function.
6470 (pass_value_profile_transformations): Replace declaration with that
6472 (make_pass_value_profile_transformations): ...new function.
6473 (pass_postreload_cse): Replace declaration with that of...
6474 (make_pass_postreload_cse): ...new function.
6475 (pass_gcse2): Replace declaration with that of...
6476 (make_pass_gcse2): ...new function.
6477 (pass_split_after_reload): Replace declaration with that of...
6478 (make_pass_split_after_reload): ...new function.
6479 (pass_branch_target_load_optimize1): Replace declaration with that
6481 (make_pass_branch_target_load_optimize1): ...new function.
6482 (pass_thread_prologue_and_epilogue): Replace declaration with that
6484 (make_pass_thread_prologue_and_epilogue): ...new function.
6485 (pass_stack_adjustments): Replace declaration with that of...
6486 (make_pass_stack_adjustments): ...new function.
6487 (pass_peephole2): Replace declaration with that of...
6488 (make_pass_peephole2): ...new function.
6489 (pass_if_after_reload): Replace declaration with that of...
6490 (make_pass_if_after_reload): ...new function.
6491 (pass_regrename): Replace declaration with that of...
6492 (make_pass_regrename): ...new function.
6493 (pass_cprop_hardreg): Replace declaration with that of...
6494 (make_pass_cprop_hardreg): ...new function.
6495 (pass_reorder_blocks): Replace declaration with that of...
6496 (make_pass_reorder_blocks): ...new function.
6497 (pass_branch_target_load_optimize2): Replace declaration with that
6499 (make_pass_branch_target_load_optimize2): ...new function.
6500 (pass_leaf_regs): Replace declaration with that of...
6501 (make_pass_leaf_regs): ...new function.
6502 (pass_split_before_sched2): Replace declaration with that of...
6503 (make_pass_split_before_sched2): ...new function.
6504 (pass_compare_elim_after_reload): Replace declaration with that of...
6505 (make_pass_compare_elim_after_reload): ...new function.
6506 (pass_sched2): Replace declaration with that of...
6507 (make_pass_sched2): ...new function.
6508 (pass_stack_regs): Replace declaration with that of...
6509 (make_pass_stack_regs): ...new function.
6510 (pass_stack_regs_run): Replace declaration with that of...
6511 (make_pass_stack_regs_run): ...new function.
6512 (pass_df_finish): Replace declaration with that of...
6513 (make_pass_df_finish): ...new function.
6514 (pass_compute_alignments): Replace declaration with that of...
6515 (make_pass_compute_alignments): ...new function.
6516 (pass_duplicate_computed_gotos): Replace declaration with that of...
6517 (make_pass_duplicate_computed_gotos): ...new function.
6518 (pass_variable_tracking): Replace declaration with that of...
6519 (make_pass_variable_tracking): ...new function.
6520 (pass_free_cfg): Replace declaration with that of...
6521 (make_pass_free_cfg): ...new function.
6522 (pass_machine_reorg): Replace declaration with that of...
6523 (make_pass_machine_reorg): ...new function.
6524 (pass_cleanup_barriers): Replace declaration with that of...
6525 (make_pass_cleanup_barriers): ...new function.
6526 (pass_delay_slots): Replace declaration with that of...
6527 (make_pass_delay_slots): ...new function.
6528 (pass_split_for_shorten_branches): Replace declaration with that of...
6529 (make_pass_split_for_shorten_branches): ...new function.
6530 (pass_split_before_regstack): Replace declaration with that of...
6531 (make_pass_split_before_regstack): ...new function.
6532 (pass_convert_to_eh_region_ranges): Replace declaration with that
6534 (make_pass_convert_to_eh_region_ranges): ...new function.
6535 (pass_shorten_branches): Replace declaration with that of...
6536 (make_pass_shorten_branches): ...new function.
6537 (pass_set_nothrow_function_flags): Replace declaration with that of...
6538 (make_pass_set_nothrow_function_flags): ...new function.
6539 (pass_dwarf2_frame): Replace declaration with that of...
6540 (make_pass_dwarf2_frame): ...new function.
6541 (pass_final): Replace declaration with that of...
6542 (make_pass_final): ...new function.
6543 (pass_rtl_seqabstr): Replace declaration with that of...
6544 (make_pass_rtl_seqabstr): ...new function.
6545 (pass_release_ssa_names): Replace declaration with that of...
6546 (make_pass_release_ssa_names): ...new function.
6547 (pass_early_inline): Replace declaration with that of...
6548 (make_pass_early_inline): ...new function.
6549 (pass_inline_parameters): Replace declaration with that of...
6550 (make_pass_inline_parameters): ...new function.
6551 (pass_update_address_taken): Replace declaration with that of...
6552 (make_pass_update_address_taken): ...new function.
6553 (pass_convert_switch): Replace declaration with that of...
6554 (make_pass_convert_switch): ...new function.
6555 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
6556 to a subclass of simple_ipa_opt_pass along with...
6557 (pass_data_ipa_tree_profile): ...new pass_data instance and...
6558 (make_pass_ipa_tree_profile): ...new function.
6559 * tree-sra.c (pass_sra_early): Convert from a global struct to a
6560 subclass of gimple_opt_pass along with...
6561 (pass_data_sra_early): ...new pass_data instance and...
6562 (make_pass_sra_early): ...new function.
6563 (pass_sra): Convert from a global struct to a subclass of
6564 gimple_opt_pass along with...
6565 (pass_data_sra): ...new pass_data instance and...
6566 (make_pass_sra): ...new function.
6567 (pass_early_ipa_sra): Convert from a global struct to a subclass of
6568 gimple_opt_pass along with...
6569 (pass_data_early_ipa_sra): ...new pass_data instance and...
6570 (make_pass_early_ipa_sra): ...new function.
6571 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
6572 subclass of gimple_opt_pass along with...
6573 (pass_data_ccp): ...new pass_data instance and...
6574 (make_pass_ccp): ...new function.
6575 (pass_fold_builtins): Convert from a global struct to a subclass of
6576 gimple_opt_pass along with...
6577 (pass_data_fold_builtins): ...new pass_data instance and...
6578 (make_pass_fold_builtins): ...new function.
6579 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
6580 subclass of gimple_opt_pass along with...
6581 (pass_data_copy_prop): ...new pass_data instance and...
6582 (make_pass_copy_prop): ...new function.
6583 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
6584 global struct to a subclass of gimple_opt_pass along with...
6585 (pass_data_rename_ssa_copies): ...new pass_data instance and...
6586 (make_pass_rename_ssa_copies): ...new function.
6587 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
6588 subclass of gimple_opt_pass along with...
6589 (pass_data_dce): ...new pass_data instance and...
6590 (make_pass_dce): ...new function.
6591 (pass_dce_loop): Convert from a global struct to a subclass of
6592 gimple_opt_pass along with...
6593 (pass_data_dce_loop): ...new pass_data instance and...
6594 (make_pass_dce_loop): ...new function.
6595 (pass_cd_dce): Convert from a global struct to a subclass of
6596 gimple_opt_pass along with...
6597 (pass_data_cd_dce): ...new pass_data instance and...
6598 (make_pass_cd_dce): ...new function.
6599 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
6600 subclass of gimple_opt_pass along with...
6601 (pass_data_dominator): ...new pass_data instance and...
6602 (make_pass_dominator): ...new function.
6603 (pass_phi_only_cprop): Convert from a global struct to a subclass of
6604 gimple_opt_pass along with...
6605 (pass_data_phi_only_cprop): ...new pass_data instance and...
6606 (make_pass_phi_only_cprop): ...new function.
6607 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
6608 subclass of gimple_opt_pass along with...
6609 (pass_data_dse): ...new pass_data instance and...
6610 (make_pass_dse): ...new function.
6611 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
6612 a subclass of gimple_opt_pass along with...
6613 (pass_data_forwprop): ...new pass_data instance and...
6614 (make_pass_forwprop): ...new function.
6615 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
6616 struct to a subclass of gimple_opt_pass along with...
6617 (pass_data_tree_ifcombine): ...new pass_data instance and...
6618 (make_pass_tree_ifcombine): ...new function.
6619 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
6620 subclass of gimple_opt_pass along with...
6621 (pass_data_ch): ...new pass_data instance and...
6622 (make_pass_ch): ...new function.
6623 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
6624 subclass of gimple_opt_pass along with...
6625 (pass_data_tree_loop): ...new pass_data instance and...
6626 (make_pass_tree_loop): ...new function.
6627 (pass_tree_loop_init): Convert from a global struct to a subclass of
6628 gimple_opt_pass along with...
6629 (pass_data_tree_loop_init): ...new pass_data instance and...
6630 (make_pass_tree_loop_init): ...new function.
6631 (pass_lim): Convert from a global struct to a subclass of
6632 gimple_opt_pass along with...
6633 (pass_data_lim): ...new pass_data instance and...
6634 (make_pass_lim): ...new function.
6635 (pass_tree_unswitch): Convert from a global struct to a subclass of
6636 gimple_opt_pass along with...
6637 (pass_data_tree_unswitch): ...new pass_data instance and...
6638 (make_pass_tree_unswitch): ...new function.
6639 (pass_predcom): Convert from a global struct to a subclass of
6640 gimple_opt_pass along with...
6641 (pass_data_predcom): ...new pass_data instance and...
6642 (make_pass_predcom): ...new function.
6643 (pass_vectorize): Convert from a global struct to a subclass of
6644 gimple_opt_pass along with...
6645 (pass_data_vectorize): ...new pass_data instance and...
6646 (make_pass_vectorize): ...new function.
6647 (pass_graphite): Convert from a global struct to a subclass of
6648 gimple_opt_pass along with...
6649 (pass_data_graphite): ...new pass_data instance and...
6650 (make_pass_graphite): ...new function.
6651 (pass_graphite_transforms): Convert from a global struct to a subclass
6652 of gimple_opt_pass along with...
6653 (pass_data_graphite_transforms): ...new pass_data instance and...
6654 (make_pass_graphite_transforms): ...new function.
6655 (pass_check_data_deps): Convert from a global struct to a subclass of
6656 gimple_opt_pass along with...
6657 (pass_data_check_data_deps): ...new pass_data instance and...
6658 (make_pass_check_data_deps): ...new function.
6659 (pass_iv_canon): Convert from a global struct to a subclass of
6660 gimple_opt_pass along with...
6661 (pass_data_iv_canon): ...new pass_data instance and...
6662 (make_pass_iv_canon): ...new function.
6663 (pass_scev_cprop): Convert from a global struct to a subclass of
6664 gimple_opt_pass along with...
6665 (pass_data_scev_cprop): ...new pass_data instance and...
6666 (make_pass_scev_cprop): ...new function.
6667 (pass_record_bounds): Convert from a global struct to a subclass of
6668 gimple_opt_pass along with...
6669 (pass_data_record_bounds): ...new pass_data instance and...
6670 (make_pass_record_bounds): ...new function.
6671 (pass_complete_unroll): Convert from a global struct to a subclass of
6672 gimple_opt_pass along with...
6673 (pass_data_complete_unroll): ...new pass_data instance and...
6674 (make_pass_complete_unroll): ...new function.
6675 (pass_complete_unrolli): Convert from a global struct to a subclass of
6676 gimple_opt_pass along with...
6677 (pass_data_complete_unrolli): ...new pass_data instance and...
6678 (make_pass_complete_unrolli): ...new function.
6679 (pass_parallelize_loops): Convert from a global struct to a subclass
6680 of gimple_opt_pass along with...
6681 (pass_data_parallelize_loops): ...new pass_data instance and...
6682 (make_pass_parallelize_loops): ...new function.
6683 (pass_loop_prefetch): Convert from a global struct to a subclass of
6684 gimple_opt_pass along with...
6685 (pass_data_loop_prefetch): ...new pass_data instance and...
6686 (make_pass_loop_prefetch): ...new function.
6687 (pass_iv_optimize): Convert from a global struct to a subclass of
6688 gimple_opt_pass along with...
6689 (pass_data_iv_optimize): ...new pass_data instance and...
6690 (make_pass_iv_optimize): ...new function.
6691 (pass_tree_loop_done): Convert from a global struct to a subclass of
6692 gimple_opt_pass along with...
6693 (pass_data_tree_loop_done): ...new pass_data instance and...
6694 (make_pass_tree_loop_done): ...new function.
6695 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
6696 struct to a subclass of gimple_opt_pass along with...
6697 (pass_data_cse_reciprocals): ...new pass_data instance and...
6698 (make_pass_cse_reciprocals): ...new function.
6699 (pass_cse_sincos): Convert from a global struct to a subclass of
6700 gimple_opt_pass along with...
6701 (pass_data_cse_sincos): ...new pass_data instance and...
6702 (make_pass_cse_sincos): ...new function.
6703 (pass_optimize_bswap): Convert from a global struct to a subclass of
6704 gimple_opt_pass along with...
6705 (pass_data_optimize_bswap): ...new pass_data instance and...
6706 (make_pass_optimize_bswap): ...new function.
6707 (pass_optimize_widening_mul): Convert from a global struct to a
6708 subclass of gimple_opt_pass along with...
6709 (pass_data_optimize_widening_mul): ...new pass_data instance and...
6710 (make_pass_optimize_widening_mul): ...new function.
6711 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
6712 subclass of gimple_opt_pass along with...
6713 (pass_data_phiopt): ...new pass_data instance and...
6714 (make_pass_phiopt): ...new function.
6715 (pass_cselim): Convert from a global struct to a subclass of
6716 gimple_opt_pass along with...
6717 (pass_data_cselim): ...new pass_data instance and...
6718 (make_pass_cselim): ...new function.
6719 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
6720 subclass of gimple_opt_pass along with...
6721 (pass_data_phiprop): ...new pass_data instance and...
6722 (make_pass_phiprop): ...new function.
6723 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
6724 subclass of gimple_opt_pass along with...
6725 (pass_data_pre): ...new pass_data instance and...
6726 (make_pass_pre): ...new function.
6727 (pass_fre): Convert from a global struct to a subclass of
6728 gimple_opt_pass along with...
6729 (pass_data_fre): ...new pass_data instance and...
6730 (make_pass_fre): ...new function.
6731 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
6732 subclass of gimple_opt_pass along with...
6733 (pass_data_reassoc): ...new pass_data instance and...
6734 (make_pass_reassoc): ...new function.
6735 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
6736 subclass of gimple_opt_pass along with...
6737 (pass_data_sink_code): ...new pass_data instance and...
6738 (make_pass_sink_code): ...new function.
6739 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
6740 subclass of gimple_opt_pass along with...
6741 (pass_data_strlen): ...new pass_data instance and...
6742 (make_pass_strlen): ...new function.
6743 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
6744 struct to a subclass of gimple_opt_pass along with...
6745 (pass_data_build_alias): ...new pass_data instance and...
6746 (make_pass_build_alias): ...new function.
6747 (pass_build_ealias): Convert from a global struct to a subclass of
6748 gimple_opt_pass along with...
6749 (pass_data_build_ealias): ...new pass_data instance and...
6750 (make_pass_build_ealias): ...new function.
6751 (pass_ipa_pta): Convert from a global struct to a subclass of
6752 simple_ipa_opt_pass along with...
6753 (pass_data_ipa_pta): ...new pass_data instance and...
6754 (make_pass_ipa_pta): ...new function.
6755 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
6756 subclass of gimple_opt_pass along with...
6757 (pass_data_uncprop): ...new pass_data instance and...
6758 (make_pass_uncprop): ...new function.
6759 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
6760 global struct to a subclass of gimple_opt_pass along with...
6761 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
6762 (make_pass_late_warn_uninitialized): ...new function.
6763 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
6764 to a subclass of gimple_opt_pass along with...
6765 (pass_data_init_datastructures): ...new pass_data instance and...
6766 (make_pass_init_datastructures): ...new function.
6767 (pass_early_warn_uninitialized): Convert from a global struct to a
6768 subclass of gimple_opt_pass along with...
6769 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
6770 (make_pass_early_warn_uninitialized): ...new function.
6771 (pass_update_address_taken): Convert from a global struct to a
6772 subclass of gimple_opt_pass along with...
6773 (pass_data_update_address_taken): ...new pass_data instance and...
6774 (make_pass_update_address_taken): ...new function.
6775 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
6776 struct to a subclass of gimple_opt_pass along with...
6777 (pass_data_release_ssa_names): ...new pass_data instance and...
6778 (make_pass_release_ssa_names): ...new function.
6779 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
6780 subclass of gimple_opt_pass along with...
6781 (pass_data_stdarg): ...new pass_data instance and...
6782 (make_pass_stdarg): ...new function.
6783 * tree-switch-conversion.c (pass_convert_switch): Convert from a
6784 global struct to a subclass of gimple_opt_pass along with...
6785 (pass_data_convert_switch): ...new pass_data instance and...
6786 (make_pass_convert_switch): ...new function.
6787 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
6788 to a subclass of gimple_opt_pass along with...
6789 (pass_data_tail_recursion): ...new pass_data instance and...
6790 (make_pass_tail_recursion): ...new function.
6791 (pass_tail_calls): Convert from a global struct to a subclass of
6792 gimple_opt_pass along with...
6793 (pass_data_tail_calls): ...new pass_data instance and...
6794 (make_pass_tail_calls): ...new function.
6795 * tree-vect-generic.c (pass_lower_vector): Convert from a global
6796 struct to a subclass of gimple_opt_pass along with...
6797 (pass_data_lower_vector): ...new pass_data instance and...
6798 (make_pass_lower_vector): ...new function.
6799 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
6800 gimple_opt_pass along with...
6801 (pass_data_lower_vector_ssa): ...new pass_data instance and...
6802 (make_pass_lower_vector_ssa): ...new function.
6803 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
6804 to a subclass of gimple_opt_pass along with...
6805 (pass_data_slp_vectorize): ...new pass_data instance and...
6806 (make_pass_slp_vectorize): ...new function.
6807 (pass_ipa_increase_alignment): Convert from a global struct to a
6808 subclass of simple_ipa_opt_pass along with...
6809 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
6810 (make_pass_ipa_increase_alignment): ...new function.
6811 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
6812 gimple_opt_pass along with...
6813 (pass_data_vrp): ...new pass_data instance and...
6814 (make_pass_vrp): ...new function.
6815 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
6816 subclass of simple_ipa_opt_pass along with...
6817 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
6818 (make_pass_ipa_free_lang_data): ...new function.
6819 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
6820 gimple_opt_pass along with...
6821 (pass_data_tsan): ...new pass_data instance and...
6822 (make_pass_tsan): ...new function.
6823 (pass_tsan_O0): Convert from a global struct to a subclass of
6824 gimple_opt_pass along with...
6825 (pass_data_tsan_O0): ...new pass_data instance and...
6826 (make_pass_tsan_O0): ...new function.
6827 * var-tracking.c (pass_variable_tracking): Convert from a global
6828 struct to a subclass of rtl_opt_pass along with...
6829 (pass_data_variable_tracking): ...new pass_data instance and...
6830 (make_pass_variable_tracking): ...new function.
6831 * web.c (pass_web): Convert from a global struct to a subclass of
6832 rtl_opt_pass along with...
6833 (pass_data_web): ...new pass_data instance and...
6834 (make_pass_web): ...new function.
6835 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
6836 declaration with that of...
6837 (make_pass_mode_switch_use): ...new function.
6838 (pass_resolve_sw_modes): Replace declaration with that of...
6839 (make_pass_resolve_sw_modes): ...new function.
6840 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
6841 from a global struct to a subclass of rtl_opt_pass along with...
6842 (pass_data_mode_switch_use): ...new pass_data instance and...
6843 (make_pass_mode_switch_use): ...new function.
6844 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
6845 from a global struct to a subclass of rtl_opt_pass along with...
6846 (pass_data_resolve_sw_modes): ...new pass_data instance and...
6847 (make_pass_resolve_sw_modes): ...new function.
6848 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
6849 struct to a subclass of rtl_opt_pass along with...
6850 (pass_data_insert_vzeroupper): ...new pass_data instance and...
6851 (make_pass_insert_vzeroupper): ...new function.
6852 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
6853 global struct to a subclass of rtl_opt_pass along with...
6854 (pass_data_work_around_errata): ...new pass_data instance and...
6855 (make_pass_work_around_errata): ...new function.
6856 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
6857 struct to a subclass of rtl_opt_pass along with...
6858 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
6859 (make_pass_mips_machine_reorg2): ...new function.
6861 2013-08-05 David Malcolm <dmalcolm@redhat.com>
6863 * passes.c (pass_manager::operator new): New.
6865 2013-08-05 David Malcolm <dmalcolm@redhat.com>
6867 Handwritten part of conversion of passes to C++ classes.
6869 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
6870 (toplev.o): Add dep on PASS_MANAGER_H.
6871 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
6872 of early local pases to reflect this moving from a global to a
6873 member of gcc::pass_manager.
6874 (cgraph_add_new_function): Likewise.
6875 * lto-cgraph.c (lto_output_node): Update for conversion of
6876 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
6877 * passes.c (opt_pass::clone): New.
6878 (opt_pass::gate): New.
6879 (opt_pass::execute): New.
6880 (opt_pass::opt_pass): New.
6881 (pass_manager::execute_early_local_passes): New.
6882 (pass_manager::execute_pass_mode_switching): new.
6883 (finish_optimization_passes): Convert to...
6884 (pass_manager::finish_optimization_passes): ...this.
6885 (finish_optimization_passes): Update for conversion of passes to
6887 (register_dump_files_1): Use has_gate since we cannot portably
6888 check a vtable entry against NULL.
6889 (dump_one_pass): Likewise.
6890 (ipa_write_summaries_2): Likewise.
6891 (ipa_write_optimization_summaries_1): Likewise.
6892 (ipa_read_summaries_1): Likewise.
6893 (ipa_read_optimization_summaries_1): Likewise.
6894 (execute_ipa_stmt_fixups): Likewise.
6895 (pass_manager::pass_manager): Rewrite pass-creation, invoking
6896 pass-creation functions rather than wiring up globals, and
6897 storing the results in fields of pass_manager generated using
6899 (pass_manager::dump_profile_report): Update for conversion of
6900 passes to C++ classes.
6901 (pass_manager::execute_ipa_summary_passes): Likewise.
6902 (execute_one_ipa_transform_pass): Likewise.
6903 (execute_one_pass): Use has_gate and has_execute since we cannot
6904 portably check a vtable entry against NULL.
6905 * pass_manager.h (pass_manager::finish_optimization_passes): New.
6906 (pass_manager): Use pass-instances.def to add fields for the
6907 various pass instances.
6908 * toplev.c (finalize): Update for move of
6909 finish_optimization_passes to a method of gcc::pass_manager.
6910 * toplev.h (finish_optimization_passes): Move to method of class
6912 * tree-pass.h (struct pass_data): New.
6913 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
6914 (opt_pass::gate): Convert to virtual function.
6915 (opt_pass::~opt_pass): New.
6916 (opt_pass::clone): New.
6917 (opt_pass::execute): Convert to virtual function.
6918 (opt_pass::opt_pass): New.
6919 (opt_pass::ctxt_): new.
6920 (gimple_opt_pass): Convert to subclass of opt_pass.
6921 (gimple_opt_pass::gimple_opt_pass): New.
6922 (rtl_opt_pass): Convert to subclass of opt_pass.
6923 (rtl_opt_pass::rtl_opt_pass): New.
6924 (ipa_opt_pass_d): Convert to subclass of opt_pass.
6925 (ipa_opt_pass_d::ipa_opt_pass_d): New.
6926 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
6927 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
6928 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
6929 invocation of pass_mode_switching to reflect this moving from a
6930 global to a member of gcc::pass_manager.
6931 (ix86_option_override): Rework how pass_insert_vzeroupper is
6932 added to the pass_manager to reflect autogenerated changes.
6933 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
6935 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
6937 PR rtl-optimization/57708
6938 * recog.c (peep2_find_free_register): Validate all regs in a
6941 2013-08-05 Jan Hubicka <jh@suse.cz>
6944 * cgraph.c (verify_cgraph_node): Accept local flags from other
6946 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
6947 (function_and_variable_visibility): Likewise.
6948 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
6950 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6952 * graph.c (init_graph_slim_pretty_print): Remove.
6953 (print_graph_cfg): Do not call it. Use local pretty printer.
6954 (start_graph_dump): Likewise.
6956 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6958 * gimple-pretty-print.c (buffer): Remove.
6959 (initialized): Likewise.
6960 (maybe_init_pretty_print): Likewise.
6961 (print_gimple_stmt): Do not call it. Use non-static local
6962 pretty_printer variable.
6963 (print_gimple_expr): Likewise.
6964 (print_gimple_seq): Likewise.
6965 (gimple_dump_bb): Likewise.
6967 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6969 * asan.c (asan_pp): Remove.
6970 (asan_pp_initialized): Likewise.
6971 (asan_pp_initialize): Likewise.
6972 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
6973 (asan_emit_stack_protection): Tidy. Use local pretty printer.
6974 (asan_add_global): Likewise.
6976 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
6978 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
6979 * diagnostic.h (diagnostic_flush_buffer): Adjust.
6980 * pretty-print.c (pp_formatted_text_data): Likewise.
6981 (pp_indent): Rename from pp_base_indent.
6982 (pp_format): Rename from pp_base_format.
6983 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
6984 (pp_format_verbatim): Rename from pp_base_format_verbatim.
6985 (pp_flush): Rename from pp_base_flush.
6986 (pp_set_line_maximum_length): Rename from
6987 pp_base_set_line_maximum_length.
6988 (pp_clear_output_area): Rename from pp_base_clear_output_area.
6989 (pp_set_prefix): Rename from pp_base_set_prefix.
6990 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
6991 (pp_emit_prefix): Rename from pp_base_emit_prefix.
6992 (pp_append_text): Rename from pp_base_append_text.
6993 (pp_formatted_text): Rename from pp_base_formatted_text.
6994 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
6995 (pp_remaining_character_count_for_line): Rename from
6996 pp_base_remaining_character_count_for_line.
6997 (pp_newline): Rename from pp_base_newline.
6998 (pp_character): Rename from pp_base_character.
6999 (pp_string): Rename from pp_base_string.
7000 (pp_maybe_space): Rename from pp_base_maybe_space.
7001 * asan.c (asan_pp_string): Adjust.
7002 (asan_emit_stack_protection): Likewise.
7003 (asan_add_global): Likewise.
7004 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
7005 * tree-mudflap.c (mf_varname_tree): Likewise.
7006 * tree-pretty-print.c (pp_tree_identifier): Rename from
7007 pp_base_tree_identifier.
7008 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
7009 Declare as function.
7011 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
7013 * pretty-print.h (pp_bar_bar): New.
7014 (pp_ampersand_ampersand): Likewise.
7015 (pp_less_equal): Likewise.
7016 (pp_greater_equal): Likewise.
7017 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
7018 printer functions instead of pp_string or operators and punctuators.
7019 (dump_gimple_call): Likewise.
7020 (dump_gimple_omp_for): Likewise.
7021 (dump_gimple_transaction): Likewise.
7022 (dump_gimple_phi): Likewise.
7023 (pp_gimple_stmt_1): Likewise.
7024 * sched-vis.c (print_insn): Likewise.
7025 * tree-mudflap.c (mf_varname_tree): Likewise.
7026 * tree-pretty-print.c (dump_block_node): Likewise.
7027 (dump_generic_node): Likewise.
7029 2013-08-02 Jan Hubicka <jh@suse.cz>
7031 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
7033 * lto-streamer-out.c (tree_is_indexable): Results decls and
7034 parm decls are not indexable.
7035 (DFS_write_tree_body): Do not follow args and results.
7036 (hash_tree): Likewise.
7037 (output_functions): Rearrange so struct function is needed
7038 only when real body is output; be able to also ouptut abstract
7039 functions; output DECL_ARGUMENTS and DECL_RESULT.
7040 (lto_output): When not in WPA, ale store abstract functions.
7041 (write_symbol): Do not care about RESULT_DECL.
7042 (output_symbol_p): Handle correctly sbtract decls.
7043 * lto-streamer-in.c (input_function): Rearrange so struct
7044 function can be NULL at entry; allow streaming of
7045 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
7046 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
7047 sanity check during LTO.
7048 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
7049 RESULT_DECl and DECL_ARGUMENTS.
7050 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
7053 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
7055 * pretty-print.h (pp_underscore): New.
7057 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
7058 printer functions instead of pp_character.
7059 (dump_binary_rhs): Likewise.
7060 (dump_ternary_rhs): Likewise.
7061 (dump_gimple_call_args): Likewise.
7062 (pp_points_to_solution): Likewise.
7063 (dump_gimple_call): Likewise.
7064 (dump_gimple_switch): Likewise.
7065 (dump_gimple_cond): Likewise.
7066 (dump_gimple_bind): Likewise.
7067 (dump_gimple_try): Likewise.
7068 (dump_gimple_omp_for): Likewise.
7069 (dump_gimple_omp_continue): Likewise.
7070 (dump_gimple_omp_single): Likewise.
7071 (dump_gimple_omp_sections): Likewise.
7072 (dump_gimple_omp_block): Likewise.
7073 (dump_gimple_omp_critical): Likewise.
7074 (dump_gimple_transaction): Likewise.
7075 (dump_gimple_asm): Likewise.
7076 (dump_gimple_phi): Likewise.
7077 (dump_gimple_omp_parallel): Likewise.
7078 (dump_gimple_omp_task): Likewise.
7079 (dump_gimple_omp_atomic_load): Likewise.
7080 (dump_gimple_omp_atomic_store): Likewise.
7081 (dump_gimple_mem_ops): Likewise.
7082 (pp_gimple_stmt_1): Likewise.
7083 (pp_cfg_jump): Likewise.
7084 (dump_implicit_edges): Likewise.
7085 (gimple_dump_bb_for_graph): Likewise.
7086 * graph.c (draw_cfg_node): Likewise.
7087 * langhooks.c (lhd_print_error_function): Likewise.
7088 * sched-vis.c (print_exp): Likewise.
7089 (print_value): Likewise.
7090 (print_pattern): Likewise.
7091 (print_insn): Likewise.
7092 (rtl_dump_bb_for_graph): Likewise.
7093 * tree-pretty-print.c (dump_function_declaration): Likewise.
7094 (dump_array_domain): Likewise.
7095 (dump_omp_clause): Likewise.
7096 (dump_location): Likewise.
7097 (dump_generic_node): Likewise.
7098 (print_struct_decl): Likewise.
7099 * diagnostic.c (diagnostic_show_locus): Use pp_space.
7101 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
7103 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
7104 candidate table when replacing a candidate statement.
7105 (replace_rhs_if_not_dup): Likewise.
7106 (replace_one_candidate): Likewise.
7108 2013-08-02 Jan Hubicka <jh@suse.cz>
7109 Martin Liska <marxin.liska@gmail.com>
7111 * cgraphunit.c (add_new_function): Fix logic when adding from
7113 (assemble_thunk): Rename to ...
7114 (expand_thunk); .. this one; export; get it working with
7115 general functions; make produced gimple valid.
7116 * cgraph.h (expand_thunk): Declare.
7118 2013-08-02 Jan Hubicka <jh@suse.cz>
7120 * ipa-cp.c (gather_context_independent_values): Use
7121 ipa_get_param_move_cost.
7122 (get_replacement_map): Remove PARAM; move parameter folding
7124 (create_specialized_node): Update.
7125 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
7126 assert that we have gimple body; update move_cost.
7127 (count_formal_params): Assert that we have gimple body.
7128 (ipa_dump_param): New function.
7129 (ipa_alloc_node_params): Break out from ...
7130 (ipa_initialize_node_params): ... here.
7131 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
7132 (ipa_write_node_info): Stream move costs.
7133 (ipa_read_node_info): Read move costs.
7134 (ipa_update_after_lto_read): Do not recompute node params.
7135 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
7136 (ipa_get_param): Check we are not in WPA.
7137 (ipa_get_param_move_cost): New.
7138 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
7139 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
7140 parm numbers to be present.
7142 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
7144 PR rtl-optimization/58048
7145 * lra-constraints.c (process_alt_operands): Don't check asm
7146 operand on register.
7148 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
7150 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
7151 the implied StoreLoad barrier for atomic operations if before.
7153 2013-08-02 Jan Hubicka <jh@suse.cz>
7154 Martin Liska <marxin.liska@gmail.com>
7156 * cgraph.c (cgraph_function_body_availability): Do not check
7158 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
7159 symtab_node_availability): Declare.
7160 * ipa.c (can_replace_by_local_alias): New.
7161 (function_and_variable_visibility): Use it.
7162 * symtab.c (symtab_for_node_and_aliases,
7163 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
7165 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
7167 PR rtl-optimization/57963
7168 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
7169 (lra_constraints): Use them.
7171 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
7173 * config/arm/types.md (define_attr "type"): Add "load_acq"
7175 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
7177 (cortex_a53_store1): Likewise.
7179 2013-08-01 Jan Hubicka <jh@suse.cz>
7181 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
7182 partitions are not needed.
7184 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
7186 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
7187 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
7188 MAYBE_NON_Q_CLASS_P where appropriate.
7190 2013-08-01 Jan Hubicka <jh@suse.cz>
7192 * cgraph.h (release_function_body): Declare.
7193 * tree.c (free_lang_data_in_decl): Free, parameters and return values
7194 of unused delcarations.
7196 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7198 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
7199 RTL form when subtracting a constant.
7201 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7203 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
7204 Generate canonical plus rtx with negated immediate instead of minus
7206 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
7208 2013-08-01 Jan Hubicka <jh@suse.cz>
7210 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
7211 (cgraph_release_function_body): Likewise.
7212 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
7213 * cgraph.h (cgrpah_node): Rename abstract_and_needed
7214 to used_as_abstract_origin.
7215 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
7216 symbols used as abstract origins.
7217 * cgraphunit.c (analyze_functions): Update.
7218 * ipa.c (symtab_remove_unreachable_nodes): Recompute
7219 used_as_abstract_origin.
7220 * tree-inline.c (tree_function_versioning): Update
7221 used_as_abstract_origin; be ready for DECL_RESULT and
7222 DECL_ARGUMENTS to be NULL.
7224 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
7225 for abstract functions.
7226 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
7229 2013-08-01 Jan Hubicka <jh@suse.cz>
7231 * profile.c (compute_value_histograms): Fix thinko.
7233 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
7235 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
7236 aarch-common-protos.h to extra_headers.
7237 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7238 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
7239 * config/aarch64/t-aarch64 (aarch-common.o): Define.
7241 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
7243 * config/aarch64/aarch64.md (define_attr "type"): Delete.
7244 Include "../arm/types.md". Define "type" attribute for all patterns.
7245 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
7248 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7250 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
7251 to support power8 load fusion.
7252 (fusion_gpr_mem_load): Likewise.
7254 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
7256 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
7257 declarations for power8 load fusion.
7258 (emit_fusion_gpr_load): Likewise.
7260 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
7261 tuning for power8, turn on fusion mode by default. Turn on sign
7262 extending fusion mode if normal fusion mode is on, and we are at
7264 (fusion_gpr_load_p): New function, return true if we can fuse an
7265 addis instruction with a dependent load to a GPR.
7266 (emit_fusion_gpr_load): Emit the instructions for power8 load
7269 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
7270 (VSX load fusion peepholes): New peepholes to fuse together an
7271 addi instruction with a VSX load instruction.
7273 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
7274 peepholes to fuse an addis instruction with a load to a GPR base
7275 register. If we are supporting sign extending fusions, convert
7276 sign extending loads to zero extending loads and add an explicit
7279 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
7281 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
7282 aarch-common-protos.h to extra_headers.
7283 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7284 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
7285 (arm_early_store_addr_dep): Likewise.
7286 (arm_no_early_alu_shift_dep): Likewise.
7287 (arm_no_early_alu_shift_value_dep): Likewise.
7288 (arm_no_early_mul_dep): Likewise.
7289 (arm_no_early_store_addr_dep): Likewise.
7290 (arm_mac_accumulator_is_mul_result): Likewise.
7291 (arm_mac_accumulator_is_result): Likewise.
7292 * config/arm/aarch-common.c: ... here. New file.
7293 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
7295 (arm_early_store_addr_dep): Likewise.
7296 (arm_no_early_alu_shift_dep): Likewise.
7297 (arm_no_early_alu_shift_value_dep): Likewise.
7298 (arm_no_early_mul_dep): Likewise.
7299 (arm_no_early_store_addr_dep): Likewise.
7300 (arm_mac_accumulator_is_mul_result): Likewise.
7301 (arm_mac_accumulator_is_result): Likewise.
7302 * config/arm/aarch-common-protos.h: ... here. New file.
7303 * config/arm/t-arm (aarch-common.o): Define.
7305 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
7307 * config/arm/arm.md: Include new file "types.md".
7308 (define_attr "type"): Move from here to ...
7309 (define_attr "mul32"): Likewise.
7310 (define_attr "mul64"): Likewise.
7311 * config/arm/types.md: ... here. New file.
7313 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
7315 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
7316 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
7318 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7320 * gen-pass-instances.awk: Fix offset of substr().
7322 2013-07-31 David Malcolm <dmalcolm@redhat.com>
7324 * Makefile.in (pass-instances.def): New.
7325 (passes.o): Replace dependency on passes.def with one on
7328 * gen-pass-instances.awk: New.
7330 * passes.c (pass_manager::pass_manager): Use pass-instances.def
7331 rather than passes.def, updating local definition of NEXT_PASS
7332 macro to add an extra NUM parameter (currently unused).
7334 2013-07-30 David Malcolm <dmalcolm@redhat.com>
7336 * Makefile.in (PASS_MANAGER_H): New.
7337 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
7338 (passes.o): Likewise.
7339 (statistics.o): Likewise.
7340 (cgraphunit.o): Likewise.
7341 (context.o): Depend on PASS_MANAGER_H.
7343 * pass_manager.h: New.
7345 * cgraphunit.c (cgraph_add_new_function): Update for moves
7346 of globals to fields of pass_manager.
7347 (analyze_function): Likewise.
7348 (expand_function): Likewise.
7349 (ipa_passes): Likewise.
7350 (compile): Likewise.
7352 * context.c (context::context): New.
7353 * context.h (context::context): New.
7354 (context::get_passes): New.
7355 (context::passes_): New.
7357 * lto-cgraph.c (input_node): Update for moves of globals to
7358 fields of pass_manager.
7360 * passes.c (all_passes): Remove, in favor of a field of the
7361 same name within the new class pass_manager.
7362 (all_small_ipa_passes): Likewise.
7363 (all_lowering_passes): Likewise.
7364 (all_regular_ipa_passes): Likewise.
7365 (all_late_ipa_passes): Likewise.
7366 (all_lto_gen_passes): Likewise.
7367 (passes_by_id): Likewise.
7368 (passes_by_id_size): Likewise.
7369 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7370 the new class pass_manager.
7371 (set_pass_for_id): Convert to...
7372 (pass_manager::set_pass_for_id): ...method.
7373 (get_pass_for_id): Convert to...
7374 (pass_manager::get_pass_for_id): ...method.
7375 (register_one_dump_file): Move body of implementation into...
7376 (pass_manager::register_one_dump_file): ...here.
7377 (register_dump_files_1): Convert to...
7378 (pass_manager::register_dump_files_1): ...method.
7379 (register_dump_files): Convert to...
7380 (pass_manager::register_dump_files): ...method.
7381 (create_pass_tab): Update for moves of globals to fields of
7383 (dump_passes): Move body of implementation into...
7384 (pass_manager::dump_passes): ...here.
7385 (register_pass): Move body of implementation into...
7386 (pass_manager::register_pass): ...here.
7387 (init_optimization_passes): Convert into...
7388 (pass_manager::pass_manager): ...constructor for new
7389 pass_manager class, and initialize the pass_lists array.
7390 (check_profile_consistency): Update for moves of globals to
7391 fields of pass_manager.
7392 (dump_profile_report): Move body of implementation into...
7393 (pass_manager::dump_profile_report): ...here.
7394 (ipa_write_summaries_1): Update for moves of pass lists from
7395 being globals to fields of pass_manager.
7396 (ipa_write_optimization_summaries): Likewise.
7397 (ipa_read_summaries): Likewise.
7398 (ipa_read_optimization_summaries): Likewise.
7399 (execute_all_ipa_stmt_fixups): Likewise.
7401 * statistics.c (statistics_fini): Update for moves of globals to
7402 fields of pass_manager.
7404 * toplev.c (general_init): Replace call to
7405 init_optimization_passes with construction of the pass_manager
7408 * tree-pass.h (all_passes): Remove, in favor of a field of the
7409 same name within the new class pass_manager.
7410 (all_small_ipa_passes): Likewise.
7411 (all_lowering_passes): Likewise.
7412 (all_regular_ipa_passes): Likewise.
7413 (all_lto_gen_passes): Likewise.
7414 (all_late_ipa_passes): Likewise.
7415 (passes_by_id): Likewise.
7416 (passes_by_id_size): Likewise.
7417 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7418 the new class pass_manager.
7419 (get_pass_for_id): Remove.
7421 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
7423 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
7426 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
7428 * arm.md (mulhi3): New expand pattern.
7430 2013-07-30 Jan Hubicka <jh@suse.cz>
7431 Martin Liska <marxin.liska@gmail.com>
7433 * profile.c (compute_value_histograms): Do not ICE when
7434 there is mismatch only on some counters.
7436 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7438 PR rtl-optimization/57637
7439 * function.c (move_insn_for_shrink_wrap): Also check the
7440 GEN set of the LIVE problem for the liveness analysis
7441 if it exists, otherwise give up.
7443 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
7445 PR tree-optimization/57993
7446 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
7447 replaced statement in the candidate table.
7448 (phi_add_costs): Return infinite cost when the hidden basis does
7449 not dominate all phis on which the candidate is dependent.
7450 (replace_one_candidate): Record replaced statement in the
7453 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
7455 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
7456 (ashlv2si3): New expander.
7457 (*ashlv2si3_i): New define_insn_and_split.
7458 * predicates.md (float_operation): Allow patterns with three
7461 PR rtl-optimization/58021
7462 * mode-switching.c (create_pre_exit): Always split off preceding
7463 insns if we are not at the basic block head.
7465 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
7467 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
7468 (UCLIBC_DYNAMIC_LINKER): New macro.
7469 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
7471 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
7472 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
7474 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
7475 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
7476 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
7477 for SF and DF modes. Use ieee_quad_format for TF mode.
7478 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
7479 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
7480 (mips_option_override): Handle `-mnan=legacy'.
7481 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
7482 `-mabs=2008' and `-mnan=2008'.
7483 (OPTION_DEFAULT_SPECS): Add "nan" default.
7484 (ASM_SPEC): Handle `-mnan='.
7485 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
7486 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
7487 comment accordingly.
7488 (neg<mode>2): Likewise.
7489 * config/mips/mips.opt (mabs, mnan): New options.
7490 * doc/install.texi (Configuration): Document `--with-nan=' option.
7491 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
7493 (MIPS Options): Document them.
7494 * config.gcc <mips*-*-*>: Handle `--with-nan='.
7495 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
7496 * configure: Regenerate.
7497 * config.in: Regenerate.
7499 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
7501 * config/i386/i386.md (float post-reload splitters): Do not check
7502 for subregs of SSE registers.
7504 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
7505 H.J. Lu <hongjiu.lu@intel.com>
7509 * config/i386/i386.md (post-reload splitter
7510 to avoid partial SSE reg dependency stalls): New pattern.
7512 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
7514 * config/s390/s390.md ("movcc"): Swap load and store instructions.
7516 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
7518 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
7519 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
7521 2013-07-26 Cary Coutant <ccoutant@google.com>
7523 * dwarf2out.c (die_checksum_ordered): Don't include template
7524 instantiations in signature.
7525 (is_template_parameter): New function.
7526 (is_template_instantiation): New function.
7527 (generate_skeleton_bottom_up): Don't include template instantiations
7529 (generate_skeleton): Likewise.
7530 (break_out_comdat_types): Move recursive call to break out nested
7532 (prune_unused_types_mark_generic_parms_dies): Call
7533 is_template_parameter.
7535 2013-07-26 Ian Bolton <ian.bolton@arm.com>
7537 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
7538 uses vector registers.
7539 * config/aarch64/iterators.md: Add attributes rtn and vas.
7541 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7542 Richard Earnshaw <richard.earnshaw@arm.com>
7544 * combine.c (simplify_comparison): Re-canonicalize operands
7546 * config/arm/arm.md (movcond_addsi): New splitter.
7548 2013-07-25 Sterling Augustine <saugustine@google.com>
7550 * dwarf2out.c (size_of_pubnames): Move code to...
7551 (include_pubname_in_output): ...here. New.
7552 (want_pubnames): Rearrange.
7553 (output_pubnames): Call include_pubname_in_output. Move assertion.
7555 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
7557 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
7559 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
7562 * doc/extend.texi: Remove obsolete builtins. Fix
7563 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
7565 2013-07-25 Jan Hubicka <jh@suse.cz>
7567 * cgraph.c (release_function_body): Break out from ...
7568 (cgraph_release_function_body): ... this one; also release DECL_RESULT
7570 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
7571 old_tree in the map.
7572 (create_specialized_node): Update.
7573 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
7575 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
7576 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
7577 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
7578 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
7581 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7583 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
7584 addsi3_carryin_alt2_<optab>): Correct output template.
7586 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7588 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
7589 Adjust for arm_restrict_it.
7590 Remove trailing whitespace.
7592 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
7594 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
7595 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
7597 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
7599 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
7601 PR rtl-optimization/57960
7602 * lra-constraints.c (process_alt_operands): Use the right mode
7603 when checking strict_low.
7605 2013-07-25 Jan Hubicka <jh@suse.cz>
7607 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
7608 * cgraph.c (cgraph_remove_node): Do not release function body
7609 when in cgraph streaming.
7610 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
7611 in other partitions are not considered reachable; fix handling of
7614 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7616 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
7618 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7623 * config/arm/arm.md ("*sibcall_insn): Replace use of
7624 Ss with US. Adjust output for v5 and v4t.
7625 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
7627 * config/arm/constraints.md ("Ss"): Rename to US.
7629 2013-07-25 Terry Guo <terry.guo@arm.com>
7631 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
7632 shift_add/shift_sub0/shift_sub1 RTXs.
7634 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
7635 Anton Blanchard <anton@au1.ibm.com>
7637 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
7638 (altivec_vpks<VI_char>ss): Likewise.
7639 (altivec_vpks<VI_char>us): Likewise.
7640 (altivec_vpku<VI_char>us): Likewise.
7641 (altivec_vpku<VI_char>um): Likewise.
7643 2013-07-24 David Malcolm <dmalcolm@redhat.com>
7645 Introduce context class.
7647 * Makefile.in (CONTEXT_H): New.
7648 (OBJS): Add context.o.
7649 (toplev.o): Add CONTEXT_H to dependencies.
7652 * toplev.c (general_init): Create the singleton gcc::context instance.
7658 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
7660 PR rtl-optimization/57968
7661 * mode-switching.c (create_pre_exit): Allow instructions that
7662 don't set a return register to need a non-exit mode.
7664 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
7665 Anton Blanchard <anton@au1.ibm.com>
7667 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
7668 operands to vperm for little endian.
7669 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
7670 of lvsl to create the control mask for a vperm for little endian.
7672 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7673 Anton Blanchard <anton@au1.ibm.com>
7675 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
7676 two operands for little-endian.
7678 2013-07-23 Steve Ellcey <sellcey@mips.com>
7680 * config/mips/mips.c (mips_case_values_threshold): New.
7681 (TARGET_CASE_VALUES_THRESHOLD): Define.
7683 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7684 Anton Blanchard <anton@au1.ibm.com>
7686 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
7687 selection of field for vector splat in little endian mode.
7689 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
7691 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
7692 expanders to rs6000.md.
7693 (ior<mode>3): Likewise.
7694 (and<mode>3): Likewise.
7695 (one_cmpl<mode>2): Likewise.
7696 (nor<mode>3): Likewise.
7697 (andc<mode>3): Likewise.
7698 (eqv<mode>3): Likewise.
7699 (nand<mode>3): Likewise.
7700 (orc<mode>3): Likewise.
7702 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
7705 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
7706 to split multi-word logical operations.
7707 (rs6000_split_logical_di): Likewise.
7708 (rs6000_split_logical): Likewise.
7710 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
7711 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
7712 and allow TImode operations in 32-bit.
7713 (vsx_and<mode>3_64bit): Likewise.
7714 (vsx_ior<mode>3_32bit): Likewise.
7715 (vsx_ior<mode>3_64bit): Likewise.
7716 (vsx_xor<mode>3_32bit): Likewise.
7717 (vsx_xor<mode>3_64bit): Likewise.
7718 (vsx_one_cmpl<mode>2_32bit): Likewise.
7719 (vsx_one_cmpl<mode>2_64bit): Likewise.
7720 (vsx_nor<mode>3_32bit): Likewise.
7721 (vsx_nor<mode>3_64bit): Likewise.
7722 (vsx_andc<mode>3_32bit): Likewise.
7723 (vsx_andc<mode>3_64bit): Likewise.
7724 (vsx_eqv<mode>3_32bit): Likewise.
7725 (vsx_eqv<mode>3_64bit): Likewise.
7726 (vsx_nand<mode>3_32bit): Likewise.
7727 (vsx_nand<mode>3_64bit): Likewise.
7728 (vsx_orc<mode>3_32bit): Likewise.
7729 (vsx_orc<mode>3_64bit): Likewise.
7731 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
7732 logical types in GPRs.
7734 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
7735 logical insns to rs6000.md, and allow TImode operations in
7737 (altivec_ior<mode>3): Likewise.
7738 (altivec_xor<mode>3): Likewise.
7739 (altivec_one_cmpl<mode>2): Likewise.
7740 (altivec_nor<mode>3): Likewise.
7741 (altivec_andc<mode>3): Likewise.
7743 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
7744 attributes for moving the 128-bit logical operations into
7746 (BOOL_REGS_OUTPUT): Likewise.
7747 (BOOL_REGS_OP1): Likewise.
7748 (BOOL_REGS_OP2): Likewise.
7749 (BOOL_REGS_UNARY): Likewise.
7750 (BOOL_REGS_AND_CR0): Likewise.
7751 (one_cmpl<mode>2): Add support for DI logical operations on
7752 32-bit, splitting the operations to 32-bit.
7756 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
7757 changes to combine the 32/64-bit code, allow logical operations on
7758 TI mode in 32-bit, and to use similar match_operator patterns like
7759 scalar mode uses. Combine the Altivec and VSX code for logical
7760 operations, and move it here.
7761 (ior<mode>3, 128-bit types): Likewise.
7762 (xor<mode>3, 128-bit types): Likewise.
7763 (one_cmpl<mode>3, 128-bit types): Likewise.
7764 (nor<mode>3, 128-bit types): Likewise.
7765 (andc<mode>3, 128-bit types): Likewise.
7766 (eqv<mode>3, 128-bit types): Likewise.
7767 (nand<mode>3, 128-bit types): Likewise.
7768 (orc<mode>3, 128-bit types): Likewise.
7769 (and<mode>3_internal): Likewise.
7770 (bool<mode>3_internal): Likewise.
7771 (boolc<mode>3_internal1): Likewise.
7772 (boolc<mode>3_internal2): Likewise.
7773 (boolcc<mode>3_internal1): Likewise.
7774 (boolcc<mode>3_internal2): Likewise.
7775 (eqv<mode>3_internal1): Likewise.
7776 (eqv<mode>3_internal2): Likewise.
7777 (one_cmpl1<mode>3_internal): Likewise.
7779 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
7781 * config/microblaze/microblaze.c (microblaze_expand_prologue):
7782 Rename flag_stack_usage to flag_stack_usage_info.
7784 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
7786 * config/microblaze/sync.md: New file.
7787 * config/microblaze/microblaze.md: Include sync.md
7788 * config/microblaze/microblaze.c: Add print_operand 'y'.
7789 * config/microblaze/constraints.md: Add memory_contraint
7790 'Q' which is a single register.
7792 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
7794 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
7796 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
7798 * reload.c (find_reloads): Exit loop once we find this operand
7799 cannot be reloaded somehow for this alternative.
7801 * reload.c (find_reloads): Exit loop once we find a hard register.
7803 * rtlanal.c (computed_jump_p): Exit loop once we find label
7806 * i386.c (ix86_pad_returns): Exit loop after setting replace.
7808 * cfgloopmanip.c (remove_path): Exit loop after setting
7811 * gensupport.c (subst_dup): Avoid loop if code is not
7812 MATCH_DUP nor MATCH_OP_DUP.
7814 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
7816 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
7818 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
7820 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
7821 true for SP_REGNUM if mode == ptr_mode.
7822 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
7823 with value R0_REGNUM + 31.
7825 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
7827 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
7828 pad pointer-typed argument downward.
7830 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
7832 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
7833 and __ILP32__ when the ILP32 model is in use.
7835 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
7837 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
7838 (aarch64_load_symref_appropriately): In the case of
7839 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
7840 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
7841 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
7842 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
7843 if 'mode' doesn't equal to 'ptr_mode'.
7844 (aarch64_output_mi_thunk): Add an assertion on the alignment of
7845 'vcall_offset'; change to call aarch64_emit_move differently depending
7846 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
7847 to calculate the upper bound of 'vcall_offset'.
7848 (aarch64_cannot_force_const_mem): Change to also return true if
7850 (aarch64_legitimize_reload_address): In the case of large
7851 displacements, add new local variable 'xmode' and an assertion
7852 based on it; change to use 'xmode' to generate the new rtx and
7854 (aarch64_asm_trampoline_template): Change to generate the template
7855 differently depending on TARGET_ILP32 or not; change to use
7856 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
7857 (aarch64_trampoline_size): Removed.
7858 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
7859 and replace immediate literals with it. Change to use 'ptr_mode'
7860 instead of 'DImode' and call convert_memory_address if the mode
7861 of 'fnaddr' doesn't equal to 'ptr_mode'.
7862 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
7864 (aarch64_elf_asm_destructor): Likewise.
7865 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
7866 on TARGET_ILP32 instead of aarch64_trampoline_size.
7867 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
7868 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
7869 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
7870 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
7871 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
7872 (storewb_pair<GPI:mode>_<P:mode>): ... this.
7873 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
7874 depending on the value of 'mode'.
7875 (add_losym_<mode>): New.
7876 (ldr_got_small_<mode>): New, based on ldr_got_small.
7877 (ldr_got_small): Remove.
7878 (ldr_got_small_sidi): New.
7879 * config/aarch64/iterators.md (P): New.
7880 (PTR): Change to 'ptr_mode' in the condition.
7882 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
7884 * config.gcc (aarch64*-*-*): Support --with-abi.
7885 (aarch64*-*-elf): Support --with-multilib-list.
7886 (aarch64*-*-linux*): Likewise.
7887 (supported_defaults): Add abi to aarch64*-*-*.
7888 * configure.ac: Mention AArch64 for --with-multilib-list.
7889 * configure: Re-generated.
7890 * config/aarch64/biarchilp32.h: New file.
7891 * config/aarch64/biarchlp64.h: New file.
7892 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
7894 (MULTILIB_DEFAULTS): Ditto.
7895 (DRIVER_SELF_SPECS): Ditto.
7896 (ASM_SPEC): Update to also substitute -mabi.
7897 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
7898 file whose name depends on -mabi= and -mbig-endian.
7899 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
7901 (POINTER_SIZE): New define.
7902 (POINTERS_EXTEND_UNSIGNED): Ditto.
7903 (enum aarch64_abi_type): New enumeration tag.
7904 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
7905 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
7906 (TARGET_ILP32): New define.
7907 * config/aarch64/aarch64.opt (mabi): New.
7909 (ilp32, lp64): New values for -mabi.
7910 * config/aarch64/t-aarch64 (comma): New define.
7911 (MULTILIB_OPTIONS): Ditto.
7912 (MULTILIB_DIRNAMES): Ditto.
7913 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
7914 * doc/invoke.texi: Document -mabi for AArch64.
7916 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
7918 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
7920 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
7921 Anton Blanchard <anton@au1.ibm.com>
7923 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
7924 endianness when selecting field to splat.
7926 2013-07-22 Eric Christopher <echristo@gmail.com>
7928 * dwarf2out.c (die_odr_checksum): New function to use
7929 CHECKSUM_ macros and ULEB128 for DIE tag.
7930 (generate_type_signature): Use.
7932 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
7934 * config.gcc (sparc*-*-*): Accept leon3 processor.
7935 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
7936 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
7937 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
7938 * config/sparc/sparc.opt (enum processor_type): Add leon3.
7939 (mfix-ut699): Adjust comment.
7940 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
7941 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
7942 (CPP_CPU_SPEC): Likewise.
7943 (ASM_CPU_SPEC): Likewise.
7944 * config/sparc/sparc.c (leon3_cost): New constant.
7945 (sparc_option_override): Add leon3 support.
7946 (mem_ref): New function.
7947 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
7948 (sparc_do_work_around_errata): Look into the instruction in the delay
7949 slot and adjust accordingly. Add fix for the data cache nullify issues
7950 of the UT699. Change insertion position for the NOP.
7951 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
7952 (leon3_load): New reservation.
7953 (leon_store): Bump latency to 2.
7954 (grfpu): New automaton.
7955 (grfpu_alu): New unit.
7956 (grfpu_ds): Likewise.
7957 (leon_fp_alu): Adjust.
7958 (leon_fp_mult): Delete.
7959 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
7960 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
7961 * config/sparc/sparc.md (cpu): Add leon3.
7962 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
7964 (atomic_test_and_set): Likewise.
7967 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
7969 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
7970 default for R5900 targets.
7971 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
7972 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
7973 * config/mips/mips.c (mips_option_override): Report an error for
7974 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
7975 for -march=r5900 -mhard-float.
7977 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
7979 * df-problems.c (can_move_insns_across): Exit loop once we
7980 find a non-fixed, non-global register.
7982 * ipa-pure-const.c (propagate_nothrow): Exit loop after
7985 * omega.c (omega_eliminate_red): Break after setting red_found.
7986 (omega_problem_has_red_equations): Similarly after setting found.
7987 (omega_query_variable): Similarly after setting coupled.
7989 2013-07-22 Marek Polacek <polacek@redhat.com>
7991 * gimplify.c: Don't include gimple.h twice.
7993 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7995 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
7996 instead of TARGET_THUMB1.
7997 (Pz): New constraint.
7998 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
8000 (compare_negsi_si): Likewise.
8001 (compare_addsi2_op0): Likewise.
8002 (compare_addsi2_op1): Likewise.
8003 (addsi3_carryin_<optab>): Likewise.
8004 (addsi3_carryin_alt2_<optab>): Likewise.
8005 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
8006 for arm_restrict_it.
8007 (subsi3_carryin): Likewise.
8008 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
8009 (minmax_arithsi): Disable for arm_restrict_it.
8010 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
8011 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
8012 (satsi_<SAT:code>_shift): Likewise.
8013 (arm_shiftsi3): Add alternative for 16-bit encoding.
8014 (arm32_movhf): Disable for arm_restrict_it.
8015 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
8016 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
8018 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
8020 * config/arm/arm.md (attribute "insn"): Delete.
8021 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
8022 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
8023 (not_shiftsi): Update for attribute change.
8024 (not_shiftsi_compare0): Likewise.
8025 (not_shiftsi_compare0_scratch): Likewise.
8026 (arm_one_cmplsi2): Likewise.
8027 (thumb1_one_cmplsi2): Likewise.
8028 (notsi_compare0): Likewise.
8029 (notsi_compare0_scratch): Likewise.
8030 (thumb1_movdi_insn): Likewise.
8031 (arm_movsi_insn): Likewise.
8032 (movhi_insn_arch4): Likewise.
8033 (movhi_bytes): Likewise.
8034 (arm_movqi_insn): Likewise.
8035 (thumb1_movqi_insn): Likewise.
8036 (arm32_movhf): Likewise.
8037 (thumb1_movhf): Likewise.
8038 (arm_movsf_soft_insn): Likewise.
8039 (thumb1_movsf_insn): Likewise.
8040 (thumb_movdf_insn): Likewise.
8041 (movsicc_insn): Likewise.
8042 (movsfcc_soft_insn): Likewise.
8043 (and_scc): Likewise.
8044 (cond_move): Likewise.
8045 (if_move_not): Likewise.
8046 (if_not_move): Likewise.
8047 (if_shift_move): Likewise.
8048 (if_move_shift): Likewise.
8049 (if_shift_shift): Likewise.
8050 (if_not_arith): Likewise.
8051 (if_arith_not): Likewise.
8052 (cond_move_not): Likewise.
8053 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
8054 (neon_mov<mode>): Likewise.
8055 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
8056 (thumb2_movsi_vfp): Likewise.
8057 (movsf_vfp): Likewise.
8058 (thumb2_movsf_vfp): Likewise.
8059 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8061 (cortexa7_older_only): Likewise.
8062 (cortexa7_younger): Likewise.
8063 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
8064 (1020alu_shift_op): Likewise.
8065 (1020alu_shift_reg_op): Likewise.
8066 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
8067 (alu_shift_op): Likewise.
8068 (alu_shift_reg_op): Likewise.
8069 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
8070 (11_alu_shift_op): Likewise.
8071 (11_alu_shift_reg_op): Likewise.
8072 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
8073 (9_alu_shift_reg_op): Likewise.
8074 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8076 (cortex_a15_alu_shift): Likewise.
8077 (cortex_a15_alu_shift_reg): Likewise.
8078 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
8079 (cortex_a5_alu_shift): Likewise.
8080 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
8082 (cortex_a53_alu_shift): Likewise.
8083 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8085 (cortex_a7_alu_reg): Likewise.
8086 (cortex_a7_alu_shift): Likewise.
8087 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8088 (cortex_a8_alu_shift): Likewise.
8089 (cortex_a8_alu_shift_reg): Likewise.
8090 (cortex_a8_mov): Likewise.
8091 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
8092 (cortex_a9_dp_shift): Likewise.
8093 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
8094 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
8095 (cortex_r4_mov): Likewise.
8096 (cortex_r4_alu_shift): Likewise.
8097 (cortex_r4_alu_shift_reg): Likewise.
8098 * config/arm/fa526.md (526_alu_op): Update for attribute change.
8099 (526_alu_shift_op): Likewise.
8100 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
8101 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
8102 (626te_alu_shift_op): Likewise.
8103 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
8104 (726te_alu_op): Likewise.
8105 (726te_alu_shift_op): Likewise.
8106 (726te_alu_shift_reg_op): Likewise.
8107 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
8108 (mp626_alu_shift_op): Likewise.
8109 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
8110 (pj4_alu_e1_conds): Likewise.
8111 (pj4_alu): Likewise.
8112 (pj4_alu_conds): Likewise.
8113 (pj4_shift): Likewise.
8114 (pj4_shift_conds): Likewise.
8115 (pj4_alu_shift): Likewise.
8116 (pj4_alu_shift_conds): Likewise.
8118 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8120 * config/arm/predicates.md (shiftable_operator_strict_it):
8122 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
8123 Disable cond_exec version for arm_restrict_it.
8124 (thumb2_smaxsi3): Convert to generate cond_exec.
8125 (thumb2_sminsi3): Likewise.
8126 (thumb32_umaxsi3): Likewise.
8127 (thumb2_uminsi3): Likewise.
8128 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
8129 (thumb2_neg_abssi2): Likewise.
8130 (thumb2_mov_scc): Add alternative for 16-bit encoding.
8131 (thumb2_movsicc_insn): Adjust alternatives.
8132 (thumb2_mov_negscc): Disable for arm_restrict_it.
8133 (thumb2_mov_negscc_strict_it): New pattern.
8134 (thumb2_mov_notscc_strict_it): New pattern.
8135 (thumb2_mov_notscc): Disable for arm_restrict_it.
8136 (thumb2_ior_scc): Likewise.
8137 (thumb2_ior_scc_strict_it): New pattern.
8138 (thumb2_cond_move): Adjust for arm_restrict_it.
8139 (thumb2_cond_arith): Disable for arm_restrict_it.
8140 (thumb2_cond_arith_strict_it): New pattern.
8141 (thumb2_cond_sub): Adjust for arm_restrict_it.
8142 (thumb2_movcond): Likewise.
8143 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
8144 (thumb2_zero_extendhisi2_v6): Likewise.
8145 (thumb2_zero_extendqisi2_v6): Likewise.
8146 (orsi_notsi_si): Likewise.
8147 (orsi_not_shiftsi_si): Likewise.
8149 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
8151 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
8152 instruction sequence is 1 byte shorter.
8154 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
8156 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
8157 it is not needed after split.
8159 2013-07-20 Iain Sandoe <iain@codesourcery.com>
8162 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
8163 second label for nonlocal goto receivers. Don't output pic base labels
8164 unless we're producing PIC; mark that action unreachable().
8165 (ix86_save_reg): If the function contains a nonlocal label, save the
8167 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
8168 * config/darwin.c (emitted_pic_label_num): New GTY.
8169 (update_pic_label_number_if_needed): New.
8170 (machopic_output_function_base_name): Adjust for nonlocal receiver
8172 (machopic_should_output_picbase_label): New.
8173 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
8174 (nonlocal_goto_receiver): New insn and split.
8176 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
8178 * config/aarch64/aarch64-builtins.c
8179 (aarch64_fold_builtin): Fold abs in all modes.
8180 * config/aarch64/aarch64-simd-builtins.def
8181 (abs): Enable for all modes.
8182 * config/aarch64/arm_neon.h
8183 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
8184 (vabs_f64): Add missing intrinsic.
8186 2013-07-19 Ian Bolton <ian.bolton@arm.com>
8188 * config/aarch64/arm_neon.h (vabs_s64): New function
8190 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
8193 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
8194 * config/avr/avr.md (adjust_len): Add `round'.
8195 * config/avr/avr-protos.h (avr_out_round): New prototype.
8196 (avr_out_plus): Add `out_label' argument.
8197 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
8198 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
8199 Handle the case where `insn' is just a pattern.
8200 (avr_out_bitop): Handle the case where `insn' is just a pattern.
8201 (avr_out_round): New function.
8202 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
8204 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
8206 * config/microblaze/microblaze.c (microblaze_expand_prologue):
8207 Add check for flag_stack_usage to handle -fstack-usage support
8209 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
8211 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
8212 interaction for new Power8 flags and VSX.
8214 2013-07-18 Sriraman Tallam <tmsriram@google.com>
8217 * tree-inline.c (expand_call_inline): Emit errors during
8218 early_inlining only if optimization is not turned on.
8220 2013-07-18 David Malcolm <dmalcolm@redhat.com>
8224 * passes.c (init_optimization_passes): Move the construction of
8225 the pass hierarchy into a new passes.def file.
8227 * Makefile.in (passes.o): Add dependency on passes.def.
8229 2013-07-18 David Malcolm <dmalcolm@redhat.com>
8231 * passes.c (init_optimization_passes): Introduce macros for
8232 constructing the tree of passes (INSERT_PASSES_AFTER,
8233 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
8234 TERMINATE_PASS_LIST).
8236 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
8237 Wei Mi <wmi@google.com>
8239 PR rtl-optimization/57878
8240 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
8242 (reload_pseudo_compare_func): Check nregs first for reload
8245 2013-07-18 David Malcolm <dmalcolm@redhat.com>
8247 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
8249 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
8251 * read-rtl.c (validate_const_int): Once an invalid character is
8252 seen, quit the loop.
8254 * gengtype.c (write_roots): Similarly once we find the "deletable"
8255 or "if_marked" option.
8257 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
8259 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
8260 "xtab" and "sat". Move value "clz" from here to ...
8261 (attriubte "type"): ... here.
8262 (satsi_<SAT:code>): Delete "insn" attribute.
8263 (satsi_<SAT:code>_shift): Likewise.
8264 (arm_zero_extendqisi2addsi): Likewise.
8265 (arm_extendqisi2addsi): Likewise.
8266 (clzsi2): Update for attribute changes.
8267 (rbitsi2): Likewise.
8268 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
8270 (arm_usatsihi): Likewise.
8271 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8273 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
8275 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
8276 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
8277 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
8278 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
8279 in alphabetical order.
8280 (attribute "core_cycles"): Update for attribute changes.
8281 (arm_addsi3): Likewise.
8282 (addsi3_compare0): Likewise.
8283 (addsi3_compare0_scratch): Likewise.
8284 (addsi3_compare_op1): Likewise.
8285 (addsi3_compare_op2): Likewise.
8286 (compare_addsi2_op0): Likewise.
8287 (compare_addsi2_op1): Likewise.
8288 (addsi3_carryin_shift_<optab>): Likewise.
8289 (subsi3_carryin_shift): Likewise.
8290 (rsbsi3_carryin_shift): Likewise.
8291 (arm_subsi3_insn): Likewise.
8292 (subsi3_compare0): Likewise.
8293 (subsi3_compare): Likewise.
8294 (arm_andsi3_insn): Likewise.
8295 (thumb1_andsi3_insn): Likewise.
8296 (andsi3_compare0): Likewise.
8297 (andsi3_compare0_scratch): Likewise.
8298 (zeroextractsi_compare0_scratch
8299 (andsi_not_shiftsi_si): Likewise.
8300 (iorsi3_insn): Likewise.
8301 (iorsi3_compare0): Likewise.
8302 (iorsi3_compare0_scratch): Likewise.
8303 (arm_xorsi3): Likewise.
8304 (thumb1_xorsi3_insn): Likewise.
8305 (xorsi3_compare0): Likewise.
8306 (xorsi3_compare0_scratch): Likewise.
8307 (satsi_<SAT:code>_shift): Likewise.
8309 (arm_shiftsi3): Likewise.
8310 (shiftsi3_compare0): Likewise.
8311 (not_shiftsi): Likewise.
8312 (not_shiftsi_compare0): Likewise.
8313 (not_shiftsi_compare0_scratch): Likewise.
8314 (arm_one_cmplsi2): Likewise.
8315 (thumb_one_complsi2): Likewise.
8316 (notsi_compare0): Likewise.
8317 (notsi_compare0_scratch): Likewise.
8318 (thumb1_zero_extendhisi2): Likewise.
8319 (arm_zero_extendhisi2): Likewise.
8320 (arm_zero_extendhisi2_v6): Likewise.
8321 (arm_zero_extendhisi2addsi): Likewise.
8322 (thumb1_zero_extendqisi2): Likewise.
8323 (thumb1_zero_extendqisi2_v6): Likewise.
8324 (arm_zero_extendqisi2): Likewise.
8325 (arm_zero_extendqisi2_v6): Likewise.
8326 (arm_zero_extendqisi2addsi): Likewise.
8327 (thumb1_extendhisi2): Likewise.
8328 (arm_extendhisi2): Likewise.
8329 (arm_extendhisi2_v6): Likewise.
8330 (arm_extendqisi): Likewise.
8331 (arm_extendqisi_v6): Likewise.
8332 (arm_extendqisi2addsi): Likewise.
8333 (thumb1_extendqisi2): Likewise.
8334 (thumb1_movdi_insn): Likewise.
8335 (arm_movsi_insn): Likewise.
8336 (movsi_compare0): Likewise.
8337 (movhi_insn_arch4): Likewise.
8338 (movhi_bytes): Likewise.
8339 (arm_movqi_insn): Likewise.
8340 (thumb1_movqi_insn): Likewise.
8341 (arm32_movhf): Likewise.
8342 (thumb1_movhf): Likewise.
8343 (arm_movsf_soft_insn): Likewise.
8344 (thumb1_movsf_insn): Likewise.
8345 (movdf_soft_insn): Likewise.
8346 (thumb_movdf_insn): Likewise.
8347 (arm_cmpsi_insn): Likewise.
8348 (cmpsi_shiftsi): Likewise.
8349 (cmpsi_shiftsi_swp): Likewise.
8350 (arm_cmpsi_negshiftsi_si): Likewise.
8351 (movsicc_insn): Likewise.
8352 (movsfcc_soft_insn): Likewise.
8353 (arith_shiftsi): Likewise.
8354 (arith_shiftsi_compare0
8355 (arith_shiftsi_compare0_scratch
8356 (sub_shiftsi): Likewise.
8357 (sub_shiftsi_compare0
8358 (sub_shiftsi_compare0_scratch
8359 (and_scc): Likewise.
8360 (cond_move): Likewise.
8361 (if_plus_move): Likewise.
8362 (if_move_plus): Likewise.
8363 (if_move_not): Likewise.
8364 (if_not_move): Likewise.
8365 (if_shift_move): Likewise.
8366 (if_move_shift): Likewise.
8367 (if_shift_shift): Likewise.
8368 (if_not_arith): Likewise.
8369 (if_arith_not): Likewise.
8370 (cond_move_not): Likewise.
8371 (thumb1_ashlsi3): Set type attribute.
8372 (thumb1_ashrsi3): Likewise.
8373 (thumb1_lshrsi3): Likewise.
8374 (thumb1_rotrsi3): Likewise.
8375 (shiftsi3_compare0_scratch): Likewise.
8376 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
8377 (neon_mov<mode>): Likewise.
8378 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
8380 (thumb2_movsi_insn): Likewise.
8381 (thumb2_cmpsi_neg_shiftsi): Likewise.
8382 (thumb2_extendqisi_v6): Likewise.
8383 (thumb2_zero_extendhisi2_v6): Likewise.
8384 (thumb2_zero_extendqisi2_v6): Likewise.
8385 (thumb2_shiftsi3_short): Likewise.
8386 (thumb2_addsi3_compare0_scratch): Likewise.
8387 (orsi_not_shiftsi_si): Likewise.
8388 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
8389 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
8391 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
8392 (1020alu_shift_op): Likewise.
8393 (1020alu_shift_reg_op): Likewise.
8394 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
8395 (alu_shift_op): Likewise.
8396 (alu_shift_reg_op): Likewise.
8397 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
8398 (11_alu_shift_op): Likewise.
8399 (11_alu_shift_reg_op): Likewise.
8400 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
8401 (9_alu_shift_reg_op): Likewise.
8402 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8404 (cortex_a15_alu_shift): Likewise.
8405 (cortex_a15_alu_shift_reg): Likewise.
8406 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
8408 (cortex_a5_alu_shift): Likewise.
8409 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
8411 (cortex_a53_alu_shift): Likewise.
8412 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8414 (cortex_a7_alu_reg): Likewise.
8415 (cortex_a7_alu_shift): Likewise.
8416 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
8418 (cortex_a8_alu_shift): Likewise.
8419 (cortex_a8_alu_shift_reg): Likewise.
8420 (cortex_a8_mov): Likewise.
8421 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
8422 (cortex_a9_dp_shift): Likewise.
8423 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
8425 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
8427 (cortex_r4_mov): Likewise.
8428 (cortex_r4_alu_shift): Likewise.
8429 (cortex_r4_alu_shift_reg): Likewise.
8430 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
8431 (526_alu_shift_op): Likewise.
8432 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
8433 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
8434 (626te_alu_shift_op): Likewise.
8435 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
8436 (726te_alu_op): Likewise.
8437 (726te_alu_shift_op): Likewise.
8438 (726te_alu_shift_reg_op): Likewise.
8439 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
8440 (mp626_alu_shift_op): Likewise.
8441 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
8442 (pj4_alu_e1_conds): Likewise.
8443 (pj4_alu): Likewise.
8444 (pj4_alu_conds): Likewise.
8445 (pj4_shift): Likewise.
8446 (pj4_shift_conds): Likewise.
8447 (pj4_alu_shift): Likewise.
8448 (pj4_alu_shift_conds): Likewise.
8449 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8451 (cortexa7_older_only): Likewise.
8452 (cortexa7_younger): Likewise.
8454 2013-07-18 David Malcolm <dmalcolm@redhat.com>
8456 * ipa-pure-const.c (generate_summary): Rename to...
8457 (pure_const_generate_summary): ... this.
8459 2013-07-17 Iain Sandoe <iain@codesourcery.com>
8461 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
8463 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
8466 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
8469 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8471 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
8472 enabled without -march=zEC12.
8473 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
8476 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
8478 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
8479 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
8480 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
8481 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
8482 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
8483 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
8484 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
8486 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
8487 and ISA_HAS_NMADD3_NMSUB3.
8488 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
8489 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
8490 (nmsub4<mode>, nmsub3<mode>): Likewise.
8491 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
8493 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
8495 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
8496 TARGET_MIPS5400 checking.
8498 2013-07-16 Jakub Jelinek <jakub@redhat.com>
8499 Peter Bergner <bergner@vnet.ibm.com>
8501 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
8502 registers in the comment.
8503 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
8504 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
8505 rather than FIRST_PSEUDO_REGISTERS.
8507 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
8509 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
8510 enable extra ISA flags with TARGET_HTM.
8512 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
8514 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
8517 2013-07-15 Cong Hou <congh@google.com>
8519 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
8520 in compare function for sorting.
8522 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
8524 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
8525 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
8526 * config/rs6000/rs6000.opt: Add -mhtm option.
8527 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
8528 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
8529 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8530 __HTM__ if the HTM instructions are available.
8531 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
8532 htm_spr_reg_operand): New define_predicates.
8533 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
8534 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
8536 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
8537 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
8538 HTM builtin functions.
8539 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
8540 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
8541 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
8542 (rs6000_builtin_mask_calculate): Likewise.
8543 (rs6000_option_override_internal): Likewise.
8544 (bdesc_htm): Add new HTM builtin support.
8545 (htm_spr_num): New function.
8546 (htm_spr_regno): Likewise.
8547 (rs6000_htm_spr_icode): Likewise.
8548 (htm_expand_builtin): Likewise.
8549 (htm_init_builtins): Likewise.
8550 (rs6000_expand_builtin): Add support for HTM builtin functions.
8551 (rs6000_init_builtins): Likewise.
8552 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
8554 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
8555 (TARGET_HTM, MASK_HTM): Define macros.
8556 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
8557 (FIXED_REGISTERS): Likewise.
8558 (CALL_USED_REGISTERS): Likewise.
8559 (CALL_REALLY_USED_REGISTERS): Likewise.
8560 (REG_ALLOC_ORDER): Likewise.
8561 (enum reg_class): Likewise.
8562 (REG_CLASS_NAMES): Likewise.
8563 (REG_CLASS_CONTENTS): Likewise.
8564 (REGISTER_NAMES): Likewise.
8565 (ADDITIONAL_REGISTER_NAMES): Likewise.
8566 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
8567 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
8568 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
8569 * config/rs6000/htm.md: New file.
8570 * config/rs6000/htmintrin.h: New file.
8571 * config/rs6000/htmxlintrin.h: New file.
8573 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
8575 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8576 Define SYMBOL_TINY_GOT, update comment.
8577 * config/aarch64/aarch64.c
8578 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
8579 (aarch64_expand_mov_immediate): Likewise.
8580 (aarch64_print_operand): Likewise.
8581 (aarch64_classify_symbol): Likewise.
8582 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
8583 (ldr_got_tiny): Define.
8585 2013-07-13 Tobias Grosser <tobias@grosser.es>
8587 PR tree-optimization/54094
8588 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
8589 scheduling dimension for the parallelism check from the polyhedral
8590 information in the AST.
8591 * graphite-dependences.c (carries_deps): Do not assume the schedule is
8594 2013-07-13 Jason Merrill <jason@redhat.com>
8596 * print-tree.c (debug_vec_tree): Use debug_raw.
8597 (debug_raw (vec<tree, va_gc> &)): New.
8598 (debug_raw (vec<tree, va_gc> *)): New.
8599 * tree.h: Declare them.
8601 2013-07-13 Bin Cheng <bin.cheng@arm.com>
8603 * ifcvt.c (ifcvt_after_combine): New static variable.
8604 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
8606 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
8607 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
8608 rest_of_handle_if_after_reload): Pass new argument for if_convert.
8610 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
8612 * config/mips/mips.c (mips_expand_call): Remove empty statement.
8614 2013-07-12 Michael Matz <matz@suse.de>
8617 * convert.c (convert_to_real): Reject non-float inner types.
8619 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
8621 * config/aarch64/aarch64-protos.h
8622 (aarch64_simd_immediate_valid_for_move): Remove.
8623 * config/aarch64/aarch64.c (simd_immediate_info): New member.
8624 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
8626 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
8628 2013-07-11 Steve Ellcey <sellcey@mips.com>
8630 * config/mips/mips.c (mips_conditional_register_usage): Do not
8631 use t[0-7] registers in MIPS16 mode when optimizing for size.
8633 2013-07-11 Sriraman Tallam <tmsriram@google.com>
8635 * config/i386/i386.c (dispatch_function_versions): Fix array
8636 indexing of function_version_info to match actual_versions.
8638 2013-07-11 Teresa Johnson <tejohnson@google.com>
8640 * vec.h (struct va_gc): Move release out-of-line.
8641 (va_gc::release): Call ggc_free on released vec.
8643 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
8645 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
8646 Require GOT register as additional operand in UNSPEC.
8647 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
8648 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8649 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8650 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
8651 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
8652 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
8653 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
8655 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
8658 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
8659 name seen by assembler/linker rather if available.
8661 2013-07-11 Andreas Schwab <schwab@suse.de>
8663 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
8665 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
8667 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
8669 2013-07-10 Joseph Myers <joseph@codesourcery.com>
8671 * doc/tm.texi.in: Move hook documentation to ....
8672 * target.def: ... here.
8674 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
8676 * doc/tm.texi: Regenerate.
8678 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
8681 * doc/invoke.texi: Document -Wconditionally-supported.
8683 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
8686 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
8689 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
8692 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
8693 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
8694 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
8695 Remove duplicate devices.
8696 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
8697 * config/avr/t-multilib: Regenerate.
8698 * config/avr/avr-tables.opt: Regenerate.
8699 * doc/avr-mmcu.texi: Regenerate.
8701 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
8704 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
8706 2013-07-10 Graham Stott <graham.stott@btinternet.com>
8708 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
8709 the cost of MULT when optimizing for size.
8711 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
8713 * config/cr16/cr16-protos.h: Don't include target.h.
8715 2013-07-09 Joseph Myers <joseph@codesourcery.com>
8717 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
8718 adjust register size for TDmode and TFmode for VSX registers.
8720 2013-07-08 Kai Tietz <ktietz@redhat.com>
8723 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
8724 hook_bool_const_tree_true.
8726 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8728 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
8729 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
8730 * config/s390/s390.md: Define FPR*_REGNUM constants.
8731 Fix FPR2_REGNUM constant (18 -> 17).
8732 ("*trunc<BFP:mode><DFP_ALL:mode>2")
8733 ("*trunc<DFP_ALL:mode><BFP:mode>2")
8734 ("trunc<BFP:mode><DFP_ALL:mode>2")
8735 ("trunc<DFP_ALL:mode><BFP:mode>2")
8736 ("*extend<BFP:mode><DFP_ALL:mode>2")
8737 ("*extend<DFP_ALL:mode><BFP:mode>2")
8738 ("extend<BFP:mode><DFP_ALL:mode>2")
8739 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
8742 2013-07-08 Graham Stott <graham.stott@btinternet.com>
8744 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
8746 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8748 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
8749 and cfun_fpr_bit_p to cfun_fpr_save_p.
8750 (s390_frame_area, s390_register_info, s390_frame_info)
8751 (s390_emit_prologue, s390_emit_epilogue)
8752 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
8754 * config/s390/s390.h: Define *_REGNUM macros for floating point
8757 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
8759 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
8761 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
8763 PR rtl-optimization/57786
8764 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
8765 and break out of the loop when it is set to false.
8767 2013-07-08 Jakub Jelinek <jakub@redhat.com>
8770 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
8771 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
8772 (const_int 63)) 0)).
8773 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
8774 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
8775 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
8777 PR rtl-optimization/57829
8778 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
8779 mask bits outside of mode are just sign-extension from mode to HWI.
8781 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
8783 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
8784 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
8785 adjust_address instead of change_address to keep info about alignment.
8786 (emit_strmov): Remove.
8787 (emit_memmov): New function.
8788 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
8789 (expand_movmem_epilogue): Likewise and return updated rtx for
8791 (expand_constant_movmem_prologue): Likewise and return updated rtx for
8792 destination and source.
8793 (decide_alignment): Refactor, handle vector_loop.
8794 (ix86_expand_movmem): Likewise.
8795 (ix86_expand_setmem): Likewise.
8796 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
8798 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
8800 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
8801 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
8803 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
8805 * config/i386/sse.md (sse_movlhps): Change alternative 3
8806 of operand 2 to "m".
8808 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
8811 * config/i386/sse.md (iptr): New mode attribute.
8812 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
8813 (<sse>_vm<plusminus_insn><mode>3): Ditto.
8814 (<sse>_vmmul<mode>3): Ditto.
8815 (<sse>_vmdiv<mode>3): Ditto.
8816 (sse_vmrcpv4sf2): Ditto.
8817 (<sse>_vmsqrt<mode>2): Ditto.
8818 (sse_vmrsqrtv4sf2): Ditto.
8819 (<sse>_vm<code><mode>3): Ditto.
8820 (avx_vmcmp<mode>3): Ditto.
8821 (<sse>_vmmaskcmp<mode>3): Ditto.
8822 (<sse>_comi): Ditto.
8823 (<sse>_ucomi): Ditto.
8824 (*xop_vmfrcz_<mode>): Ditto.
8825 (*fmai_fmadd_<mode>): Ditto.
8826 (*fmai_fmsub_<mode>): Ditto.
8827 (*fmai_fnmadd_<mode>): Ditto.
8828 (*fmai_fnmsub_<mode>): Ditto.
8829 (*fma4i_vmfmadd_<mode>): Ditto.
8830 (*fma4i_vmfmsub_<mode>): Ditto.
8831 (*fma4i_vmfnmadd_<mode>): Ditto.
8832 (*fma4i_vmfnmsub_<mode>): Ditto.
8833 (*xop_vmfrcz_<mode>): Ditto.
8834 (sse_cvtps2pi): Ditto.
8835 (sse_cvttps2pi): Ditto.
8836 (sse_cvtss2si): Ditto.
8837 (sse_cvtss2si_2): Ditto.
8838 (sse_cvtss2siq_2): Ditto.
8839 (sse_cvttss2si): Ditto.
8840 (sse_cvttss2siq): Ditto.
8841 (sse_cvtsd2si): Ditto.
8842 (sse_cvtsd2si_2): Ditto.
8843 (sse_cvtsd2siq_2): Ditto.
8844 (sse_cvttsd2si): Ditto.
8845 (sse_cvttsd2siq): Ditto.
8846 (sse_cvtsd2ss): Ditto.
8847 (sse_cvtss2sd): Ditto.
8848 (avx2_pbroadcast<mode>): Ditto.
8849 (avx2_pbroadcast<mode>_1): Ditto.
8850 (*avx_vperm_broadcast_v4sf): Ditto.
8852 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
8853 (sse_movlhps): Ditto.
8854 (sse_storehps): Ditto.
8855 (sse_loadhps): Ditto.
8856 (sse_storelps): Ditto.
8857 (sse_loadlps): Ditto.
8858 (*vec_concatv4sf): Ditto.
8859 (*vec_interleave_highv2df): Ditto.
8860 (*vec_interleave_lowv2df): Ditto.
8861 (*vec_extractv2df_1_sse): Ditto.
8862 (*vec_extractv2df_0_sse): Ditto.
8863 (sse2_storelpd): Ditto.
8864 (sse2_loadlpd): Ditto.
8865 (sse2_movsd): Ditto.
8866 (*vec_concatv4si): Ditto.
8867 (vec_concatv2di): Ditto.
8869 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
8870 for Intel asm dialect.
8871 (mmx_punpcklwd): Ditto.
8872 (mmx_punpckldq): Ditto.
8874 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
8875 for intel assembler dialect.
8877 2013-07-06 Jakub Jelinek <jakub@redhat.com>
8880 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
8881 for BUILT_IN_C{LZ,LRSB}*.
8882 * tree.h (CASE_INT_FN): Add FN##IMAX case.
8883 * tree-vrp.c (extract_range_basic): Handle
8884 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
8885 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
8886 fall thru to code calling set_value*.
8887 * builtins.c (expand_builtin): Remove *IMAX cases.
8888 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
8889 if width is bigger than 2*HWI.
8891 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
8893 PR rtl-optimization/55342
8894 * lra-int.h (lra_subreg_reload_pseudos): New.
8895 * lra.c: Add undoing optional reloads to the block diagram.
8896 (lra_subreg_reload_pseudos): New.
8897 (lra_optional_reload_pseudos): Change comments.
8898 (lra): Init and clear lra_subreg_reload_pseudos. Clear
8899 lra_optional_reload_pseudos after undo transformations.
8900 * lra-assigns.c (pseudo_prefix_title): New.
8901 (lra_setup_reg_renumber): Use it.
8902 (spill_for): Ditto. Check subreg reload pseudos too.
8903 (assign_by_spills): Consider subreg reload pseudos too.
8904 * lra-constraints.c (simplify_operand_subreg): Use
8905 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
8906 (curr_insn_transform): Recognize and do optional reloads.
8907 (undo_optional_reloads): New.
8908 (lra_undo_inheritance): Call undo_optional_reloads.
8910 2013-07-05 Thomas Quinot <quinot@adacore.com>
8912 * tree-complex.c (expand_complex_operations_1): Fix typo.
8914 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
8916 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
8917 (tune_params): New member 'const vec_costs'.
8918 * config/aarch64/aarch64.c (generic_vector_cost): New.
8919 (generic_tunings): New member 'generic_vector_cost'.
8920 (aarch64_builtin_vectorization_cost): New.
8921 (aarch64_add_stmt_cost): New.
8922 (TARGET_VECTORIZE_ADD_STMT_COST): New.
8923 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
8925 2013-07-03 Jakub Jelinek <jakub@redhat.com>
8928 * config/i386/predicates.md (vsib_address_operand): Disallow
8929 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
8931 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
8934 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
8935 expand_builtin_setjmp_receiver.
8936 (expand_label): Adjust, call expand_builtin_setjmp_receiver
8937 with NULL for the label parameter.
8938 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
8939 the frame-pointer. Adjust comments.
8940 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
8941 only if LABEL is non-NULL.
8943 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
8945 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
8946 (ARM_ABI_AAPCS64): Ditto.
8948 (ARM_DEFAULT_ABI): Ditto.
8950 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8952 * config/aarch64/aarch64-builtins.c
8953 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
8954 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
8956 * config/aarch64/aarch64-simd.md
8957 (aarch64_ld1<VALL:mode>): New.
8958 (aarch64_st1<VALL:mode>): Likewise.
8959 * config/aarch64/arm_neon.h
8960 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
8962 2013-07-02 Sriraman Tallam <tmsriram@google.com>
8964 * config/i386/i386.c (gate_insert_vzeroupper): Check if
8966 (ix86_option_override_internal):Turn on all -mavx target flags by
8967 default as they are dependent on AVX anyway.
8969 2013-07-02 Cary Coutant <ccoutant@google.com>
8971 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
8973 (loc_checksum_ordered): Likewise.
8974 (hash_loc_operands): Remove inline keyword.
8976 2013-07-02 Jakub Jelinek <jakub@redhat.com>
8978 PR tree-optimization/57741
8979 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
8980 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
8981 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
8982 Allow REAL_CST step_exprs if flag_associative_math.
8983 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
8985 2013-07-02 Ian Bolton <ian.bolton@arm.com>
8987 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
8989 2013-07-02 Ian Bolton <ian.bolton@arm.com>
8991 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
8993 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8995 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
8997 (iorsi3_insn): Likewise.
8998 (arm_xorsi3): Likewise.
9000 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
9002 * arm.md (attribute "wtype"): Delete. Move attribute values from here
9004 (attribute "type"): ... here, and prefix with "wmmx_".
9005 (attribute "core_cycles"): Update for attribute changes.
9006 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
9007 (tbcstv4hi): Likewise.
9008 (tbcstv2si): Likewise.
9009 (iwmmxt_iordi3): Likewise.
9010 (iwmmxt_xordi3): Likewise.
9011 (iwmmxt_anddi3): Likewise.
9012 (iwmmxt_nanddi3): Likewise.
9013 (iwmmxt_arm_movdi): Likewise.
9014 (iwmmxt_movsi_insn): Likewise.
9015 (mov<mode>_internal): Likewise.
9016 (and<mode>3_iwmmxt): Likewise.
9017 (ior<mode>3_iwmmxt): Likewise.
9018 (xor<mode>3_iwmmxt): Likewise.
9019 (add<mode>3_iwmmxt): Likewise.
9020 (ssaddv8qi3): Likewise.
9021 (ssaddv4hi3): Likewise.
9022 (ssaddv2si3): Likewise.
9023 (usaddv8qi3): Likewise.
9024 (usaddv4hi3): Likewise.
9025 (usaddv2si3): Likewise.
9026 (sub<mode>3_iwmmxt): Likewise.
9027 (sssubv8qi3): Likewise.
9028 (sssubv4hi3): Likewise.
9029 (sssubv2si3): Likewise.
9030 (ussubv8qi3): Likewise.
9031 (ussubv4hi3): Likewise.
9032 (ussubv2si3): Likewise.
9033 (mulv4hi3_iwmmxt): Likewise.
9034 (smulv4hi3_highpart): Likewise.
9035 (umulv4hi3_highpart): Likewise.
9036 (iwmmxt_wmacs): Likewise.
9037 (iwmmxt_wmacsz): Likewise.
9038 (iwmmxt_wmacu): Likewise.
9039 (iwmmxt_wmacuz): Likewise.
9040 (iwmmxt_clrdi): Likewise.
9041 (iwmmxt_clrv8qi): Likewise.
9042 (iwmmxt_clr4hi): Likewise.
9043 (iwmmxt_clr2si): Likewise.
9044 (iwmmxt_uavgrndv8qi3): Likewise.
9045 (iwmmxt_uavgrndv4hi3): Likewise.
9046 (iwmmxt_uavgv8qi3): Likewise.
9047 (iwmmxt_uavgv4hi3): Likewise.
9048 (iwmmxt_tinsrb): Likewise.
9049 (iwmmxt_tinsrh): Likewise.
9050 (iwmmxt_tinsrw): Likewise.
9051 (iwmmxt_textrmub): Likewise.
9052 (iwmmxt_textrmsb): Likewise.
9053 (iwmmxt_textrmuh): Likewise.
9054 (iwmmxt_textrmsh): Likewise.
9055 (iwmmxt_textrmw): Likewise.
9056 (iwmxxt_wshufh): Likewise.
9057 (eqv8qi3): Likewise.
9058 (eqv4hi3): Likewise.
9059 (eqv2si3): Likewise.
9060 (gtuv8qi3): Likewise.
9061 (gtuv4hi3): Likewise.
9062 (gtuv2si3): Likewise.
9063 (gtv8qi3): Likewise.
9064 (gtv4hi3): Likewise.
9065 (gtv2si3): Likewise.
9066 (smax<mode>3_iwmmxt): Likewise.
9067 (umax<mode>3_iwmmxt): Likewise.
9068 (smin<mode>3_iwmmxt): Likewise.
9069 (umin<mode>3_iwmmxt): Likewise.
9070 (iwmmxt_wpackhss): Likewise.
9071 (iwmmxt_wpackwss): Likewise.
9072 (iwmmxt_wpackdss): Likewise.
9073 (iwmmxt_wpackhus): Likewise.
9074 (iwmmxt_wpackwus): Likewise.
9075 (iwmmxt_wpackdus): Likewise.
9076 (iwmmxt_wunpckihb): Likewise.
9077 (iwmmxt_wunpckihh): Likewise.
9078 (iwmmxt_wunpckihw): Likewise.
9079 (iwmmxt_wunpckilb): Likewise.
9080 (iwmmxt_wunpckilh): Likewise.
9081 (iwmmxt_wunpckilw): Likewise.
9082 (iwmmxt_wunpckehub): Likewise.
9083 (iwmmxt_wunpckehuh): Likewise.
9084 (iwmmxt_wunpckehuw): Likewise.
9085 (iwmmxt_wunpckehsb): Likewise.
9086 (iwmmxt_wunpckehsh): Likewise.
9087 (iwmmxt_wunpckehsw): Likewise.
9088 (iwmmxt_wunpckelub): Likewise.
9089 (iwmmxt_wunpckeluh): Likewise.
9090 (iwmmxt_wunpckeluw): Likewise.
9091 (iwmmxt_wunpckelsb): Likewise.
9092 (iwmmxt_wunpckelsh): Likewise.
9093 (iwmmxt_wunpckelsw): Likewise.
9094 (ror<mode>3): Likewise.
9095 (ashr<mode>3_iwmmxt): Likewise.
9096 (lshr<mode>3_iwmmxt): Likewise.
9097 (ashl<mode>3_iwmmxt): Likewise.
9098 (ror<mode>3_di): Likewise.
9099 (ashr<mode>3_di): Likewise.
9100 (lshr<mode>3_di): Likewise.
9101 (ashl<mode>3_di): Likewise.
9102 (iwmmxt_wmadds): Likewise.
9103 (iwmmxt_wmaddu): Likewise.
9104 (iwmmxt_tmia): Likewise.
9105 (iwmmxt_tmiaph): Likewise.
9106 (iwmmxt_tmiabb): Likewise.
9107 (iwmmxt_tmiatb): Likewise.
9108 (iwmmxt_tmiabt): Likewise.
9109 (iwmmxt_tmiatt): Likewise.
9110 (iwmmxt_tmovmskb): Likewise.
9111 (iwmmxt_tmovmskh): Likewise.
9112 (iwmmxt_tmovmskw): Likewise.
9113 (iwmmxt_waccb): Likewise.
9114 (iwmmxt_wacch): Likewise.
9115 (iwmmxt_waccw): Likewise.
9116 (iwmmxt_waligni): Likewise.
9117 (iwmmxt_walignr): Likewise.
9118 (iwmmxt_walignr0): Likewise.
9119 (iwmmxt_walignr1): Likewise.
9120 (iwmmxt_walignr2): Likewise.
9121 (iwmmxt_walignr3): Likewise.
9122 (iwmmxt_wsadb): Likewise.
9123 (iwmmxt_wsadh): Likewise.
9124 (iwmmxt_wsadbz): Likewise.
9125 (iwmmxt_wsadhz): Likewise.
9126 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
9127 (iwmmxt_wabsdiffb): Likewise.
9128 (iwmmxt_wabsdiffh): Likewise.
9129 (iwmmxt_wabsdiffw): Likewise.
9130 (iwmmxt_waddsubhx): Likewise
9131 (iwmmxt_wsubaddhx): Likewise.
9132 (addc<mode>3): Likewise.
9133 (iwmmxt_avg4): Likewise.
9134 (iwmmxt_avg4r): Likewise.
9135 (iwmmxt_wmaddsx): Likewise.
9136 (iwmmxt_wmaddux): Likewise.
9137 (iwmmxt_wmaddsn): Likewise.
9138 (iwmmxt_wmaddun): Likewise.
9139 (iwmmxt_wmulwsm): Likewise.
9140 (iwmmxt_wmulwum): Likewise.
9141 (iwmmxt_wmulsmr): Likewise.
9142 (iwmmxt_wmulumr): Likewise.
9143 (iwmmxt_wmulwsmr): Likewise.
9144 (iwmmxt_wmulwumr): Likewise.
9145 (iwmmxt_wmulwl): Likewise.
9146 (iwmmxt_wqmulm): Likewise.
9147 (iwmmxt_wqmulwm): Likewise.
9148 (iwmmxt_wqmulmr): Likewise.
9149 (iwmmxt_wqmulwmr): Likewise.
9150 (iwmmxt_waddbhusm): Likewise.
9151 (iwmmxt_waddbhusl): Likewise.
9152 (iwmmxt_wqmiabb): Likewise.
9153 (iwmmxt_wqmiabt): Likewise.
9154 (iwmmxt_wqmiatb): Likewise.
9155 (iwmmxt_wqmiatt): Likewise.
9156 (iwmmxt_wqmiabbn): Likewise.
9157 (iwmmxt_wqmiabtn): Likewise.
9158 (iwmmxt_wqmiatbn): Likewise.
9159 (iwmmxt_wqmiattn): Likewise.
9160 (iwmmxt_wmiabb): Likewise.
9161 (iwmmxt_wmiabt): Likewise.
9162 (iwmmxt_wmiatb): Likewise.
9163 (iwmmxt_wmiatt): Likewise.
9164 (iwmmxt_wmiabbn): Likewise.
9165 (iwmmxt_wmiabtn): Likewise.
9166 (iwmmxt_wmiatbn): Likewise.
9167 (iwmmxt_wmiattn): Likewise.
9168 (iwmmxt_wmiawbb): Likewise.
9169 (iwmmxt_wmiawbt): Likewise.
9170 (iwmmxt_wmiawtb): Likewise.
9171 (iwmmxt_wmiawtt): Likewise.
9172 (iwmmxt_wmiawbbn): Likewise.
9173 (iwmmxt_wmiawbtn): Likewise.
9174 (iwmmxt_wmiawtbn): Likewise.
9175 (iwmmxt_wmiawttn): Likewise.
9176 (iwmmxt_wmerge): Likewise.
9177 (iwmmxt_tandc<mode>3): Likewise.
9178 (iwmmxt_torc<mode>3): Likewise.
9179 (iwmmxt_torvsc<mode>3): Likewise.
9180 (iwmmxt_textrc<mode>3): Likewise.
9181 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
9182 (wmmxt_pack): Likewise.
9183 (wmmxt_mult_c1): Likewise.
9184 (wmmxt_mult_c2): Likewise.
9185 (wmmxt_alu_c1): Likewise.
9186 (wmmxt_alu_c2): Likewise.
9187 (wmmxt_alu_c3): Likewise.
9188 (wmmxt_transfer_c1): Likewise.
9189 (wmmxt_transfer_c2): Likewise.
9190 (wmmxt_transfer_c3): Likewise.
9191 (marvell_f_iwmmxt_wstr): Likewise.
9192 (marvell_f_iwmmxt_wldr): Likewise.
9194 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
9196 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
9198 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
9201 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
9202 * lra-constraints.c (need_for_split_p): Check call used hard regs
9203 living through calls.
9205 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
9206 call used regs for call insn.
9208 2013-06-28 Jakub Jelinek <jakub@redhat.com>
9211 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
9212 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
9214 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
9216 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
9218 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
9220 * lra-constraints.c (need_for_split_p): Check call used hard regs
9221 living through calls.
9223 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9226 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
9227 to tie with any other modes. Eliminate Altivec vector mode tests,
9228 since these are a subset of ALTIVEC or VSX vector modes. Simplify
9229 code, to return 0 if testing MODE2 for a condition, if we've
9230 already tested MODE1 for the same condition.
9232 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
9234 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
9237 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
9239 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
9240 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
9242 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
9244 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
9246 (aarch64_symbolic_constant_p): Remove.
9247 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
9248 static. Fix line length and white space.
9249 (aarch64_symbolic_constant_p): Remove.
9250 * config/aarch64/predicates.md (aarch64_valid_symref):
9251 Use aarch64_classify_symbol_expression.
9253 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9255 * config/arm/constraints.md (Ts): New constraint.
9256 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
9258 (compare_scc): Use "Ts" constraint for operand 0.
9259 (ior_scc_scc): Likewise.
9260 (and_scc_scc): Likewise.
9261 (and_scc_scc_nodom): Likewise.
9262 (ior_scc_scc_cmp): Likewise for operand 7.
9263 (and_scc_scc_cmp): Likewise.
9264 * config/arm/thumb2.md (thumb2_movsi_insn):
9265 Add alternatives for 16-bit encodings.
9266 (thumb2_movhi_insn): Likewise.
9267 (thumb2_movsicc_insn): Likewise.
9268 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
9269 (thumb2_negscc): Use "Ts" constraint.
9270 Move mvn instruction outside cond_exec block.
9271 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
9272 for 16-bit encodings.
9274 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9276 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
9278 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
9279 (mulsi3subsi): Likewise.
9280 (mulsidi3adddi): Likewise.
9281 (mulsidi3_v6): Likewise.
9282 (umulsidi3_v6): Likewise.
9283 (umulsidi3adddi_v6): Likewise.
9284 (smulsi3_highpart_v6): Likewise.
9285 (umulsi3_highpart_v6): Likewise.
9286 (mulhisi3tb): Likewise.
9287 (mulhisi3bt): Likewise.
9288 (mulhisi3tt): Likewise.
9289 (maddhisi4): Likewise.
9290 (maddhisi4tb): Likewise.
9291 (maddhisi4tt): Likewise.
9292 (maddhidi4): Likewise.
9293 (maddhidi4tb): Likewise.
9294 (maddhidi4tt): Likewise.
9295 (zeroextractsi_compare0_scratch): Likewise.
9296 (insv_zero): Likewise.
9297 (insv_t2): Likewise.
9298 (anddi_notzesidi_di): Likewise.
9299 (anddi_notsesidi_di): Likewise.
9300 (andsi_notsi_si): Likewise.
9301 (iordi_zesidi_di): Likewise.
9302 (xordi_zesidi_di): Likewise.
9303 (andsi_iorsi3_notsi): Likewise.
9305 (smax_m1): Likewise.
9307 (not_shiftsi): Likewise.
9308 (unaligned_loadsi): Likewise.
9309 (unaligned_loadhis): Likewise.
9310 (unaligned_loadhiu): Likewise.
9311 (unaligned_storesi): Likewise.
9312 (unaligned_storehi): Likewise.
9313 (extv_reg): Likewise.
9314 (extzv_t2): Likewise.
9316 (udivsi3): Likewise.
9317 (arm_zero_extendhisi2addsi): Likewise.
9318 (arm_zero_extendqisi2addsi): Likewise.
9319 (compareqi_eq0): Likewise.
9320 (arm_extendhisi2_v6): Likewise.
9321 (arm_extendqisi2addsi): Likewise.
9322 (arm_movt): Likewise.
9323 (thumb2_ldrd): Likewise.
9324 (thumb2_ldrd_base): Likewise.
9325 (thumb2_ldrd_base_neg): Likewise.
9326 (thumb2_strd): Likewise.
9327 (thumb2_strd_base): Likewise.
9328 (thumb2_strd_base_neg): Likewise.
9329 (arm_negsi2): Add alternative for 16-bit encoding.
9330 (arm_one_cmplsi2): Likewise.
9332 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9334 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
9335 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
9336 (movdfcc): Likewise.
9337 * config/arm/vfp.md (*thumb2_movsf_vfp):
9338 Disable predication for arm_restrict_it.
9339 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
9340 (*thumb2_movdfcc_vfp): Likewise.
9341 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
9342 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
9343 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
9344 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
9345 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
9346 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
9347 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
9348 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
9349 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
9350 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
9351 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
9352 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
9353 Disable predication for arm_restrict_it.
9355 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
9357 * config/i386/bmiintrin.h (_bextr_u32): New.
9358 (_bextr_u64): Ditto.
9360 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
9362 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
9363 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
9364 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
9365 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
9366 * config/mips/n32-elf.h: ...this new file.
9368 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
9371 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
9372 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
9374 2013-06-27 Catherine Moore <clm@codesourcery.com>
9376 * config/mips/mips-tables.opt: Regenerate.
9377 * config/mips/mips-cpus.def: Add m14ke and m14kec.
9378 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
9379 * doc/invoke.texi: Add -m14kc.
9381 2013-06-27 Jakub Jelinek <jakub@redhat.com>
9384 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
9385 constraints of operand 1 and 2.
9388 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
9389 to match RTL canonicalization. Swap predicates and
9390 constraints of operand 1 and 2.
9392 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
9394 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
9395 Process OP_INOUT regs for splitting too.
9397 2013-06-27 Jakub Jelinek <jakub@redhat.com>
9399 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
9400 decl before the loop, initialize to NULL.
9401 (vectorizable_load): Initialize ptr_incr to NULL.
9403 2013-06-27 Martin Jambor <mjambor@suse.cz>
9406 * ipa-ref.h (ipa_maybe_record_reference): Declare.
9407 * ipa-ref.c (ipa_maybe_record_reference): New function.
9408 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
9409 * ipa-cp.c (create_specialized_node): Record potential references from
9411 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
9413 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
9415 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
9416 parameter 'mode' of type 'enum machine_mode mode'; change to pass
9417 'mode' to force_reg.
9418 (aarch64_add_offset): Update calls to aarch64_force_temporary.
9419 (aarch64_expand_mov_immediate): Likewise.
9421 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
9423 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
9424 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
9426 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9428 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
9429 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9430 (struct machine_function): Add tbegin_p.
9431 (s390_canonicalize_comparison): Fold CC mode compares to
9432 conditional jump if possible.
9433 (s390_emit_jump): Return the emitted jump.
9434 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
9435 Handle CCRAWmode compares.
9436 (s390_option_override): Default to -mhtm if available.
9437 (s390_reg_clobbered_rtx): Handle floating point regs as well.
9438 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
9439 FPRs instead of df_regs_ever_live_p.
9440 (s390_optimize_nonescaping_tx): New function.
9441 (s390_init_frame_layout): Extend clobbered_regs array to cover
9443 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
9444 (s390_expand_tbegin): New function.
9445 (enum s390_builtin): New enum definition.
9446 (code_for_builtin): New array definition.
9447 (s390_init_builtins): New function.
9448 (s390_expand_builtin): New function.
9449 (TARGET_INIT_BUILTINS): Define.
9450 (TARGET_EXPAND_BUILTIN): Define.
9451 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
9452 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
9453 (s390_alc_comparison): Likewise.
9454 * config/s390/s390-modes.def: Add CCRAWmode.
9455 * config/s390/s390.h (processor_flags): Add PF_TX.
9456 (TARGET_CPU_HTM): Define macro.
9457 (TARGET_HTM): Define macro.
9458 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
9459 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
9460 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9461 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
9462 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
9463 (TBEGIN_MASK, TBEGINC_MASK): New constants.
9464 ("*cc_to_int"): Move up.
9465 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
9466 constants other than 0.
9467 ("*ccraw_to_int"): New insn and splitter definition.
9468 ("tbegin", "tbegin_nofloat", "tbegin_retry")
9469 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
9470 ("tx_assist"): New expander.
9471 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
9472 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
9473 * config/s390/s390.opt: Add -mhtm option.
9474 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
9475 * config/s390/htmxlintrin.h: New file.
9476 * config/s390/htmintrin.h: New file.
9477 * config/s390/s390intrin.h: New file.
9478 * doc/extend.texi: Document htm builtins.
9479 * config.gcc: Add the new header files to extra_headers.
9481 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
9483 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
9484 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
9486 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9487 Pat Haugen <pthaugen@us.ibm.com>
9488 Peter Bergner <bergner@vnet.ibm.com>
9490 * config/rs6000/power8.md: New.
9491 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
9492 setting for power8 entry.
9493 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
9494 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
9495 test for Power4/Power5 only.
9496 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
9498 (force_new_group): Adjust comment.
9499 * config/rs6000/rs6000.md: Include power8.md.
9501 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
9503 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
9504 * config/arm/arm-protos.h (arm_max_conditional_execute): New
9506 (tune_params): Update comment.
9507 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
9508 (arm_max_conditional_execute): New function.
9509 (thumb2_final_prescan_insn): Use max_insn_skipped and
9510 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
9512 2013-06-25 Jakub Jelinek <jakub@redhat.com>
9514 PR tree-optimization/57705
9515 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
9516 SSA_NAME step, provided that it is not defined inside the loop.
9517 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
9518 (get_initial_def_for_induction): Handle SSA_NAME IV step.
9520 2013-06-25 Martin Jambor <mjambor@suse.cz>
9523 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
9524 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
9526 (ipa_note_param_call): Initialize member_ptr flag.
9527 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
9528 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
9529 (ipa_write_indirect_edge_info): Stream member_ptr flag.
9530 (ipa_read_indirect_edge_info): Likewise.
9532 2013-06-25 Richard Biener <rguenther@suse.de>
9535 * passes.c (init_optimization_passes): Move pass_fold_builtins
9536 and pass_dce earlier with -Og.
9538 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
9540 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
9541 <BIT_FIELD_REF>: Remove trailing TAB.
9542 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
9545 2013-06-24 Martin Jambor <mjambor@suse.cz>
9547 PR tree-optimization/57358
9548 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
9549 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
9550 (ipa_analyze_params_uses): Generate pessimistic info when true.
9552 2013-06-24 Martin Jambor <mjambor@suse.cz>
9554 PR tree-optimization/57539
9555 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
9556 global.inlined_to of the new node to it. All callers changed.
9557 * ipa-inline-transform.c (clone_inlined_nodes): New variable
9558 inlining_into, pass it to cgraph_clone_node.
9559 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
9560 ipa_free_edge_args_substructures.
9561 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
9562 rdesc linked list. Do not assert rdesc edges have inlined caller.
9563 Assert we have found an rdesc in the rdesc list.
9565 2013-06-24 Richard Biener <rguenther@suse.de>
9567 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
9568 (pointer_set_lookup): Declare.
9569 (class pointer_map): New template class implementing a
9570 generic pointer to T map.
9571 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
9572 pointer_map<T>::contains, pointer_map<T>::insert,
9573 pointer_map<T>::traverse): New functions.
9574 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
9575 (pointer_set_lookup): New function.
9576 (pointer_set_contains): Use pointer_set_lookup.
9577 (pointer_set_insert): Likewise.
9578 (insert_aux): Remove.
9579 (struct pointer_map_t): Embed a pointer_set_t.
9580 (pointer_map_create): Adjust.
9581 (pointer_map_destroy): Likewise.
9582 (pointer_map_contains): Likewise.
9583 (pointer_map_insert): Likewise.
9584 (pointer_map_traverse): Likewise.
9585 * tree-streamer.h (struct streamer_tree_cache_d): Use a
9586 pointer_map<unsigned> instead of a pointer_map_t.
9587 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
9588 (streamer_tree_cache_lookup): Likewise.
9589 (streamer_tree_cache_create): Likewise.
9590 (streamer_tree_cache_delete): Likewise.
9591 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
9592 pointer_map<unsigned> instead of a pointer_map_t.
9593 (lto_init_tree_ref_encoder): Adjust.
9594 (lto_destroy_tree_ref_encoder): Likewise.
9595 * lto-section-out.c (lto_output_decl_index): Likewise.
9596 (lto_record_function_out_decl_state): Likewise.
9597 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
9599 2013-06-24 Richard Biener <rguenther@suse.de>
9601 PR tree-optimization/57488
9602 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
9604 2013-06-24 Alan Modra <amodra@gmail.com>
9606 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
9607 (gen_easy_altivec_constant): Likewise.
9608 * config/rs6000/predicates.md (easy_vector_constant_add_self,
9609 easy_vector_constant_msb): Likewise.
9611 2013-06-23 Jakub Jelinek <jakub@redhat.com>
9614 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
9615 add missing return true.
9617 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
9620 * config/sh/predicates.md (general_extend_operand): Invoke
9621 general_movsrc_operand for memory operands.
9622 (general_movsrc_operand): Allow reg+reg addressing, do not use
9623 general_operand for memory operands.
9625 2013-06-23 Sriraman Tallam <tmsriram@google.com>
9627 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
9628 when current target options does not apply.
9629 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
9630 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
9631 * config/i386/bmiintrin.h: Pass appropriate target
9632 attributes to header.
9633 * config/i386/mmintrin.h: Ditto.
9634 * config/i386/nmmintrin.h: Ditto.
9635 * config/i386/avx2intrin.h: Ditto.
9636 * config/i386/fxsrintrin.h: Ditto.
9637 * config/i386/tbmintrin.h: Ditto.
9638 * config/i386/xsaveintrin.h: Ditto.
9639 * config/i386/f16cintrin.h: Ditto.
9640 * config/i386/xtestintrin.h: Ditto.
9641 * config/i386/xsaveoptintrin.h: Ditto.
9642 * config/i386/bmi2intrin.h: Ditto.
9643 * config/i386/lzcntintrin.h: Ditto.
9644 * config/i386/smmintrin.h: Ditto.
9645 * config/i386/wmmintrin.h: Ditto.
9646 * config/i386/x86intrin.h: Remove all header include guards.
9647 * config/i386/prfchwintrin.h: Ditto.
9648 * config/i386/pmmintrin.h: Ditto.
9649 * config/i386/tmmintrin.h: Ditto.
9650 * config/i386/xmmintrin.h: Ditto.
9651 * config/i386/popcntintrin.h: Ditto.
9652 * config/i386/rdseedintrin.h: Ditto.
9653 * config/i386/ammintrin.h: Ditto.
9654 * config/i386/emmintrin.h: Ditto.
9655 * config/i386/immintrin.h: Remove all header include guards.
9656 * config/i386/fma4intrin.h: Ditto.
9657 * config/i386/lwpintrin.h: Ditto.
9658 * config/i386/xopintrin.h: Ditto.
9659 * config/i386/ia32intrin.h: Ditto.
9660 * config/i386/avxintrin.h: Ditto.
9661 * config/i386/rtmintrin.h: Ditto.
9662 * config/i386/fmaintrin.h: Ditto.
9663 * config/i386/mm3dnow.h: Ditto.
9665 2013-06-22 Sriraman Tallam <tmsriram@google.com>
9667 * common/config/i386/i386-common.c: Handle LZCNT.
9669 2013-06-22 Andi Kleen <ak@linux.intel.com>
9671 * doc/extend.texi: Use __atomic_store_n instead of
9672 __atomic_store in HLE example.
9674 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
9676 * config/sh/sh.c: Remove <cstdlib> workaround.
9678 2013-06-21 Andi Kleen <ak@linux.intel.com>
9680 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
9682 2013-06-21 Andi Kleen <ak@linux.intel.com>
9684 * doc/extend.texi: Document that __atomic_clear and
9685 __atomic_test_and_set should only be used with bool.
9687 2013-06-20 Jan Hubicka <jh@suse.cz>
9689 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
9691 * tree.c (decls_same_for_odr): New function.
9692 (same_for_edr): New function.
9693 (types_same_for_odr): New function.
9694 (get_binfo_at_offset): Use it.
9695 * tree.h (types_same_for_odr): Declare.
9697 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
9698 Jason Merrill <jason@redhat.com>
9700 * system.h: Include <cstdlib> as well as <stdlib.h>.
9702 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
9705 * config/i386/i386.c (construct_container): Report error if
9706 long double is used with disabled x87 float returns.
9708 2013-06-20 Jan Hubicka <jh@suse.cz>
9710 * lto-cgraph.c (input_symtab): Do not set cgraph state.
9712 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
9714 PR rtl-optimization/57425
9715 PR rtl-optimization/57569
9716 * alias.c (write_dependence_p): Remove parameters mem_mode and
9717 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
9718 Changed all callers.
9719 (canon_anti_dependence): Get comments and semantics in sync.
9720 Add parameter mem_canonicalized. Changed all callers.
9721 * rtl.h (canon_anti_dependence): Update prototype.
9723 2013-06-20 Richard Biener <rguenther@suse.de>
9725 * data-streamer-in.c (streamer_read_uhwi): Optimize single
9726 byte case, inline streamer_read_uchar and defer section
9729 2013-06-20 Richard Biener <rguenther@suse.de>
9731 PR tree-optimization/57584
9732 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
9733 SSA names into the expanded expression that take part in
9734 abnormal coalescing.
9736 2013-06-19 Sharad Singhai <singhai@google.com>
9738 * gcov.c (print_usage): Handle new option.
9739 (process_args): Ditto.
9740 (get_gcov_intermediate_filename): New function.
9741 (output_intermediate_file): New function.
9742 (output_gcov_file): New function
9743 (generate_results): Handle new option.
9744 (release_function): Relase demangled name.
9745 (read_graph_file): Handle demangled name.
9746 (output_lines): Ditto.
9747 * doc/gcov.texi: Document gcov intermediate format.
9749 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
9752 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
9753 (lra_emit_add): Use the functions. Add comment about Y as an
9756 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
9759 * collect2.c (collect_atexit): New.
9760 (collect_exit): Delete.
9761 (main): Register collect_atexit with atexit.
9762 (collect_wait): Change collect_exit to exit.
9764 * collect2.h (collect_exit): Delete.
9765 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
9767 2013-06-19 Wei Mi <wmi@google.com>
9769 PR rtl-optimization/57518
9770 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
9771 if regno is used in paradoxical subreg.
9772 (update_equiv_regs): Check pdx_subregs[regno] before
9773 set a reg to be equivalent with a mem.
9775 2013-06-19 Matthias Klose <doko@ubuntu.com>
9778 * file-find.h (find_a_file): Add a mode parameter.
9779 * file-find.c (find_a_file): Likewise.
9780 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
9781 with X_OK for the executables.
9782 * collect2.c (main): Call find_a_file with X_OK.
9784 2013-06-19 Steve Ellcey <sellcey@mips.com>
9787 * config/mips/mips.md (casesi_internal_mips16_<mode>):
9788 Use NEXT_INSN instead of next_real_insn.
9790 2013-06-19 Jan Hubicka <jh@suse.cz>
9792 * cgraph.h (const_value_known_p): Replace by ...
9793 (ctor_for_folding): .. this one.
9794 * cgraphunit.c (process_function_and_variable_attributes): Use it.
9795 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
9796 * expr.c (expand_expr_real_1): Likewise.
9797 (string_constant): Likewise.
9798 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
9799 * ipa.c (process_references): Likewise.
9800 (symtab_remove_unreachable_nodes): Likewise.
9801 * ipa-inline-analysis.c (param_change_prob): Likewise.
9802 * gimple-fold.c (canonicalize_constructor_val): Likewise.
9803 (get_base_constructor): Likwise.
9804 * varpool.c (varpool_remove_node): Likewise.
9805 (varpool_remove_initializer): LIkewise.
9806 (dump_varpool_node): LIkwise.
9807 (const_value_known_p): Rewrite to ...
9808 (ctor_for_folding): ... this one.
9810 2013-06-19 Jakub Jelinek <jakub@redhat.com>
9813 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
9814 PERSONALITY in $PATH derived prefixes.
9816 2013-06-19 Jeff Law <law@redhat.com>
9818 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
9821 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
9822 (simplify_bitwise_binary): Use it to simpify certain binary ops on
9825 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
9827 * config/arm/vfp.md: Move VFP instruction classification documentation
9829 * config/arm/arm.md: ... here. Update instruction classification
9832 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
9834 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
9835 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
9836 pattern. Use more efficient sequences on ARMv5 and Thumb2.
9838 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
9841 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9842 with NEXT_INSN. Use tablejump_p to check for jump table data
9845 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
9848 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
9849 that now in C++ the value is correct per the C++ standards.
9851 2013-06-19 Richard Biener <rguenther@suse.de>
9853 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
9856 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9859 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9862 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9863 with next_active_insn.
9865 2013-06-18 Sriraman Tallam <tmsriram@google.com>
9867 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
9868 functions are inlined during failures to flag an error.
9869 * tree-inline.c (expand_call_inline): Allow the error to be flagged
9870 in early inline pass.
9872 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
9874 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
9877 2013-06-18 Julian Brown <julian@codesourcery.com>
9879 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
9880 Permit virtual register pre-reload if !strict.
9881 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
9883 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
9885 * config/arm/neon.md (movmisalign<mode>): Use
9886 neon_perm_struct_or_reg_operand instead of
9887 neon_struct_or_register_operand.
9888 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
9889 neon_permissive_struct_operand instead of neon_struct_operand.
9890 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
9891 neon_vector_mem_operand.
9892 * config/arm/predicates.md (neon_struct_operand): Adjust call to
9893 neon_vector_mem_operand.
9894 (neon_permissive_struct_operand): New.
9895 (neon_struct_or_register_operand): Rename to...
9896 (neon_perm_struct_or_reg_operand): This. Adjust call to
9897 neon_vector_mem_operand.
9899 2013-06-18 Richard Biener <rguenther@suse.de>
9901 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
9902 * lto-streamer.h: Include pointer-set.h.
9903 (struct lto_decl_slot): Remove.
9904 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
9905 Remove next_index entry.
9906 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9907 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9908 (lto_init_tree_ref_encoder): Adjust.
9909 (lto_destroy_tree_ref_encoder): Likewise.
9910 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9911 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9912 (lto_output_decl_index): Adjust.
9913 (lto_new_out_decl_state): Likewise.
9914 (lto_record_function_out_decl_state): Likewise.
9915 * lto-streamer-out.c (copy_function): Likewise.
9917 2013-06-18 Richard Biener <rguenther@suse.de>
9919 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
9920 * cgraphunit.c: Include cfgloop.h.
9921 (init_lowered_empty_function): Initialize the loop tree.
9922 (assemble_thunk): Insert new BBs into loops.
9924 2013-06-18 Richard Biener <rguenther@suse.de>
9926 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
9927 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
9928 the map from cache entry to cache index optional.
9929 (streamer_tree_cache_replace_tree): Adjust accordingly.
9930 (streamer_tree_cache_append): Likewise.
9931 (streamer_tree_cache_delete): Likewise.
9932 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
9933 streamer cache map from cache entry to cache index.
9934 * lto-streamer-out.c (create_output_block): Adjust.
9936 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
9938 * config/arm/arm.md (attribute "insn"): Move multiplication and
9939 division attributes to...
9940 (attribute "type"): ... here. Remove mult.
9941 (attribute "mul32"): New attribute.
9942 (attribute "mul64"): Add umaal.
9943 (*arm_mulsi3): Update attributes.
9944 (*arm_mulsi3_v6): Likewise.
9945 (*thumb_mulsi3): Likewise.
9946 (*thumb_mulsi3_v6): Likewise.
9947 (*mulsi3_compare0): Likewise.
9948 (*mulsi3_compare0_v6): Likewise.
9949 (*mulsi_compare0_scratch): Likewise.
9950 (*mulsi_compare0_scratch_v6): Likewise.
9951 (*mulsi3addsi): Likewise.
9952 (*mulsi3addsi_v6): Likewise.
9953 (*mulsi3addsi_compare0): Likewise.
9954 (*mulsi3addsi_compare0_v6): Likewise.
9955 (*mulsi3addsi_compare0_scratch): Likewise.
9956 (*mulsi3addsi_compare0_scratch_v6): Likewise.
9957 (*mulsi3subsi): Likewise.
9958 (*mulsidi3adddi): Likewise.
9959 (*mulsi3addsi_v6): Likewise.
9960 (*mulsidi3adddi_v6): Likewise.
9961 (*mulsidi3_nov6): Likewise.
9962 (*mulsidi3_v6): Likewise.
9963 (*umulsidi3_nov6): Likewise.
9964 (*umulsidi3_v6): Likewise.
9965 (*umulsidi3adddi): Likewise.
9966 (*umulsidi3adddi_v6): Likewise.
9967 (*smulsi3_highpart_nov6): Likewise.
9968 (*smulsi3_highpart_v6): Likewise.
9969 (*umulsi3_highpart_nov6): Likewise.
9970 (*umulsi3_highpart_v6): Likewise.
9971 (mulhisi3): Likewise.
9972 (*mulhisi3tb): Likewise.
9973 (*mulhisi3bt): Likewise.
9974 (*mulhisi3tt): Likewise.
9975 (maddhisi4): Likewise.
9976 (*maddhisi4tb): Likewise.
9977 (*maddhisi4tt): Likewise.
9978 (maddhidi4): Likewise.
9979 (*maddhidi4tb): Likewise.
9980 (*maddhidi4tt): Likewise.
9982 (udivsi3): Likewise.
9983 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
9984 (thumb2_mulsi_short_compare0): Likewise.
9985 (thumb2_mulsi_short_compare0_scratch): Likewise.
9986 * config/arm/arm1020e.md (1020mult1): Update attribute change.
9987 (1020mult2): Likewise.
9988 (1020mult3): Likewise.
9989 (1020mult4): Likewise.
9990 (1020mult5): Likewise.
9991 (1020mult6): Likewise.
9992 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
9994 (cortex_a15_mult64): Likewise.
9995 (cortex_a15_sdiv): Likewise.
9996 (cortex_a15_udiv): Likewise.
9997 * config/arm/arm1026ejs.md (mult1): Update attribute change.
10003 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
10004 (pj4_ir_div): Likewise.
10005 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
10006 (11_mult2): Likewise.
10007 (11_mult3): Likewise.
10008 (11_mult4): Likewise.
10009 (11_mult5): Likewise.
10010 (11_mult6): Likewise.
10011 (11_mult7): Likewise.
10012 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
10013 (cortex_a8_mla): Likewise.
10014 (cortex_a8_mull): Likewise.
10015 (cortex_a8_smulwy): Likewise.
10016 (cortex_a8_smlald): Likewise.
10017 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
10018 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
10019 (cortex_r4_mul_3): Likewise.
10020 (cortex_r4_mla_4): Likewise.
10021 (cortex_r4_mla_3): Likewise.
10022 (cortex_r4_smlald): Likewise.
10023 (cortex_r4_mull): Likewise.
10024 (cortex_r4_sdiv): Likewise.
10025 (cortex_r4_udiv): Likewise.
10026 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
10027 (cortex_a7_idiv): Likewise.
10028 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
10029 (9_mult2): Likewise.
10030 (9_mult3): Likewise.
10031 (9_mult4): Likewise.
10032 (9_mult5): Likewise.
10033 (9_mult6): Likewise.
10034 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
10035 (cortex_a53_sdiv): Likewise.
10036 (cortex_a53_udiv): Likewise.
10037 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
10038 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
10039 (mp626_mult2): Likewise.
10040 (mp626_mult3): Likewise.
10041 (mp626_mult4): Likewise.
10042 * config/arm/fa526.md (526_mult1): Update attribute change.
10043 (526_mult2): Likewise.
10044 * config/arm/arm-generic.md (mult): Update attribute change.
10045 (mult_ldsched_strongarm): Likewise.
10046 (mult_ldsched): Likewise.
10047 (multi_cycle): Likewise.
10048 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
10049 * config/arm/fa606te.md (606te_mult1): Update attribute change.
10050 (606te_mult2): Likewise.
10051 (606te_mult3): Likewise.
10052 (606te_mult4): Likewise.
10053 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
10054 (cortex_a9_mac16): Likewise.
10055 (cortex_a9_multiply): Likewise.
10056 (cortex_a9_mac): Likewise.
10057 (cortex_a9_multiply_long): Likewise.
10058 * config/arm/fa626te.md (626te_mult1): Update attribute change.
10059 (626te_mult2): Likewise.
10060 (626te_mult3): Likewise.
10061 (626te_mult4): Likewise.
10063 2013-06-18 Richard Biener <rguenther@suse.de>
10066 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
10068 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10071 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10072 with next_active_insn.
10074 2013-06-18 Alan Modra <amodra@gmail.com>
10076 * config/rs6000/rs6000.h (enum data_align): New.
10077 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
10078 (DATA_ABI_ALIGNMENT): Define.
10079 (CONSTANT_ALIGNMENT): Correct comment.
10080 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
10081 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
10083 2013-06-17 David Malcolm <dmalcolm@redhat.com>
10085 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
10086 ATTRIBUTE_UNUSED marking.
10088 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
10090 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
10091 alternative and update.
10092 (aarch64_dup_lanedi): Delete.
10093 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
10094 * config/aarch64/aarch64-simd-builtins.def: Update.
10096 2013-06-17 Richard Biener <rguenther@suse.de>
10098 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
10099 (lto_input_scc): Declare.
10100 (lto_input_tree_1): Likewise.
10101 (struct lto_stats_d): Add num_tree_bodies_output and
10102 num_pickle_refs_output.
10103 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
10104 (lto_read_tree_1): Split out from ...
10105 (lto_read_tree): ... this.
10106 (lto_input_scc): New function.
10107 (lto_input_tree_1): Split out from ...
10108 (lto_input_tree): ... this. Handle LTO_tree_scc.
10109 (lto_data_in_create): Create the streamer cache without hashes.
10110 * lto-streamer-out.c (create_output_block): Create the streamer
10111 cache with hashes when not doing WPA.
10112 (lto_write_tree_1): Split out from ...
10113 (lto_write_tree): ... this.
10114 (get_symbol_initial_value): New function.
10115 (lto_output_tree_1): Split out from ...
10116 (lto_output_tree): ... this. Write trees as series of SCCs
10117 using a DFS walk via DFS_write_tree.
10118 (struct sccs, struct scc_entry): New types.
10119 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
10120 (DFS_write_tree_body): New function.
10121 (DFS_write_tree): Likewise.
10122 (hash_tree): Likewise.
10123 (scc_entry_compare): Likewise.
10124 (hash_scc): Likewise.
10125 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
10126 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
10127 TREE_CHAIN as regular reference.
10128 (streamer_read_integer_cst): Remove.
10129 (streamer_get_pickled_tree): Adjust.
10130 * tree-streamer-out.c (streamer_write_chain): Disable streaming
10131 of DECL_EXTERNALs in BLOCK_VARS for now.
10132 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
10134 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
10135 Add hash value argument and record that if hashes are recorded
10137 (streamer_tree_cache_insert_1): Adjust.
10138 (streamer_tree_cache_insert): Likewise.
10139 (streamer_tree_cache_insert_at): Rename to ...
10140 (streamer_tree_cache_replace_tree): ... this and adjust.
10141 (streamer_tree_cache_append): Adjust.
10142 (record_common_node): Likewise.
10143 (streamer_tree_cache_create): Add argument whether to
10144 record hash values together with trees.
10145 (streamer_tree_cache_delete): Adjust.
10146 * tree-streamer.h (struct streamer_tree_cache_d): Add
10148 (streamer_read_integer_cst): Remove.
10149 (streamer_tree_cache_insert): Adjust.
10150 (streamer_tree_cache_append): Likewise.
10151 (streamer_tree_cache_insert_at): Rename to ...
10152 (streamer_tree_cache_replace_tree): ... this and adjust.
10153 (streamer_tree_cache_create): Add argument whether to record hashes.
10154 (streamer_tree_cache_get): Rename to ...
10155 (streamer_tree_cache_get_tree): ... this.
10156 (streamer_tree_cache_get_hash): New function.
10157 * tree.c (cache_integer_cst): New function.
10158 * tree.h (cache_integer_cst): Declare.
10159 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
10160 * lto-symtab.c (lto_varpool_replace_node): Only release
10161 DECL_INITIAL of non-prevailing decls.
10162 * varpool.c (varpool_remove_initializer): Do not release
10163 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
10165 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
10167 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
10168 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
10169 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
10170 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
10171 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
10172 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
10173 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
10174 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
10175 instead of TARGET_64BIT.
10176 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
10177 Require ISA_HAS_<D>DIV.
10179 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
10181 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
10182 (mips*-*-linux*): Move default with_llsc setting to where other
10184 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
10186 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
10187 Likewise. Remove default with_tune setting. Move default float
10188 setting to its own block. Handle with_llsc in the same block as above.
10190 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
10192 PR rtl-optimization/57425
10193 PR rtl-optimization/57569
10194 * alias.c (write_dependence_p): Add new parameters mem_mode,
10195 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
10196 Changed all callers.
10197 (canon_anti_dependence): New function.
10198 * cse.c (check_dependence): Use canon_anti_dependence.
10199 * cselib.c (cselib_invalidate_mem): Likewise.
10200 * rtl.h (canon_anti_dependence): Declare.
10202 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
10204 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
10205 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
10206 ".set mips3" for 64-bit targets.
10208 2013-06-15 Dehao Chen <dehao@google.com>
10210 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
10211 * gimple-low.c (gimple_check_call_matching_types): Likewise.
10212 (gimple_check_call_args): Likewise.
10213 * value-prof.c (check_ic_target): Likewise.
10214 * ipa-inline.c (early_inliner): Likewise.
10215 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
10216 * cgraph.c (cgraph_create_edge_1): Likewise.
10217 (cgraph_make_edge_direct): Likewise.
10219 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10222 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
10223 rs6000_output_move_128bit to handle emitting quad memory
10224 operations. Set attribute length to 8 bytes.
10226 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
10228 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
10230 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
10231 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
10232 (aarch64_<su>mlsl<mode>): Likewise.
10234 2013-06-14 Mike Stump <mikestump@comcast.net>
10236 * Makefile.in (TARGET_H): Add insn-codes.h.
10238 2013-06-14 Alan Modra <amodra@gmail.com>
10240 PR middle-end/57134
10241 PR middle-end/57586
10242 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
10243 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
10244 bitfield expansion when EXPAND_MEMORY.
10245 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
10247 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
10249 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
10250 test for clearing quad memory on 32-bit later.
10252 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
10254 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
10255 (fold_negate_expr): Likewise.
10256 (fold_real_zero_addition_p): Handle vectors.
10257 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
10259 2013-06-14 Alan Modra <amodra@gmail.com>
10261 * varasm.c (force_const_mem): Revert 2013-06-07 change.
10263 2013-06-13 Jan Hubicka <jh@suse.cz>
10265 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
10266 Local comdats are not externally visible.
10267 * symtab.c (dump_symtab_base): Dump externally visible.
10268 (verify_symtab_base): Verify back links in the symtab hash.
10270 2013-06-13 Bin Cheng <bin.cheng@arm.com>
10272 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
10273 CONVERT_EXPR as equal nodes.
10275 2013-06-13 Bin Cheng <bin.cheng@arm.com>
10277 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
10279 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
10281 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
10282 Generalize to complex and vector.
10283 * tree.c (build_all_ones_cst): New function.
10284 * tree.h (build_all_ones_cst): Declare it.
10286 2013-06-13 Alan Modra <amodra@gmail.com>
10288 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
10289 * config/rs6000/rs6000.md (signbittf2): New insn.
10290 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
10291 (abstf2_internal, cmptf_internal2): Likewise.
10292 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
10294 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10295 Pat Haugen <pthaugen@us.ibm.com>
10296 Peter Bergner <bergner@vnet.ibm.com>
10298 * config/rs6000/rs6000.c (emit_load_locked): Add support for
10299 power8 byte, half-word, and quad-word atomic instructions.
10300 (emit_store_conditional): Likewise.
10301 (rs6000_expand_atomic_compare_and_swap): Likewise.
10302 (rs6000_expand_atomic_op): Likewise.
10304 * config/rs6000/sync.md (larx): Add new modes for power8.
10306 (AINT): New mode iterator to include TImode as well as normal
10307 integer modes on power8.
10308 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
10309 that VSX registers are not considered. Use AINT mode iterator
10310 instead of INT1 to allow inclusion of quad word atomic operations
10312 (load_locked<mode>): Likewise.
10313 (store_conditional<mode>): Likewise.
10314 (atomic_compare_and_swap<mode>): Likewise.
10315 (atomic_exchange<mode>): Likewise.
10316 (atomic_nand<mode>): Likewise.
10317 (atomic_fetch_<fetchop_name><mode>): Likewise.
10318 (atomic_nand_fetch<mode>): Likewise.
10319 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
10321 (ATOMIC): On power8, add QImode, HImode modes.
10322 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
10323 modes that promote to SImode.
10324 (load_lockedti): Convert TImode arguments to PTImode, so that we
10325 get a guaranteed even/odd register pair.
10326 (load_lockedpti): Likewise.
10327 (store_conditionalti): Likewise.
10328 (store_conditionalpti): Likewise.
10330 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
10331 atomic load/store instructions.
10334 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
10336 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
10338 (insn_count): New attribute, with most cases extracted from...
10339 (length): ...here. Redefine most cases in terms of insn_count.
10340 (single_insn): Delete.
10341 (can_delay): Use insn_count to check for single instructions.
10342 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
10343 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
10344 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
10345 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
10346 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
10347 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
10348 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
10349 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
10350 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
10351 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
10352 rather than "length".
10353 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
10354 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
10355 Use "insn_count" rather than "length".
10356 * config/mips/mips-dsp.md
10357 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
10358 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
10361 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
10363 PR tree-optimization/57361
10364 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
10366 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
10368 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
10370 (aarch64_simd_combine<mode>): New instruction expansion.
10371 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
10372 function prototype.
10373 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
10374 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
10376 2013-06-12 Jan Hubicka <jh@suse.cz>
10378 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
10379 decl has when in streaming stage.
10380 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
10381 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
10383 2013-06-12 Roland Stigge <stigge@antcom.de>
10386 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
10388 2013-06-12 Jakub Jelinek <jakub@redhat.com>
10390 PR tree-optimization/57537
10391 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
10392 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
10394 2013-06-12 Richard Biener <rguenther@suse.de>
10396 * data-streamer.h (streamer_write_char_stream): CSE
10397 obs->current_pointer.
10398 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
10399 streamer_write_char_stream manually and optimize the resulting loop.
10400 (streamer_write_hwi_stream): Likewise.
10402 2013-06-12 Jan Hubicka <jh@suse.cz>
10404 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
10405 * cgraph.h (varpool_create_empty_node): Declare.
10406 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
10408 * symtab.c (symtab_unregister_node): Be lax about missin entries
10410 (symtab_get_node): Update comment.
10411 * varpool.c (varpool_create_empty_node): Break out from ...
10412 (varpool_node_for_decl): ... here.
10413 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
10415 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
10417 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
10418 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
10419 part. Use straight-line flow at the end.
10420 <COMPONENT_REF>: Remove superfluous else.
10421 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
10423 2013-06-12 Jakub Jelinek <jakub@redhat.com>
10426 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
10427 target hook even for !TREE_PUBLIC decls. If no resolution info
10428 is available, return false for common and external decls.
10430 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
10432 * config/rl78/constraints.md (U): New constraint.
10433 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
10436 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
10439 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
10440 to allow returning address to AT_PLATFORM name.
10442 2013-06-11 Jan Hubicka <jh@suse.cz>
10444 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
10445 * cgraph.h (symtab_node_base): Add weakref flag.
10446 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
10447 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
10448 (output_weakrefs): Use weakref flag.
10449 * fold-const.c (simple_operand_p): Handle WEAK.
10450 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
10451 * ipa.c (varpool_externally_visible_p): Drop weakref.
10452 (function_and_variable_visibility): Update comment; fix weakref
10453 sanity checks; do not clear DECL_WEAK on them.
10454 * lto-cgraph.c (lto_output_node): update.
10455 (lto_output_varpool_node): Update.
10456 (input_overwrite_node): Update.
10457 (input_node): Update.
10458 (input_varpool_node): Update.
10459 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
10460 (lto_symtab_merge_symbols): Add sanity check.
10461 (lto_symtab_prevailing_decl): Do not special case weakrefs.
10462 * passes.c (rest_of_decl_compilation): Set static flag, too.
10463 * symtab.c (dump_symtab_base): Dump weakref.
10464 (verify_symtab_base): Sanity check weakrefs.
10465 (symtab_make_decl_local): Remove duplicated code.
10466 (symtab_alias_ultimate_target): Simplify.
10467 * varpool.c (varpool_create_variable_alias): Set weakref flag.
10469 2013-06-11 Tom de Vries <tom@codesourcery.com>
10471 * genautomata.c (gen_regexp_sequence): Handle els_num == -1. Handle
10472 sequence_vect == NULL.
10474 2013-06-11 DJ Delorie <dj@redhat.com>
10476 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
10477 (rl78_unwind_word_mode): New.
10479 2013-06-11 David Malcolm <dmalcolm@redhat.com>
10481 * final.c (debug_prefix_maps): Make static.
10483 2013-06-11 David Malcolm <dmalcolm@redhat.com>
10485 * function.c (initial_trampoline): Remove stray copy.
10487 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
10489 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
10491 2013-06-11 Martin Jambor <mjambor@suse.cz>
10493 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
10494 within bounds at the beginning of the function.
10496 2013-06-11 Alan Modra <amodra@gmail.com>
10498 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
10500 (get_named_section): Don't NULL !DECL_P decl.
10502 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
10504 * doc/invoke.texi (core-avx2): Document.
10506 (atom): Updated with MOVBE.
10508 2013-06-11 Richard Biener <rguenther@suse.de>
10510 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
10512 2013-06-11 Anton Blanchard <anton@samba.org>
10514 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
10515 correct shift value in little-endian mode.
10517 2013-06-11 Jakub Jelinek <jakub@redhat.com>
10520 * varasm.c (get_variable_align): Move #endif to the right place.
10522 2013-06-10 Cary Coutant <ccoutant@google.com>
10524 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
10525 for hash so that hash table traversal order is deterministic.
10527 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
10528 Pat Haugen <pthaugen@us.ibm.com>
10529 Peter Bergner <bergner@vnet.ibm.com>
10531 * config/rs6000/vector.md (GPR move splitter): Do not split moves
10532 of vectors in GPRS if they are direct moves or quad word load or
10535 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
10537 (direct_move_p): Likewise.
10538 (quad_load_store_p): Likewise.
10540 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
10541 classes into bins based on the physical register type.
10542 (reg_class_to_reg_type): Likewise.
10543 (IS_STD_REG_TYPE): Likewise.
10544 (IS_FP_VECT_REG_TYPE): Likewise.
10545 (reload_fpr_gpr): Arrays to determine what insn to use if we can
10546 use direct move instructions.
10547 (reload_gpr_vsx): Likewise.
10548 (reload_vsx_gpr): Likewise.
10549 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
10550 information that is a simplification of register classes. Also
10551 precalculate direct move reload helpers.
10552 (direct_move_p): New function to return true if the operation can
10553 be done as a direct move instruciton.
10554 (quad_load_store_p): New function to return true if the operation
10555 is a quad memory operation.
10556 (rs6000_legitimize_address): If quad memory, only allow register
10557 indirect for TImode addresses.
10558 (rs6000_legitimate_address_p): Likewise.
10559 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
10560 (rs6000_reload_register_type): Likewise.
10561 (register_to_reg_type): Return register type.
10562 (rs6000_secondary_reload_simple_move): New helper function for
10563 secondary reload and secondary memory needed to identify anything
10564 that is a simple move, and does not need reloading.
10565 (rs6000_secondary_reload_direct_move): New helper function for
10566 secondary reload to identify cases that can be done with several
10567 instructions via the direct move instructions.
10568 (rs6000_secondary_reload_move): New helper function for secondary
10569 reload to identify moves between register types that can be done.
10570 (rs6000_secondary_reload): Add support for quad memory operations
10571 and for direct move.
10572 (rs6000_secondary_memory_needed): Likewise.
10573 (rs6000_debug_secondary_memory_needed): Change argument names.
10574 (rs6000_output_move_128bit): New function to return the move to
10575 use for 128-bit moves, including knowing about the various
10576 limitations of quad memory operations.
10578 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
10579 memory operations. call rs6000_output_move_128bit for the actual
10580 instruciton(s) to generate.
10581 (vsx_movti_64bit): Likewise.
10583 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
10584 (UNSPEC_P8V_MTVSRWZ): Likewise.
10585 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
10586 (UNSPEC_P8V_MTVSRD): Likewise.
10587 (UNSPEC_P8V_XXPERMDI): Likewise.
10588 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
10589 (UNSPEC_FUSION_GPR): Likewise.
10590 (FMOVE128_GPR): New iterator for direct move.
10591 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
10592 (f32_sv): Likewise.
10593 (f32_dm): Likewise.
10594 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
10595 loads and direct move instructions.
10596 (zero_extendsidi2_lfiwzx): Likewise.
10597 (extendsidi2_lfiwax): Likewise.
10598 (extendsidi2_nocell): Likewise.
10599 (floatsi<mode>2_lfiwax): Likewise.
10600 (lfiwax): Likewise.
10601 (floatunssi<mode>2_lfiwzx): Likewise.
10602 (lfiwzx): Likewise.
10603 (fix_trunc<mode>_stfiwx): Likewise.
10604 (fixuns_trunc<mode>_stfiwx): Likewise.
10605 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
10606 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
10607 (parity<mode>2_cmpb): Set length/type attr.
10608 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
10609 for 'mr.' to fast_compare.
10610 (bpermd_<mode>): Change type attr to popcnt.
10611 (p8_fmrgow_<mode>): New insns for power8 direct move support.
10612 (p8_mtvsrwz_1): Likewise.
10613 (p8_mtvsrwz_2): Likewise.
10614 (reload_fpr_from_gpr<mode>): Likewise.
10615 (p8_mtvsrd_1): Likewise.
10616 (p8_mtvsrd_2): Likewise.
10617 (p8_xxpermdi_<mode>): Likewise.
10618 (reload_vsx_from_gpr<mode>): Likewise.
10619 (reload_vsx_from_gprsf): Likewise.
10620 (p8_mfvsrd_3_<mode>): LIkewise.
10621 (reload_gpr_from_vsx<mode>): Likewise.
10622 (reload_gpr_from_vsxsf): Likewise.
10623 (p8_mfvsrd_4_disf): Likewise.
10624 (multi-word GPR splits): Do not split direct moves or quad memory
10627 2013-06-10 David Malcolm <dmalcolm@redhat.com>
10629 * tree-into-ssa.c (interesting_blocks): Make static.
10631 2013-06-10 Jakub Jelinek <jakub@redhat.com>
10634 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
10635 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
10636 Use DATA_ABI_ALIGNMENT for that case instead if defined.
10637 (get_variable_align): New function.
10638 (get_variable_section, emit_bss, emit_common,
10639 assemble_variable_contents, place_block_symbol): Use
10640 get_variable_align instead of DECL_ALIGN.
10641 (assemble_noswitch_variable): Add align argument, use it
10642 instead of DECL_ALIGN.
10643 (assemble_variable): Adjust caller. Use get_variable_align
10644 instead of DECL_ALIGN.
10645 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
10647 (DATA_ABI_ALIGNMENT): Define.
10648 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
10649 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
10650 opt is false, only return the psABI mandated alignment increase.
10651 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
10652 (DATA_ABI_ALIGNMENT): ... this.
10653 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
10654 (DATA_ABI_ALIGNMENT): ... this.
10655 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
10656 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
10657 (DATA_ABI_ALIGNMENT): ... this.
10658 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
10659 * doc/tm.texi: Regenerated.
10661 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
10663 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
10664 cmp_code to construct REG_EQUAL note.
10666 2013-06-09 Jakub Jelinek <jakub@redhat.com>
10669 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
10670 that operands[2] doesn't overlap with operands[0].
10672 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
10673 Jan Hubicka <jh@suse.cz>
10675 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
10676 hack to mark symbols as used.
10678 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
10680 PR rtl-optimization/57559
10681 * lra-constraints.c (process_alt_operands): Don't discourage
10682 memory with known offset for offsetable memory constraint.
10683 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
10685 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
10687 * varasm.c (struct oc_local_state): Reorder fields.
10688 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
10689 and adjust accordingly.
10690 (output_constructor): Reorder initialization code and adjust call to
10691 output_constructor_bitfield.
10693 2013-06-07 Jan Hubicka <jh@suse.cz>
10695 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
10697 2013-06-07 David Malcolm <dmalcolm@redhat.com>
10699 * tree-object-size.c (unknown): Make const.
10701 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10703 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
10704 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
10705 for last alternative in the cpu_facility attribute.
10707 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10710 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
10711 (xordi3): Change operand 2 constraint to arm_xordi_operand.
10712 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
10713 * config/arm/constraints.md (Dg): New constraint.
10714 * config/arm/neon.md (xordi3_neon): Remove.
10715 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
10716 * config/arm/predicates.md (arm_xordi_operand): New predicate.
10718 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10720 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
10721 Clean up alternatives.
10723 2013-06-07 Alan Modra <amodra@gmail.com>
10725 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
10728 2013-06-07 Alan Modra <amodra@gmail.com>
10730 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
10732 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10734 * config/arm/constraints.md (Df): New constraint.
10735 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
10736 Correct length attribute for last two alternatives.
10738 2013-06-07 Alan Modra <amodra@gmail.com>
10740 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
10741 override user -mfp-in-toc.
10742 (offsettable_ok_by_alignment): Consider just the current access
10743 rather than the whole object, unless BLKmode. Handle
10744 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
10745 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
10746 for -mcmodel=medium.
10747 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
10748 override user -mfp-in-toc or -msum-in-toc. Default to
10749 -mno-fp-in-toc for -mcmodel=medium.
10751 2013-06-06 DJ Delorie <dj@redhat.com>
10753 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
10754 TARGET_VALID_POINTER_MODE.
10756 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
10757 Pat Haugen <pthaugen@us.ibm.com>
10758 Peter Bergner <bergner@vnet.ibm.com>
10760 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10761 Document new power8 builtins.
10763 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
10764 condition code register, to allow 128-bit logical operations to be
10765 done in the VSX or GPR registers.
10766 (nor<mode>3): Use the canonical form for nor.
10767 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
10768 vclz*, and vpopcnt* vector instructions.
10769 (nand<mode>3): Likewise.
10770 (orc<mode>3): Likewise.
10771 (clz<mode>2): LIkewise.
10772 (popcount<mode>2): Likewise.
10774 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
10775 that only the GPRs are recognized.
10777 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10778 support for new power8 builtins.
10780 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
10782 (xscvdpspn): Likewise.
10788 (vpopcnt): Likewise.
10789 (vpopcntb): Likewise.
10790 (vpopcnth): Likewise.
10791 (vpopcntw): Likewise.
10792 (vpopcntd): Likewise.
10794 (vmrgew): Likewise.
10795 (vmrgow): Likewise.
10797 (eqv_v16qi3): Likewise.
10798 (eqv_v8hi3): Likewise.
10799 (eqv_v4si3): Likewise.
10800 (eqv_v2di3): Likewise.
10801 (eqv_v4sf3): Likewise.
10802 (eqv_v2df3): Likewise.
10804 (nand_v16qi3): Likewise.
10805 (nand_v8hi3): Likewise.
10806 (nand_v4si3): Likewise.
10807 (nand_v2di3): Likewise.
10808 (nand_v4sf3): Likewise.
10809 (nand_v2df3): Likewise.
10811 (orc_v16qi3): Likewise.
10812 (orc_v8hi3): Likewise.
10813 (orc_v4si3): Likewise.
10814 (orc_v2di3): Likewise.
10815 (orc_v4sf3): Likewise.
10816 (orc_v2df3): Likewise.
10818 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
10819 allow power8 quad mode in 64-bit.
10820 (rs6000_builtin_vectorized_function): Add support to vectorize
10821 ISA 2.07 count leading zeros, population count builtins.
10822 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
10823 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
10824 (builtin_function_type): Add vgbbd builtin function which takes an
10826 (altivec_expand_vec_perm_const): Add support for new power8 merge
10829 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
10830 that does not include TImdoe for use with 32-bit.
10831 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
10833 (UNSPEC_VSX_CVDPSPN): Likewise.
10834 (vsx_xscvdpspn): Likewise.
10835 (vsx_xscvspdpn): Likewise.
10836 (vsx_xscvdpspn_scalar): Likewise.
10837 (vsx_xscvspdpn_directmove): Likewise.
10838 (vsx_and<mode>3): Split logical operations into 32-bit and
10839 64-bit. Add support to do logical operations on TImode as well as
10840 VSX vector types. Allow logical operations to be done in either
10841 VSX registers or in general purpose registers in 64-bit mode. Add
10842 splitters if GPRs were used. For AND, add clobber of CCmode to
10843 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
10845 (vsx_and<mode>3_32bit): Likewise.
10846 (vsx_and<mode>3_64bit): Likewise.
10847 (vsx_ior<mode>3): Likewise.
10848 (vsx_ior<mode>3_32bit): Likewise.
10849 (vsx_ior<mode>3_64bit): Likewise.
10850 (vsx_xor<mode>3): Likewise.
10851 (vsx_xor<mode>3_32bit): Likewise.
10852 (vsx_xor<mode>3_64bit): Likewise.
10853 (vsx_one_cmpl<mode>2): Likewise.
10854 (vsx_one_cmpl<mode>2_32bit): Likewise.
10855 (vsx_one_cmpl<mode>2_64bit): Likewise.
10856 (vsx_nor<mode>3): Likewise.
10857 (vsx_nor<mode>3_32bit): Likewise.
10858 (vsx_nor<mode>3_64bit): Likewise.
10859 (vsx_andc<mode>3): Likewise.
10860 (vsx_andc<mode>3_32bit): Likewise.
10861 (vsx_andc<mode>3_64bit): Likewise.
10862 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
10863 and xxlorc instructions.
10864 (vsx_eqv<mode>3_64bit): Likewise.
10865 (vsx_nand<mode>3_32bit): Likewise.
10866 (vsx_nand<mode>3_64bit): Likewise.
10867 (vsx_orc<mode>3_32bit): Likewise.
10868 (vsx_orc<mode>3_64bit): Likewise.
10870 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
10872 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
10874 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
10875 (p8_vmrgow): Likewise.
10876 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
10877 GPRs to be split under VSX.
10878 (p8v_clz<mode>2): Add power8 count leading zero support.
10879 (p8v_popcount<mode>2): Add power8 population count support.
10880 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
10883 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
10886 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
10888 (vec_nand): Likewise.
10889 (vec_vclz): Likewise.
10890 (vec_vclzb): Likewise.
10891 (vec_vclzd): Likewise.
10892 (vec_vclzh): Likewise.
10893 (vec_vclzw): Likewise.
10894 (vec_vgbbd): Likewise.
10895 (vec_vmrgew): Likewise.
10896 (vec_vmrgow): Likewise.
10897 (vec_vpopcnt): Likewise.
10898 (vec_vpopcntb): Likewise.
10899 (vec_vpopcntd): Likewise.
10900 (vec_vpopcnth): Likewise.
10901 (vec_vpopcntw): Likewise.
10903 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
10905 PR rtl-optimization/57468
10906 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
10909 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
10911 PR rtl-optimization/57459
10912 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
10913 type when setting live regs.
10915 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
10917 * config/s390/s390.opt (mlra): New option.
10918 * config/s390/s390.c (s390_decompose_address): Check displacement
10919 for all registers for LRA.
10920 (s390_secondary_reload): Don't used secondary reloads for LRA.
10921 (s390_lra_p): New function.
10922 (TARGET_LRA_P): Define.
10923 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
10924 of attribute cpu_facility to zarch for the last alternative.
10925 (*cmpmem_short): Ditto.
10927 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
10929 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
10930 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
10931 (arm_expand_prologue): Likewise.
10933 2013-06-06 Teresa Johnson <tejohnson@google.com>
10936 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
10937 as this is now done by redirect_edge_and_branch_force.
10938 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
10939 barriers, and fix interaction with splitting.
10940 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
10941 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
10942 reflect changes made in the routine.
10943 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
10944 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
10945 since this is called in cfglayout mode, and replace partition fixup
10946 with assert as that is now done by force_nonfallthru_and_redirect.
10947 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
10948 already be marked with region crossing note.
10949 (insert_section_boundary_note): Make non-static, gate on flag
10950 has_bb_partition, rewrite to also check for multiple partitions.
10951 (rest_of_handle_reorder_blocks): Remove call to
10952 insert_section_boundary_note, now done later during free_cfg.
10953 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
10954 * bb-reorder.h (insert_section_boundary_note): Declare.
10955 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
10956 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
10957 invoke insert_section_boundary_note.
10958 (try_redirect_by_replacing_jump): Remove unnecessary
10959 check for region crossing note.
10960 (fixup_partition_crossing): New function.
10961 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
10962 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
10963 in non-cfglayout mode.
10964 (force_nonfallthru_and_redirect): Fixup partition boundaries,
10965 remove old code that tried to do this. Emit barrier correctly
10966 when we are in cfglayout mode.
10967 (last_bb_in_partition): New function.
10968 (rtl_split_edge): Correctly fixup partition boundaries.
10969 (commit_one_edge_insertion): Remove old code that tried to
10970 fixup region crossing edge since this is now handled in
10971 split_block, and set up insertion point correctly since
10972 block may now end in a jump.
10973 (verify_hot_cold_block_grouping): Guard against checking when not in
10974 linearized RTL mode.
10975 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
10977 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
10978 rtl_verify_flow_info, so not called in cfglayout mode.
10979 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
10980 (fixup_reorder_chain): Remove old code that attempted to fixup region
10981 crossing note as this is now handled in force_nonfallthru_and_redirect.
10982 (duplicate_insn_chain): Don't duplicate switch section notes.
10983 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
10985 * basic-block.h (emit_barrier_after_bb): Declare.
10987 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10989 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
10990 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
10991 arm_usatsihi): Adjust alternatives for arm_restrict_it.
10993 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10995 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
10997 * config/arm/ldmstm.md: Regenerate.
10999 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11001 * config/arm/sync.md (atomic_loaddi_1):
11002 Disable predication for arm_restrict_it.
11003 (arm_load_exclusive<mode>): Likewise.
11004 (arm_load_exclusivesi): Likewise.
11005 (arm_load_exclusivedi): Likewise.
11006 (arm_load_acquire_exclusive<mode>): Likewise.
11007 (arm_load_acquire_exclusivesi): Likewise.
11008 (arm_load_acquire_exclusivedi): Likewise.
11009 (arm_store_exclusive<mode>): Likewise.
11010 (arm_store_exclusive<mode>): Likewise.
11011 (arm_store_release_exclusivedi): Likewise.
11012 (arm_store_release_exclusive<mode>): Likewise.
11014 2013-06-06 Richard Biener <rguenther@suse.de>
11016 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
11018 (lto_tag_is_tree_code_p): Adjust.
11019 (lto_tag_is_gimple_code_p): Likewise.
11020 (lto_gimple_code_to_tag): Likewise.
11021 (lto_tag_to_gimple_code): Likewise.
11022 (lto_tree_code_to_tag): Likewise.
11023 (lto_tag_to_tree_code): Likewise.
11024 * data-streamer.h (streamer_write_hwi_in_range): Use
11025 uhwi streaming to stream the normalized range.
11026 (streamer_read_hwi_in_range): Likewise.
11028 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11030 * config/arm/arm.md (enabled_for_depr_it): New attribute.
11031 (predicable_short_it): Likewise.
11032 (predicated): Likewise.
11033 (enabled): Handle above.
11034 (define_cond_exec): Set predicated attribute to yes.
11036 2013-06-05 Mike Stump <mikestump@comcast.net>
11038 * gdbinit.in (__FUNCTION__): Add.
11040 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
11042 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
11043 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
11045 2013-06-05 Jan Hubicka <jh@suse.cz>
11047 * varasm.c (mark_decl_referenced): Revert the removal until targets
11050 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
11052 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
11053 instead of mark_decl_referenced.
11055 2013-06-05 Jan Hubicka <jh@suse.cz>
11057 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
11058 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
11059 and symtab_used_from_object_file_p.
11060 (cgraph_make_node_local_1): Clear forced_by_abi.
11061 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
11062 * cgraph.h (symtab_node_base): Add forced_by_abi.
11063 (decide_is_variable_needed): Remove.
11064 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
11065 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
11066 (decide_is_symbol_needed): ... this one; handle symbols in general;
11067 always analyze virtuals; honnor forced_by_abi.
11068 (cgraph_finalize_function): Update.
11069 (varpool_finalize_decl): Update.
11070 (symbol_defined_and_needed): Remove.
11071 (analyze_functions): Update.
11072 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
11073 output_refs, input_overwrite_node): Handle forced_by_abi.
11074 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
11075 (address_taken_from_non_vtable_p): ... this one.
11076 (comdat_can_be_unshared_p_1): New function.
11077 (cgraph_comdat_can_be_unshared_p): Rename to ...
11078 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
11079 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
11080 (function_and_variable_visibility): Clear forced_by_abi as needed.
11081 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
11082 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
11083 * symtab.c (dump_symtab_base): Dump forced_by_abi.
11084 * varpool.c (decide_is_variable_needed): Remove.
11086 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11088 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
11089 (arm_option_override): Override arm_restrict_it where appropriate.
11090 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
11091 * config/arm/arm.opt (mrestrict-it): New command-line option.
11092 * doc/invoke.texi: Document -mrestrict-it.
11094 2013-06-05 David Malcolm <dmalcolm@redhat.com>
11096 * tsan.c (tsan_atomic_table): Make const.
11098 2013-06-05 Richard Biener <rguenther@suse.de>
11100 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
11101 index associated with the tree we are supposed to replace.
11102 * tree-streamer-out.c (pack_ts_base_value_fields): Output
11103 TREE_ASM_WRITTEN as zero for everything but SSA names.
11105 2013-06-05 David Malcolm <dmalcolm@redhat.com>
11107 * tree-ssa-structalias.c (call_stmt_vars): Make static.
11109 2013-06-04 Jan Hubicka <jh@suse.cz>
11111 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
11112 (input_node, input_varpool_node): Handle correctly external same
11114 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
11115 nodes at ltrans stage.
11117 2013-06-04 Jan Hubicka <jh@suse.cz>
11119 * ipa-inline.c (update_caller_keys): Fix availability test.
11120 (update_callee_keys): Likewise.
11121 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
11122 to follow ELF standard.
11124 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
11126 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
11127 (mips64r5900el-*-elf*): New configurations.
11128 * config/mips/mips-cpus.def (r5900): New processor.
11129 * config/mips/mips-tables.opt: Regenerate.
11130 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
11131 (mips_issue_rate): Handle PROCESSOR_R5900.
11132 (mips_reorg_process_insns): Force reorder mode for the R5900.
11133 * config/mips/mips.h (TARGET_MIPS5900): Define.
11134 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
11136 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
11138 * config/mips/mips.md (processor): Add r5900.
11139 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
11141 2013-06-04 Ian Bolton <ian.bolton@arm.com>
11143 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
11144 into function to generate MOVI instruction.
11145 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
11146 (aarch64_preferred_simd_mode): Turn into wrapper.
11147 (aarch64_output_scalar_simd_mov_immediate): New function.
11148 * config/aarch64/aarch64-protos.h: Add prototype for above.
11150 2013-06-04 Ian Bolton <ian.bolton@arm.com>
11152 * config/aarch64/aarch64.c (simd_immediate_info): Remove
11153 element_char member.
11154 (sizetochar): Return signed char.
11155 (aarch64_simd_valid_immediate): Remove elchar and other
11156 unnecessary variables.
11157 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
11158 Calculate element_char as required.
11159 * config/aarch64/aarch64-protos.h: Update and move prototype
11160 for aarch64_output_simd_mov_immediate.
11161 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11164 2013-06-04 Ian Bolton <ian.bolton@arm.com>
11166 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
11167 information completed by aarch64_simd_valid_immediate.
11168 (aarch64_legitimate_constant_p): Update arguments.
11169 (aarch64_simd_valid_immediate): Work with struct rather than many
11171 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
11172 (aarch64_simd_make_constant): Update arguments.
11173 (aarch64_output_simd_mov_immediate): Work with struct rather than
11174 many pointers. Output immediate directly rather than as operand.
11175 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11177 * config/aarch64/constraints.md (Dn): Update arguments.
11179 2013-06-04 Ian Bolton <ian.bolton@arm.com>
11181 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
11183 (aarch64_simd_immediate_valid_for_move): Remove.
11184 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
11185 (aarch64_simd_make_constant): Update call.
11186 (aarch64_output_simd_mov_immediate): Update call.
11187 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11189 * config/aarch64/constraints.md (Dn): Update call.
11191 2013-06-04 Ian Bolton <ian.bolton@arm.com>
11193 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
11194 return type to bool for prototype.
11195 (aarch64_legitimate_constant_p): Check for true instead of not -1.
11196 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
11197 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
11199 2013-06-04 Catherine Moore <clm@codesourcery.com>
11201 * config/mips/mips.opt (meva): New.
11202 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
11203 (ASM_SPEC): Handle -meva.
11204 * doc/invoke.texi (meva): Document.
11206 2013-06-04 Alan Modra <amodra@gmail.com>
11208 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
11211 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11213 * rtl.def: Add extra fourth optional field to define_cond_exec.
11214 * gensupport.c (process_one_cond_exec): Process attributes from
11216 * doc/md.texi: Document fourth field in define_cond_exec.
11218 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
11220 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
11221 out the processing order as in store_bit_field_1.
11223 2013-06-04 Jan Hubicka <jh@suse.cz>
11225 PR middle-end/57500
11226 * cgraphunit.c (cgraph_process_same_body_aliases): Create
11227 non-VAR_DECL node if it does not exist yet.
11229 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
11231 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
11232 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
11233 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
11234 target_cpu_default setting.
11236 2013-06-03 Teresa Johnson <tejohnson@google.com>
11238 * dumpfile.c (opt_info_switch_p): Change -fopt-info
11239 default to -fopt-info=optimized instead of all.
11240 * doc/invoke.texi: Ditto.
11241 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
11242 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
11243 (execute_vect_slp): Emit BB vectorization success under
11244 MSG_OPTIMIZED_LOCATIONS.
11245 * tree-vect-slp.c (vect_slp_transform_bb): Change
11246 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
11247 * tree-vect-loop.c (vect_transform_loop): Ditto.
11249 2013-06-03 Jason Merrill <jason@redhat.com>
11252 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11253 Use TARGET_EXPR for C++.
11255 2013-06-03 Jakub Jelinek <jakub@redhat.com>
11257 PR rtl-optimization/57268
11258 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
11259 if DEBUG_INSN_P (insn).
11262 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
11264 PR rtl-optimization/57268
11265 * sched-deps.c (sched_analyze_2): Flush dependence lists if
11266 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11268 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
11270 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
11271 (ix86_avoid_lea_for_addr): Likewise.
11272 (exact_dependency_1): Likewise.
11273 (ix86_adjust_cost): Likewise.
11274 (swap_top_of_ready_list): Fix formatting and !reload_completed check
11276 (do_reorder_for_imul): Fix typo, formatting and
11277 !reload_completed check removed.
11278 (ix86_sched_reorder): Fix typo and formatting.
11279 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
11282 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
11284 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
11286 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
11288 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
11289 <STRING_CST>: Likewise.
11290 <VECTOR_CST>: Likewise.
11292 2013-06-01 Janus Weil <janus@gcc.gnu.org>
11293 Mikael Morin <mikael@gcc.gnu.org>
11295 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
11296 * config.in: Regenerated.
11297 * configure: Regenerated.
11299 2013-06-01 Jan Hubicka <jh@suse.cz>
11301 PR middle-end/57366
11302 * cgraphunit.c (compile): When weakref is not supported,
11303 set up transparent aliases before final output pass.
11304 * varasm.c (assemble_alias): Do not try to do it here.
11306 2013-06-01 Jan Hubicka <jh@suse.cz>
11308 PR middle-end/57467
11309 * passes.c (for_per_function): Skip unanalyzed functions.
11311 2013-06-01 Jan Hubicka <jh@suse.cz>
11313 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
11314 (lto_symtab_merge_symbols_1): ... this one.
11315 (lto_symtab_merge_cgraph_nodes): Rename to ...
11316 (lto_symtab_merge_symbols): ... this one; simplify.
11317 * cgraph.c (same_body_aliases_done): Rename to ...
11318 (cpp_implicit_aliases_done): ... this one.
11319 (cgraph_create_function_alias): Update.
11320 (cgraph_same_body_alias): Update.
11321 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
11322 (verify_edge_corresponds_to_fndecl): Simplify.
11323 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
11324 (cgraph_node): Remove same_body_alias.
11325 (varpool_node): Remove alias_of and extra_name_alias.
11326 (same_body_aliases_done): Rename to ..
11327 (cpp_implicit_aliases_done): ... this one.
11328 (symtab_alias_ultimate_target): Add default parameter.
11329 (symtab_resolve_alias): New function.
11330 (fixup_same_cpp_alias_visibility): Declare.
11331 (cgraph_function_node): Add default parameter.
11332 (cgraph_node_asm_name): Likewise.
11333 (cgraph_function_or_thunk_node): Add default parameter; do
11334 not ICE when it is NULL.
11335 (varpool_variable_node): Likewise.
11336 * tree-emutls.c (create_emultls_var): Update.
11337 (ipa_lower_emutls): Update.
11338 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
11339 (cgraph_reset_node): Reset alias info.
11340 (cgraph_finalize_function): Update.
11341 (fixup_same_cpp_alias_visibility): Move to symtab.c.
11342 (analyze_function): Simplify.
11343 (cgraph_process_same_body_aliases): Simplify.
11344 (analyze_functions): Fixup same body aliases.
11345 (handle_alias_pairs): Simplify.
11346 (assemble_thunk): Update.
11347 (assemble_thunks_and_aliases): Update.
11348 (output_weakrefs): Rewrite.
11349 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
11350 (lto_output_varpool_node): Likewise.
11351 (compute_ltrans_boundary): Remve assert.
11352 (get_alias_symbol): New functoin.
11353 (input_node): Rewrite alias handling.
11354 (input_varpool_node): Likewise.
11355 * ipa-pure-const.c (propagate_pure_const): Fix formating.
11356 * ipa.c (process_references): Handle weakrefs correctly.
11357 (symtab_remove_unreachable_nodes): Likewise.
11358 * trans-mem.c (get_cg_data): Update.
11359 (ipa_tm_create_version_alias): Update.
11360 (ipa_tm_execute): Update.
11361 * symtab.c (dump_symtab_base): Dump aliases.
11362 (verify_symtab_base): Verify aliases.
11363 (symtab_node_availability): New function.
11364 (symtab_alias_ultimate_target): Simplify.
11365 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
11366 handle all the fixup cases.
11367 (symtab_resolve_alias): New function.
11368 * passes.c (ipa_write_summaries): Handle weakrefs.
11369 * varpool.c (varpool_analyze_node): Simplify.
11370 (assemble_aliases): Update.
11371 (varpool_create_variable_alias): Simplify.
11372 (varpool_extra_name_alias): Simplify.
11373 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
11374 (lto_symtab_merge_symbols): ... this one.
11376 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
11379 PR rtl-optimization/57268
11380 * sched-deps.c (sched_analyze_2): Flush dependence lists if
11381 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11383 2013-06-01 Tobias Burnus <burnus@net-b.de>
11385 Partially reverted:
11386 2013-05-31 Tobias Burnus <burnus@net-b.de>
11388 PR middle-end/57073
11389 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
11392 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
11394 PR rtl-optimization/57268
11395 * sched-deps.c (sched_analyze_2): Flush dependence lists if
11396 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11398 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
11400 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
11401 unordered comparison operators when -fno-trapping-math is in effect
11403 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
11404 and implement unordered comparison operators properly on the e500.
11406 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
11408 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
11409 for constant scalar integers.
11410 (simplify_relational_operation_1): Likewise.
11412 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11414 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
11415 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
11418 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
11419 Igor Zamyatin <igor.zamyatin@intel.com>
11421 Silvermont (SLM) architecture performance tuning.
11422 * config/i386/i386.h (enum ix86_tune_indices): Add
11423 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
11424 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
11426 * config/i386/i386.c (initial_ix86_tune_features)
11427 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
11428 (ix86_lea_outperforms): Handle Silvermont tuning.
11429 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
11431 (ix86_use_lea_for_mov): Likewise.
11432 (ix86_avoid_lea_for_addr): Likewise.
11433 (ix86_lea_for_add_ok): Likewise.
11434 (exact_dependency_1): New function.
11435 (exact_store_load_dependency): Likewise.
11436 (ix86_adjust_cost): Handle Silvermont tuning.
11437 (do_reoder_for_imul): Likewise.
11438 (swap_top_of_ready_list): New function.
11439 (ix86_sched_reorder): Changed to handle Silvermont tuning.
11441 * config/i386/i386.md (peepholes that split memory operand in fp
11444 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
11446 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11447 Remove un-necessary braces.
11449 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
11451 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11452 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
11454 2013-05-31 Tobias Burnus <burnus@net-b.de>
11456 PR middle-end/57073
11457 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
11459 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11462 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
11463 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
11464 * config/arm/neon.md (iordi3_neon): Remove.
11465 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
11466 * config/arm/predicates.md (imm_for_neon_logic_operand):
11467 Move to earlier in the file.
11468 (neon_logic_op2): Likewise.
11469 (arm_iordi_operand_neon): New predicate.
11471 2013-05-31 Richard Biener <rguenther@suse.de>
11473 PR tree-optimization/57478
11474 PR tree-optimization/57453
11475 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
11478 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
11480 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
11481 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
11483 2013-05-30 Tobias Burnus <burnus@net-b.de>
11484 Thomas Koenig <tkoenig@gcc.gnu.org>
11486 PR middle-end/57073
11487 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
11488 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
11490 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
11492 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
11494 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
11496 * target.def (register_usage_leveling_p): New hook.
11497 * targhooks.c (default_register_usage_leveling_p): New.
11498 * targhooks.h (default_register_usage_leveling_p): New prototype.
11499 * lra-assigns.c (register_usage_leveling_p): Use the hook.
11500 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
11501 * doc/tm.texi: Update.
11502 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
11504 2013-05-30 Ian Bolton <ian.bolton@arm.com>
11506 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
11507 (*insv_reg<mode>): New define_insn.
11509 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
11511 PR rtl-optimization/57439
11512 * postreload.c (move2add_valid_value_p): Check that we have
11513 a zero subreg_regno_offset when accessing the register in
11514 the requested mode.
11516 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
11517 Igor Zamyatin <igor.zamyatin@intel.com>
11519 Silvermont (SLM) architecture pipeline model, tuning and
11521 * config.gcc: Add slm config options and target.
11523 * config/i386/slm.md: New.
11525 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
11527 * config/i386/i386-c.c (ix86_target_macros_internal): New case
11529 (ix86_target_macros_internal): Likewise.
11531 * config/i386/i386.c (slm_cost): New cost.
11532 (m_SLM): New macro flag.
11533 (initial_ix86_tune_features): Set m_SLM.
11534 (x86_accumulate_outgoing_args): Likewise.
11535 (x86_arch_always_fancy_math_387): Likewise.
11536 (processor_target_table): Add slm cost.
11537 (cpu_names): Add slm cpu name.
11538 (x86_option_override_internal): Set SLM ISA.
11539 (ix86_issue_rate): New case PROCESSOR_SLM.
11540 (ia32_multipass_dfa_lookahead): Likewise.
11541 (fold_builtin_cpu): Add slm.
11543 * config/i386/i386.h (TARGET_SLM): New target macro.
11544 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
11545 (processor_type): Add PROCESSOR_SLM.
11547 * config/i386/i386.md (cpu): Add new value "slm".
11548 (slm.md): Include slm.md.
11550 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
11551 Zhenqiang Chen <zhenqiang.chen@linaro.org>
11553 * config/arm/arm-protos.h: Add and update function protos.
11554 * config/arm/arm.c (use_simple_return_p): New added.
11555 (thumb2_expand_return): Check simple_return flag.
11556 * config/arm/arm.md: Add simple_return and conditional simple_return.
11557 * config/arm/iterators.md: Add iterator for return and simple_return.
11559 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
11561 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
11562 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
11563 (arm_emit_vfp_multi_reg_pop): Likewise.
11564 (thumb2_emit_ldrd_pop): Likewise.
11565 (arm_expand_epilogue): Add misc REG_CFA notes.
11566 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
11568 2013-05-29 Lawrence Crowl <crowl@google.com>
11570 * config/arm/t-arm: Update for below.
11572 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
11573 Change type to hash_table. Update dependent calls and types.
11575 * config/i386/t-cygming: Update for below.
11577 * config/i386/t-interix: Update for below.
11579 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
11580 Change type to hash_table. Update dependent calls and types.
11581 (i386_find_on_wrapper_list::wrappers): Likewise.
11583 * config/ia64/t-ia64: Update for below.
11585 * config/ia64/ia64.c (bundle_state_table):
11586 Change type to hash_table. Update dependent calls and types.
11588 * config/mips/mips.c (mips_reorg_process_insns::htab):
11589 Change type to hash_table. Update dependent calls and types.
11591 * config/sol2.c (solaris_comdat_htab):
11592 Change type to hash_table. Update dependent calls and types.
11594 * config/t-sol2: Update for above.
11596 2013-05-29 Teresa Johnson <tejohnson@google.com>
11598 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
11599 functions are not yet marked as defined.
11601 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11602 Pat Haugen <pthaugen@us.ibm.com>
11603 Peter Bergner <bergner@vnet.ibm.com>
11605 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
11609 (vrotl<mode>3): Likewise.
11610 (vashl<mode>3): Likewise.
11611 (vlshr<mode>3): Likewise.
11612 (vashr<mode>3): Likewise.
11614 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11615 support for power8 V2DI builtins.
11617 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
11618 power8 V2DI builtins.
11619 (vupkhsw): Likewise.
11620 (vupklsw): Likewise.
11621 (vaddudm): Likewise.
11622 (vminsd): Likewise.
11623 (vmaxsd): Likewise.
11624 (vminud): Likewise.
11625 (vmaxud): Likewise.
11626 (vpkudum): Likewise.
11627 (vpksdss): Likewise.
11628 (vpkudus): Likewise.
11629 (vpksdus): Likewise.
11634 (vsubudm): Likewise.
11635 (vcmpequd): Likewise.
11636 (vcmpgtsd): Likewise.
11637 (vcmpgtud): Likewise.
11638 (vcmpequd_p): Likewise.
11639 (vcmpgtsd_p): Likewise.
11640 (vcmpgtud_p): Likewise.
11641 (vupkhsw): Likewise.
11642 (vupklsw): Likewise.
11643 (vaddudm): Likewise.
11644 (vmaxsd): Likewise.
11645 (vmaxud): Likewise.
11646 (vminsd): Likewise.
11647 (vminud): Likewise.
11648 (vpksdss): Likewise.
11649 (vpksdus): Likewise.
11650 (vpkudum): Likewise.
11651 (vpkudus): Likewise.
11656 (vsubudm): Likewise.
11658 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
11659 support for power8 V2DI instructions.
11661 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
11662 power8 V2DI instructions. Combine pack and unpack insns to use an
11663 iterator for each mode. Check whether a particular mode supports
11664 Altivec instructions instead of just checking TARGET_ALTIVEC.
11665 (UNSPEC_VPKUWUM): Likewise.
11666 (UNSPEC_VPKSHSS): Likewise.
11667 (UNSPEC_VPKSWSS): Likewise.
11668 (UNSPEC_VPKUHUS): Likewise.
11669 (UNSPEC_VPKSHUS): Likewise.
11670 (UNSPEC_VPKUWUS): Likewise.
11671 (UNSPEC_VPKSWUS): Likewise.
11672 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
11673 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
11674 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
11675 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
11676 (UNSPEC_VUPKHSB): Likewise.
11677 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
11678 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
11679 (UNSPEC_VUPKHSH): Likewise.
11680 (UNSPEC_VUPKLSB): Likewise.
11681 (UNSPEC_VUPKLSH): Likewise.
11683 (VI_char): Likewise.
11684 (VI_scalar): Likewise.
11685 (VI_unit): Likewise.
11687 (VP_small): Likewise.
11688 (VP_small_lc): Likewise.
11689 (VU_char): Likewise.
11690 (add<mode>3): Likewise.
11691 (altivec_vaddcuw): Likewise.
11692 (altivec_vaddu<VI_char>s): Likewise.
11693 (altivec_vadds<VI_char>s): Likewise.
11694 (sub<mode>3): Likewise.
11695 (altivec_vsubcuw): Likewise.
11696 (altivec_vsubu<VI_char>s): Likewise.
11697 (altivec_vsubs<VI_char>s): Likewise.
11698 (altivec_vavgs<VI_char>): Likewise.
11699 (altivec_vcmpbfp): Likewise.
11700 (altivec_eq<mode>): Likewise.
11701 (altivec_gt<mode>): Likewise.
11702 (altivec_gtu<mode>): Likewise.
11703 (umax<mode>3): Likewise.
11704 (smax<mode>3): Likewise.
11705 (umin<mode>3): Likewise.
11706 (smin<mode>3): Likewise.
11707 (altivec_vpkuhum): Likewise.
11708 (altivec_vpkuwum): Likewise.
11709 (altivec_vpkshss): Likewise.
11710 (altivec_vpkswss): Likewise.
11711 (altivec_vpkuhus): Likewise.
11712 (altivec_vpkshus): Likewise.
11713 (altivec_vpkuwus): Likewise.
11714 (altivec_vpkswus): Likewise.
11715 (altivec_vpks<VI_char>ss): Likewise.
11716 (altivec_vpks<VI_char>us): Likewise.
11717 (altivec_vpku<VI_char>us): Likewise.
11718 (altivec_vpku<VI_char>um): Likewise.
11719 (altivec_vrl<VI_char>): Likewise.
11720 (altivec_vsl<VI_char>): Likewise.
11721 (altivec_vsr<VI_char>): Likewise.
11722 (altivec_vsra<VI_char>): Likewise.
11723 (altivec_vsldoi_<mode>): Likewise.
11724 (altivec_vupkhsb): Likewise.
11725 (altivec_vupkhs<VU_char>): Likewise.
11726 (altivec_vupkls<VU_char>): Likewise.
11727 (altivec_vupkhsh): Likewise.
11728 (altivec_vupklsb): Likewise.
11729 (altivec_vupklsh): Likewise.
11730 (altivec_vcmpequ<VI_char>_p): Likewise.
11731 (altivec_vcmpgts<VI_char>_p): Likewise.
11732 (altivec_vcmpgtu<VI_char>_p): Likewise.
11733 (abs<mode>2): Likewise.
11734 (vec_unpacks_hi_v16qi): Likewise.
11735 (vec_unpacks_hi_v8hi): Likewise.
11736 (vec_unpacks_lo_v16qi): Likewise.
11737 (vec_unpacks_hi_<VP_small_lc>): Likewise.
11738 (vec_unpacks_lo_v8hi): Likewise.
11739 (vec_unpacks_lo_<VP_small_lc>): Likewise.
11740 (vec_pack_trunc_v8h): Likewise.
11741 (vec_pack_trunc_v4si): Likewise.
11742 (vec_pack_trunc_<mode>): Likewise.
11744 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
11746 (vec_vmaxsd): Likewise.
11747 (vec_vmaxud): Likewise.
11748 (vec_vminsd): Likewise.
11749 (vec_vminud): Likewise.
11750 (vec_vpksdss): Likewise.
11751 (vec_vpksdus): Likewise.
11752 (vec_vpkudum): Likewise.
11753 (vec_vpkudus): Likewise.
11754 (vec_vrld): Likewise.
11755 (vec_vsld): Likewise.
11756 (vec_vsrad): Likewise.
11757 (vec_vsrd): Likewise.
11758 (vec_vsubudm): Likewise.
11759 (vec_vupkhsw): Likewise.
11760 (vec_vupklsw): Likewise.
11762 2013-05-29 Jan Hubicka <jh@suse.cz>
11764 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
11765 flags; reorder rest of fields in more consistent way.
11766 (varpool_node): Remove analyzed, finalized and alias.
11767 (cgraph_ndoe): Likewise.
11768 (symtab_alias_ultimate_target): New function.
11769 (cgraph_function_node): Move offline.
11770 (cgraph_reset_node): Declare.
11771 (cgraph_comdat_can_be_unshared_p): Remove.
11772 (varpool_remove_initializer): Declare.
11773 (varpool_first_defined_variable, varpool_next_defined_variable
11774 cgraph_first_defined_function, cgraph_next_defined_function): Update.
11775 (cgraph_function_with_gimple_body_p): Update.
11776 (varpool_all_refs_explicit_p): Update.
11777 (symtab_alias_target): New function.
11778 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
11779 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
11780 (cgraph_function_or_thunk_node): Simplify using
11781 symtab_alias_ultimate_target.
11782 (varpool_variable_node): Likewise.
11783 * cgraph.c (cgraph_create_function_alias): Update.
11784 (cgraph_add_thunk): Update.
11785 (cgraph_remove_node): Update.
11786 (dump_cgraph_node): Do not dump removed flags.
11787 (cgraph_function_body_availability): Update.
11788 (cgraph_propagate_frequency): Update.
11789 (verify_cgraph_node): Check sanity of local flag.
11790 (cgraph_function_node): Move here from cgraph.h; revamp for
11791 cgraph_function_or_thunk_node.
11792 * lto-symtab.c (lto_varpool_replace_node): Update.
11793 (lto_symtab_resolve_can_prevail_p): Update.
11794 (lto_symtab_merge_cgraph_nodes): Update.
11795 * ipa-cp.c (determine_versionability, initialize_node_lattices,
11796 propagate_constants_accross_call, devirtualization_time_bonus,
11797 ipcp_propagate_stage): Update.
11798 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
11799 * ipa-inline-transform.c (clone_inlined_nodes,
11800 preserve_function_body_p): Update.
11801 * ipa-reference.c (propagate): Update.
11802 (write_node_summary_p): Update.
11803 * toplev.c (wrapup_global_declaration_2): Update.
11804 * cgraphunit.c (cgraph_analyze_function): Rename to ...
11805 (analyze_function) ... this one.
11806 (cgraph_process_new_functions): Update.
11807 (cgraph_reset_node): Export.
11808 (cgraph_finalize_function): Update.
11809 (cgraph_add_new_function): Update.
11810 (process_function_and_variable_attributes): Update.
11811 (varpool_finalize_decl): Update.
11812 (symbol_finalized): Remove.
11813 (symbol_finalized_and_needed): Rename to ...
11814 (symbol_defined_and_needed): ... update.
11815 (cgraph_analyze_functions): Update.
11816 (handle_alias_pairs): Update.
11817 (mark_functions_to_output): Update.
11818 (assemble_thunk): Update.
11819 (output_in_order): Update.
11820 (output_weakrefs): Update.
11821 (finalize_compilation_unit): Update.
11822 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
11823 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
11824 input_node, input_varpool_node): Update.
11825 * dbxout.c (dbxout_expand_expr): Update.
11826 * cgraphclones.c (cgraph_clone_node): Update.
11827 (cgraph_copy_node_for_versioning): Update.
11828 (cgraph_materialize_clone): Update.
11829 (cgraph_materialize_all_clones): Update.
11830 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
11831 propagate_pure_const, propagate_nothrow): Update.
11832 * lto-streamer-out.c (lto_output, write_symbol): Update.
11833 * ipa-utils.c (ipa_reverse_postorder): Update.
11834 * ipa-inline.c (can_inline_edge_p): Update.
11835 (update_caller_keys, ipa_inline): Update.
11836 * dwarf2out.c (reference_to_unused,
11837 premark_types_used_by_global_vars_helper): Update.
11838 * tree-eh.c (tree_could_trap_p): Update.
11839 * ipa-split.c (consider_split, execute_split_functions): Update.
11840 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
11841 has_addr_references_p): Update; move ahead in file for better
11843 (process_references): Simplify.
11844 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
11845 bodies are removed.
11846 (cgraph_comdat_can_be_unshared_p): Make static.
11847 (cgraph_externally_visible_p): Update.
11848 (varpool_externally_visible_p): Update.
11849 (function_and_variable_visibility): Update.
11850 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
11851 ipa_tm_mark_force_output_node): Update.
11852 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
11853 estimate_edge_devirt_benefit, inline_generate_summary,
11854 inline_write_summary): Update.
11855 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
11856 * ipa-prop.c (ipa_compute_jump_functions): Update.
11857 (ipa_print_node_params, ipa_prop_read_section,
11858 ipa_update_after_lto_read, read_replacements_section): Update.
11859 * varasm.c (mark_decl_referenced): Update.
11860 (assemble_alias, dump_tm_clone_pairs): Update.
11861 * tree-inline.c (copy_bb): Update.
11862 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
11864 * symtab.c (dump_symtab_base): Print new flags.
11865 (verify_symtab_base): Verify new flags.
11866 (symtab_alias_ultimate_target): New function.
11867 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
11868 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
11870 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
11872 * i386.c (ix86_get_function_versions_dispatcher,
11873 ix86_generate_version_dispatcher_body): Update.
11874 (fold_builtin_cpu): Use varpool_add_new_variable.
11875 * varpool.c (varpool_remove_initializer): Break out from ...
11876 (varpool_remove_node): ... this one.
11877 (dump_varpool_node, varpool_node_for_asm,
11878 cgraph_variable_initializer_availability, varpool_analyze_node,
11879 varpool_assemble_decl, varpool_remove_unreferenced_decls,
11880 varpool_finalize_named_section_flags, varpool_create_variable_alias):
11883 2013-05-29 Jan Hubicka <jh@suse.cz>
11885 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
11887 2013-05-29 Easwaran Raman <eraman@google.com>
11889 PR tree-optimization/57442
11890 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
11891 when control exits the main loop.
11893 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
11895 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
11897 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
11898 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
11899 * rx/t-rx: Add rx100 under multi library matches option for nofpu
11902 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11904 PR tree-optimization/57441
11905 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
11906 Don't limit size of incr_vec to number of candidates.
11908 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
11910 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
11911 and mips16 directories.
11912 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
11913 (MULTILIB_DIRNAMES): Ditto.
11914 (MULTILIB_EXCEPTIONS): Add new exceptions.
11915 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
11916 (MULTILIB_DIRNAMES): Ditto.
11917 (MULTILIB_EXCEPTIONS): Add new exceptions.
11919 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11920 Marcus Shawcroft <marcus.shawcroft@arm.com>
11922 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
11923 SYMBOL_TINY_ABSOLUTE.
11924 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
11925 SYMBOL_TINY_ABSOLUTE.
11926 (aarch64_expand_mov_immediate): Likewise.
11927 (aarch64_classify_symbol): Likewise.
11928 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
11929 Permit SYMBOL_TINY_ABSOLUTE.
11930 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
11932 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11933 Marcus Shawcroft <marcus.shawcroft@arm.com>
11935 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
11936 Refactor if/switch. Replace gcc_assert with if.
11938 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
11940 * config/i386/i386.c (initial_ix86_tune_features): Enable
11941 FP Reassociation for AMD bdver1 and bdver2.
11943 2013-05-29 Martin Jambor <mjambor@suse.cz>
11945 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
11946 and IMAGPART_EXPR do not occur within other handled_components.
11948 2013-05-29 Richard Biener <rguenther@suse.de>
11950 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
11951 access on whether the use is in the BB we currently try to
11953 (vect_bb_vectorization_profitable_p): Pass the BB we currently
11954 vectorize to vect_bb_slp_scalar_cost.
11956 2013-05-29 Richard Biener <rguenther@suse.de>
11958 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
11959 computing scalar cost offsetted by stmts that are kept live
11961 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
11962 for computation of scalar cost.
11964 2013-05-28 Steve Ellcey <sellcey@mips.com>
11966 * config/mips/mips-cpus.def (mips32r2): Change processor type.
11968 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
11970 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
11971 array notation built-in reduction functions.
11972 * doc/passes.texi (Passes): Added documentation about changes done
11974 * doc/invoke.texi (C Dialect Options): Added documentation about
11975 the -fcilkplus flag.
11976 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
11977 (BUILTINS_DEF): Depend on cilkplus.def.
11978 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
11979 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
11980 * cilkplus.def: New file.
11982 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
11984 PR rtl-optimization/57439
11985 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
11987 2013-05-28 Easwaran Raman <eraman@google.com>
11989 PR tree-optimization/57337
11990 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
11991 (find_insert_point): Correctly identify the insertion point
11992 when two statements with the same UID is compared.
11994 2013-05-28 Richard Biener <rguenther@suse.de>
11996 PR tree-optimization/56787
11997 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
11998 from the list of data references.
11999 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
12001 (vect_analyze_loop_operations): Likewise.
12002 (vect_transform_loop): Remove clobbers.
12004 2013-05-28 Martin Jambor <mjambor@suse.cz>
12006 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
12007 and REALPART_EXPRs have scalar type.
12009 2013-05-28 Richard Biener <rguenther@suse.de>
12011 PR tree-optimization/57411
12012 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
12014 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
12015 virtual operand propagation.
12017 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
12019 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
12020 destination register for bmasksi_vis.
12021 (vector_init_bshuffle): Likewise.
12022 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
12024 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
12026 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
12027 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
12028 mode if the instruction isn't available in the original mode.
12029 * config/sparc/sparc.opt (mfix-ut699): New option.
12030 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
12031 (divdf3): Turn into expander.
12032 (divdf3_nofix): New insn.
12033 (divdf3_fix): Likewise.
12034 (divsf3): Disable if -mfix-ut699.
12035 (sqrtdf2): Turn into expander.
12036 (sqrtdf2_nofix): New insn.
12037 (sqrtdf2_fix): Likewise.
12038 (sqrtsf2): Disable if -mfix-ut699.
12040 2013-05-27 Richard Biener <rguenther@suse.de>
12042 PR middle-end/57412
12043 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
12044 block for the new loop.
12046 2013-05-27 Richard Biener <rguenther@suse.de>
12048 PR tree-optimization/57343
12049 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
12050 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
12051 (number_of_iterations_cond): Do not build the folded tree.
12053 2013-05-27 Richard Biener <rguenther@suse.de>
12056 PR middle-end/57381
12057 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12058 OEP_CONSTANT_ADDRESS_OF retained.
12060 PR tree-optimization/57417
12061 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
12062 for unchanged base.
12063 (set_ssa_val_to): Compare addresses using
12064 get_addr_base_and_unit_offset.
12066 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
12068 PR rtl-optimization/56833
12069 * postreload.c (move2add_record_mode): New function.
12070 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
12071 (move2add_use_add2_insn): Use move2add_record_sym_value.
12072 (move2add_use_add3_insn): Likewise.
12073 (reload_cse_move2add): Use move2add_valid_value_p and
12074 move2add_record_mode. Invalidate call-clobbered and REG_INC
12075 affected regs by setting reg_mode to VOIDmode.
12076 (move2add_note_store): Don't pretend the inside of a SUBREG is
12077 the actual destination. Invalidate single/leading registers by
12078 setting reg_mode to VOIDmode.
12079 Use move2add_record_sym_value, move2add_valid_value_p and
12080 move2add_record_mode.
12082 2013-05-27 Richard Biener <rguenther@suse.de>
12084 PR tree-optimization/57396
12085 * tree-affine.c (double_int_constant_multiple_p): Properly
12086 return false for val == 0 and div != 0.
12088 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
12090 * config/mips/mips.h: Use #elif in preprocessor conditions.
12092 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
12095 * config/mips/constraints.md (kl): New constraint.
12096 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
12097 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
12098 constraint for operand 0. Split after CSE for MIPS16. Emit a move
12099 from LO for MIPS16.
12100 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
12102 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
12105 * config/mips/mips.c (mips_can_inline_p): New function.
12106 (TARGET_CAN_INLINE_P): Define.
12108 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
12110 * sched-int.h (ds_t, dw_t): Make unsigned int.
12111 Fix documentation that describes how all the ds_t bits are used.
12112 Reserve the last bit for delayed-branch scheduling.
12113 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
12114 (BITS_PER_DEP_WEAK): Fix definition and documentation.
12115 (gen_dep_weak_1): Remove prototype.
12116 * sched-deps.c (get_dep_weak_1): Make static.
12117 * target.def (speculate_insn, needs_block_p, gen_spec_check,
12118 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
12119 * doc/tm.texi: Regenerate.
12120 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
12122 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
12125 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
12127 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
12128 Sandra Loosemore <sandra@codesourcery.com>
12130 * config.gcc (powerpc-*): Allow native for with-cpu.
12132 2013-05-24 Jeff Law <law@redhat.com>
12134 PR tree-optimization/57124
12135 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
12136 conversion feeding a condition if the range has an overflow
12137 if -fstrict-overflow. Add warnings for when we do make the
12140 2013-05-24 Dehao Chen <dehao@google.com>
12142 * tree-cfg.c (locus_discrim_map): Fix the typo.
12143 (locus_discrim_hasher): Likewise.
12144 (locus_discrim_hasher::hash): Likewise.
12145 (locus_discrim_hasher::equal): Likewise.
12147 2013-05-24 Martin Jambor <mjambor@suse.cz>
12149 PR tree-optimization/57294
12150 * cgraph.h (ipa_record_stmt_references): Declare.
12151 * cgraphbuild.c (ipa_record_stmt_references): New function.
12152 (build_cgraph_edges): Use ipa_record_stmt_references.
12153 (rebuild_cgraph_edges): Likewise.
12154 (cgraph_rebuild_references): Likewise.
12155 * ipa-prop.c (ipa_modify_call_arguments): Discard references
12156 associated with the old statement and build references from the
12157 newly built statements.
12158 * ipa-ref.c (ipa_remove_stmt_references): New function.
12159 * ipa-ref.h (ipa_remove_stmt_references): Declare.
12161 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
12163 * lra-constraints.c (emit_spill_move): Use smaller mode for
12165 (check_and_process_move): Consider mem-reg moves for secondary
12167 (curr_insn_transform): Don't lose insns emitted before for
12168 secondary memory moves.
12169 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
12170 reg set up in the current insn.
12172 2013-05-24 Dehao Chen <dehao@google.com>
12174 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
12176 (locus_descrim_hasher::equal): Likewise.
12177 (build_gimple_cfg): New discriminator assignment algorithm.
12178 (make_edges): Likewise.
12179 (next_discriminator_for_locus): Likewise.
12180 (same_line_p): Likewise.
12181 (assign_discriminators): Likewise.
12182 (make_cond_expr_edges): Likewise.
12183 (make_gimple_switch_edges): Likewise.
12184 (make_goto_expr_edges): Likewise.
12185 (make_gimple_asm_edges): Likewise.
12187 2013-05-24 Ian Bolton <ian.bolton@arm.com>
12189 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
12190 X format specifier to only display bottom 16 bits.
12191 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
12192 immediate to match for operand 2, since it will be masked.
12194 2013-05-24 Richard Biener <rguenther@suse.de>
12196 PR tree-optimization/57287
12197 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
12198 all SSA names that occur in abnormal PHIs.
12200 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
12203 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
12204 that index is not negative.
12206 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
12208 PR rtl-optimization/55177
12209 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
12210 (simplify_byte_swapping_operation): New.
12211 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
12212 (simplify_relational_operation_1): Deal with BSWAP.
12214 2013-05-23 Richard Henderson <rth@redhat.com>
12217 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
12218 (ix86_reorg): Call it.
12220 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
12223 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
12224 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
12225 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
12227 2013-05-23 Christian Bruel <christian.bruel@st.com>
12230 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
12232 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
12233 Marcus Shawcroft <marcus.shawcroft@arm.com>
12235 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
12236 * config/aarch64/constraints.md (Usa): Remove.
12237 * doc/md.texi (AArch64 Usa): Remove.
12239 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
12240 Marcus Shawcroft <marcus.shawcroft@arm.com>
12242 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
12243 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
12244 * config/aarch64/predicates.md (aarch64_const_address): Remove.
12245 (aarch64_mov_operand): Use aarch64_mov_operand_p.
12247 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
12249 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
12250 instruction (AdvSIMD).
12251 * config/aarch64/aarch64-builtins.c
12252 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
12253 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
12255 2013-05-23 Martin Jambor <mjambor@suse.cz>
12257 PR middle-end/57347
12258 * tree.h (contains_bitfld_component_ref_p): Declare.
12259 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
12260 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
12262 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
12263 not access a bit-field. Assert all final offsets are byte-aligned.
12265 2013-05-23 Richard Biener <rguenther@suse.de>
12267 PR tree-optimization/57380
12268 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
12269 least one invariant or re-used load.
12270 * passes.c (init_optimization_passes): Move pass_phiprop before
12273 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
12275 * config/aarch64/aarch64-simd.md
12276 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
12278 2013-05-23 Richard Biener <rguenther@suse.de>
12280 PR middle-end/57381
12281 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12282 OEP_CONSTANT_ADDRESS_OF retained.
12284 2013-05-23 Jakub Jelinek <jakub@redhat.com>
12286 PR middle-end/57344
12287 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
12288 don't lower unit. Handle unit not being always BITS_PER_WORD.
12290 2013-05-23 Richard Biener <rguenther@suse.de>
12292 PR rtl-optimization/57341
12293 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
12294 instead of true_dependence.
12296 2013-05-22 David Malcolm <dmalcolm@redhat.com>
12298 * bb-reorder.c (branch_threshold): Make const.
12299 (exec_threshold): Ditto.
12301 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12302 Pat Haugen <pthaugen@us.ibm.com>
12303 Peter Bergner <bergner@vnet.ibm.com>
12305 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
12306 documentation for the power8 crypto builtins.
12308 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
12310 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
12311 macros for defining power8 builtin functions.
12312 (BU_P8V_AV_2): Likewise.
12313 (BU_P8V_AV_P): Likewise.
12314 (BU_P8V_VSX_1): Likewise.
12315 (BU_P8V_OVERLOAD_1): Likewise.
12316 (BU_P8V_OVERLOAD_2): Likewise.
12317 (BU_CRYPTO_1): Likewise.
12318 (BU_CRYPTO_2): Likewise.
12319 (BU_CRYPTO_3): Likewise.
12320 (BU_CRYPTO_OVERLOAD_1): Likewise.
12321 (BU_CRYPTO_OVERLOAD_2): Likewise.
12322 (XSCVSPDP): Fix typo, point to the correct instruction.
12323 (VCIPHER): Add power8 crypto builtins.
12324 (VCIPHERLAST): Likewise.
12325 (VNCIPHER): Likewise.
12326 (VNCIPHERLAST): Likewise.
12327 (VPMSUMB): Likewise.
12328 (VPMSUMH): Likewise.
12329 (VPMSUMW): Likewise.
12330 (VPERMXOR_V2DI): Likewise.
12331 (VPERMXOR_V4SI: Likewise.
12332 (VPERMXOR_V8HI: Likewise.
12333 (VPERMXOR_V16QI: Likewise.
12334 (VSHASIGMAW): Likewise.
12335 (VSHASIGMAD): Likewise.
12336 (VPMSUM): Likewise.
12337 (VPERMXOR): Likewise.
12338 (VSHASIGMA): Likewise.
12340 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12341 __CRYPTO__ if the crypto instructions are available.
12342 (altivec_overloaded_builtins): Add support for overloaded power8
12345 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
12346 support for power8 crypto builtins.
12347 (builtin_function_type): Likewise.
12348 (altivec_init_builtins): Add support for builtins that take vector
12349 long long (V2DI) arguments.
12351 * config/rs6000/crypto.md: New file, define power8 crypto
12354 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12355 Pat Haugen <pthaugen@us.ibm.com>
12356 Peter Bergner <bergner@vnet.ibm.com>
12358 * doc/invoke.texi (Option Summary): Add power8 options.
12359 (RS/6000 and PowerPC Options): Likewise.
12361 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
12362 constraints.md instead of rs6000.h. Reorder w* constraints. Add
12363 wm, wn, wr documentation.
12365 * config/rs6000/constraints.md (wm): New constraint for VSX
12366 registers if direct move instructions are enabled.
12367 (wn): New constraint for no registers.
12368 (wq): New constraint for quad word even GPR registers.
12369 (wr): New constraint if 64-bit instructions are enabled.
12370 (wv): New constraint if power8 vector instructions are enabled.
12371 (wQ): New constraint for quad word memory locations.
12373 * config/rs6000/predicates.md (const_0_to_15_operand): New
12374 constraint for 0..15 for crypto instructions.
12375 (gpc_reg_operand): If VSX allow registers in VSX registers as well
12376 as GPR and floating point registers.
12377 (int_reg_operand): New predicate to match only GPR registers.
12378 (base_reg_operand): New predicate to match base registers.
12379 (quad_int_reg_operand): New predicate to match even GPR registers
12380 for quad memory operations.
12381 (vsx_reg_or_cint_operand): New predicate to allow vector logical
12382 operations in both GPR and VSX registers.
12383 (quad_memory_operand): New predicate for quad memory operations.
12384 (reg_or_indexed_operand): New predicate for direct move support.
12386 * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
12387 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
12388 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
12389 (POWERPC_MASKS): Add power8 options.
12390 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
12393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
12394 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
12396 * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
12397 (-mpower8-fusion): New power8 options.
12398 (-mpower8-fusion-sign): Likewise.
12399 (-mpower8-vector): Likewise.
12400 (-mcrypto): Likewise.
12401 (-mdirect-move): Likewise.
12402 (-mquad-memory): Likewise.
12404 * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
12405 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
12407 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
12408 (rs6000_debug_vector_unit): Add p8_vector.
12409 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
12410 definitions. Also print fusion state.
12411 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
12412 (rs6000_builtin_mask_calculate): Add power8 builtin support.
12413 (rs6000_option_override_internal): Add support for power8.
12414 (rs6000_common_init_builtins): Add debugging for skipped builtins
12415 if -mdebug=builtin.
12416 (rs6000_adjust_cost): Add power8 support.
12417 (rs6000_issue_rate): Likewise.
12418 (insn_must_be_first_in_group): Likewise.
12419 (insn_must_be_last_in_group): Likewise.
12420 (force_new_group): Likewise.
12421 (rs6000_register_move_cost): Likewise.
12422 (rs6000_opt_masks): Likewise.
12424 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
12425 power8 capable assembler, default to power7 options.
12426 (TARGET_DIRECT_MOVE): Likewise.
12427 (TARGET_CRYPTO): Likewise.
12428 (TARGET_P8_VECTOR): Likewise.
12429 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
12430 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
12431 (VECTOR_MEM_P8_VECTOR_P): Likewise.
12432 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
12433 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
12434 (TARGET_XSCVDPSPN): Likewise.
12435 (TARGET_XSCVSPDPN): Likewsie.
12436 (TARGET_SYNC_HI_QI): Likewise.
12437 (TARGET_SYNC_TI): Likewise.
12438 (MASK_CRYPTO): Likewise.
12439 (MASK_DIRECT_MOVE): Likewise.
12440 (MASK_P8_FUSION): Likewise.
12441 (MASK_P8_VECTOR): Likewise.
12442 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
12443 temporary used by some of the direct move instructions to get two FP
12444 temporary registers does not force creation of a stack frame.
12445 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
12446 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
12447 that any VSX registers are tieable, even if they are also an
12448 Altivec vector mode.
12449 (r6000_reg_class_enum): Add wm, wr, wv constraints.
12450 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
12451 (RS6000_BTM_CRYPTO): Likewise.
12452 (RS6000_BTM_COMMON): Likewise.
12454 * config/rs6000/rs6000.md (cpu attribute): Add power8.
12455 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
12456 (enum rs6000_vector): Add power8 vector support.
12458 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12462 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
12463 (any_sibcall_could_use_r3): this and handle indirect calls.
12464 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
12466 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12468 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
12470 2013-05-22 Richard Biener <rguenther@suse.de>
12472 PR middle-end/57349
12473 * profile.c (branch_prob): Do not split blocks that are
12474 abnormally receiving from ECF_RETURNS_TWICE functions.
12476 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
12478 * recog.c (offsettable_address_addr_space_p): Fix calculation of
12479 address mode. Move pointer mode initialization to the same place.
12481 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
12483 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
12484 while it has any effect.
12486 2013-05-21 Easwaran Raman <eraman@google.com>
12488 PR tree-optimization/57322
12489 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
12490 UID of the statement added to the BB to be 1.
12492 2013-05-21 Jakub Jelinek <jakub@redhat.com>
12494 PR tree-optimization/57331
12495 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
12496 of conversion from pointer type to integral type with integer.
12498 2013-05-21 Martin Jambor <mjambor@suse.cz>
12501 * ipa-prop.c (ipa_read_node_info): Process param_used and
12502 controlled_uses in the same order as when writing.
12504 2013-05-21 Magnus Granberg <baldrick@free.fr>
12507 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
12509 2013-05-21 Richard Biener <rguenther@suse.de>
12511 PR tree-optimization/57318
12512 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
12513 estimate stmts with side-effects as likely eliminated.
12515 2013-05-21 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/57330
12518 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
12519 preserve the call stmts fntype.
12521 2013-05-21 Richard Biener <rguenther@suse.de>
12523 PR tree-optimization/57303
12524 * tree-ssa-sink.c (statement_sink_location): Improve killing
12525 stmt detection and properly handle self-assignments.
12527 2013-05-21 Christian Bruel <christian.bruel@st.com>
12529 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
12530 spanning registers. LEAF_REG_REMAP is supported only for contiguous
12531 registers. Set register size out of the PARALLEL loop.
12533 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
12536 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
12537 (*fmasf4, *fmasf4_media): New insns.
12539 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
12541 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
12542 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
12543 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
12544 (mips_idiv_insns): Update the comments to say that the returned
12545 instruction counts are in units of BASE_INSN_LENGTH.
12546 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
12547 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
12548 using 2 rather than 4 as the length of indirect MIPS16 and
12549 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
12550 length of a NOP. Don't divide MIPS16 lengths by 2.
12551 (mips16_split_long_branches): Assume a branch is long if the
12552 length is greater than 4 rather than 8.
12553 * config/mips/mips.md (length): Give MIPS16 lengths directly,
12554 rather than multiplying them by 2. Multiply instruction counts
12555 by BASE_INSN_LENGTH rather than 4.
12556 (*jump_mips16, tls_get_tp_mips16_<mode>)
12557 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
12559 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
12561 * config/mips/mips.md (extended_mips16): Remove branch case.
12562 (length): Remove duplicated extended_mips16 test.
12564 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
12566 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
12568 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
12570 * recog.h (Recog_data): Rename to...
12571 (recog_data_d): ...this.
12572 (recog_data): Update accordingly.
12573 * recog.c (recog_data): Likewise.
12574 * reload.c (save_recog_data): Likewise.
12575 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
12576 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
12578 2013-05-17 Julian Brown <julian@codesourcery.com>
12580 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
12581 found in a REG_EQUAL note, invalidate it.
12583 2013-05-17 Easwaran Raman <eraman@google.com>
12585 * tree-ssa-reassoc.c (find_insert_point): New function.
12586 (insert_stmt_after): Likewise.
12587 (get_def_stmt): Likewise.
12588 (ensure_ops_are_available): Likewise.
12589 (not_dominated_by): Likewise.
12590 (rewrite_expr_tree): Do not move statements beyond what is
12591 necessary. Remove call to swap_ops_for_binary_stmt...
12592 (reassociate_bb): ... and move it here.
12593 (build_and_add_sum): Assign UIDs for new statements.
12594 (linearize_expr): Likewise.
12595 (do_reassoc): Renumber gimple statement UIDs.
12597 2013-05-17 Jan Hubicka <jh@suse.cz>
12599 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
12601 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
12602 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
12604 (output_weakrefs): Update.
12606 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
12607 Martin Jambor <mjambor@suse.cz>
12609 PR middle-end/57276
12610 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
12611 value that corresponds to the given aggval is found in values vector.
12613 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
12615 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
12616 sse, sse2, sse3, ssse3 and sse4a flags to options.
12618 2013-05-17 David Malcolm <dmalcolm@redhat.com>
12620 * gengtype-state.c: (s_expr_writer): New class, to handle
12621 prettifying of output layout of s-expressions.
12622 (state_writer): New class, to write out gtype.state.
12623 (state_written_type_count): Move this variable into member data of
12625 (s_expr_writer::s_expr_writer): New code: constructor for new class
12626 (state_writer::state_writer(): ditto
12627 (s_expr_writer::write_new_line): New function
12628 (s_expr_writer::write_any_indent): ditto
12629 (s_expr_writer::begin_s_expr): ditto
12630 (s_expr_writer::end_s_expr): ditto
12631 (write_state_fileloc): convert to method of state_writer...
12632 (state_writer:: write_state_fileloc): ...and use methods of
12633 s_expr_writer to write indentation into the gtype.state output file
12634 to visually represent the hierarchical structure of the list
12636 (write_state_fields): ditto, renaming to...
12637 (state_writer::write_state_fields)
12638 (write_state_a_string): ditto, renaming to...
12639 (state_writer::write_state_a_string)
12640 (write_state_string_option): ditto, renaming to...
12641 (state_writer::write_state_string_option)
12642 (write_state_type_option): ditto, renaming to...
12643 (state_writer::write_state_type_option)
12644 (write_state_nested_option): ditto, renaming to...
12645 (state_writer::write_state_nested_option)
12646 (write_state_option): ditto, renaming to...
12647 (state_writer::write_state_option)
12648 (write_state_options): ditto, renaming to...
12649 (state_writer::write_state_options)
12650 (write_state_lang_bitmap): ditto, renaming to...
12651 (state_writer::write_state_lang_bitmap)
12652 (write_state_version): ditto, renaming to...
12653 (state_writer::write_state_version)
12654 (write_state_scalar_type): ditto, renaming to...
12655 (state_writer::write_state_scalar_type)
12656 (write_state_string_type): ditto, renaming to...
12657 (state_writer::write_state_string_type)
12658 (write_state_undefined_type): ditto, renaming to...
12659 (state_writer::write_state_undefined_type)
12660 (write_state_struct_union_type): ditto, renaming to...
12661 (state_writer::write_state_struct_union_type)
12662 (write_state_struct_type): ditto, renaming to...
12663 (state_writer::write_state_struct_type)
12664 (write_state_user_struct_type): ditto, renaming to...
12665 (state_writer::write_state_user_struct_type)
12666 (write_state_lang_struct_type): ditto, renaming to...
12667 (state_writer::write_state_lang_struct_type)
12668 (write_state_param_struct_type): ditto, renaming to...
12669 (state_writer::write_state_param_struct_type)
12670 (write_state_pointer_type): ditto, renaming to...
12671 (state_writer::write_state_pointer_type)
12672 (write_state_array_type): ditto, renaming to...
12673 (state_writer::write_state_array_type)
12674 (write_state_gc_used): ditto, renaming to...
12675 (state_writer::write_state_gc_used)
12676 (write_state_common_type_content): ditto, renaming to...
12677 (state_writer::write_state_common_type_content)
12678 (write_state_type): ditto, renaming to...
12679 (state_writer::write_state_type)
12680 (write_state_pair_list): ditto, renaming to...
12681 (state_writer::write_state_pair_list)
12682 (write_state_pair): ditto, renaming to...
12683 (state_writer::write_state_pair)
12684 (write_state_typedefs): ditto, renaming to...
12685 (state_writer::write_state_typedefs)
12686 (write_state_structures): ditto, renaming to...
12687 (state_writer::write_state_structures)
12688 (write_state_param_structs): ditto, renaming to...
12689 (state_writer::write_state_param_structs)
12690 (write_state_variables): ditto, renaming to...
12691 (state_writer::write_state_variables)
12692 (write_state_srcdir): ditto, renaming to...
12693 (state_writer::write_state_srcdir)
12694 (write_state_files_list): ditto, renaming to...
12695 (state_writer::write_state_files_list)
12696 (write_state_languages): ditto, renaming to...
12697 (state_writer::write_state_languages)
12698 (write_state): create a state_writer instance and use it when
12699 writing out the state file
12701 2013-05-17 Mike Stump <mikestump@comcast.net>
12703 PR rtl-optimization/57304
12704 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
12705 accessing DF_REF_REAL_LOC.
12707 2013-05-17 Jakub Jelinek <jakub@redhat.com>
12709 PR rtl-optimization/57281
12710 PR rtl-optimization/57300
12711 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
12712 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
12713 what the other splitter did if the registers are dead.
12715 2013-05-17 Richard Biener <rguenther@suse.de>
12717 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
12720 2013-05-17 Jakub Jelinek <jakub@redhat.com>
12722 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
12725 2013-05-17 Marek Polacek <polacek@redhat.com>
12727 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
12728 length when doing non-zero store of storing '\0' to '\0'.
12730 2013-05-17 Jakub Jelinek <jakub@redhat.com>
12732 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
12733 vect_external_def oprnd1 with loop_vinfo, try to emit
12734 optional cast, negation and and stmts on the loop preheader
12735 edge instead of into the pattern def seq.
12737 PR tree-optimization/57051
12738 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
12739 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
12741 2013-05-16 Nick Clifton <nickc@redhat.com>
12743 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
12744 (rl78_is_naked_func): New function.
12745 (rl78_expand_prologue): Skip prologue generation for naked functions.
12746 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
12747 * doc/extend.texi (naked): Add RL78 to the list of processors
12748 that supports this attribute.
12750 2013-05-16 Jeff Law <law@redhat.com>
12752 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
12754 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
12756 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
12757 cache parameters using detect_caches_amd also for CYRIX,
12758 NSC and TM2 signatures.
12760 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
12761 Dzianis Kahanovich <mahatma@eu.by>
12765 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12766 VIA/Centaur processors and determine their cache parameters
12767 using detect_caches_amd.
12769 2013-05-16 Teresa Johnson <tejohnson@google.com>
12771 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
12772 (rtl_verify_edges): New function.
12773 (rtl_verify_bb_insns): Ditto.
12774 (rtl_verify_bb_pointers): Ditto.
12775 (rtl_verify_bb_insn_chain): Ditto.
12776 (rtl_verify_fallthru): Ditto.
12777 (rtl_verify_bb_layout): Ditto.
12778 (rtl_verify_flow_info_1): Outline checks into new functions.
12779 (rtl_verify_flow_info): Ditto.
12781 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
12783 * cfghooks.c (copy_bbs): Add update_dominance argument.
12784 * cfghooks.h (copy_bbs): Update prototype.
12785 * tree-cfg.c (gimple_duplicate_sese_region):
12786 Add update_dominance argument.
12787 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
12788 * tree-ssa-loop-ch.c (copy_loop_headers): Update
12789 gimple_duplicate_sese_region call.
12790 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
12791 Update copy_bbs call.
12792 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
12793 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
12795 2013-05-16 Jakub Jelinek <jakub@redhat.com>
12797 * tree-vectorizer.h (NUM_PATTERNS): Increment.
12798 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
12799 vect_recog_rotate_pattern.
12800 (vect_recog_rotate_pattern): New function.
12802 2013-05-16 Jason Merrill <jason@redhat.com>
12804 * Makefile.in (LLINKER): New variable.
12805 (mostlyclean): Remove link mutex.
12806 * configure.ac: Handle --enable-link-mutex.
12807 * lock-and-run.sh: New script.
12809 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12812 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
12815 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12817 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
12819 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
12821 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
12822 * config/arm/arm.c (next_consecutive_mem): New function.
12823 (gen_movmem_ldrd_strd): Likewise.
12824 * config/arm/arm.md (movmemqi): Update condition and code.
12825 (unaligned_loaddi, unaligned_storedi): New patterns.
12827 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12829 * config.gcc: Obsolete *-*-solaris2.9*.
12830 * doc/install.texi (Specific, *-*-solaris2*): Document it.
12832 2013-05-16 Richard Biener <rguenther@suse.de>
12834 * passes.c (init_optimization_passes): Move pass_parallelize_loops
12835 earlier, after GRAPHITE transforms and IV canonicalization.
12837 2013-05-16 Jakub Jelinek <jakub@redhat.com>
12839 * omp-low.c (extract_omp_for_data): For collapsed loops,
12840 if at least one of the loops is known at compile time to
12841 iterate zero times, set count to 0.
12842 (expand_omp_regimplify_p): New function.
12843 (expand_omp_for_generic): For collapsed loops, if at least
12844 one of the loops isn't known to iterate at least once,
12845 add runtime check with setting count to 0.
12846 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
12847 For unsigned types if it isn't known at compile time that
12848 the loop will iterate at least once, add runtime check to bypass
12849 the whole loop if initial condition isn't true.
12851 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
12853 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
12855 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
12857 PR middle-end/57286
12858 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
12859 transformations to avoid an infinite loop.
12861 2013-05-16 Marek Polacek <polacek@redhat.com>
12863 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
12865 2013-05-15 Leif Ekblad <leif@rdos.net>
12867 * config/i386/i386.c (ix86_decompose_address): Use
12868 DEFAULT_TLS_SEG_REG to access TLS segment register.
12869 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
12870 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
12871 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
12873 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
12876 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
12877 sibling calls to functions that would normally be lazily bound,
12878 unless $gp is call-clobbered.
12880 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
12882 * config/i386/i386.c (ix86_option_override_internal): Update
12883 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
12884 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
12885 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
12886 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
12888 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
12890 2013-05-15 Andreas Schwab <schwab@suse.de>
12892 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
12893 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
12896 2013-05-15 Teresa Johnson <tejohnson@google.com>
12898 * loop-unroll.c (report_unroll_peel): Check decision before
12899 emitting unroll/peel message.
12901 2013-05-15 Teresa Johnson <tejohnson@google.com>
12903 * function.h (has_bb_partition): New rtl_data flag.
12904 (bb_reorder_complete): Ditto.
12905 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
12906 instead of flag_reorder_blocks_and_partition.
12907 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
12908 with some enhancements.
12909 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
12910 * bb-reorder.c (connect_traces): Check for has_bb_partition
12911 instead of flag_reorder_blocks_and_partition.
12912 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
12913 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
12914 verify_hot_cold_block_grouping.
12915 (partition_hot_cold_basic_blocks): Set has_bb_partition.
12917 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12920 * config/arm/predicates.md (call_insn_operand): New predicate.
12921 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
12922 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
12923 if insn is not a tail call.
12924 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
12926 * config/arm/arm.h (enum reg_class): New caller save register class.
12927 (REG_CLASS_NAMES): Likewise.
12928 (REG_CLASS_CONTENTS): Likewise.
12929 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
12932 2013-05-15 Richard Biener <rguenther@suse.de>
12934 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
12935 of MSG_OPTIMIZED_LOCATIONS.
12936 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
12937 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
12939 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
12940 of MSG_OPTIMIZED_LOCATIONS.
12941 (execute_vect_slp): Likewise.
12942 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
12943 (vect_create_cond_for_alias_checks): Likewise.
12944 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12945 (vect_recog_widen_mult_pattern): Likewise.
12946 (vect_recog_widen_sum_pattern): Likewise.
12947 (vect_recog_over_widening_pattern): Likewise.
12948 (vect_recog_widen_shift_pattern): Likewise.
12949 (vect_recog_vector_vector_shift_pattern): Likewise.
12950 (vect_recog_divmod_pattern): Likewise.
12951 (vect_recog_mixed_size_cond_pattern): Likewise.
12952 (vect_recog_bool_pattern): Likewise.
12953 (vect_pattern_recog_1): Likewise.
12955 2013-05-15 Martin Jambor <mjambor@suse.cz>
12957 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
12958 non-functions to builtin_unreachable.
12959 * ipa-inline-transform.c (inline_call): Do not assert estimates were
12960 correct when new direct edges were discovered.
12962 2013-05-15 Martin Jambor <mjambor@suse.cz>
12964 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
12965 header, print symbol order instead of node uid, print more information
12966 about indirect edge targets.
12967 (ipa_make_edge_direct_to_target): Print symbol order instead of node
12969 (ipa_make_edge_direct_to_target): Likewise.
12970 (remove_described_reference): Likewise.
12971 (propagate_controlled_uses): Likewise.
12972 (ipa_print_node_params): Also print symbol order.
12973 (ipcp_transform_function): Print symbol order instead of node uids.
12974 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
12975 (cgraph_get_create_real_symbol_node): Likewise.
12976 * ipa-cp.c (print_lattice): Likewise.
12977 (print_all_lattices): Likewise.
12978 (determine_versionability): Likewise.
12979 (initialize_node_lattices): Likewise.
12980 (estimate_local_effects): Likewise.
12981 (update_profiling_info): Likewise.
12982 (create_specialized_node): Likewise.
12983 (perhaps_add_new_callers): Likewise.
12984 (decide_about_value): Likewise.
12985 (decide_whether_version_node): Likewise.
12986 (identify_dead_nodes): Likewise.
12987 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
12988 (dump_inline_summary): Likewise.
12989 (estimate_node_size_and_time): Likewise.
12990 (inline_analyze_function): Likewise.
12991 * ipa-inline.c (report_inline_failed_reason): Likewise.
12992 (want_early_inline_function_p): Likewise.
12993 (edge_badness): Likewise.
12994 (update_edge_key): Likewise.
12995 (inline_small_functions): Likewise. Add dumping of order to two other
12997 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
12998 instead of node uids.
12999 (propagate_pure_const): Likewise.
13000 (propagate_pure_const): Likewise.
13001 * ipa-utils.c (dump_cgraph_node_set): Likewise.
13002 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
13003 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
13005 * tree-pretty-print.c (dump_function_header): Likewise.
13006 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
13007 Print symbol order instead of node uids.
13009 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13011 * config/s390/s390.c (s390_register_move_cost): Don't impose the
13012 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
13014 2013-05-15 Richard Biener <rguenther@suse.de>
13016 PR tree-optimization/57275
13017 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
13018 return value for fail to do runtime alias checks for gather loads.
13020 2013-05-15 Jan Hubicka <jh@suse.cz>
13024 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
13025 weakrefs are not external.
13026 (lto_symtab_merge_decls): Fix thinko when dealing with
13027 non-lto_symtab decls.
13028 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
13029 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
13030 * varpool.c (dump_varpool_node): Dump more flags.
13032 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
13034 * config/i386/i386.c (processor_alias_table): Add instruction
13035 FSGSBASE for AMD bdver3 architecture.
13037 2013-05-14 Jakub Jelinek <jakub@redhat.com>
13039 * tree.c (warn_deprecated_use): Print file:line using locus color.
13040 * diagnostic.c (diagnostic_report_current_module): Print file:line
13041 and file:line:column using locus color.
13043 2013-05-14 Mike Stump <mikestump@comcast.net>
13045 * gdbinit.in: Add __null.
13047 2013-05-14 Mike Stump <mikestump@comcast.net>
13049 * recog.h: Rename struct recog_data to Recog_data.
13050 * recog.c: Likewise.
13051 * reload.c (can_reload_into): Likewise.
13052 * config/picochip/picochip.c: Likewise.
13054 2013-05-14 Mike Stump <mikestump@comcast.net>
13056 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
13058 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
13060 * resource.h (struct resources): Remove unch_memory member.
13061 (CLEAR_RESOURCE): Don't clear unch_memory.
13062 * resource.c (mark_referenced_resources): Don't set it.
13063 (mark_set_resources): Likewise.
13064 (mark_target_live_regs): Don't clear it.
13065 (init_resource_info): Likewise.
13066 * reorg.c (resource_conflicts_p): Don't compare it.
13067 (redundant_insn): Don't set it.
13069 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
13071 * emit-rtl.c (next_label): Remove unused function.
13072 (skip_consecutive_labels, link_cc0_insns): Move to ...
13073 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
13074 only place where these functions are used, and make them static.
13076 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
13078 * fold-const.c (fold_negate_expr): Handle vectors.
13079 (fold_truth_not_expr): Make it static.
13080 (fold_invert_truthvalue): New static function.
13081 (invert_truthvalue_loc): Handle vectors. Do not call
13082 fold_truth_not_expr directly.
13083 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
13084 <TRUTH_NOT_EXPR>: Do not cast to boolean.
13085 (fold_comparison): Handle vector constants.
13086 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
13087 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
13088 * tree.h (fold_truth_not_expr): Remove declaration.
13090 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
13092 * config/aarch64/aarch64-simd.md
13093 (aarch64_vcond_internal<mode>): Rename to...
13094 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
13095 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
13096 float modes. Clarify all iterator modes.
13097 (vcond<mode><mode>): Use new name for vcond expanders.
13098 (vcond<v_cmp_result><mode>): Likewise.
13099 (vcondu<mode><mode>: Likewise.
13100 * config/aarch64/iterators.md (VDQF_COND): New.
13102 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
13105 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
13106 variable for the shift amount. Check that we shift by non-negative
13109 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
13112 * config/arm/arm.h (EPILOGUE_USES): Only return true
13113 for LR_REGNUM after epilogue_completed.
13115 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
13117 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
13118 is error_mark_node.
13120 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13123 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
13124 and Solaris 11+/x86 with gld.
13125 * configure: Regenerate.
13127 2013-05-14 Jakub Jelinek <jakub@redhat.com>
13129 * expmed.c (expand_shift_1): Canonicalize rotates by
13130 constant bitsize / 2 to bitsize - 1.
13131 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
13132 case ROTATERT>: Likewise.
13135 2013-05-10 Jakub Jelinek <jakub@redhat.com>
13137 * config/i386/i386.md (rotateinv): New code attr.
13138 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13139 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13140 roll $31, %eax, etc.
13142 2013-05-14 Richard Biener <rguenther@suse.de>
13144 PR middle-end/57235
13145 * tree-eh.c (sink_clobbers): Give up for successors with
13146 multiple predecessors and no virtual uses.
13148 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
13150 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
13151 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
13153 2013-05-14 Jakub Jelinek <jakub@redhat.com>
13155 PR middle-end/57251
13156 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
13157 the case when both op0 and op1 have VOIDmode.
13159 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
13161 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
13162 in multiply-accumulate mode.
13164 2013-05-13 Guozhi Wei <carrot@google.com>
13166 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
13168 2013-05-13 Kai Tietz <ktietz@redhat.com>
13171 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
13172 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
13173 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
13174 * config/i386/i386.c (ix86_option_override_internal): Likewise.
13175 (ix86_expand_prologue): Likewise.
13176 (ix86_expand_split_stack_prologue): Likewise.
13177 (legitimate_pic_address_disp_p): Likewise.
13178 (legitimize_pic_address): Likewise.
13179 (legitimize_tls_address): Likewise.
13180 (legitimize_pe_coff_symbol): Likewise.
13181 (output_pic_addr_const): Likewise.
13182 (construct_plt_address): Likewise.
13183 (ix86_expand_call): Likewise.
13184 (x86_output_mi_thunk): Likewise.
13185 (x86_function_profiler): Likewise.
13187 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
13189 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
13190 similar switch cases.
13191 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
13192 (aarch64_simd_mov_to_<mode>low): Delete.
13193 (aarch64_simd_mov_to_<mode>high): Delete.
13194 (move_lo_quad_<mode>): Add w<-r alternative.
13195 (aarch64_simd_move_hi_quad_<mode>): Likewise.
13196 (aarch64_simd_mov_from_*): Update type attribute.
13197 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
13200 2013-05-13 Jan Hubicka <jh@suse.cz>
13202 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
13203 * config/i386/i386.c (ix86_compute_frame_layout,
13204 ix86_expand_epilogue, emit_i387_cw_initialization,
13205 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
13206 ix86_local_alignment): Fix use of size/speed predicates.
13208 2013-05-13 Jakub Jelinek <jakub@redhat.com>
13210 PR tree-optimization/45216
13211 PR tree-optimization/57157
13212 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
13213 the (-Y) & (B - 1) variant if OP is |.
13214 * expmed.c (expand_shift_1): For rotations by const0_rtx just
13215 return shifted. Use (-op1) & (prec - 1) as other_amount
13216 instead of prec - op1.
13218 2013-05-13 Martin Jambor <mjambor@suse.cz>
13220 PR middle-end/42371
13221 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
13222 (ipa_constant_data): New type.
13223 (ipa_jump_func): Use ipa_constant_data to hold information about
13224 constant jump functions.
13225 (ipa_get_jf_constant): Adjust to jump function type changes.
13226 (ipa_get_jf_constant_rdesc): New function.
13227 (ipa_param_descriptor): New field controlled_uses.
13228 (ipa_get_controlled_uses): New function.
13229 (ipa_set_controlled_uses): Likewise.
13230 * ipa-ref.h (ipa_find_reference): Declare.
13231 * ipa-prop.c (ipa_cst_ref_desc): New type.
13232 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
13234 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
13235 New parameter cs. Adjust all callers.
13236 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
13237 (remove_described_reference): New function.
13238 (jfunc_rdesc_usable): Likewise.
13239 (try_make_edge_direct_simple_call): Decrement controlled use count,
13240 attempt to remove reference if it hits zero.
13241 (combine_controlled_uses_counters): New function.
13242 (propagate_controlled_uses): Likewise.
13243 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
13244 (ipa_edge_duplication_hook): Duplicate reference descriptions.
13245 (ipa_print_node_params): Print described use counter.
13246 (ipa_write_jump_function): Adjust to jump function type changes.
13247 (ipa_read_jump_function): New parameter CS, pass it to
13248 ipa_set_jf_constant. Adjust caller.
13249 (ipa_write_node_info): Stream controlled use count
13250 (ipa_read_node_info): Likewise.
13251 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
13253 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
13254 count. Remove cloning-added reference if it reaches zero.
13255 * ipa-ref.c (ipa_find_reference): New function.
13257 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
13259 * config/i386/i386.c (processor_target_table): Modified default
13260 alignment values for AMD BD and BT architectures.
13262 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
13264 * tree-vect-generic.c (uniform_vector_p): Move ...
13265 * tree.c (uniform_vector_p): ... here.
13266 * tree.h (uniform_vector_p): Declare it.
13267 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
13270 2013-05-13 Jakub Jelinek <jakub@redhat.com>
13272 PR tree-optimization/57230
13273 * tree-ssa-strlen.c (handle_char_store): Record length for
13274 array store from STRING_CST.
13276 PR tree-optimization/57230
13277 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
13280 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
13282 * config/epiphany/epiphany.c (epiphany_init): Check size of
13283 NUM_MODES_FOR_MODE_SWITCHING.
13284 (epiphany_expand_prologue):
13285 Remove CONFIG_REGNUM initial value handling code.
13286 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
13287 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
13288 (emit_set_fp_mode, epiphany_mode_after): Likewise.
13289 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
13290 Don't return 1 for FP_MODE_NONE.
13291 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
13292 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
13293 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
13294 * config/epiphany/epiphany.md (save_config): New pattern.
13296 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
13298 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
13300 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
13302 * config/i386/i386.md (memory): Handle sseishft1.
13303 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
13304 (*vec_extractv2di_1): Ditto.
13306 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
13308 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
13311 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
13313 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
13314 Add mthumb/march=armv7-a multilib.
13315 Add mthumb/march=armv7-r multilib.
13316 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
13318 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
13320 * config/v850/t-rtems: Add more multilibs.
13322 2013-05-10 Richard Biener <rguenther@suse.de>
13324 PR tree-optimization/57214
13325 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
13326 not propagate from SSA names that occur in abnormal PHI nodes.
13328 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
13330 * stor-layout.c (element_precision): New function.
13331 * machmode.h (element_precision): Declare it.
13332 * tree.c (build_minus_one_cst): New function.
13333 (element_precision): Likewise.
13334 * tree.h (build_minus_one_cst): Declare new function.
13335 (element_precision): Likewise.
13336 * fold-const.c (operand_equal_p): Use element_precision.
13337 (fold_binary_loc): Handle vector types.
13338 * convert.c (convert_to_integer): Use element_precision.
13339 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
13342 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
13344 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
13345 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
13346 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
13347 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13348 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
13349 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
13350 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
13351 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13352 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
13353 (Uuw8): New constraints.
13354 (Usb4): Move into alphabetical order.
13355 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
13356 (sd8_operand, ub8_operand, uw8_operand): New predicates.
13357 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
13358 previously unnamed patterns.
13359 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
13360 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
13361 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
13362 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
13363 of set_attr_alternative/if_then_else. Use extended_mips16 instead
13364 of specific lengths.
13366 2013-05-10 Jakub Jelinek <jakub@redhat.com>
13368 * config/i386/i386.md (rotateinv): New code attr.
13369 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13370 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13371 roll $31, %eax, etc.
13373 PR tree-optimization/45216
13374 PR tree-optimization/57157
13375 * tree-ssa-forwprop.c (simplify_rotate): New function.
13376 (ssa_forward_propagate_and_combine): Call it.
13378 2013-05-10 Richard Biener <rguenther@suse.de>
13380 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
13381 disable peeling when we version for aliasing.
13382 (vector_alignment_reachable_p): Honor explicit user alignment.
13383 (vect_supportable_dr_alignment): Likewise.
13384 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
13385 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
13386 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
13387 then peeling to arrange for the cost-model check to come first.
13389 2013-05-10 Alan Modra <amodra@gmail.com>
13391 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
13392 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
13393 * configure: Regenerate.
13395 2013-05-10 Alan Modra <amodra@gmail.com>
13398 * varasm.c (default_elf_select_section): Move !DECL_P check..
13399 (get_named_section): ..to here before calling get_section_name.
13401 (default_section_type_flags): Add DECL_P check.
13402 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
13403 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
13405 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
13407 * config/epiphany/epiphany.c (epiphany_expand_prologue):
13408 When using gen_stack_adjust_str with a register offset, add a
13409 REG_FRAME_RELATED_EXPR note.
13411 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
13413 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
13414 (*vec_extractv4si_zext_mem): Ditto.
13415 (*vec_extractv2di): Add 0->x and x->x alternatives.
13416 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
13417 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
13419 2013-05-09 Jason Merrill <jason@redhat.com>
13421 N3639 C++1y VLA support
13422 * gimplify.c (gimplify_vla_decl): Don't touch an existing
13425 * tree.c (build_constructor_va): New.
13426 * tree.h: Declare it.
13428 2013-05-09 Martin Jambor <mjambor@suse.cz>
13431 * gimple-fold.c (canonicalize_constructor_val): Call
13432 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
13434 2013-05-09 Jan Hubicka <jh@suse.cz>
13435 Richard Biener <rguenther@suse.de>
13438 * symtab.c (symtab_make_decl_local): Do not add private names.
13440 2013-05-09 Jan Hubicka <jh@suse.cz>
13443 * symtab.c (insert_to_assembler_name_hash): Handle clones.
13444 (unlink_from_assembler_name_hash): Likewise.
13445 (symtab_prevail_in_asm_name_hash, symtab_register_node,
13446 symtab_unregister_node, symtab_initialize_asm_name_hash,
13447 change_decl_assembler_name): Update.
13449 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
13451 * config/aarch64/aarch64.md: New movtf split.
13452 (*movtf_aarch64): Update.
13453 (aarch64_movdi_tilow): Handle TF modes and rename to
13454 aarch64_movdi_<mode>low.
13455 (aarch64_movdi_tihigh): Handle TF modes and rename to
13456 aarch64_movdi_<mode>high
13457 (aarch64_movtihigh_di): Handle TF modes and rename to
13458 aarch64_mov<mode>high_di
13459 (aarch64_movtilow_di): Handle TF modes and rename to
13460 aarch64_mov<mode>low_di
13461 (aarch64_movtilow_tilow): Remove spurious whitespace.
13462 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
13464 (aarch64_print_operand): Update.
13466 2013-05-09 Alan Modra <amodra@gmail.com>
13468 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
13470 * configure: Regenerate.
13472 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
13474 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
13475 splitter preparation statements.
13476 * config/i386/sse.md (*vec_extract* splitters): Ditto.
13477 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
13480 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13482 * gimple-ssa-strength-reduction.c (count_candidates): Change
13483 return value to int.
13484 (analyze_candidates_and_replace): Change type of length to int.
13486 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
13488 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
13489 (*vec_extract<mode>): Use VI12_128 mode iterator.
13490 (*vec_extract<mode>_mem): Ditto.
13491 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
13494 2013-05-08 Diego Novillo <dnovillo@google.com>
13499 2012-08-17 Diego Novillo <dnovillo@google.com>
13502 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
13503 * config.in: Regenerate.
13504 * configure: Regenerate.
13505 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
13507 2013-05-08 Jan Hubicka <jh@suse.cz>
13510 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
13511 * cgraph.h (symtab_node_base): Add unique_name.
13512 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
13513 input_overwrite_node, input_varpool_node): Stream unique_name.
13514 * cgraphclones.c (cgraph_create_virtual_clone,
13515 cgraph_function_versioning): Set unique_name.
13516 * ipa.c (function_and_variable_visibility): Set unique_name.
13518 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13520 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
13521 (alloc_cand_and_find_basis): Restrict conditional candidate
13522 processing to CAND_MULTs.
13524 2013-05-08 Jan Hubicka <jh@suse.cz>
13527 lto-symtab.c (lto_symtab_symbol_p): New function.
13528 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
13529 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
13530 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
13531 Skip static symbols.
13533 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
13535 PR tree-optimization/57200
13536 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
13537 Only call inform if the preceding warning_at returns true.
13539 2013-05-07 Han Shen <shenhan@google.com>
13541 * cfgexpand.c (record_or_union_type_has_array_p): New function.
13542 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
13543 * common.opt (fstack-protector-strong): New option.
13544 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
13545 * doc/invoke.texi (Optimization Options): Document
13546 "-fstack-protector-strong".
13547 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
13549 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
13551 * config/mips/mips.c (mips_machine_reorg2): Return 0.
13553 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
13555 * ira.c (update_equiv_regs): Add insn having equiv memory even if
13556 it is not lhs of the insn.
13557 (setup_reg_equiv): Remove insn having equiv memory which it is not
13559 * lra-constraints.c (process_address): Try to improve generation
13560 code for address base + disp.
13561 (lra_constraints): Make correct the code for checking insn setting
13562 up backward equivalence. Remove insn only if it is in the init
13564 * lra-eliminations.c (update_reg_eliminate): Change return value.
13565 (lra_eliminate): Use the result.
13567 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
13569 * config/i386/sse.md (ssescalarnummask): New mode attribute.
13570 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
13571 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
13572 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
13573 register target operands.
13574 (*vec_extractv8hi_sse2): New pattern.
13575 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
13576 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
13577 (*vec_extract<mode>_mem): New insn and split pattern.
13579 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
13581 * config/arm/arm.c (arm_asan_shadow_offset): New function.
13582 (TARGET_ASAN_SHADOW_OFFSET): Define.
13583 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
13584 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
13586 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13588 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
13589 (incr_vec_index): Return -1 if increment not found.
13590 (create_add_on_incoming_edge): Assert if increment not found.
13591 (record_increment): Limit number of increments recorded.
13592 (all_phi_incrs_profitable): Return false if an increment not found.
13593 (replace_profitable_candidates): Don't process increments that were
13595 (analyze_candidates_and_replace): Limit size of incr_vec.
13597 2013-05-07 Richard Biener <rguenther@suse.de>
13599 * calls.c (special_function_p): setjmp-like functions are leaf.
13600 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
13601 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
13603 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
13605 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
13606 (aarch64_simd_mov<mode>): New expander.
13607 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
13608 (aarch64_simd_mov_to_<mode>high): Likewise.
13609 (aarch64_simd_mov_from_<mode>low): Likewise.
13610 (aarch64_simd_mov_from_<mode>high): Likewise.
13611 (aarch64_dup_lane<mode>): Update.
13612 (aarch64_dup_lanedi): New instruction pattern.
13613 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
13614 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
13616 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13618 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
13619 (replace_mult_candidate): Remove unnecessary argument; remove
13620 unnecessary parameter from call to introduce_cast_before_cand.
13621 (replace_unconditional_candidate): Remove unnecessary parameter
13622 from call to replace_mult_candidate.
13623 (replace_conditional_candidate): Likewise.
13624 (insert_initializers): Use make_temp_ssa_name.
13625 (introduce_cast_before_cand): Remove unnecessary argument; use
13626 make_temp_ssa_name.
13627 (replace_one_candidate): Remove unnecessary argument; remove
13628 unnecessary parameter from calls to introduce_cast_before_cand.
13629 (replace_profitable_candidates): Remove unnecessary parameters
13630 from calls to replace_one_candidate.
13632 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13634 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13635 phi def as possibly hiding a basis for a CAND_ADD whose operands
13636 have been commuted in the analysis.
13637 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
13639 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13641 * config/aarch64/aarch64.md
13642 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
13643 shift value between 0-4.
13645 2013-05-07 Richard Biener <rguenther@suse.de>
13647 * double-int.h (rshift): New overload.
13648 * double-int.c (rshift): New function.
13649 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
13650 (create_reference_ops_from_ref): Remove.
13651 (vn_reference_insert): Use shared ops for constructing the
13652 reference and copy it.
13654 2013-05-07 Richard Biener <rguenther@suse.de>
13656 PR middle-end/57190
13657 * tree-eh.c (sink_clobbers): Properly propagate
13658 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
13660 2013-05-07 Jakub Jelinek <jakub@redhat.com>
13662 PR tree-optimization/57149
13663 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
13664 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
13665 collect_phi_def_edges, execute_late_warn_uninitialized): Use
13666 uninit_undefined_value_p instead of ssa_undefined_value_p.
13669 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
13670 for modifier == EXPAND_INITIALIZER.
13672 2013-05-07 Anton Blanchard <anton@samba.org>
13674 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
13675 for powerpc64 little endian.
13676 * configure: Regenerate.
13678 2013-05-06 Graham Stott <grahams@btinternet.com>
13680 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
13681 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
13682 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
13683 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
13685 2013-05-06 Graham Stott <grahams@btinternet.com>
13687 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
13688 codes which allow non-lvalues.
13690 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
13692 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
13693 components are all 1s.
13694 (integer_minus_onep): New function.
13695 * tree.h (integer_minus_onep): Declare it.
13696 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
13697 integer_minus_onep instead of integer_all_onesp.
13699 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
13702 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
13703 variations of these patterns.
13705 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
13707 * config/i386/i386.md (isa): Add x64_sse4 member.
13708 (enabled): Handle x64_sse4.
13709 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
13710 instruction for 64bit SSE4_1 targets. Update insn attributes.
13711 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
13712 instruction for SSE4_1 targets. Update insn attributes.
13713 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13714 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
13715 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
13717 (*vec_extractv4si): Rename from *sse4_1_pextrd.
13718 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
13719 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
13721 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
13724 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
13726 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
13728 * final.c (do_assembler_dialects): Don't handle curly braces and
13729 vertical bar escaped by % as dialect delimiters.
13730 (output_asm_insn): Print curly braces and vertical bar if escaped
13731 by % and ASSEMBLER_DIALECT defined.
13732 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
13733 * doc/tm.texi: Regenerated.
13735 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
13737 * config/mips/mips.c: Include tree-pass.h.
13738 (mips_reorg): Split in pre- and post-dbr_schedule parts.
13739 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
13740 (pass_mips_machine_reorg2): New machine specific pass.
13741 (insert_pass_mips_machine_reorg2): New pass plugin definition.
13742 (mips_option_override): Register the new pass.
13743 * rtl.h (cleanup_barriers): Remove prototype.
13744 (dbr_schedule): Likewise.
13745 * jump.c (cleanup_barriers): Make static.
13746 * reorg.c (dbr_schedule): Likewise.
13748 2013-05-06 Richard Biener <rguenther@suse.de>
13750 PR tree-optimization/57185
13751 * tree-parloops.c (add_field_for_reduction): Handle anonymous
13752 SSA names properly.
13754 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
13757 * config/i386/i386.c (add_parameter_dependencies): Add dependence
13758 between "first_arg" and "insn", not "last" and "insn".
13760 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13762 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
13763 (find_candidates_in_block): Re-enable slsr_process_phi.
13764 (create_phi_basis): Fix double counting of candidate adjustment.
13766 2013-05-06 Richard Biener <rguenther@suse.de>
13768 PR middle-end/57147
13769 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
13770 the edge is also fallthru, preserve it and just clear the
13772 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
13773 also complex, preserve that and just clear the fallthru flag.
13774 * tree-inline.c (update_ssa_across_abnormal_edges): Also
13775 update virtual operands.
13777 2013-05-06 Alan Modra <amodra@gmail.com>
13779 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
13780 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
13781 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
13782 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
13783 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13784 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
13786 2013-05-06 Alan Modra <amodra@gmail.com>
13788 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
13789 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
13790 (DEFAULT_ASM_ENDIAN): Define.
13791 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13792 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
13793 Update -K PIC clause from sysv4.h.
13794 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
13795 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
13797 2013-05-06 Alan Modra <amodra@gmail.com>
13799 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
13800 twice for little-endian.
13801 (ashrdi3_no_power, ashrdi3): Support little-endian.
13803 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
13806 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
13807 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
13809 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
13810 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
13811 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
13814 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
13815 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
13817 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
13818 * config/pa/pa.opt: Make mbig-switch a no-op.
13819 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
13820 (CASE_VECTOR_MODE): Always return SImode.
13821 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
13822 for the !TARGET_BIG_SWITCH case.
13823 * config/pa/pa-linux.h: Likewise.
13824 * config/pa/pa-openbsd.h: Likewise.
13825 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
13826 * config/pa/pa.md (short_jump): Remove define_insn.
13827 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
13828 (casesi0): Remove define_insn.
13829 (type): Remove btable_branch.
13830 (pa_combine_type): Likewise.
13831 (in_nullified_branch_delay): Likewise.
13832 (in_call_delay): Likewise.
13833 (define_delay): Likewise.
13834 (define_insn_reservation "Z3"): Likewise.
13835 (define_insn_reservation "Z4"): Likewise.
13836 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
13837 (pa_adjust_insn_length): Remove adjustment for btable branches.
13838 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
13841 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
13843 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13844 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
13845 Add m->r,x alternatives.
13846 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
13847 splitters using SWI48x mode iterator.
13848 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
13849 TARGET_64BIT. Add m->x alternative.
13850 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
13851 Add o->x alternative. Enable for TARGET_SSE.
13852 (sse_storeq): Remove expander.
13853 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
13854 with memory input operand.
13855 (*vec_extractv2di_1 splitter): New.
13856 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
13857 * config/i386/i386.md (ssevecmodelower): New mode attribute.
13859 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13861 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
13862 (INT_LOWPART): Delete.
13863 (extract_MB): Adjust.
13864 (extract_ME): Adjust.
13865 (print_operand): Adjust.
13867 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13869 * config/rs6000/predicates.md (reg_or_add_cint_operand,
13870 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13871 (reg_or_logical_cint_operand, easy_fp_constant,
13872 logical_const_operand): Delete "CONST_DOUBLE" case.
13873 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
13874 "HOST_BITS_PER_WIDE_INT == 64" test.
13875 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
13876 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
13877 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
13878 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
13880 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
13881 CONST_DOUBLE DImode/VOIDmode case.
13882 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
13883 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
13884 CONST_DOUBLE VOIDmode case.
13885 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13886 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
13887 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
13888 Delete CONST_DOUBLE case.
13889 (splitters for mov FMOVE64 const_double): Delete
13890 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
13891 "HOST_BITS_PER_WIDE_INT >= 64" test.
13892 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
13894 (mov DI const_double): Delete.
13896 2013-05-04 Jakub Jelinek <jakub@redhat.com>
13898 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
13899 on op shows all bits zero in mode of a lowpart subreg, return zero.
13901 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13904 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
13905 to save TFmode registers and DImode to save TImode registers for
13906 caller save operations.
13907 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
13908 mark being partially clobbered since they only use the first
13911 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
13912 and TDmode only use the upper 64-bits of each VSX register.
13914 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13916 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
13917 (find_candidates_in_block): Disable slsr_process_phi.
13919 2013-05-03 Guozhi Wei <carrot@google.com>
13921 * coverage.c (coverage_obj_init): Move the construction of gcov
13923 (build_init_ctor): ... here.
13925 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13927 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
13928 (slsr_cand_d): Redefine def_phi.
13929 (stride_status, phi_adjust_status, count_phis_status): New enums.
13930 (find_phi_def): New.
13931 (find_basis_for_base_expr): New.
13932 (find_basis_for_candidate): Handle hidden bases.
13933 (alloc_cand_and_find_basis): Handle phi candidates.
13934 (slsr_process_phi): New.
13935 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
13936 (create_mul_imm_cand): Likewise.
13937 (create_add_ssa_cand): Exclude phi base candidates.
13938 (create_add_imm_cand): Likewise.
13939 (slsr_process_cast): Likewise.
13940 (slsr_process_copy): Likewise.
13941 (find_candidates_in_block): Handle phi candidates.
13942 (dump_candidate): Likewise.
13943 (unconditional_cands): Delete.
13944 (unconditional_cands_with_known_stride_p): Delete.
13945 (phi_dependent_cand_p): New.
13946 (cand_increment): Handle phi-dependent candidates.
13947 (replace_dependent): Delete.
13948 (replace_mult_candidate): New.
13949 (replace_unconditional_candidate): New.
13950 (incr_vec_index): Move to avoid forward reference.
13951 (create_add_on_incoming_edge): New.
13952 (create_phi_basis): New.
13953 (replace_dependents): Delete.
13954 (replace_conditional_candidate): New.
13955 (phi_add_costs): New.
13956 (replace_uncond_cands_and_profitable_phis): New.
13957 (record_increment): Handle phi adjustments.
13958 (record_phi_increments): New.
13959 (record_increments): Handle phi adjustments.
13960 (phi_incr_cost): New.
13961 (lowest_cost_path): Handle phis.
13962 (total_savings): Likewise.
13963 (analyze_increments): Likewise.
13964 (ncd_with_phi): New.
13965 (ncd_of_cand_and_phis): New.
13966 (nearest_common_dominator_for_cands): Handle phi increments.
13967 (all_phi_incrs_profitable): New.
13968 (replace_profitable_candidates): Handle phi-dependent candidates.
13969 (analyze_candidates_and_replace): Likewise.
13971 2013-05-03 Teresa Johnson <tejohnson@google.com>
13974 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
13975 do not exceed REG_BR_PROB_BASE.
13977 2013-05-03 Jeff Law <law@redhat.com>
13979 PR tree-optimization/57144
13980 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
13981 operand of the condition will bit into the new type when eliminating
13982 a cast feeding a condition.
13984 2013-05-03 Jakub Jelinek <jakub@redhat.com>
13986 PR rtl-optimization/57130
13987 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
13988 of COMPARE as in_code to the recursive call if needed.
13990 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
13992 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
13993 (enabled): Handle new members.
13994 * config/i386/sse.md (*vec_concatv2si): Merge from
13995 *vec_concatv2si_sse2 and vec_concatv2si_sse.
13996 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
13998 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
14000 PR tree-optimization/57027
14001 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
14002 for fnms opportunity, check we got the prerequisite kind
14003 of tree / gimple before using accessor functions.
14005 2013-05-03 Richard Biener <rguenther@suse.de>
14007 * double-int.h (lshift): New overload without precision
14008 and arith argument.
14009 (operator *=, operator +=, operator -=): Move ...
14010 * double-int.c (operator *=, operator +=, operator -=): ... here
14011 and implement more efficiently.
14012 (mul_double_with_sign): Remove.
14013 (lshift_double): Adjust to take unsinged shift argument, push
14014 dispatching code to callers.
14015 (mul_double_wide_with_sign): Add early out for callers that
14016 are not interested in high parts or overflow.
14017 (lshift): New function.
14018 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
14019 dispatch code here.
14020 (lrotate, rrotate): Use logical shifts.
14021 * expr.c (get_inner_reference): Use lshift.
14022 * fixed-value.c (do_fixed_divide): Likewise.
14023 * tree-dfa.c (get_ref_base_and_extent): Likewise.
14024 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
14025 (indirect_refs_may_alias_p): Likewise.
14026 (stmt_kills_ref_p_1): Likewise.
14028 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
14030 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
14032 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
14034 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
14035 scalar form of FABD instruction.
14037 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
14039 * lra-constraints.c (process_alt_operands): Add checking alt
14040 number to choose the best alternative.
14042 2013-05-02 Richard Biener <rguenther@suse.de>
14044 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
14045 bitmap and its handling.
14046 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
14048 2013-05-02 Richard Biener <rguenther@suse.de>
14050 PR middle-end/57140
14051 * tree-inline.c (copy_loops): Properly handle removed loops.
14052 (copy_cfg_body): Mark destination loops for fixup if source
14053 loops needed fixup.
14055 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
14058 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
14059 generating simple_return for naked functions.
14061 2013-05-02 Martin Jambor <mjambor@suse.cz>
14063 PR middle-end/56988
14064 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
14065 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
14067 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
14068 ipa_agg_replacement_value structures.
14069 (known_aggs_to_agg_replacement_list): Likewise.
14070 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
14071 (read_agg_replacement_chain): Likewise.
14072 (ipcp_transform_function): Also check that by_ref flags match.
14074 2013-05-02 Richard Biener <rguenther@suse.de>
14076 * graphds.h (struct graph): Add obstack member.
14077 * graphds.c (new_graph): Initialize obstack and allocate
14079 (add_edge): Allocate edge from the obstack.
14080 (free_graph): Free the obstack instead of all edges and vertices.
14082 2013-05-02 Teresa Johnson <tejohnson@google.com>
14084 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
14086 * cfg.c (update_bb_profile_for_threading): Ditto.
14087 * tree-inline.c (copy_bb): Ditto.
14088 (copy_edges_for_bb): Ditto.
14089 (initialize_cfun): Ditto.
14090 (copy_cfg_body): Ditto.
14091 (expand_call_inline): Ditto.
14092 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
14093 (estimate_node_size_and_time): Ditto.
14094 (inline_merge_summary): Ditto.
14095 * cgraphclones.c (cgraph_clone_edge): Ditto.
14096 (cgraph_clone_node): Ditto.
14097 * sched-rgn.c (compute_dom_prob_ps): Ditto.
14098 (compute_trg_info): Ditto.
14100 2013-05-02 Ian Bolton <ian.bolton@arm.com>
14102 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
14103 S reg when fp attribute set.
14104 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
14106 2013-05-02 Ian Bolton <ian.bolton@arm.com>
14108 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
14110 (*and_one_cmplsi3_compare0_uxtw): Likewise.
14111 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
14112 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
14114 2013-05-02 Richard Biener <rguenther@suse.de>
14116 * tree-scalar-evolution.c (scev_info_hasher): Remove.
14117 (struct instantiate_cache_entry): New type.
14118 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
14119 (struct instantiate_cache_type): New type.
14120 (set_instantiated_value, get_instantiated_value): Remove.
14121 (get_instantiated_value_entry): New function.
14122 (instantiate_scev_name): Use the new cache and adjust.
14123 (instantiate_scev_poly): Adjust.
14124 (instantiate_scev_binary): Likewise.
14125 (instantiate_array_ref): Likewise.
14126 (instantiate_scev_convert): Likewise.
14127 (instantiate_scev_not): Likewise.
14128 (instantiate_scev_3): Likewise.
14129 (instantiate_scev_2): Likewise.
14130 (instantiate_scev_r): Likewise.
14131 (instantiate_scev): Likewise.
14132 (resolve_mixers): Likewise.
14134 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
14137 * lra-constraints.c (best_small_class_operands_num): Remove.
14138 (process_alt_operands): Remove small_class_operands_num. Take
14139 small classes operands into losers and only if the operand is not
14140 matched. Modify debugging output.
14141 (curr_insn_transform): Remove best_small_class_operands_num.
14144 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14146 * config/aarch64/aarch64-builtins.c
14147 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
14148 * config/aarch64/aarch64-simd-builtins.def
14149 (reduc_splus_): Add new modes.
14150 (reduc_uplus_): New.
14151 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
14152 (reduc_uplus_v4sf): Likewise.
14153 (reduc_splus_v4sf): Likewise.
14154 (aarch64_addv<mode>): Likewise.
14155 (reduc_uplus_<mode>): Likewise.
14156 (reduc_splus_<mode>): Likewise.
14157 (aarch64_addvv2di): Likewise.
14158 (reduc_uplus_v2di): Likewise.
14159 (reduc_splus_v2di): Likewise.
14160 (aarch64_addvv2si): Likewise.
14161 (reduc_uplus_v2si): Likewise.
14162 (reduc_splus_v2si): Likewise.
14163 (reduc_<sur>plus_<mode>): New.
14164 (reduc_<sur>plus_v2di): Likewise.
14165 (reduc_<sur>plus_v2si): Likewise.
14166 (reduc_<sur>plus_v4sf): Likewise.
14167 (aarch64_addpv4sf): Likewise.
14168 * config/aarch64/arm_neon.h
14169 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
14170 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
14171 add UNSPEC_SADDV, UNSPEC_UADDV.
14173 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
14175 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14177 * config/aarch64/arm_neon.h
14178 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
14180 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14182 * config/aarch64/aarch64-builtins
14183 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
14185 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14187 * config/aarch64/aarch64-simd-builtins.def
14188 (reduc_smax_): New.
14189 (reduc_smin_): Likewise.
14190 (reduc_umax_): Likewise.
14191 (reduc_umin_): Likewise.
14192 (reduc_smax_nan_): Likewise.
14193 (reduc_smin_nan_): Likewise.
14196 (smax): Update for V2SF, V4SF and V2DF modes.
14199 (smin_nan): Likewise.
14200 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
14201 (<su><maxmin><mode>3): ...This, refactor.
14202 (s<maxmin><mode>3): New.
14203 (<maxmin_uns><mode>3): Likewise.
14204 (reduc_<maxmin_uns>_<mode>): Refactor.
14205 (reduc_<maxmin_uns>_v4sf): Likewise.
14206 (reduc_<maxmin_uns>_v2si): Likewise.
14207 (aarch64_<fmaxmin><mode>: Remove.
14208 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
14210 (vmin<q>_f<32,64>): Likewise.
14211 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14213 (su): Add mappings for smax, smin, umax, umin.
14215 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14216 (FMAXMIN): Rename as...
14217 (FMAXMIN_UNS): ...This.
14219 (fmaxminv): Likewise.
14220 (fmaxmin): Likewise.
14222 (maxmin_uns_op): Likewise.
14224 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14226 * config/aarch64/arm_neon.h
14227 (vac<ge, gt><sd>_f<32, 64>): Rename to...
14228 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
14229 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
14231 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14233 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
14234 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
14236 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14238 * config/aarch64/aarch64-simd.md
14239 (vcond<mode>_internal): Handle special cases for constant masks.
14240 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
14241 (vcondu<mode><mode>): Likewise.
14242 (vcond<v_cmp_result><mode>): New.
14244 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14246 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
14247 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
14248 * config/aarch64/aarch64-simd-builtins.def
14249 (cmeq): Update to BUILTIN_VALLDI.
14254 * config/aarch64/arm_neon.h
14255 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
14256 to builtins or C as appropriate.
14258 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
14260 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
14262 (cmhi): Rename to...
14264 * config/aarch64/aarch64-simd.md
14265 (simd_mode): Add SF.
14266 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
14267 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
14268 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
14269 (cstore<mode>_neg): ...This.
14270 * config/aarch64/iterators.md
14272 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
14273 (COMPARISONS): New.
14274 (UCOMPARISONS): Likewise.
14275 (optab): Add missing comparisons.
14281 (VCMP_S): Likewise.
14282 (VCMP_U): Likewise.
14283 (V_cmp_result): Add DF, SF modes.
14284 (v_cmp_result): Likewise.
14286 (vmtype): Likewise.
14287 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
14289 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
14291 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
14292 define_insn to define_insn_and_split.
14293 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
14294 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
14295 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
14296 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
14297 (thumb2_negscc): Likewise.
14299 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
14301 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
14303 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
14305 * config/arm/thumb2.md: Remove trailing whitespaces.
14307 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
14309 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
14310 Use gen_int_mode rather than GEN_INT.
14312 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
14314 * value-prof.c (stream_in_histogram_value): Remove the strayed
14317 2013-04-30 Richard Biener <rguenther@suse.de>
14319 PR middle-end/57122
14320 * cfghooks.c (split_edge): Properly check for the loop latch edge.
14322 2013-04-30 Richard Biener <rguenther@suse.de>
14324 PR middle-end/57107
14325 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
14327 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
14329 PR rtl-optimization/56957
14330 PR rtl-optimization/57105
14331 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
14332 variable. Use just INSN_UID for determining whether an insn
14333 should be only disconnected from the insn stream.
14334 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
14336 2013-04-30 Jakub Jelinek <jakub@redhat.com>
14338 PR tree-optimization/57104
14339 * tsan.c (instrument_expr): Don't instrument accesses to
14340 DECL_HARD_REGISTER VAR_DECLs.
14342 2013-04-30 Richard Biener <rguenther@suse.de>
14344 * function.h (loops_for_fn): New inline function.
14345 (set_loops_for_fn): Likewise.
14346 * cfgloop.h (place_new_loop): Add struct function parameter.
14347 (get_loop): Likewise.
14348 (get_loops): Likewise.
14349 (number_of_loops): Likewise.
14350 (fel_next): Adjust.
14351 (fel_init): Likewise.
14352 * cfg.c (get_loop_copy): Adjust.
14353 * cfgloop.c (flow_loops_dump): Likewise.
14354 (record_loop_exits): Likewise.
14355 (verify_loop_structure): Likewise.
14356 * cfgloopanal.c (mark_irreducible_loops): Likewise.
14357 (estimate_reg_pressure_cost): Likewise.
14358 (mark_loop_exit_edges): Likewise.
14359 * cfgloopmanip.c (place_new_loop): Likewise.
14360 (add_loop): Likewise.
14361 (duplicate_loop): Likewise.
14362 * graph.c (draw_cfg_nodes): Likewise.
14363 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
14364 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
14365 (extract_affine_chrec): Likewise.
14366 (build_scop_iteration_domain): Likewise.
14367 * graphite.c (graphite_initialize): Likewise.
14368 * ira-build.c (create_loop_tree_nodes): Likewise.
14369 (more_one_region_p): Likewise.
14370 (rebuild_regno_allocno_maps): Likewise.
14371 (mark_loops_for_removal): Likewise.
14372 (mark_all_loops_for_removal): Likewise.
14373 (remove_unnecessary_regions): Likewise.
14374 (ira_build): Likewise.
14375 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
14376 * loop-init.c (fix_loop_structure): Likewise.
14377 (gate_rtl_move_loop_invariants): Likewise.
14378 (gate_rtl_unswitch): Likewise.
14379 (gate_rtl_unroll_and_peel_loops): Likewise.
14380 (rtl_doloop): Likewise.
14381 * lto-streamer-in.c (input_cfg): Likewise.
14382 * lto-streamer-out.c (output_cfg): Likewise.
14383 * modulo-sched.c (sms_schedule): Likewise.
14384 * predict.c (tree_estimate_probability): Likewise.
14385 (tree_estimate_probability_driver): Likewise.
14386 (estimate_loops): Likewise.
14387 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
14388 (move_sese_region_to_fn): Likewise.
14389 (debug_loop_num): Likewise.
14390 * tree-chrec.c (chrec_evaluate): Likewise.
14391 (hide_evolution_in_other_loops_than_loop): Likewise.
14392 (chrec_component_in_loop_num): Likewise.
14393 (reset_evolution_in_loop): Likewise.
14394 (evolution_function_is_invariant_rec_p): Likewise.
14395 * tree-if-conv.c (main_tree_if_conversion): Likewise.
14396 * tree-inline.c (copy_loops): Likewise.
14397 (copy_cfg_body): Likewise.
14398 (tree_function_versioning): Likewise.
14399 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
14400 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
14402 (add_to_evolution_1): Likewise.
14403 (scev_const_prop): Likewise.
14404 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
14405 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
14406 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
14407 (tree_ssa_lim_initialize): Likewise.
14408 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
14409 (verify_loop_closed_ssa): Likewise.
14410 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
14411 (tree_ssa_loop_im): Likewise.
14412 (tree_ssa_loop_unswitch): Likewise.
14413 (tree_vectorize): Likewise.
14414 (check_data_deps): Likewise.
14415 (tree_ssa_loop_ivcanon): Likewise.
14416 (tree_ssa_loop_bounds): Likewise.
14417 (tree_complete_unroll): Likewise.
14418 (tree_complete_unroll_inner): Likewise.
14419 (tree_parallelize_loops): Likewise.
14420 (tree_ssa_loop_prefetch): Likewise.
14421 (tree_ssa_loop_ivopts): Likewise.
14422 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
14423 * tree-vectorizer.c (vectorize_loops): Likewise.
14425 2013-04-29 Mike Frysinger <vapier@gentoo.org>
14427 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
14428 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
14429 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
14430 with EABI_LINK_SPEC.
14432 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
14435 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
14438 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
14441 * lra-constraints.c (process_alt_operands): Discourage a bit more
14442 using memory for pseudos. Print cost dump for alternatives.
14443 Modify cost values for conflicts with early clobbers.
14444 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
14446 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
14449 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
14451 2013-04-29 Ian Bolton <ian.bolton@arm.com>
14453 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
14454 from/to S register.
14455 (movdi_aarch64): Support LDR/STR from/to D register.
14457 2013-04-29 Ian Bolton <ian.bolton@arm.com>
14459 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
14460 or higher by default.
14462 2013-04-29 Richard Biener <rguenther@suse.de>
14464 PR middle-end/57075
14465 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
14466 even if not adding abnormal edges for calls that can make
14469 2013-04-29 Richard Biener <rguenther@suse.de>
14471 PR middle-end/57103
14472 * tree-cfg.c (move_stmt_op): Fix condition under which to update
14474 (move_stmt_r): Remove redundant checking.
14476 2013-04-29 Teresa Johnson <tejohnson@google.com>
14479 * basic-block.h (apply_scale): New function.
14480 (apply_probability): Use apply_scale.
14481 * gimple-streamer-in.c (input_bb): Ditto.
14482 * lto-streamer-in.c (input_cfg): Ditto.
14483 * lto-cgraph.c (merge_profile_summaries): Ditto.
14484 * tree-optimize.c (execute_fixup_cfg): Ditto.
14485 * tree-inline.c (copy_bb): Update comment to use apply_scale.
14486 (copy_edges_for_bb): Ditto.
14487 (copy_cfg_body): Ditto.
14489 2013-04-29 Tom de Vries <tom@codesourcery.com>
14491 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
14492 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
14493 (tail_merge_optimize): Handle current_loops == NULL.
14495 2013-04-26 Jeff Law <law@redhat.com>
14497 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
14498 (simplify_cond_using_ranges): Generalize code to simplify
14499 COND_EXPRs where one argument is a constant and the other
14500 is an SSA_NAME created by an integral type conversion.
14502 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14504 * config/arm/arm.md (store_minmaxsi): Use only when
14505 optimize_insn_for_size_p.
14507 2013-04-29 Christian Bruel <christian.bruel@st.com>
14510 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
14512 2013-04-29 Richard Biener <rguenther@suse.de>
14514 PR middle-end/57089
14515 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
14516 loop tree make sure to schedule a fixup for the child as well.
14517 (expand_omp_for_generic): Properly add loops.
14518 (expand_omp_for_static_nochunk): Likewise.
14519 (expand_omp_for_static_chunk): Likewise.
14520 (expand_omp_for): For the degenerate case fixup loops.
14521 (expand_omp_sections): Fix default bb placement in loops.
14522 (expand_omp_atomic_pipeline): Properly add loops.
14524 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14526 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
14528 2013-04-29 Tom de Vries <tom@codesourcery.com>
14530 * tree-ssa-tail-merge.c: Update header comment.
14532 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14534 * config/aarch64/arm_neon.h
14535 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
14536 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
14537 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
14538 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
14539 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
14540 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
14541 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
14542 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
14544 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14546 * config/aarch64/aarch64-simd.md
14547 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
14548 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
14549 fix_trunc, fixuns_trunc.
14550 (ftrunc<VDQF:mode>2): New.
14551 * config/aarch64/iterators.md (optab): Add fix, fixuns.
14552 (fix_trunc_optab): New.
14554 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14556 * config/aarch64/aarch64-builtins.c
14557 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
14558 iceilf, lround, iroundf.
14560 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
14563 * config/i386/i386.h (enum ix86_tune_indices)
14564 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
14565 New, split from X86_TUNE_INTER_UNIT_MOVES.
14566 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
14567 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
14568 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
14569 (TARGET_INTER_UNIT_MOVES): Remove.
14570 * config/i386/i386.c (initial_ix86_tune_features): Update.
14571 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
14572 (ix86_expand_convert_uns_didf_sse): Use
14573 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14574 (ix86_expand_vector_init_one_nonzero): Ditto.
14575 (ix86_expand_vector_init_interleave): Ditto.
14576 (inline_secondary_memory_needed): Return true for moves from SSE class
14577 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
14578 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14579 * config/i386/constraints.md (Yi, Ym): Depend on
14580 TARGET_INTER_UNIT_MOVES_TO_VEC.
14581 (Yj, Yn): New constraints.
14582 * config/i386/i386.md (*movdi_internal): Change constraints of
14583 operand 1 from Yi to Yj and from Ym to Yn.
14584 (*movsi_internal): Ditto.
14585 (*movdf_internal): Ditto.
14586 (*movsf_internal): Ditto.
14587 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
14588 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14589 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
14590 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
14591 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
14592 * config/i386/sse.md (movdi_to_sse): Ditto.
14593 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
14594 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
14595 TARGET_INTER_UNIT_MOVES.
14596 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
14597 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
14598 instead of TARGET_INTER_UNIT_MOVES.
14599 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
14600 operand 1 from Yi to Yj and from Ym to Yn.
14602 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14604 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
14605 (float_truncate_hi_): Likewise.
14606 (float_extend_lo_): Likewise.
14607 (float_truncate_lo_): Likewise.
14608 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
14609 (aarch64_float_extend_lo_v2df): Likewise.
14610 (vec_unpacks_hi_v4sf): Likewise.
14611 (aarch64_float_truncate_lo_v2sf): Likewise.
14612 (aarch64_float_truncate_hi_v4sf): Likewise.
14613 (vec_pack_trunc_v2df): Likewise.
14614 (vec_pack_trunc_df): Likewise.
14616 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14618 * config/aarch64/aarch64-builtins.c
14619 (aarch64_fold_builtin): Fold float conversions.
14620 * config/aarch64/aarch64-simd-builtins.def
14621 (floatv2si, floatv4si, floatv2di): New.
14622 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
14623 * config/aarch64/aarch64-simd.md
14624 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
14625 * config/aarch64/iterators.md (FLOATUORS): New.
14626 (optab): Add float, floatuns.
14627 (su_optab): Likewise.
14629 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14631 * config/aarch64/aarch64-builtins.c
14632 (aarch64_builtin_vectorized_function): Use new names for
14634 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
14635 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
14636 (fcvtzu): Split as...
14637 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
14638 (fcvtas): Split as...
14639 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
14640 (fcvtau): Split as...
14641 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
14642 (fcvtps): Split as...
14643 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
14644 (fcvtpu): Split as...
14645 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
14646 (fcvtms): Split as...
14647 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
14648 (fcvtmu): Split as...
14649 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
14650 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
14651 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
14652 (lfrintnusf, lfrintnudf): Likewise.
14653 * config/aarch64/aarch64-simd.md
14654 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
14656 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
14657 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
14658 (fcvt_pattern): Likewise.
14660 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14662 * config/aarch64/aarch64-simd.md
14663 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
14664 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
14666 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14668 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
14669 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
14670 (vrnd<a,m,n,p>_f32): Implement using builtins.
14671 (vrnd<i,x><q>_f<32, 64>): New.
14673 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
14675 * config/aarch64/aarch64-builtins.c
14676 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
14677 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
14678 (frintz): Rename to...
14680 (frintp): Rename to...
14682 (frintm): Rename to...
14684 (frinti): Rename to...
14685 (nearbyint): ...this.
14686 (frintx): Rename to...
14688 (frinta): Rename to...
14690 * config/aarch64/aarch64-simd.md
14691 (aarch64_frint<frint_suffix><mode>): Delete.
14692 (<frint_pattern><mode>2): Convert to insn.
14693 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
14694 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
14695 (frint_pattern): Likewise.
14696 (frint_suffix): Likewise.
14698 2013-04-29 Richard Biener <rguenther@suse.de>
14700 PR tree-optimization/57081
14701 * loop-init.c: Include tree-flow.h.
14702 (loop_optimizer_finalize): Free number of iteration estimates.
14703 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
14705 2013-04-29 Jakub Jelinek <jakub@redhat.com>
14707 PR tree-optimization/57083
14708 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
14709 non-singleton shift count range, zero extend low_bound for uns case.
14711 * config/i386/predicates.md (general_vector_operand): New predicate.
14712 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
14713 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
14714 if they aren't nonimmediate operands. If their original values
14715 satisfy const_vector_equal_evenodd_p, don't shift them.
14716 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
14717 predicates. For the SSE4.1 case force operands[{1,2}] into registers
14718 if not nonimmediate_operand.
14719 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
14720 instead of register_operand.
14721 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
14723 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
14725 * stor-layout.c (finalize_size_functions): Allocate a structure and
14726 reset cfun before dumping the functions.
14728 2013-04-27 Jakub Jelinek <jakub@redhat.com>
14730 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
14733 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
14734 use xop_pmacsdqh if uns_p.
14735 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
14736 the immediate rotate count.
14738 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
14740 * rtl.h (struct rtx_def): Add comment for field jump.
14741 (LRA_SUBREG_P): New macro.
14742 * recog.c (register_operand): Check LRA_SUBREG_P.
14743 * lra.c (lra): Add note at the end of RTL code. Align non-empty
14745 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14746 (lra_final_code_change): Skip subreg change for operators.
14747 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14748 if there are no operand changes.
14749 * lra-constraints.c (curr_insn_set): New.
14750 (match_reload): Set LRA_SUBREG_P.
14751 (emit_spill_move): Ditto.
14752 (check_and_process_move): Use curr_insn_set. Process only single
14753 set insns. Don't initialize sec_mem_p and change_p.
14754 (simplify_operand_subreg): Use LRA_SUBREG_P.
14755 (reg_in_class_p): New function.
14756 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
14757 of #ifdef. Add code to remove cycling.
14758 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14759 non-null disp. Reload inner instead of disp when base and index
14760 are null. Try to put lo_sum into register.
14761 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14762 (check_and_process_move): Move code for move cost check to
14763 simple_move_p. Remove equiv_substitution.
14764 (simple_move_p): New function.
14765 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
14766 curr_insn_set. Call check_and_process_move only for single set
14767 insns. Use the new function. Move call of check_and_process_move
14768 after operand equiv substitution and address process.
14770 2013-04-26 Jakub Jelinek <jakub@redhat.com>
14773 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
14774 with nonlocal goto receivers or returns twice calls, ignore
14775 unininitialized values from abnormal edges to nl goto receiver
14776 or returns twice call.
14778 2013-04-26 Jakub Jelinek <jakub@redhat.com>
14780 PR tree-optimization/57051
14781 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
14782 and VEC_RSHIFT_EXPR if shift count is a multiple of element
14785 2013-04-26 Richard Biener <rguenther@suse.de>
14787 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
14788 (expand_omp_taskreg): Likewise. Mark loops for fixup.
14789 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
14790 (fixup_loop_arrays_after_move): New function.
14791 (move_sese_region_to_fn): Properly outline the loop tree parts
14792 of the SESE region.
14794 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
14796 * config/i386/i386.md (type, unit): Fix long lines.
14798 2013-04-26 Richard Biener <rguenther@suse.de>
14800 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
14801 (lto-streamer-out.o): Likewise.
14802 * cfgloop.c (init_loops_structure): Export, add struct function
14803 argument and adjust.
14804 (flow_loops_find): Adjust.
14805 * cfgloop.h (enum loop_estimation): Add EST_LAST.
14806 (init_loops_structure): Declare.
14807 * lto-streamer-in.c: Include cfgloop.h.
14808 (input_cfg): Input the loop tree.
14809 * lto-streamer-out.c: Include cfgloop.h.
14810 (output_cfg): Output the loop tree.
14811 (output_struct_function_base): Do not drop PROP_loops.
14813 2013-03-26 Richard Biener <rguenther@suse.de>
14815 * tree-cfg.c (execute_build_cfg): Build the loop tree.
14816 (pass_build_cfg): Provide PROP_loops.
14817 (move_sese_region_to_fn): Remove loops that are outlined into fn
14819 * tree-inline.c: Include cfgloop.h.
14820 (initialize_cfun): Do not drop PROP_loops.
14821 (copy_loops): New function.
14822 (copy_cfg_body): Copy loop structure.
14823 (tree_function_versioning): Initialize destination loop tree.
14824 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
14825 (pass_parallelize_loops): Do IL verification.
14826 * loop-init.c (loop_optimizer_init): Fixup loops if required.
14827 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
14828 the CFG make sure we fixup loops as well.
14829 * tree-ssa-tail-merge.c: Include cfgloop.h.
14830 (replace_block_by): When merging loop latches mark loops for fixup.
14831 * lto-streamer-out.c (output_struct_function_base): Drop
14832 PROP_loops for now.
14833 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
14834 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
14835 * ipa-split.c: Include cfgloop.h.
14836 (split_function): Add the new return block to the loop tree root.
14837 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
14838 whether we have removed the forwarder block.
14839 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
14840 * cfgloop.h (place_new_loop): Declare.
14841 * cfgloopmanip.c (place_new_loop): Export.
14842 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
14843 (tree-switch-conversion.o): Likewise.
14844 (tree-complex.o): Likewise.
14845 (tree-inline.o): Likewise.
14846 (tree-ssa-tailmerge.o): Likewise.
14847 (ipa-split.o): Likewise.
14848 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
14849 (tree-ssa-copy.o): Likewise.
14850 * tree-switch-conversion.c: Include cfgloop.h
14851 (process_switch): If we emit a bit-test cascade, schedule loops
14853 * tree-complex.c: Include cfgloop.h.
14854 (expand_complex_div_wide): Properly add new basic-blocks to loops.
14855 * asan.c: Include cfgloop.h.
14856 (create_cond_insert_point): Properly add new basic-blocks to
14857 loops, schedule loop fixup.
14858 * cfgloop.c (verify_loop_structure): Check that looks are not
14860 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
14862 (expand_omp_for_generic): Likewise.
14863 (expand_omp_sections): Likewise.
14864 (expand_omp_atomic_pipeline): Schedule loops for fixup.
14865 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
14866 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
14867 is initialized, not when loops are present.
14868 * tree-parloops.c (parallelize_loops): Remove checking here.
14869 * passes.c (init_optimization_passes): Schedule a copy-propagation
14870 pass before complete unrolling of inner loops.
14872 2013-04-26 Jakub Jelinek <jakub@redhat.com>
14874 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
14875 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
14876 (colorize_init): Add argument to _WIN32 version.
14877 * toplev.c: Include diagnostic-color.h.
14878 (process_options): Default to -fdiagnostics-color=auto if
14879 GCC_COLORS env var is in the environment.
14880 * common.opt (fdiagnostics-color=): Add Var and Init.
14881 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
14882 env var is in the environment, the default is auto rather than never.
14884 * diagnostic.h (file_name_as_prefix): Add context argument.
14885 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
14886 the string as locus.
14887 * langhooks.c (lhd_print_error_function): Adjust caller.
14889 2013-04-25 Lawrence Crowl <crowl@google.com>
14891 * var-tracking.c (shared_hash_def::htab):
14892 Change type to hash_table. Update dependent calls and types.
14894 2013-04-25 Lawrence Crowl <crowl@google.com>
14896 * Makefile.in: Update as needed below.
14898 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
14899 Move declaration to after the type's method definitons.
14901 * attribs.c (htab_t scoped_attributes::attribute_hash):
14902 Change type to hash_table. Update dependent calls and types.
14904 * bitmap.c (htab_t bitmap_desc_hash):
14905 Change type to hash_table. Update dependent calls and types.
14907 * cselib.c (htab_t cselib_hash_table):
14908 Change type to hash_table. Update dependent calls and types.
14910 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
14911 (hash_string_slot_node): Move implementation into lto-streamer.h
14912 struct string_slot_hasher.
14913 (eq_string_slot_node): Likewise.
14915 * data-streamer-out.c: Update output_block::string_hash_table
14916 dependent calls and types.
14918 * dwarf2cfi.c (htab_t trace_index):
14919 Change type to hash_table. Update dependent calls and types.
14921 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
14922 Change type to hash_table. Update dependent calls and types.
14923 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
14924 (htab_t optimize_external_refs::map): Likewise.
14925 (htab_t output_comp_unit::extern_map): Likewise.
14926 (htab_t output_comdat_type_unit::extern_map): Likewise.
14927 (htab_t output_macinfo::macinfo_htab): Likewise.
14928 (htab_t optimize_location_lists::htab): Likewise.
14929 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
14931 * except.c (htab_t ehspec_hash_type):
14932 Change type to hash_table. Update dependent calls and types.
14933 (assign_filter_values::ttypes): Likewise.
14934 (assign_filter_values::ehspec): Likewise.
14935 (sjlj_assign_call_site_values::ar_hash): Likewise.
14936 (convert_to_eh_region_ranges::ar_hash): Likewise.
14938 * gcse.c (htab_t pre_ldst_table):
14939 Change type to hash_table. Update dependent calls and types.
14941 * ggc-common.c (htab_t saving_htab):
14942 Change type to hash_table. Update dependent calls and types.
14943 (htab_t loc_hash): Likewise.
14944 (htab_t ptr_hash): Likewise.
14945 (call_count): Rename ggc_call_count.
14946 (call_alloc): Rename ggc_call_alloc.
14947 (loc_descriptor): Rename make_loc_descriptor.
14948 (add_statistics): Rename ggc_add_statistics.
14950 * ggc-common.c (saving_htab):
14951 Change type to hash_table. Update dependent calls and types.
14953 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
14954 (push_gimplify_context): Likewise.
14955 (pop_gimplify_context): Likewise.
14956 (struct gimple_temp_hash_elt): Added.
14957 (struct gimplify_hasher): Likewise.
14958 (struct gimplify_ctx.temp_htab):
14959 Change type to hash_table. Update dependent calls and types.
14961 * gimple-fold.c: Include gimplify-ctx.h.
14963 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
14964 Change type to hash_table. Update dependent calls and types.
14965 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
14966 avoid potential global name collision.
14968 * gimplify.c: Include gimplify-ctx.h.
14969 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
14970 (htab_t gimplify_ctx::temp_htab):
14971 Update dependent calls and types for new type hash_table.
14972 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
14973 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
14975 * gimplify-ctx.h: New.
14976 (struct gimple_temp_hash_elt): Move from gimplify.c.
14977 (class gimplify_hasher): New.
14978 (struct gimplify_ctx): Move from gimple.h.
14979 (htab_t gimplify_ctx::temp_htab):
14980 Change type to hash_table. Update dependent calls and types.
14982 * graphite-clast-to-gimple.c: Include graphite-htab.h.
14983 (htab_t ivs_params::newivs_index):
14984 Change type to hash_table. Update dependent calls and types.
14985 (htab_t ivs_params::params_index): Likewise.
14986 (htab_t print_generated_program::params_index): Likewise.
14987 (htab_t gloog::newivs_index): Likewise.
14988 (htab_t gloog::params_index): Likewise.
14990 * graphite.c: Include graphite-htab.h.
14991 4htab_t graphite_transform_loops::bb_pbb_mapping):
14992 Change type to hash_table. Update dependent calls and types.
14994 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
14995 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
14996 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
14998 * graphite-dependences.c: Include graphite-htab.h.
14999 (loop_is_parallel_p): Change hash table type of parameter.
15001 * graphite-htab.h: New.
15002 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
15003 (extern find_pbb_via_hash): Move from graphite-poly.h.
15004 (extern loop_is_parallel_p): Move from graphite-poly.h.
15005 (extern get_loop_body_pbbs): Move from graphite-poly.h.
15007 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
15008 (extern loop_is_parallel_p): Move to graphite-htab.h.
15009 (extern get_loop_body_pbbs): Move to graphite-htab.h.
15011 * haifa-sched.c (htab_t delay_htab):
15012 Change type to hash_table. Update dependent calls and types.
15013 (htab_t delay_htab_i2): Likewise.
15015 * ira-color.c (htab_t allocno_hard_regs_htab):
15016 Change type to hash_table. Update dependent calls and types.
15018 * ira-costs.c (htab_t cost_classes_htab):
15019 Change type to hash_table. Update dependent calls and types.
15021 * loop-invariant.c (htab_t merge_identical_invariants::eq):
15022 Change type to hash_table. Update dependent calls and types.
15024 * loop-iv.c (htab_t bivs):
15025 Change type to hash_table. Update dependent calls and types.
15027 * loop-unroll.c (htab_t opt_info::insns_to_split):
15028 Change type to hash_table. Update dependent calls and types.
15029 (htab_t opt_info::insns_with_var_to_expand): Likewise.
15031 * lto-streamer.h (struct string_slot): Move from data-streamer.h
15032 (struct string_slot_hasher): New.
15033 (htab_t output_block::string_hash_table):
15034 Change type to hash_table. Update dependent calls and types.
15036 * lto-streamer-in.c (freeing_string_slot_hasher): New.
15037 (htab_t file_name_hash_table):
15038 Change type to hash_table. Update dependent calls and types.
15040 * lto-streamer-out.c: Update output_block::string_hash_table dependent
15043 * lto-streamer.c (htab_t tree_htab):
15044 Change type to hash_table. Update dependent calls and types.
15046 * omp-low.c: Include gimplify-ctx.h.
15048 * passes.c (htab_t name_to_pass_map):
15049 Change type to hash_table. Update dependent calls and types.
15050 (pass_traverse): Rename to passes_pass_traverse.
15052 * plugin.c (htab_t event_tab):
15053 Change type to hash_table. Update dependent calls and types.
15055 * postreload-gcse.c (htab_t expr_table):
15056 Change type to hash_table. Update dependent calls and types.
15057 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
15059 * sese.c (debug_rename_map_1): Make extern.
15060 (htab_t copy_bb_and_scalar_dependences::rename_map):
15061 Change type to hash_table. Update dependent calls and types.
15063 * sese.h (extern debug_rename_map): Move to .c file.
15065 * store-motion.c (htab_t store_motion_mems_table):
15066 Change type to hash_table. Update dependent calls and types.
15068 * trans-mem.c (htab_t tm_new_mem_hash):
15069 Change type to hash_table. Update dependent calls and types.
15071 * tree-browser.c (htab_t TB_up_ht):
15072 Change type to hash_table. Update dependent calls and types.
15074 * tree-cfg.c (htab_t discriminator_per_locus):
15075 Change type to hash_table. Update dependent calls and types.
15077 * tree-complex.c: Include tree-hasher.h
15078 (htab_t complex_variable_components):
15079 Change type to hash_table. Update dependent calls and types.
15081 * tree-eh.c (htab_t finally_tree):
15082 Change type to hash_table. Update dependent calls and types.
15084 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
15085 struct int_tree_hasher.
15086 (extern int_tree_map_eq): Likewise.
15087 (uid_decl_map_hash): Removed.
15088 (extern decl_tree_map_eq): Likewise.
15090 * tree-hasher.h: New.
15091 (struct int_tree_hasher): New.
15092 (typedef int_tree_htab_type): New.
15094 * tree-inline.c: Include gimplify-ctx.h.
15096 * tree-mudflap.c: Include gimplify-ctx.h.
15098 * tree-parloops.c: Include tree-hasher.h.
15099 (htab_t eliminate_local_variables_stmt::decl_address):
15100 Change type to hash_table. Update dependent calls and types.
15101 (htab_t separate_decls_in_region::decl_copies): Likewise.
15103 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
15104 Change type to hash_table. Update dependent calls and types.
15106 * tree-sra.c (candidates):
15107 Change type to hash_table. Update dependent calls and types.
15109 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
15111 (int_tree_map_hash): Likewise.
15113 * tree-ssa-dom.c (htab_t avail_exprs):
15114 Change type to hash_table. Update dependent calls and types.
15116 * tree-ssa-live.c (var_map_base_init::tree_to_index):
15117 Change type to hash_table. Update dependent calls and types.
15119 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
15120 Change type to hash_table. Update dependent calls and types.
15122 * tree-ssa-phiopt.c (seen_ssa_names):
15123 Change type to hash_table. Update dependent calls and types.
15125 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
15126 Change type to hash_table. Update dependent calls and types.
15128 * tree-ssa-uncprop.c (equiv):
15129 Change type to hash_table. Update dependent calls and types.
15131 2013-04-25 Jakub Jelinek <jakub@redhat.com>
15133 PR rtl-optimization/57003
15134 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
15135 call note_stores with kill_clobbered_value callback again after
15136 killing regs_invalidated_by_call.
15138 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
15140 * config/aarch64/aarch64-simd.md
15141 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
15142 (aarch64_simd_bsl<mode>): Likewise.
15143 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
15145 2013-04-25 Marek Polacek <polacek@redhat.com>
15147 PR tree-optimization/57066
15148 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
15150 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
15152 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
15154 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
15156 * config/aarch64/aarch64-builtins.c
15157 (aarch64_fold_builtin): New.
15158 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
15159 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
15160 * config/aarch64/aarch64-simd-builtins.def (abs): New.
15161 * config/aarch64/arm_neon.h
15162 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
15164 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
15165 Tejas Belagod <tejas.belagod@arm.com>
15167 * config/aarch64/aarch64-builtins.c
15168 (aarch64_gimple_fold_builtin): New.
15169 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
15170 * config/aarch64/aarch64-simd-builtins.def (addv): New.
15171 * config/aarch64/aarch64-simd.md (addpv4sf): New.
15172 (addvv4sf): Update.
15173 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
15175 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15177 * config/aarch64/aarch64.md
15178 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
15180 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15182 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
15183 (*ngcsi_uxtw): New pattern.
15185 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15186 Julian Brown <julian@codesourcery.com>
15188 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
15189 (TB_DREG): Add T_V4HF.
15190 (v4hf_UP): New macro.
15191 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15192 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15193 Handle initialisation of V4HF. Adjust initialisation of reinterpret
15195 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15196 (arm_vector_mode_supported_p): Handle V4HF.
15197 (arm_mangle_map): Handle V4HFmode.
15198 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
15199 * config/arm/arm_neon_builtins.def: Add entries for
15200 vcvtv4hfv4sf, vcvtv4sfv4hf.
15201 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
15202 (neon_vcvtv4hfv4sf): Likewise.
15203 * config/arm/neon-gen.ml: Handle half-precision floating point
15205 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
15206 * config/arm/arm_neon.h: Regenerate.
15207 * config/arm/neon.ml (type elts): Add F16.
15208 (type vectype): Add T_float16x4, T_floatHF.
15209 (type vecmode): Add V4HF.
15210 (type features): Add Requires_FP_bit feature.
15211 (elt_width): Handle F16.
15212 (elt_class): Likewise.
15213 (elt_of_class_width): Likewise.
15214 (mode_of_elt): Refactor.
15215 (type_for_elt): Handle F16, fix error messages.
15216 (vectype_size): Handle T_float16x4.
15217 (vcvt_sh): New function.
15218 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
15219 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
15220 (string_of_mode): Handle V4HF.
15221 * doc/arm-neon-intrinsics.texi: Regenerate.
15223 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
15225 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
15226 format specifier in 'X' case.
15228 2013-04-25 Alan Modra <amodra@gmail.com>
15231 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
15232 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
15233 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
15234 Repeat for many other rotate/shift and mask patterns using subregs.
15235 Name lshiftrt insns.
15236 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
15237 on WORDS_BIG_ENDIAN.
15239 2013-04-25 Alan Modra <amodra@gmail.com>
15241 * config.gcc: Support little-endian powerpc-linux targets.
15242 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
15243 (LINK_OS_LINUX_SPEC): Define.
15244 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
15245 Preserve MASK_LITTLE_ENDIAN.
15246 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
15247 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
15248 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
15249 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
15250 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
15251 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
15252 Correct fp word order for little-endian. Don't shift toc entries
15253 smaller than a word for little-endian.
15254 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
15255 (bswapdi2 splits): Correct low-part subreg for little-endian.
15256 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
15257 low/high where such is correct only for be.
15258 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
15259 little-endian for -mcall-aixdesc.
15261 2013-04-25 Alan Modra <amodra@gmail.com>
15263 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
15264 replace_equiv_address_nv.
15266 2013-04-25 Alan Modra <amodra@gmail.com>
15268 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
15270 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
15273 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
15274 * rtl.h (struct rtx_def): ...
15276 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
15278 PR rtl-optimizations/57046
15279 * lra-constraints (split_reg): Set up lra_risky_transformations_p
15280 for multi-reg splits.
15282 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
15284 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
15286 2013-04-24 Sterling Augustine <saugustine@google.com>
15288 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
15289 (comp_dir_string, debug_str_dwo_section): New.
15290 (DEBUG_STR_DWO_SECTION): Rename to ...
15291 (DEBUG_DWO_STR_SECTION): ... this.
15292 (DEBUG_NORM_STR_SECTION): Delete.
15293 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
15294 (DEBUG_STR_DWO_SECTION_FLAGS): New.
15295 (find_AT_string): Move most logic to ...
15296 (find_AT_string_in_table): ... here. New.
15297 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
15298 add_skeleton_AT_string. Delete logic.
15299 (output_skeleton_debug_sections): Remove call to
15300 add_top_level_skeleton_die_attrs.
15301 (add_comp_dir_attribute): Move logic to comp_dir_string.
15302 (dwarf2out_init): Initialize debug_str_dwo_section.
15303 (output_indirect_string): Call find_string_form.
15304 (output_indirect_strings): Rewrite.
15305 (prune_unused_types): Empty skeleton_debug_str_hash.
15306 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
15307 (dwarf2out_finish): Call output_indirect_strings.
15309 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
15311 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
15313 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
15315 * rtl.h (struct rtx_def): Add comment for field jump.
15316 (LRA_SUBREG_P): New macro.
15317 * recog.c (register_operand): Check LRA_SUBREG_P.
15318 * lra.c (lra): Add note at the end of RTL code. Align non-empty
15320 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
15321 (lra_final_code_change): Skip subreg change for operators.
15322 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
15323 if there are no operand changes.
15324 * lra-constraints.c (curr_insn_set): New.
15325 (match_reload): Set LRA_SUBREG_P.
15326 (emit_spill_move): Ditto.
15327 (check_and_process_move): Use curr_insn_set. Process only single
15328 set insns. Don't initialize sec_mem_p and change_p.
15329 (simplify_operand_subreg): Use LRA_SUBREG_P.
15330 (reg_in_class_p): New function.
15331 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
15332 of #ifdef. Add code to remove cycling.
15333 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
15334 non-null disp. Reload inner instead of disp when base and index
15335 are null. Try to put lo_sum into register.
15336 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
15337 (check_and_process_move): Move code for move cost check to
15338 simple_move_p. Remove equiv_substitution.
15339 (simple_move_p): New function.
15340 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
15341 curr_insn_set. Call check_and_process_move only for single set
15342 insns. Use the new function. Move call of check_and_process_move
15343 after operand equiv substitution and address process.
15345 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
15347 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
15348 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
15349 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
15351 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
15353 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
15355 2013-04-24 Marek Polacek <polacek@redhat.com>
15357 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
15358 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
15359 (select_loops_exit_conditions): Likewise.
15360 (number_of_iterations_for_all_loops): Likewise.
15361 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
15362 (scev_analysis): Likewise.
15364 2013-04-02 Catherine Moore <clm@codesourcery.com>
15365 Chao-ying Fu <fu@mips.com>
15367 * config/mips/micromips.md (jraddiusp): New pattern.
15368 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
15369 instruction if possible.
15371 2013-04-24 Alan Modra <amodra@gmail.com>
15373 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
15375 2013-04-24 Julian Brown <julian@codesourcery.com>
15376 Chung-Lin Tang <cltang@codesourcery.com>
15378 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
15379 dependency behavior in enumeration type DIE generation. Add TODO note
15380 to comments about future DW_FORM_sdata/udata re-work of related code.
15382 2013-04-23 Lawrence Crowl <crowl@google.com>
15384 * Makefile.in: Update as needed below.
15386 * hash-table.h (class hash_table):
15387 Correct many methods with parameter types compare_type to the correct
15388 value_type. (Correct code was unlikely to notice the change.)
15389 (hash_table::elements_with_deleted) New.
15390 (class hashtable::iterator): New.
15391 (hashtable::begin()): New.
15392 (hashtable::end()): New.
15393 (FOR_EACH_HASH_TABLE_ELEMENT): New.
15395 * statistics.c (statistics_hashes):
15396 Change type to hash_table. Update dependent calls and types.
15398 * tree-into-ssa.c (var_infos):
15399 Change type to hash_table. Update dependent calls and types.
15401 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
15402 Change type to hash_table. Update dependent calls and types.
15404 * tree-ssa-loop-im.c (struct mem_ref.refs):
15405 Change type to hash_table. Update dependent calls and types.
15407 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
15408 Change type to hash_table. Update dependent calls and types.
15410 * tree-ssa-sccvn.c (vn_tables_s::nary):
15411 Change type to hash_table. Update dependent calls and types.
15412 (vn_tables_s::phis): Likewise.
15413 (vn_tables_s::references): Likewise.
15415 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
15416 (vn_reference_eq): Update parameter and return types.
15418 * tree-ssa-structalias.c (pointer_equiv_class_table):
15419 Change type to hash_table. Update dependent calls and types.
15420 (location_equiv_class_table): Likewise.
15422 * tree-vect-data-refs.c: Consequential changes for making
15423 peeling a hash_table.
15425 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
15426 (destroy_loop_vec_info): Dependent hash_table update.
15428 * tree-vectorizer.h (peeling_htab):
15429 Change type to hash_table. Update dependent calls and types.
15431 2013-04-23 Shiva Chen <shiva0217@gmail.com>
15433 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
15434 to check the register content is equal or not.
15435 * lra-constraints.c (match_reload): Use lra_assign_reg_val
15436 to assign register content record.
15437 * lra-eliminations.c (update_reg_eliminate): Use
15438 lra_update_reg_val_offset to update register content offset.
15439 * lra-int.h (struct lra_reg): Add offset member.
15440 (lra_reg_val_equal_p): New static inline function.
15441 (lra_update_reg_val_offset): New static inline function.
15442 (lra_assign_reg_val): New static inline function.
15443 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
15444 to assign register content record.
15445 (initialize_lra_reg_info_element): Initial offset to zero.
15447 2013-04-23 Catherine Moore <clm@codesourcery.com>
15449 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
15450 operands. Record compression.
15452 2013-04-23 Xinliang David Li <davidxl@google.com>
15454 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
15456 2013-04-23 Richard Biener <rguenther@suse.de>
15458 PR middle-end/57036
15459 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
15460 parameter, only add abnormal goto edges from the copied body
15461 if the call could perform abnormal gotos.
15462 (copy_cfg_body): Adjust.
15464 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
15466 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
15468 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
15470 * coretypes.h (gimple_stmt_iterator): Add struct to make
15473 2013-04-23 Richard Biener <rguenther@suse.de>
15475 PR tree-optimization/57026
15476 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
15477 from SSA names occuring in abnormal PHI nodes.
15479 2013-04-22 Andi Kleen <ak@linux.intel.com>
15481 * lto/lto.c (print_lto_report_1): Fix LTO report names.
15483 2013-04-22 Andi Kleen <ak@linux.intel.com>
15485 * lto/lto.c (print_lto_report_1): Declare early.
15486 (read_cgraph_and_symbols): Call print_lto_report_1 early.
15488 2013-04-22 Andi Kleen <ak@linux.intel.com>
15490 * common.opt (-flto-report-wpa): Add.
15491 * doc/invoke.texi (-flto-report-wpa): Add.
15492 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
15495 2013-04-22 Xinliang David Li <davidxl@google.com>
15497 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
15498 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
15499 * Makefile.in: New dependency
15501 David Daney <ddaney.cavm@gmail.com>
15503 * configure.ac (gcc_cv_as_micromips_support): Use the
15504 --fatal-warnings option.
15505 * configure: Regenerate.
15507 2013-04-22 Marek Polacek <polacek@redhat.com>
15510 * tsan.c (instrument_expr): Don't instrument expression
15511 in case its size is zero.
15513 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
15517 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15519 * config/alpha/alpha.c (TARGET_LRA_P): New define.
15521 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
15523 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
15524 (gimple_stmt_iterator): New typedef.
15525 * gimple.h (gimple_stmt_iterator): Rename to...
15526 (gimple_stmt_iterator_d): ... This.
15527 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
15528 trees be valid for GIMPLE and GENERIC.
15529 (TARGET_GIMPLE_FOLD_BUILTIN): New.
15530 * gimple-fold.c (gimple_fold_call): Call target hook
15531 gimple_fold_builtin.
15532 * hooks.c (hook_bool_gsiptr_false): New.
15533 * hooks.h (hook_bool_gsiptr_false): New.
15534 * target.def (fold_stmt): New.
15535 * doc/tm.texi: Regenerate.
15537 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
15540 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
15541 a set sp if no stack realignment.
15543 2013-04-22 Nick Clifton <nickc@redhat.com>
15545 * config.gcc (tilegx-linux): Extend extra_objs rather than
15547 (tilepro-linux): Likewise.
15549 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
15551 * config/aarch64/aarch64-builtins.c
15553 (CF0, CF1, CF2, CF3, CF4, CF10): New.
15554 (VAR<1-12>): Add MAP parameter.
15555 (BUILTIN_*): Likewise.
15556 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
15557 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
15558 (aarch64_ushl_n<mode>): Likewise.
15559 (aarch64_sshr_n<mode>): Likewise.
15560 (aarch64_ushr_n<mode>): Likewise.
15561 (aarch64_<maxmin><mode>): Likewise.
15562 (aarch64_sqrt<mode>): Likewise.
15563 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
15564 (vshr<q>_n_*): Likewise.
15566 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
15568 * config/aarch64/aarch64-builtins.c
15569 (aarch64_simd_builtin_type_mode): Handle SF types.
15571 (BUILTIN_GPF): Define.
15572 (aarch64_init_simd_builtins): Handle SF types.
15573 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
15574 (frecps): Likewise.
15575 (frecpx): Likewise.
15576 * config/aarch64/aarch64-simd.md
15577 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
15578 (aarch64_frecpe<mode>): New.
15579 (aarch64_frecps<mode>): Likewise.
15580 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
15581 (v8type): Add frecp<esx>.
15582 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
15583 (aarch64_frecps<mode>): Likewise.
15584 * config/aarch64/iterators.md (FRECP): New.
15585 (frecp_suffix): Likewise.
15586 * config/aarch64/arm_neon.h
15587 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
15589 2013-04-22 Christian Bruel <christian.bruel@st.com>
15592 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
15593 (REG_CLASS_NAMES): Idem.
15594 (REG_CLASS_CONTENTS): Idem.
15595 (REGCLASS_HAS_FP_REG): Idem.
15596 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
15597 (sh_conditional_register_usage): Idem.
15599 2013-04-21 Jeff Law <law@redhat.com>
15601 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
15602 (ssa_forward_propagate_and_combine): Use it.
15604 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
15606 * lra.c: Update the flow chart diagram.
15608 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
15610 PR rtl-optimization/56847
15611 * lra-constraints.c (process_alt_operands): Discourage alternative
15612 with non-matche doffsettable memory constraint fro memory with
15615 2013-04-19 Richard Biener <rguenther@suse.de>
15617 PR tree-optimization/56982
15618 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
15620 * gimplify.c (gimplify_call_expr): Notice special calls.
15621 (gimplify_modify_expr): Likewise.
15622 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
15623 abnormal control flow receivers.
15624 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
15625 in the same way as cfun->has_nonlocal_labels.
15626 (gimple_purge_dead_abnormal_call_edges): Likewise.
15627 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
15628 receivers start a basic-block.
15630 2013-04-19 Richard Biener <rguenther@suse.de>
15632 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
15634 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
15635 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
15636 (SLP_TREE_LOAD_PERMUTATION): Add.
15637 (vect_transform_slp_perm_load): Adjust prototype.
15638 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
15639 (vect_free_slp_instance): Likewise.
15640 (vect_create_new_slp_node): Likewise.
15641 (vect_supported_slp_permutation_p): Remove.
15642 (vect_slp_rearrange_stmts): Adjust.
15643 (vect_supported_load_permutation_p): Likewise. Inline
15644 vect_supported_slp_permutation_p here.
15645 (vect_analyze_slp_instance): Compute load permutations per
15646 slp node instead of per instance.
15647 (vect_get_slp_defs): Adjust.
15648 (vect_transform_slp_perm_load): Likewise.
15649 (vect_schedule_slp_instance): Remove redundant code.
15650 (vect_schedule_slp): Remove hack for PR56270, add it ...
15651 * tree-vect-stmts.c (vectorizable_load): ... here, do not
15652 CSE loads for SLP. Adjust.
15654 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
15656 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
15657 spelling in two comments.
15659 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
15662 * config/arm/arm.c (load_multiple_sequence): Require SP
15663 as base register for loads if SP is in the register list.
15665 2013-04-19 Martin Jambor <mjambor@suse.cz>
15667 PR tree-optimization/56718
15668 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
15669 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
15670 and made public. Adjusted all callers.
15671 (ipa_intraprocedural_devirtualization): New function.
15672 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
15673 (ipa_intraprocedural_devirtualization): Likewise.
15674 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
15676 2013-04-19 Richard Biener <rguenther@suse.de>
15678 PR tree-optimization/57000
15679 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
15681 2013-04-19 Terry Guo <terry.guo@arm.com>
15683 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
15685 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
15686 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
15687 (cortex_m4_fmacs): Use new reservations.
15688 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
15690 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
15692 PR rtl-optimization/56999
15693 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
15695 (lra_coalesce): Remove split_origin_bitmap and related code.
15696 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
15697 ranges if necessary.
15699 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
15701 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
15703 (ix86_expand_call): Remove clobbered_registers array and use
15704 x86_64_ms_sysv_extra_clobbered_registers instead.
15705 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
15707 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
15709 * config/i386/i386.md (*call_rex64_ms_sysv): Use
15710 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
15711 (*call_value_rex64_ms_sysv): Ditto.
15713 2013-04-18 Cary Coutant <ccoutant@google.com>
15715 * dwarf2out.c (output_pubnames): Check die_perennial_p of
15716 parent instead of die_mark.
15718 2013-04-18 Diego Novillo <dnovillo@google.com>
15720 * gimple.c (create_gimple_tmp): New.
15721 (get_expr_type): New.
15722 (build_assign): New.
15723 (build_type_cast): New.
15724 * gimple.h (enum ssa_mode): Define.
15725 (gimple_seq_set_location): New.
15726 * asan.c (build_check_stmt): Change some gimple_build_* calls
15727 to use build_assign and build_type_cast.
15729 2013-04-18 Richard Biener <rguenther@suse.de>
15731 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
15732 handle negative step. Remove redundant checks.
15733 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
15734 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
15735 for negative step and grouped loads fail to vectorize.
15737 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
15739 * emit-rtl.c (reset_insn_used_flags): New function.
15740 (reset_all_used_flags): Use it.
15741 (verify_insn_sharing): New function.
15742 (verify_rtl_sharing): Fix verification for SEQUENCEs.
15744 2013-04-18 Jakub Jelinek <jakub@redhat.com>
15746 PR tree-optimization/56984
15747 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
15748 and (x >> M) >= N don't register any assertion if N << M is the
15751 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
15753 * lower-subreg.c (resolve_simple_move): If called self-recursive,
15754 do not delete_insn insns that have not yet been emitted, only
15755 unlink them with remove_insn.
15756 * df-scan.c (df_insn_delete): Revert r197492.
15758 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
15760 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
15761 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
15763 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
15765 * config/arm/arm.md (movsicc_insn): Convert define_insn into
15766 define_insn_and_split.
15767 (and_scc,ior_scc,negscc): Likewise.
15768 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
15770 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
15772 * config/arm/arm.c (use_return_insn): Return 0 for targets that
15773 can benefit from using a sequence of LDRD instructions in epilogue
15774 instead of a single LDM instruction.
15776 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
15779 * doc/extend.texi: Fix typo.
15781 2013-04-17 Richard Biener <rguenther@suse.de>
15783 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
15784 (vect_build_slp_tree): ... here.
15785 (vect_build_slp_tree_1): Compute which stmts of the SLP group
15786 match. Remove special-casing of mismatched complex loads.
15787 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
15788 re-try the match with swapped commutative operands.
15789 (vect_supported_load_permutation_p): Remove special-casing of
15790 mismatched complex loads.
15791 (vect_analyze_slp_instance): Adjust.
15793 2013-04-17 Richard Biener <rguenther@suse.de>
15795 PR rtl-optimization/56921
15796 * cfgloop.h (struct loop): Add simple_loop_desc member.
15797 (struct niter_desc): Mark with GTY(()).
15798 (simple_loop_desc): Do not use aux field but simple_loop_desc.
15799 * loop-iv.c (get_simple_loop_desc): Likewise.
15800 (free_simple_loop_desc): Likewise.
15803 2013-04-16 Richard Biener <rguenther@suse.de>
15805 PR rtl-optimization/56921
15806 * loop-init.c (pass_rtl_move_loop_invariants): Add
15807 TODO_do_not_ggc_collect to todo_flags_finish.
15808 (pass_rtl_unswitch): Same.
15809 (pass_rtl_unroll_and_peel_loops): Same.
15810 (pass_rtl_doloop): Same.
15812 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
15814 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
15815 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
15816 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
15818 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
15819 * tree-streamer.c (record_common_node): Adjust reference in comment.
15821 2013-04-17 Terry Guo <terry.guo@arm.com>
15823 * config/arm/cortex-m4.md: Add a new bypass.
15825 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15827 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
15829 (*subs_<optab><mode>_multp2): New pattern.
15830 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15831 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15833 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15835 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
15836 (*subs_mul_imm_<mode>): New pattern.
15838 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
15841 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
15842 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
15843 (vsx_movti_32bit): Change j->wa to O->wa.
15845 2013-04-16 Richard Biener <rguenther@suse.de>
15847 PR rtl-optimization/56921
15848 * loop-init.c (pass_rtl_move_loop_invariants): Add
15849 TODO_do_not_ggc_collect to todo_flags_finish.
15850 (pass_rtl_unswitch): Same.
15851 (pass_rtl_unroll_and_peel_loops): Same.
15852 (pass_rtl_doloop): Same.
15854 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
15856 * config/arm/arm.c (emit_multi_reg_push): New declaration
15857 for an existing function.
15858 (arm_emit_strd_push): New function.
15859 (arm_expand_prologue): Used here.
15860 (arm_emit_ldrd_pop): New function.
15861 (arm_expand_epilogue): Used here.
15862 (arm_get_frame_offsets): Update condition.
15863 (arm_emit_multi_reg_pop): Add a special case for load of a single
15864 register with writeback.
15866 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
15868 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
15871 2013-04-16 Richard Biener <rguenther@suse.de>
15873 PR tree-optimization/56756
15874 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
15875 (first_mem_ref_loc): New.
15876 (execute_sm): Place the load temporarily before a previous
15877 access instead of in the latch edge to ensure its SSA dependencies
15878 are defined at points dominating the load.
15880 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
15882 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
15883 correct fix by moving header and footer insn to the footer of
15884 the merged basic block. Clear BB_END of the merged-away block.
15886 PR middle-end/43631
15887 * emit-rtl.c (make_note_raw): New function.
15888 (link_insn_into_chain): New static inline function.
15889 (add_insn): Use it.
15890 (add_insn_before, add_insn_after): Factor insn chain linking code...
15891 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
15892 using link_insn_into_chain.
15893 (note_outside_basic_block_p): New helper function for emit_note_after
15894 and emit_note_before.
15895 (emit_note_after): Use nobb variant of add_insn_after if the note
15896 should not be contained in a basic block.
15897 (emit_note_before): Use nobb variant of add_insn_before if the note
15898 should not be contained in a basic block.
15899 (emit_note_copy): Use make_note_raw.
15900 (emit_note): Likewise.
15901 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
15902 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
15903 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
15904 the moved barrier the tail of the basic block it follows.
15905 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
15907 2013-04-15 Jakub Jelinek <jakub@redhat.com>
15909 PR tree-optimization/56962
15910 * gimple-ssa-strength-reduction.c (record_increment): Only set
15911 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
15912 either rhs1 or rhs2 is equal to c->base_expr.
15914 2013-04-15 Richard Biener <rguenther@suse.de>
15916 PR tree-optimization/56933
15917 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
15919 (GROUP_READ_WRITE_DEPENDENCE): Remove.
15920 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
15921 * tree-vect-data-refs.c (vect_analyze_group_access): Move
15922 dependence check ...
15923 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
15925 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
15926 GROUP_READ_WRITE_DEPENDENCE.
15928 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15930 * emit-rtl.c (reset_all_used_flags): New function.
15931 (verify_rtl_sharing): Call reset_all_used_flags before and after
15932 performing the checks.
15934 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15936 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
15937 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
15938 * config/arm/constraints.md (De): New constraint.
15939 * config/arm/neon.md (anddi3_neon): Delete.
15940 (neon_vand<mode>): Expand to standard anddi3 pattern.
15941 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
15942 Move earlier in the file.
15943 (neon_inv_logic_op2): Likewise.
15944 (arm_anddi_operand_neon): New predicate.
15946 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15948 * configure.ac (gcc_cv_ld_as_needed): Set
15949 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
15950 Use -z ignore, -z record on *-*-solaris2*.
15951 (HAVE_LD_AS_NEEDED): Update comment.
15952 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
15953 * configure: Regenerate.
15954 * config.in: Regenerate.
15955 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
15956 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
15957 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
15958 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
15959 equivalents. Fix markup.
15960 * doc/tm.texi: Regenerate.
15962 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
15964 * config/i386/i386.opt: New option mstack-protector-guard=.
15965 * config/i386/i386-opts.h: Add enum stack_protector_guard.
15966 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
15967 TARGET_SSP_TLS_GUARD.
15968 * config/i386/i386.c (ix86_option_override_internal): Set
15969 ix86_stack_protector_guard.
15970 * config/i386/i386.md (stack_protect_set): Enable for
15971 TARGET_SSP_TLS_GUARD only.
15972 (stack_protect_set_<mode>): Ditto.
15973 (stack_protect_test): Ditto.
15974 (stack_protect_test_<mode>): Ditto.
15975 * doc/invoke.texi (i386 Option): Document.
15977 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
15980 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
15981 (S_MODES): Set H_MODE bit.
15982 (SF_MODES): Set only S_MODE and SF_MODE bits.
15983 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
15984 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
15985 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
15986 <MODE_FLOAT>: Likewise.
15988 2013-04-15 Joey Ye <joey.ye@arm.com>
15990 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
15992 2013-04-15 Joey Ye <joey.ye@arm.com>
15994 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
15996 (thumb_far_jump_used_p): Count instruction size and set
15999 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
16001 * reorg.c (fill_simple_delay_slots): Reindent block of code.
16002 * resource.c (mark_target_live_regs): Reformat conditional block.
16004 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
16006 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
16007 notes, they are emitted only just before final.
16008 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
16010 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
16012 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
16013 * cfgrtl.c (delete_insn): Call it here instead.
16014 * lra-spills.c (lra_final_code_change): Use delete_insn.
16015 * haifa-sched.c (sched_remove_insn): Likewise.
16016 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
16017 returning to the nop pool.
16018 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
16019 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
16021 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
16023 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
16024 * doc/tm.texi: Regenerated.
16026 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
16028 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
16031 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
16033 * df-core.c (df_find_def): Compare register numbers.
16034 (df_find_use): Likewise.
16036 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
16039 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
16040 lra_in_progress for return.
16042 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
16044 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
16045 define_insn into define_insn_and_split and emit movsicc patterns.
16047 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
16049 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
16051 2013-04-12 Richard Biener <rguenther@suse.de>
16053 * tree-pass.h (TODO_do_not_ggc_collect): New.
16054 * passes.c (execute_one_ipa_transform_pass): Honor
16055 TODO_do_not_ggc_collect.
16056 (execute_one_pass): Likewise.
16059 2013-04-10 Richard Biener <rguenther@suse.de>
16061 * passes.c (init_optimization_passes): Remove reload pass.
16062 * ira.c (do_reload): Merge into ...
16064 (rest_of_handle_reload): Remove.
16065 (pass_reload): Likewise.
16066 * config/i386/i386.c (ix86_option_override): Refer to ira instead
16067 of reload for vzeroupper pass placement.
16069 2013-04-12 Jakub Jelinek <jakub@redhat.com>
16071 PR tree-optimization/56918
16072 PR tree-optimization/56920
16073 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
16074 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
16075 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
16076 use wide_mul_with_sign method.
16078 2013-04-12 Richard Biener <rguenther@suse.de>
16080 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
16081 not be considered a gimple constant.
16083 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
16085 * fold-const.c (const_binop): Handle vector shifts by a scalar.
16086 (fold_binary_loc): Call const_binop also for mixed vector-scalar
16089 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
16090 Jakub Jelinek <jakub@redhat.com>
16092 * opts.c: Include diagnostic-color.h.
16093 (common_handle_option): Handle OPT_fdiagnostics_color_.
16094 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
16095 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
16096 (diagnostic-color.o): New.
16097 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
16098 (diagnostic_color_rule): New enum.
16099 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
16100 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
16101 the location string.
16102 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
16103 either NULL, or color kind.
16104 * diagnostic-color.c: New file.
16105 * diagnostic-color.h: New file.
16106 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
16108 * doc/invoke.texi (-fdiagnostics-color): Document.
16109 * pretty-print.h (pp_show_color): Define.
16110 (struct pretty_print_info): Add show_color field.
16111 * diagnostic.c: Include diagnostic-color.h.
16112 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
16113 macros. Colorize error:, warning: etc. strings and also the location
16115 (diagnostic_show_locus): Colorize the caret line.
16116 * pretty-print.c: Include diagnostic-color.h.
16117 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
16118 inside of %< %> quotes or quoted through q format modifier.
16120 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16122 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
16124 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16126 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
16127 code in CC_NZ mode.
16128 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
16131 2013-04-11 Marek Polacek <polacek@redhat.com>
16133 PR tree-optimization/48184
16134 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
16136 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
16138 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
16139 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
16140 (skip_simple_arithmetic): Tidy up.
16141 * tree.h (skip_simple_constant_arithmetic): Declare.
16143 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16145 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
16147 2013-04-11 Richard Biener <rguenther@suse.de>
16149 * tree-vect-loop.c (get_initial_def_for_induction): Properly
16150 generate vector constants.
16152 2013-04-11 Richard Biener <rguenther@suse.de>
16154 PR tree-optimization/56878
16155 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
16156 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
16158 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16159 Prefer to align the DR with the most invariant base address.
16161 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
16163 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
16166 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
16168 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
16169 floating-point vector comparisons against 0.
16171 2013-04-11 Jakub Jelinek <jakub@redhat.com>
16173 PR tree-optimization/56899
16174 * fold-const.c (extract_muldiv_1): Apply distributive law
16175 only if TYPE_OVERFLOW_WRAPS (ctype).
16177 2013-04-11 Bin Cheng <bin.cheng@arm.com>
16180 * ira-costs.c (scan_one_insn): Check whether the source rtx of
16181 loading has side effect.
16183 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
16185 * config/sparc/sparc.c: Include tree-pass.h.
16186 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
16187 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
16188 head of file. Change return type. Split off gate function.
16189 (sparc_gate_work_around_errata): New function.
16190 (pass_work_around_errata): New pass definition.
16191 (insert_pass_work_around_errata) New pass insert definition to
16192 insert pass_work_around_errata just after delayed-branch scheduling.
16193 (sparc_option_override): Insert the pass.
16194 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
16196 2013-04-10 David S. Miller <davem@davemloft.net>
16198 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
16199 or -mcpu=hypersparc.
16201 * target.def (cstore_mode): New hook.
16202 * target.h: Include insn-codes.h
16203 * targhooks.c: Likewise.
16204 (default_cstore_mode): New function.
16205 * targhooks.h: Declare it.
16206 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
16207 * doc/tm.texi: Rebuild.
16208 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
16209 target hook, rather than inspecting the insn_data.
16210 * config/sparc/sparc.c (sparc_cstore_mode): New function.
16211 (TARGET_CSTORE_MODE): Redefine.
16212 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
16214 * config/sparc/predicates.md (cstore_result_operand): New special
16216 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
16217 Use it for operand 0.
16218 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
16219 (*snesi_special): Likewise.
16220 (*snesi_zero): Likewise.
16221 (*seqsi_zero): Likewise.
16222 (*sltu_insn): Likewise.
16223 (*sgeu_insn): Likewise.
16224 (*seqdi_special): Make operand 0 and comparison operation be of
16226 (*snedi_special): Likewise.
16227 (*snedi_special_vis3): Likewise.
16228 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
16229 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
16230 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
16231 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
16232 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
16233 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
16234 (*sltu_extend_sp64): Likewise.
16235 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
16236 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
16237 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
16238 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
16239 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
16241 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
16243 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
16244 (aarch64_start_file): Use the new function.
16246 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
16247 Jason Merrill <jason@redhat.com>
16249 * common.opt: Add -gdwarf.
16250 * opts.c (common_handle_option): Handle it.
16251 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
16253 2013-04-10 Richard Biener <rguenther@suse.de>
16255 * passes.c (execute_todo): Do not call ggc_collect conditional here.
16256 (execute_one_ipa_transform_pass): But unconditionally here.
16257 (execute_one_pass): And here.
16258 (init_optimization_passes): Remove reload pass.
16259 * tree-pass.h (TODO_ggc_collect): Remove.
16260 (pass_reload): Likewise.
16261 * ira.c (do_reload): Merge into ...
16263 (rest_of_handle_reload): Remove.
16264 (pass_reload): Likewise.
16265 * config/i386/i386.c (ix86_option_override): Refer to ira instead
16266 of reload for vzeroupper pass placement.
16267 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
16268 and todo_flags_finish of all passes.
16270 2013-04-10 Richard Biener <rguenther@suse.de>
16272 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
16273 first_const_oprnd field, rename first_def_type to first_op_type.
16274 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
16275 (vect_get_and_check_slp_defs): Always use the type of the
16276 operand. Allow mixed vect_external_def, vect_constant_def types.
16277 (vect_get_constant_vectors): Handle mixed vect_external_def,
16278 vect_constant_def types.
16280 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
16282 PR tree-optimization/55524
16283 * tree-ssa-math-opts.c
16284 (convert_mult_to_fma): Don't use an fms construct
16285 when we don't have an fms operation, but fnma, and it looks
16286 likely that we'll be able to use the latter.
16288 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
16290 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
16292 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
16293 inline fail caused by overwritable functions.
16295 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
16297 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
16298 unnecessary bits in the constant power of two case.
16300 2013-04-10 Richard Biener <rguenther@suse.de>
16302 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
16303 broken code swapping operands.
16304 (vect_build_slp_tree): Do not compute load permutations here.
16305 (vect_analyze_slp_instance): Compute load permutations here,
16306 after building the SLP tree.
16308 2013-04-09 Christian Bruel <christian.bruel@st.com>
16310 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
16311 of next/prev_real_insn.
16313 2013-04-09 Jan Hubicka <jh@suse.cz>
16315 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
16316 Drop aliased parameter.
16317 (function_and_variable_visibility): Do not handle alias pairs.
16318 * cgraph.c (varpool_externally_visible_p): Update prototype.
16319 * varpool.c (varpool_add_new_variable): Update.
16321 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16323 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
16325 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
16327 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
16329 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
16331 2013-04-09 Marek Polacek <polacek@redhat.com>
16333 PR tree-optimization/48762
16334 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
16336 2013-04-09 Richard Biener <rguenther@suse.de>
16338 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
16340 (vect_build_slp_tree): Likewise.
16341 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
16342 calculating the cost of a SLP instance.
16343 (vect_analyze_slp_instance): Use it from here, after building
16346 2013-04-09 Jakub Jelinek <jakub@redhat.com>
16348 PR middle-end/56883
16349 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
16350 expand_omp_for_static_chunk): Use simple_p = true in
16351 force_gimple_operand_gsi calls when assigning to addressable decls.
16353 2013-04-09 Jeff Law <law@redhat.com>
16355 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
16356 when the boolean was created by converting a wider object which
16357 had a boolean range.
16359 2013-04-09 Richard Biener <rguenther@suse.de>
16361 * tree-vectorizer.h (slp_void_p): Remove.
16362 (slp_tree): Typedef before _slp_tree declaration.
16363 (struct _slp_tree): Use a vector of slp_tree as children.
16364 (vect_get_place_in_interleaving_chain): Remove.
16365 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
16367 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
16369 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
16370 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
16371 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
16372 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
16373 Use slp_node instead of slp_void_p and adjust.
16375 2013-04-09 Richard Biener <rguenther@suse.de>
16377 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
16378 work that is not necessary.
16380 2013-04-09 Jakub Jelinek <jakub@redhat.com>
16382 PR tree-optimization/56854
16383 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
16384 forward into clobber stmts if it would change MEM_REF lhs into
16387 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
16389 * tree.c (type_hash_lookup, type_hash_add): Make static.
16390 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
16392 2013-04-09 Richard Biener <rguenther@suse.de>
16394 * tree.h (unsave_expr_now): Remove.
16395 * tree-inline.c (mark_local_for_remap_r): Remove.
16396 (unsave_expr_1): Likewise.
16397 (unsave_r): Likewise.
16398 (unsave_expr_now): Likewise.
16399 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
16400 (propagate_tree_value): Likewise.
16402 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
16404 * doc/rtl.texi (sequence): Rewrite documentation to match the
16405 current use of SEQUENCE rtl objects.
16406 * rtl.def (SEQUENCE): Likewise.
16408 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
16409 Update documentation.
16410 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
16411 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
16413 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
16415 2013-04-08 Teresa Johnson <tejohnson@google.com>
16417 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
16418 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
16420 (estimate_edge_size_and_time): Add comment to suggest using rounding
16422 (estimate_node_size_and_time): Ditto.
16423 (remap_edge_change_prob): Use helper rounding divide methods.
16424 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
16425 (gimple_mod_pow2_value_transform): Ditto.
16426 (gimple_mod_subtract_transform): Ditto.
16427 (gimple_ic_transform): Ditto.
16428 (gimple_stringops_transform): Ditto.
16429 * stmt.c (conditional_probability): Ditto.
16430 (emit_case_dispatch_table): Ditto.
16431 * lto-cgraph.c (merge_profile_summaries): Ditto.
16432 * tree-optimize.c (execute_fixup_cfg): Ditto.
16433 * cfgcleanup.c (try_forward_edges): Ditto.
16434 * cfgloopmanip.c (scale_loop_profile): Ditto.
16436 (duplicate_loop_to_header_edge): Ditto.
16437 (lv_adjust_loop_entry_edge): Ditto.
16438 * tree-vect-loop.c (vect_transform_loop): Ditto.
16439 * profile.c (compute_branch_probabilities): Ditto.
16440 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
16441 * lto-streamer-in.c (input_cfg): Ditto.
16442 * gimple-streamer-in.c (input_bb): Ditto.
16443 * ipa-cp.c (update_profiling_info): Ditto.
16444 (update_specialized_profile): Ditto.
16445 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
16446 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
16448 * sched-rgn.c (compute_dom_prob_ps): Ditto.
16449 (compute_trg_info): Ditto.
16450 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
16451 (purge_dead_edges): Ditto.
16452 * loop-unswitch.c (unswitch_loop): Ditto.
16453 * cgraphclones.c (cgraph_clone_edge): Ditto.
16454 (cgraph_clone_node): Ditto.
16455 * tree-inline.c (copy_bb): Ditto.
16456 (copy_edges_for_bb): Ditto.
16457 (initialize_cfun): Ditto.
16458 (copy_cfg_body): Ditto.
16459 (expand_call_inline): Ditto.
16461 2013-04-08 Kai Tietz <ktietz@redhat.com>
16463 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
16464 TARGET_CYGWIN64 by TARGET_64BIT.
16466 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
16468 * config/epiphany/epiphany.md (GPR_1): New constant.
16469 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
16470 * config/epiphany/epiphany.c (gen_compare_reg):
16471 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
16472 is already in place.
16473 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
16474 Don't require being called during rtl expansion; If y operlaps r0,
16476 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
16477 (epiphany_expand_epilogue): Likewise.
16479 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
16480 Don't use CC_FPmode for ORDERED / UNORDERED.
16481 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
16483 * config/epiphany/constraints.md (CnL): New constraint.
16484 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
16485 * config/epiphany/predicates.md (add_operand): Allow 1024.
16487 * config/epiphany/epiphany.md (logical_op): New code iterator.
16488 (op_mnc): New code attribute.
16489 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
16490 (mov_f+1, mov_f+2): New peephole2 patterns.
16492 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
16493 (cstoresi4): Also allow re-use of zero result when doing a NE
16494 comparison to a non-zero operand.
16495 Use (clobber (scratch)) for first insn if the gpr output is not needed.
16497 * config/epiphany/epiphany.md (<insn_opname>v2si3):
16498 Use gen_addsi3_i / gen_subsi3_i.
16500 2013-04-08 Jakub Jelinek <jakub@redhat.com>
16504 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
16505 contain anything but clobbers, at most one __builtin_stack_restore,
16506 optionally debug stmts and final resx, and if it has at least one
16507 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
16508 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
16509 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
16510 which isn't defaut definition, remove them.
16511 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
16512 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
16513 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
16514 with MEM_REF LHS with SSA_NAME address.
16516 2013-04-08 Jeff Law <law@redhat.com>
16518 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
16520 2013-04-08 Richard Biener <rguenther@suse.de>
16522 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
16524 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
16525 determined vector type.
16526 (vect_analyze_data_refs): Likewise.
16527 (vect_get_new_vect_var): Adjust.
16528 (vect_create_destination_var): Preserve SSA name versions.
16529 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
16530 not dump anything here.
16532 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
16534 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
16535 Add member lr_slot_known.
16536 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
16538 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
16539 Remove code that sets lr_slot_offset according to what a previous
16540 version of epiphany_emit_save_restore used to do.
16541 (epiphany_emit_save_restore): When doing an lr save or restore,
16542 set/verify lr_slot_known and lr_slot_offset.
16544 2013-04-08 Xinyu Qi <xyqi@marvell.com>
16547 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
16550 2013-04-08 Richard Biener <rguenther@suse.de>
16552 * alias.c (find_base_term): Fix thinko in previous change.
16554 2013-04-08 Jakub Jelinek <jakub@redhat.com>
16556 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
16557 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
16558 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
16559 if possible to compute val.
16560 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
16561 For QImode integers don't require anything about precision. Use
16562 const_with_all_bytes_same to find out if the constant doesn't have
16563 repeated bytes in it.
16565 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16567 * config/s390/s390.c (s390_expand_insv): Only accept insertions
16570 2013-04-08 Marek Polacek <polacek@redhat.com>
16572 PR rtl-optimization/48182
16573 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
16576 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16579 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
16580 nuses, make sure we have a label.
16582 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16585 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
16586 (rs6000_emit_swdiv_low_precision): Remove.
16587 (rs6000_emit_swdiv): Rewrite to handle between one and four
16588 iterations of Newton-Raphson generally; modify required number of
16589 iterations for some cases.
16590 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
16592 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
16594 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
16595 set-but-unused variable.
16597 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
16598 basic blocks of released function bodies garbage-collectable.
16600 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
16601 (struct rtl_opt_pass): Add TODO_df_finish.
16603 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
16605 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16607 * config/arm/constraints.md (q): New constraint.
16608 * config/arm/ldrdstrd.md: New file.
16609 * config/arm/arm.md (ldrdstrd.md) New include.
16610 (arm_movdi): Use "q" instead of "r" constraint
16611 for double-word memory access.
16612 (movdf_soft_insn): Likewise.
16613 * config/arm/vfp.md (movdi_vfp): Likewise.
16614 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
16615 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
16616 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
16617 (mem_ok_for_ldrd_strd): Likewise.
16618 (output_move_double): Update assertion.
16620 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16622 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
16624 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16626 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
16627 define_insn_and_split.
16628 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
16630 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16632 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
16633 define_insn_and_split.
16634 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
16635 (shiftsi3_compare): New pattern.
16636 (rrx): New pattern.
16637 * config/arm/unspecs.md (UNSPEC_RRX): New.
16639 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16641 * config/arm/arm.md (negdi_extendsidi): New pattern.
16642 (negdi_zero_extendsidi): Likewise.
16644 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16646 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
16647 define_insn_and_split.
16648 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
16649 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
16651 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16653 * config/arm/arm.md (arm_subdi3): Convert define_insn into
16654 define_insn_and_split.
16655 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
16656 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
16658 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16660 * config/arm/arm.md (subsi3_carryin): New pattern.
16661 (subsi3_carryin_const): Likewise.
16662 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
16663 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
16665 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16667 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
16669 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
16671 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
16672 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
16674 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16676 * config/arm/arm.c (arm_expand_builtin): Change fcode
16677 type to unsigned int.
16679 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16681 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
16683 2013-04-04 Ian Lance Taylor <iant@google.com>
16685 * doc/standards.texi (Standards): The Go frontend supports the Go 1
16688 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
16690 PR middle-end/56729
16691 * df-scan.c (df_insn_delete): Disable failing assert.
16693 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16695 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
16696 New function prototype.
16697 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
16698 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
16699 (arm_builtin_vectorized_function): New function.
16701 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16703 * config/arm/arm_neon_builtins.def: New file.
16704 * config/arm/arm.c (neon_builtin_data): Move contents to
16705 arm_neon_builtins.def.
16706 (enum arm_builtins): Include neon builtin definitions.
16707 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
16708 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
16710 2013-04-04 Marek Polacek <polacek@redhat.com>
16712 PR tree-optimization/48186
16713 * predict.c (maybe_hot_frequency_p): Return false if
16714 HOT_BB_FREQUENCY_FRACTION is 0.
16715 (cgraph_maybe_hot_edge_p): Likewise.
16717 2013-04-04 Richard Biener <rguenther@suse.de>
16719 PR tree-optimization/56826
16720 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
16723 2013-04-04 Richard Biener <rguenther@suse.de>
16725 PR tree-optimization/56213
16726 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
16727 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
16729 2013-04-04 Richard Biener <rguenther@suse.de>
16731 PR tree-optimization/56837
16732 * tree-loop-distribution.c (classify_partition): For non-zero
16733 values require that the value has the same precision as its
16734 mode to be useful as memset value.
16736 2013-04-03 Nick Clifton <nickc@redhat.com>
16738 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
16739 (fmssf4): Use fmsf.s on E3V5 architectures.
16740 (fnmasf4): Use fnmaf.s on E3V5 architectures.
16741 (fnmssf4): Use fnmsf.s on E3V5 architectures.
16743 2013-04-03 Jeff Law <law@redhat.com>
16745 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
16746 (lra-eliminations.o): Likewise.
16748 2013-04-03 Teresa Johnson <tejohnson@google.com>
16750 * gcov-io.c (compute_working_sets): Moved most of body of old
16751 compute_working_sets here from profile.c.
16752 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
16753 (gcov_working_set_t): Moved typedef here from basic-block.h
16754 (compute_working_set): Declare.
16755 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
16756 (get_working_sets): Renamed from compute_working_set,
16757 replace most of body with call to new compute_working_sets.
16758 (get_exec_counts): Replace call to compute_working_sets
16759 to get_working_sets.
16760 * profile.h (get_working_sets): Renamed from compute_working_set.
16761 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
16762 to get_working_sets.
16763 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
16764 * gcov-dump.c (dump_working_sets): New function.
16766 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
16768 * hwint.c (sext_hwi, zext_hwi): New functions.
16769 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
16770 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
16771 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
16772 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
16773 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
16774 (sext_hwi, zext_hwi): New functions.
16776 2013-04-03 Jeff Law <law@redhat.com>
16778 PR tree-optimization/56799
16779 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
16780 back test for widening conversion erroneously dropped in prior change.
16782 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16785 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
16786 instead of next_real_insn.
16788 2013-04-03 Marek Polacek <polacek@redhat.com>
16791 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
16793 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16796 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
16798 (thumb1_output_casesi): Likewise.
16799 (thumb2_output_casesi): Likewise.
16801 2013-04-03 Richard Biener <rguenther@suse.de>
16803 PR tree-optimization/56817
16804 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
16806 (tree_unroll_loops_completely_1): ... new function to manually
16807 walk the loop tree, properly defering outer loops of unrolled
16808 loops to later iterations.
16810 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
16812 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
16813 (vectorizable_load): Likewise.
16814 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
16815 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
16817 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
16819 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
16822 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
16824 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
16826 2013-04-03 Bin Cheng <bin.cheng@arm.com>
16828 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
16830 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
16832 PR tree-optimization/56790
16833 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
16836 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
16838 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
16840 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
16841 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
16844 2013-04-03 Jakub Jelinek <jakub@redhat.com>
16847 * tree.h (force_folding_builtin_constant_p): New decl.
16848 * builtins.c (force_folding_builtin_constant_p): New variable.
16849 (fold_builtin_constant_p): Fold immediately also if
16850 force_folding_builtin_constant_p.
16852 2013-04-03 Richard Biener <rguenther@suse.de>
16854 PR tree-optimization/56812
16855 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
16856 DRs of the same interleaving chain are independent.
16858 2013-04-02 Jason Merrill <jason@redhat.com>
16860 * gdbinit.in (pbb): Use debug fn.
16862 2013-04-02 Lawrence Crowl <crowl@google.com>
16864 * sese.h (struct ivtype_map_elt_s): Remove unused.
16865 (extern debug_ivtype_map): Remove unused.
16866 (extern eq_ivtype_map_elts): Remove unused.
16867 * sese.c (debug_ivtype_map): Removed unused.
16868 (debug_ivtype_map_1): Removed unused.
16869 (debug_ivtype_elt): Remove unused.
16870 (eq_ivtype_map_elts): Remove unused.
16873 2013-04-02 Kai Tietz <ktietz@redhat.com>
16876 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
16877 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
16878 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
16880 (legitimize_pe_coff_symbol): Likewise.
16881 (is_imported_p): New helper-function.
16882 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
16883 for Windows x64 targets.
16884 (ix86_expand_prologue): Optimize for pe-coff targets.
16885 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
16886 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
16887 medium/large code-model.
16888 (legitimize_pic_address): Likewise.
16889 (legitimize_tls_address): Likewise.
16890 (ix86_expand_call): Likewise.
16891 (x86_output_mi_thunk): Likewise.
16892 (get_dllimport_decl): Add new beimport argument.
16893 (construct_plt_address): Don't assert for x64 pe-coff targets.
16894 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
16896 (SYMBOL_FLAG_STUBVAR): New macro.
16897 (SYMBOL_REF_STUBVAR_P): Likewise.
16898 * config/i386/winnt.c (stub_list): New structure.
16899 (stub_head): New local variable.
16900 (i386_pe_record_stub): New function.
16901 (i386_pe_file_end): Emit refptr-stubs.
16903 2013-04-02 Jakub Jelinek <jakub@redhat.com>
16905 PR rtl-optimization/56745
16906 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
16907 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
16910 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
16911 and both of them are MEM_REFs, just compare first argument for
16912 equality and attempt to deal even with differing offsets.
16915 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
16916 of gimple_clobber_p to be MEM_REF.
16917 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
16918 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
16919 after gimplification.
16920 * asan.c (get_mem_ref_of_assignment): Don't instrument
16921 gimple_clobber_p stmts.
16922 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
16923 gimple_clobber_p stmt if they have MEM_REF lhs and
16924 are dead because of another gimple_clobber_p stmt.
16925 * tree-ssa-live.c (clear_unused_block_pointer): Treat
16926 gimple_clobber_p stmts like debug stmts.
16927 (remove_unused_locals): Remove clobbers with MEM_REF lhs
16928 that refer to unused VAR_DECLs or uninitialized values.
16929 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
16930 gimple_clobber_p stmts if they refer to removed parameters.
16931 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
16934 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
16936 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
16937 using SWI48 mode attribute.
16939 2013-04-02 Wei Mi <wmi@google.com>
16941 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
16942 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
16943 *<rotate_insn><mode>3_mask in i386.md.
16945 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
16947 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
16949 2013-04-02 Richard Biener <rguenther@suse.de>
16951 PR tree-optimization/56778
16952 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16953 Runtime alias tests are not supported for gather loads.
16954 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
16955 stmts referenced from SSA operands before updating SSA form.
16957 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
16958 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16960 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
16961 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
16962 * config/arm/cortex-a53.md: New file.
16963 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
16964 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
16965 * config/arm/arm.c (arm_issue_rate): Likewise.
16966 * config/arm/arm-tune.md: Regenerate
16967 * config/arm/arm-tables.opt: Regenerate.
16968 * config/arm/arm-cores.def: Add cortex-a53.
16970 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
16972 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
16975 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
16977 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
16978 scalar load/store operations using B/H registers.
16979 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
16981 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
16983 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
16985 * config/aarch64/aarch64.c
16986 (aarch64_simd_scalar_immediate_valid_for_move): New.
16987 * config/aarch64/aarch64-protos.h
16988 (aarch64_simd_scalar_immediate_valid_for_move): New.
16989 * config/aarch64/constraints.md (Dh, Dq): New.
16990 * config/aarch64/iterators.md (hq): New.
16992 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
16994 * reorg.c (get_branch_condition): Deal with conditional returns.
16995 (fill_simple_delay_slots): Remove dead code dealing with jumps.
16997 2013-04-01 Wei Mi <wmi@google.com>
16999 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
17000 Truncate operand 2 using %b asm operand modifier.
17001 (*<shift_insn><mode>3_mask): Ditto.
17002 (*<rotate_insn><mode>3_mask): Ditto.
17004 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
17006 PR middle-end/56798
17007 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
17009 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
17011 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
17013 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
17015 2013-03-30 Lawrence Crowl <crowl@google.com>
17017 * dse.c (clear_alias_sets): Remove never set.
17018 (disqualified_clear_alias_sets): Remove never set.
17019 (clear_alias_mode_pool): Remove never set.
17020 (dse_step0): Remove condition that is never true.
17021 (canon_address): Remove condition that is never true.
17022 (dse_step7): Remove condition that is never true.
17023 (rest_of_handle_dse): Remove condition that is never true.
17024 (rest_of_handle_dse::did_global): Remove never read from above.
17025 (dse_step2_spill): Remove never called from above.
17026 (dse_step5_spill): Remove never called from above.
17028 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
17030 * doc/md.texi (Standard Names) <casesi>: Update documentation for
17031 JUMP_TABLE_DATA changes.
17032 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
17033 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
17034 (Insns) <jump_table_data>: New entry.
17035 * doc/tm.texi: Regenerate.
17037 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
17039 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
17040 for table jump at the end of a basic block using tablejump_p.
17041 * targhooks.c (default_invalid_within_doloop): Likewise.
17042 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
17043 target hook implementation that is identical to the default hook.
17044 (rs6000_invalid_within_doloop): Remove.
17046 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
17047 unused variable from tablejump_p call.
17049 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
17050 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
17051 (INSN_DELETED_P): Likewise.
17052 (emit_jump_table_data): New prototype.
17053 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
17054 after 4th as unused.
17055 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
17056 * sched-vis.c (print_insn): Likewise.
17057 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
17058 insn for compatibility with back ends that use next_active_insn to
17059 identify jump table data.
17060 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
17061 (remove_insn): Likewise.
17062 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
17064 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
17065 (emit_jump_table_data): New function.
17067 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
17068 basic block, a JUMP_TABLE_DATA never is.
17069 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
17070 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
17071 off from code handling real insns.
17072 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
17073 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
17074 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
17075 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
17076 is not a NONDEBUG_INSN_P.
17077 * ira-costs.c (scan_one_insn): Likewise.
17078 * jump.c (mark_all_labels): Likewise.
17079 (mark_jump_label_1): Likewise.
17080 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17081 * lra.c (get_insn_freq): Expect all insns reaching here to be in
17083 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
17084 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
17085 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
17086 JUMP_TABLE_DATA_P insns.
17087 (calculate_elim_costs_all_insns): Likewise.
17088 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
17089 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
17090 (delete_output_reload): Code style fixups.
17091 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
17092 insn flags on this non-insn.
17093 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
17094 as scheduling barriers, for pre-change compatibility.
17095 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
17096 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
17098 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
17099 redundant JUMP_TABLE_DATA_P test.
17100 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
17101 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17102 (frv_for_each_packet): Likewise.
17103 * config/i386/i386.c (min_insn_size): Likewise.
17104 (ix86_avoid_jump_mispredicts): Likewise.
17105 * config/m32r/m32r.c (m32r_is_insn): Likewise.
17106 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
17107 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17108 (mips16_insn_length): Robustify.
17109 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
17110 (mips16_split_long_branches): Likewise.
17111 * config/pa/pa.c (pa_combine_instructions): Likewise.
17112 * config/rs6000/rs6000.c (get_next_active_insn): Treat
17113 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
17114 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
17115 as contributing to pool range lengths.
17116 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
17117 Remove redundant JUMP_TABLE_DATA_P test.
17118 (sh_loop_align): Likewise.
17119 (split_branches): Likewise.
17120 (sh_insn_length_adjustment): Likewise.
17121 * config/spu/spu.c (get_branch_target): Likewise.
17123 2013-03-29 Jan Hubicka <jh@suse.cz>
17125 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
17126 gcov streaming; stream hot bb threshold to ltrans.
17127 * predict.c (get_hot_bb_threshold): Break out from ....
17128 (maybe_hot_count_p): ... here.
17129 (set_hot_bb_threshold): New function.
17130 * lto-section-in.c (lto_section_name): Add profile.
17131 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
17132 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
17133 and data-streamer.h
17134 (histogram_entry): New structure.
17135 (histogram, histogram_pool): New global vars.
17136 (histogram_hash): New structure.
17137 (histogram_hash::hash): New method.
17138 (histogram_hash::equal): Likewise.
17139 (account_time_size): New function.
17140 (cmp_counts): New function.
17141 (dump_histogram): New function.
17142 (ipa_profile_generate_summary): New function.
17143 (ipa_profile_write_summary): New function.
17144 (ipa_profile_read_summary): New function.
17145 (ipa_profile): Decide on threshold.
17146 (pass_ipa_profile): Add ipa_profile_write_summary and
17147 ipa_profile_read_summary.
17148 * Makefile.in (ipa.o): Update dependencies.
17149 * lto-streamer.h (LTO_section_ipa_profile): New section.
17151 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
17153 * tree.h (VAR_P): New.
17155 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
17158 * doc/invoke.texi ([-fwhole-program]): Fix typo.
17160 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
17162 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
17163 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17164 (control_flow_insn_p): Likewise.
17165 * cfgrtl.c (duplicate_insn_chain): Likewise.
17166 * final.c (get_attr_length_1): Likewise.
17167 (shorten_branches): Likewise.
17168 (final_scan_insn): Likewise.
17169 * function.c (instantiate_virtual_regs): Likewise.
17170 * gcse.c (insert_insn_end_basic_block): Likewise.
17171 * ira-costs.c (scan_one_insn): Likewise.
17172 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17173 * lra.c (check_rtl): Likewise.
17174 * reload1.c (elimination_costs_in_insn): Likewise.
17175 * reorg.c (follow_jumps): Likewise.
17177 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
17178 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17179 (thumb_far_jump_used_p): Likewise.
17180 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
17181 (workaround_speculation): Likewise.
17182 (add_sched_insns_for_speculation): Likewise.
17183 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
17184 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17185 (frv_for_each_packet): Likewise.
17186 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
17187 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
17188 (final_emit_insn_group_barriers): Likewise.
17189 * config/m32r/m32r.c (m32r_is_insn): Likewise.
17190 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17191 (mips16_insn_length): Likewise.
17192 * config/pa/pa.c (pa_reorg): Likewise.
17193 (pa_combine_instructions): Likewise.
17194 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
17195 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
17196 (sh_reorg): Likewise.
17197 (split_branches): Likewise.
17198 * config/spu/spu.c (get_branch_target): Likewise.
17200 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
17203 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
17205 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
17206 Fix declaration name.
17208 2013-03-28 Lawrence Crowl <crowl@google.com>
17210 * graphds.h (struct graph.indicies): Remove unused.
17211 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
17212 (SCOP_ORIGINAL_PDDRS): Remove unused.
17213 * sese.h (extern insert_loop_close_phis): Removed unused.
17214 (extern insert_guard_phis): Removed unused.
17215 (extern ivtype_map_elt_info): Removed unused.
17216 (new_ivtype_map_elt): Removed unused.
17217 * sese.c (ivtype_map_elt_info): Removed unused.
17219 2013-03-28 Lawrence Crowl <crowl@google.com>
17221 * Makefile.in: Add several missing include dependences.
17223 (test-dump.o): New. This object is not added to any executable,
17224 but is present for ad-hoc testing.
17226 (debug (const bitmap_head_def &)): New.
17227 (debug (const bitmap_head_def *)): New.
17229 (extern debug (const bitmap_head_def &)): New.
17230 (extern debug (const bitmap_head_def *)): New.
17232 (debug (edge_def &)): New.
17233 (debug (edge_def *)): New.
17235 (debug (basic_block_def &)): New.
17236 (debug (basic_block_def *)): New.
17238 (dump_node (const_tree, int, FILE *)): Correct source file.
17240 (debug (die_struct &)): New.
17241 (debug (die_struct *)): New.
17243 (extern debug (die_struct &)): New.
17244 (extern debug (die_struct *)): New.
17245 * gimple-pretty-print.c
17246 (debug (gimple_statement_d &)): New.
17247 (debug (gimple_statement_d *)): New.
17248 * gimple-pretty-print.h
17249 (extern debug (gimple_statement_d &)): New.
17250 (extern debug (gimple_statement_d *)): New.
17252 (debug (ira_allocno_copy &)): New.
17253 (debug (ira_allocno_copy *)): New.
17254 (debug (ira_allocno &)): New.
17255 (debug (ira_allocno *)): New.
17257 (extern debug (ira_allocno_copy &)): New.
17258 (extern debug (ira_allocno_copy *)): New.
17259 (extern debug (ira_allocno &)): New.
17260 (extern debug (ira_allocno *)): New.
17262 (debug (live_range &)): New.
17263 (debug (live_range *)): New.
17265 (debug (lra_live_range &)): New.
17266 (debug (lra_live_range *)): New.
17268 (debug (lra_live_range &)): New.
17269 (debug (lra_live_range *)): New.
17271 (debug (omega_pb_d &)): New.
17272 (debug (omega_pb_d *)): New.
17274 (extern debug (omega_pb_d &)): New.
17275 (extern debug (omega_pb_d *)): New.
17277 (debug (const rtx_def &)): New.
17278 (debug (const rtx_def *)): New.
17280 (debug_tree (tree): Move within file.
17281 (debug_raw (const tree_node &)): New.
17282 (debug_raw (const tree_node *)): New.
17283 (dump_tree_via_hooks (const tree_node *, int)): New.
17284 (debug (const tree_node &)): New.
17285 (debug (const tree_node *)): New.
17286 (debug_verbose (const tree_node &)): New.
17287 (debug_verbose (const tree_node *)): New.
17288 (debug_head (const tree_node &)): New.
17289 (debug_head (const tree_node *)): New.
17290 (debug_body (const tree_node &)): New.
17291 (debug_body (const tree_node *)): New.
17292 (debug_vec_tree (tree): Move and reimplement in terms of dump.
17293 (debug (vec<tree, va_gc> &)): New.
17294 (debug (vec<tree, va_gc> *)): New.
17296 (extern debug (const rtx_def &)): New.
17297 (extern debug (const rtx_def *)): New.
17299 (debug_raw (simple_bitmap_def &)): New.
17300 (debug_raw (simple_bitmap_def *)): New.
17301 (debug (simple_bitmap_def &)): New.
17302 (debug (simple_bitmap_def *)): New.
17304 (extern debug (simple_bitmap_def &)): New.
17305 (extern debug (simple_bitmap_def *)): New.
17306 (extern debug_raw (simple_bitmap_def &)): New.
17307 (extern debug_raw (simple_bitmap_def *)): New.
17309 (debug (vinsn_def &)): New.
17310 (debug (vinsn_def *)): New.
17311 (debug_verbose (vinsn_def &)): New.
17312 (debug_verbose (vinsn_def *)): New.
17313 (debug (expr_def &)): New.
17314 (debug (expr_def *)): New.
17315 (debug_verbose (expr_def &)): New.
17316 (debug_verbose (expr_def *)): New.
17317 (debug (vec<rtx> &)): New.
17318 (debug (vec<rtx> *)): New.
17320 (extern debug (vinsn_def &)): New.
17321 (extern debug (vinsn_def *)): New.
17322 (extern debug_verbose (vinsn_def &)): New.
17323 (extern debug_verbose (vinsn_def *)): New.
17324 (extern debug (expr_def &)): New.
17325 (extern debug (expr_def *)): New.
17326 (extern debug_verbose (expr_def &)): New.
17327 (extern debug_verbose (expr_def *)): New.
17328 (extern debug (vec<rtx> &)): New.
17329 (extern debug (vec<rtx> *)): New.
17331 (_list_iter_cond_expr): Make inline instead of static.
17333 (debug (sreal &)): New.
17334 (debug (sreal *)): New.
17336 (extern debug (sreal &)): New.
17337 (extern debug (sreal *)): New.
17339 (extern debug_raw (const tree_node &)): New.
17340 (extern debug_raw (const tree_node *)): New.
17341 (extern debug (const tree_node &)): New.
17342 (extern debug (const tree_node *)): New.
17343 (extern debug_verbose (const tree_node &)): New.
17344 (extern debug_verbose (const tree_node *)): New.
17345 (extern debug_head (const tree_node &)): New.
17346 (extern debug_head (const tree_node *)): New.
17347 (extern debug_body (const tree_node &)): New.
17348 (extern debug_body (const tree_node *)): New.
17349 (extern debug (vec<tree, va_gc> &)): New.
17350 (extern debug (vec<tree, va_gc> *)): New.
17352 (debug (struct loop &)): New.
17353 (debug (struct loop *)): New.
17354 (debug_verbose (struct loop &)): New.
17355 (debug_verbose (struct loop *)): New.
17356 * tree-dump.c: Add header dependence.
17358 (extern debug (struct loop &)): New.
17359 (extern debug (struct loop *)): New.
17360 (extern debug_verbose (struct loop &)): New.
17361 (extern debug_verbose (struct loop *)): New.
17363 (debug (data_reference &)): New.
17364 (debug (data_reference *)): New.
17365 (debug (vec<data_reference_p> &)): New.
17366 (debug (vec<data_reference_p> *)): New.
17367 (debug (vec<ddr_p> &)): New.
17368 (debug (vec<ddr_p> *)): New.
17370 (extern debug (data_reference &)): New.
17371 (extern debug (data_reference *)): New.
17372 (extern debug (vec<data_reference_p> &)): New.
17373 (extern debug (vec<data_reference_p> *)): New.
17374 (extern debug (vec<ddr_p> &)): New.
17375 (extern debug (vec<ddr_p> *)): New.
17377 (debug (pt_solution &)): New.
17378 (debug (pt_solution *)): New.
17380 (extern debug (pt_solution &)): New.
17381 (extern debug (pt_solution *)): New.
17383 (debug (_var_map &)): New.
17384 (debug (_var_map *)): New.
17385 (debug (tree_live_info_d &)): New.
17386 (debug (tree_live_info_d *)): New.
17388 (extern debug (_var_map &)): New.
17389 (extern debug (_var_map *)): New.
17390 (extern debug (tree_live_info_d &)): New.
17391 (extern debug (tree_live_info_d *)): New.
17393 2013-03-28 Jan Hubicka <jh@suse.cz>
17395 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
17397 2013-03-28 Ian Bolton <ian.bolton@arm.com>
17399 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
17400 record only when desired or required.
17402 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
17404 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
17405 *vec_extractv2di_1_rex64. Use x64 isa attribute.
17407 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17409 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
17410 (*andsi3_compare0_uxtw): New pattern.
17411 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
17412 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
17414 2013-03-28 Jan Hubicka <jh@suse.cz>
17416 * data-streamer-in.c (streamer_read_gcov_count): New function.
17417 * gimple-streamer-out.c: Include value-prof.h.
17418 (output_gimple_stmt): Output histogram.
17419 (output_bb): Use streamer_write_gcov_count.
17420 * value-prof.c: Include data-streamer.h
17421 (dump_histogram_value): Add HIST_TYPE_MAX.
17422 (stream_out_histogram_value): New function.
17423 (stream_in_histogram_value): New function.
17424 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
17425 (stream_out_histogram_value, stream_in_histogram_value): Declare.
17426 * data-streamer-out.c (streamer_write_gcov_count): New function.
17427 (streamer_write_gcov_count_stream): New function.
17428 * lto-cgraph.c (lto_output_edge): Update counter streaming.
17429 (lto_output_node): Likewise.
17430 (input_node, input_edge): Likewise.
17431 * lto-streamer-out.c (output_cfg): Update streaming.
17432 * lto-streamer-in.c (input_cfg): Likewise.
17433 * data-streamer.h (streamer_write_gcov_count,
17434 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
17435 * gimple-streamer-in.c: Include value-prof.h
17436 (input_gimple_stmt): Input histograms.
17437 (input_bb): Update profile streaming.
17439 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
17441 * genmodes.c (emit_max_int): New function.
17442 (emit_insn_modes_h): Added call to emit_max_function.
17443 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
17445 * machmode.def: Fixed comment.
17447 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
17449 * combine.c (try_combine): Removed useless assert.
17450 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
17452 2013-03-28 Marek Polacek <polacek@redhat.com>
17453 Richard Biener <rguenther@suse.de>
17455 PR tree-optimization/56695
17456 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
17457 build signed result of a vector comparison.
17458 * tree-cfg.c (verify_gimple_comparison): Check that a result
17459 of a vector comparison has signed type.
17461 2013-03-28 Richard Biener <rguenther@suse.de>
17463 PR tree-optimization/37021
17464 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
17465 do not restrict gaps between groups.
17466 * tree-vect-stmts.c (vectorizable_load): Properly account for
17467 a gap between groups.
17469 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
17471 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
17472 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
17475 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
17477 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
17478 * doc/extend.texi (Named Address Spaces): Ditto.
17479 (Variable Attributes): Ditto.
17481 2013-03-27 Kai Tietz <ktietz@redhat.com>
17483 * config.build: Add support for cygwin x64 target.
17484 * config.gcc: Likewise.
17485 * config.host: Likewise.
17486 * configure.ac: Likewise
17487 * configure: Regenerated.
17489 2013-03-27 Kai Tietz <ktietz@redhat.com>
17491 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
17492 * config/i386/t-cygwin-w64: New file.
17493 * config/i386/cygwin-w64.h: New file.
17494 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
17495 and add support for x64-cygwin target.
17496 (CPP_SPEC): Likewise.
17497 (CXX_WRAP_SPEC_LIST): Undefine before define.
17498 (LIBGCJ_SONAME): Use 15 as version.
17500 2013-03-27 Richard Biener <rguenther@suse.de>
17502 PR tree-optimization/56716
17503 * tree-ssa-structalias.c (perform_var_substitution): Adjust
17504 dumping for ref nodes.
17506 2013-03-27 Martin Jambor <mjambor@suse.cz>
17508 PR tree-optimization/55334
17509 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
17510 restricted pointers to arrays.
17512 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
17514 * Makefile.in (.SUFFIXES): Add .cc.
17515 (.c.o): Apply same recipe for implicit rule .cc.o.
17517 2013-03-27 Richard Biener <rguenther@suse.de>
17519 PR tree-optimization/37021
17520 * tree-vect-data-refs.c (vect_check_strided_load): Allow
17521 REALPART/IMAGPART_EXPRs around the supported refs.
17522 * tree-ssa-structalias.c (find_func_aliases): Assume that
17523 floating-point values are not used to transfer pointers.
17525 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
17527 * target.def (TARGET_HAS_IFUNC_P): New target hook.
17528 * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
17529 * doc/tm.texi: Regenerate.
17530 * targhooks.h (default_has_ifunc_p): New.
17531 * targhooks.c (default_has_ifunc_p): Ditto.
17532 * config/linux-protos.h: New file.
17533 * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
17534 hook for linux which disables support of indirect functions in android.
17535 * config/linux-android.c: New file.
17536 * config/t-linux-android.c: Ditto.
17537 * config.gcc: Added new object file linux-android.o.
17538 * config/i386/i386.c (ix86_get_function_versions_dispatcher):
17539 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
17540 * varasm.c (do_assemble_alias): Likewise.
17541 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
17542 doesn't support indirect functions.
17543 * configure: Regenerate.
17545 2013-03-27 Bin Cheng <bin.cheng@arm.com>
17548 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
17549 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
17552 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17554 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
17556 2013-03-27 Terry Guo <terry.guo@arm.com>
17558 * config/arm/arm-cores.def: Added core cortex-r7.
17559 * config/arm/arm-tune.md: Regenerated.
17560 * config/arm/arm-tables.opt: Regenerated.
17561 * doc/invoke.texi: Added entry for core cortex-r7.
17563 2013-03-27 Walter Lee <walt@tilera.com>
17565 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
17566 double-decrement of next_scratch_regno.
17568 2013-03-27 Walter Lee <walt@tilera.com>
17570 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
17572 (insn_v1mulus): Ditto.
17573 (insn_v2muls): Ditto.
17575 2013-03-27 Walter Lee <walt@tilera.com>
17577 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
17578 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
17580 2013-03-27 Walter Lee <walt@tilera.com>
17582 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
17583 (*sibcall_value): Ditto.
17585 2013-03-27 Walter Lee <walt@tilera.com>
17587 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
17588 (insn_mnz_v8qi): ... this ...
17589 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
17591 (insn_v<n>mnz): Replaced by ...
17592 (insn_v1mnz): ... this ...
17593 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
17595 (insn_mz_<mode>): Replaced by ...
17596 (insn_mz_v8qi): ... this ...
17597 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
17599 (insn_v<n>mz): Replaced by ...
17600 (insn_v1mz): ... this ...
17601 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
17604 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
17606 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
17608 2013-03-26 Roland McGrath <mcgrathr@google.com>
17610 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
17611 than fprintf with a non-constant, non-format string.
17613 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
17615 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
17616 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
17617 operand 0 predicate.
17618 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
17619 attribute. Use general_x64nomem_operand as operand 1 predicate.
17620 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
17621 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
17622 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
17623 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
17624 (mov<mode>_insv_1): Remove expander. Merge insn with
17625 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
17626 Use general_x64nomem_operand as operand 1 predicate.
17627 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
17628 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
17629 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
17630 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
17631 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
17632 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
17633 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
17634 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
17635 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
17636 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
17637 (general_x64nomem_operand): Ditto.
17639 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17641 * config/rtems.opt: Add -pthread option.
17643 2013-03-26 Richard Biener <rguenther@suse.de>
17645 * alias.c (find_base_term): Avoid redundant and not used recursion.
17646 (base_alias_check): Get the initial base term from the caller.
17647 (true_dependence_1): Compute and pass base terms to base_alias_check.
17648 (write_dependence_p): Likewise.
17649 (may_alias_p): Likewise.
17651 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
17653 * config/aarch64/aarch64.c (aarch64_classify_address): Support
17654 PC-relative load in SI modes and above only.
17656 2013-03-26 Xinyu Qi <xyqi@marvell.com>
17658 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
17659 * config/arm/iwmmxt.md (WCGR0): Update.
17660 (WCGR1, WCGR2, WCGR3): Likewise.
17662 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
17664 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
17665 Use x64 and nox64 isa attributes.
17667 2013-03-26 Richard Biener <rguenther@suse.de>
17669 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
17670 alignment computations and rely on get_object_alignment_1
17671 for the !TYPE_P case.
17672 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
17674 2013-03-26 Walter Lee <walt@tilera.com>
17676 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
17677 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
17679 2013-03-25 Jeff Law <law@redhat.com>
17681 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
17682 check for INTEGRAL_TYPE_P that was missing due to checking in
17683 wrong version of prior patch.
17685 2013-03-25 Walter Lee <walt@tilera.com>
17687 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
17688 TILEGX_INSN_SHUFFLEBYTES1.
17689 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
17691 (tilegx_builtins): Ditto.
17692 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
17694 2013-03-25 Walter Lee <walt@tilera.com>
17696 * config/tilegx/tilegx.md (floatsisf2): New pattern.
17697 (floatunssisf2): New pattern.
17698 (floatsidf2): New pattern.
17699 (floatunssidf2): New pattern.
17701 2013-03-25 Walter Lee <walt@tilera.com>
17703 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
17704 tests for constraint J, K, N, P.
17706 2013-03-25 Walter Lee <walt@tilera.com>
17708 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
17709 Use indirect/pcrel encoding.
17710 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
17713 2013-03-25 Steve Ellcey <sellcey@mips.com>
17715 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
17716 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
17717 * config/mips/mips.c (mips_option_override): Set IMADD default.
17718 * config/mips/mips.h (PTF_AVOID_IMADD): New.
17719 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
17720 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
17721 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
17722 * doc/invoke.texi (-mimadd/-mno-imadd): New.
17724 2013-03-25 Jeff Law <law@redhat.com>
17726 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
17727 slightly to avoid creating and folding useless trees. Simplify
17728 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
17730 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
17732 * config/i386/i386.md (*zero_extendsidi2): Merge with
17733 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
17734 * config/i386/predicates.md (x86_64_zext_operand): Rename from
17735 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
17736 targets. Clarify comment.
17738 2013-03-25 Martin Jambor <mjambor@suse.cz>
17740 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
17741 pass-through jump functions differently.
17742 (ipa_read_jump_function): Likewise. Also use setter functions to set
17745 2013-03-25 Martin Jambor <mjambor@suse.cz>
17747 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
17748 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
17750 (ipa_get_indirect_edge_target): New function.
17751 (devirtualization_time_bonus): New parameter known_aggs, pass it to
17752 ipa_get_indirect_edge_target. Update all callers.
17753 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
17754 ipa_get_indirect_edge_target_1 instead of calling
17755 ipa_get_indirect_edge_target.
17756 (create_specialized_node): Pass aggvlas to
17757 ipcp_discover_new_direct_edges.
17759 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17761 * config/arm/arm.md (f_sels, f_seld): New types.
17762 (*cmov<mode>): New pattern.
17763 * config/arm/predicates.md (arm_vsel_comparison_operator): New
17766 2013-03-25 Kai Tietz <ktietz@redhat.com>
17768 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
17769 POSIX-printf for mingw-hosted builds.
17771 2013-03-25 Richard Biener <rguenther@suse.de>
17773 PR middle-end/56694
17774 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
17775 must-not-throw stmt location.
17777 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17779 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
17780 Emit load-acquire versions when acq is true.
17781 (arm_emit_store_exclusive): Add rel parameter.
17782 Emit store-release versions when rel is true.
17783 (arm_split_compare_and_swap): Use acquire-release instructions
17785 of barriers when appropriate.
17786 (arm_split_atomic_op): Likewise.
17787 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
17788 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
17789 (VUNSPEC_SLX): Likewise.
17790 (VUNSPEC_LDA): Likewise.
17791 (VUNSPEC_STL): Likewise.
17792 * config/arm/sync.md (atomic_load<mode>): New pattern.
17793 (atomic_store<mode>): Likewise.
17794 (arm_load_acquire_exclusive<mode>): Likewise.
17795 (arm_load_acquire_exclusivesi): Likewise.
17796 (arm_load_acquire_exclusivedi): Likewise.
17797 (arm_store_release_exclusive<mode>): Likewise.
17799 2013-03-25 Catherine Moore <clm@codesourcery.com>
17801 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
17802 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
17803 * config/mip/predicates.md (lwsp_swsp_operand,
17804 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
17805 sb16_operand, db4_operand, db7_operand, ib3_operand,
17806 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
17807 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
17808 andi16_operand): New predicates.
17809 * config/mips/mips.md (compression): New attribute.
17810 (enabled): New attribute.
17811 (length): Consider compression in computing length.
17812 (shift_compression): New code attribute.
17813 (*add<mode>3): New operands. Record compression.
17814 (sub<mode>3): Likewise.
17815 (one_cmpl<mode>2): Likewise.
17816 (*and<mode>3): Likewise.
17817 (*ior<mode>3): Likewise.
17818 (unnamed pattern for xor): Likewise.
17819 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
17820 (*<optab><mode>3): Likewise.
17821 (*mov<mode>_internal: Likewise.
17822 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
17823 (mips_unsigned_immediate_p): New.
17824 (umips_lwsp_swsp_address_p): New.
17825 (m16_based_address_p): New.
17826 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
17827 (mips_unsigned_immediate_p): New prototype.
17828 (lwsp_swsp_address_p): New prototype.
17829 (m16_based_address_p): New prototype.
17830 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
17831 (mips_signed_immediate_p): New function.
17832 (m16_based_address_p): New function.
17833 (lwsp_swsp_address_p): New function.
17834 (mips_print_operand_punctuation): Recognize short delay slot insns
17835 for microMIPS.add<mode>3"
17837 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17840 * config/arm/iterators.md (v_cmp_result): New mode attribute.
17841 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
17843 2013-03-25 Richard Biener <rguenther@suse.de>
17845 PR tree-optimization/56689
17846 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
17849 2013-03-25 Richard Biener <rguenther@suse.de>
17851 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
17853 (memory_references): Likewise.
17854 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
17855 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
17856 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
17857 (gather_mem_refs_in_loops): Fold into ...
17858 (analyze_memory_references): ... this. Move initialization
17859 to tree_ssa_lim_initialize.
17860 (fill_always_executed_in): Rename to ...
17861 (fill_always_executed_in_1): ... this.
17862 (fill_always_executed_in): Move contains_call computation to
17863 this new function from ...
17864 (tree_ssa_lim_initialize): ... here.
17865 (tree_ssa_lim): Call fill_always_executed_in.
17867 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
17869 * postreload.c (reload_combine): Fix code detecting returns.
17871 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
17873 * function.c (emit_use_return_register_into_block): On cc0 targets,
17874 do not emit the sequence between cc0 setter and user.
17876 2013-03-25 Kai Tietz <ktietz@redhat.com>
17878 * config/i386/predicates.md (local_symbolic_operand): Interpret
17879 dll-imported symbols as none-local.
17881 2013-03-25 Richard Biener <rguenther@suse.de>
17883 * tree-ssa-loop-im.c (struct depend): Remove.
17884 (struct lim_aux_data): Make depends a vec of gimples.
17885 (free_lim_aux_data): Adjust.
17886 (add_dependency): Likewise.
17887 (set_level): Likewise.
17889 2013-03-25 Richard Biener <rguenther@suse.de>
17891 PR middle-end/56434
17892 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
17893 the pointer returned by calls with ECF_MALLOC set.
17895 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
17897 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
17899 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
17901 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
17902 using MMXMODE mode iterator.
17903 (*move<mode>_internal): Merge with *movv2sf_internal and
17904 *movv2sf_internal_rex64 using MMXMODE mode iterator.
17906 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
17908 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
17909 (record_last_mem_set_info): Likewise.
17911 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
17912 of XNEWVEC followed by memset.
17913 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
17915 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
17917 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
17918 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
17919 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
17920 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
17921 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
17922 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
17923 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
17924 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
17925 BARRIER_P instead of GET_CODE.
17927 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17929 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
17930 inaccuracy in the probing code.
17932 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
17933 (ctrapdi4): Likewise.
17935 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17937 * calls.c (expand_call): Add missing guard to code handling return
17938 of non-BLKmode structures in MSB.
17939 * function.c (expand_function_end): Likewise.
17941 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17943 * combine.c (try_combine): Adjust comment. Do not add the set of
17944 insn #0 if the destination indirectly is set or dies in insn #2.
17945 Tidy up code to distribute a new note.
17947 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
17949 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
17950 also for alternatives 16 and 17.
17952 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
17954 * config/i386/sse.md (*mov<mode>_internal): Merge with
17955 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
17956 Emit insn template depending on type attribute. Use
17957 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17958 movd instead of movq mnemonic for interunit moves. Rewrite mode
17959 attribute calculation. Remove unit attribute calculation.
17960 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
17961 Set prefix_data16 attribute for DImode ssemov types.
17962 Use Ym instead of y for SSE-MMX conversion alternatives.
17963 Reorder operand constraints.
17965 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
17967 * df.h (df_insn_delete): Adjust prototype.
17968 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
17969 and let it decide whether mark the basic block dirty.
17970 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
17971 * df-scan.c (df_insn_info_delete): New helper function, split
17972 off from df_insn_delete.
17973 (df_scan_free_bb_info): Use it.
17974 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
17976 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
17977 that the insn is actually an insn and it has a non-NULL basic block.
17978 Do not mark basic block dirty if only deleting a DEBUG_INSN.
17980 2013-03-22 Richard Biener <rguenther@suse.de>
17982 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
17984 (mem_ref_alloc): Do not allocate them.
17985 (refs_independent_p): Do not query or maintain a cache.
17987 2013-03-22 Richard Biener <rguenther@suse.de>
17989 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
17990 (gather_mem_refs_in_loops): Do not compute it.
17991 (analyze_memory_references): Do not allocate it.
17992 (tree_ssa_lim_finalize): Do not free it.
17993 (for_all_locs_in_loop): Do not query all_refs_in_loop.
17995 2013-03-22 Richard Biener <rguenther@suse.de>
17997 * is-a.h (as_a): Use gcc_checking_assert.
17999 2013-03-22 Ian Bolton <ian.bolton@arm.com>
18001 * config/aarch64/aarch64.c (aarch64_print_operand): New
18002 format specifier for printing a constant in hex.
18003 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
18004 format specifier for printing second operand.
18006 2013-03-22 Richard Biener <rguenther@suse.de>
18008 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
18010 (gather_mem_refs_in_loops): Perform store accumulation here.
18011 (create_vop_ref_mapping_loop): Remove.
18012 (create_vop_ref_mapping): Likewise.
18013 (analyze_memory_references): Initialize refs_stored_in_loop.
18014 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
18015 (record_indep_loop): Remove.
18016 (record_dep_loop): New function.
18017 (ref_indep_loop_p_1): Adjust to only walk over references
18018 in the loop, not its subloops.
18019 (ref_indep_loop_p): Rename to ...
18020 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
18021 maintaining a more fine-grained cache.
18022 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
18023 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
18025 2013-03-22 Richard Biener <rguenther@suse.de>
18027 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
18028 (struct mem_ref): Make accesses_in_loop a vec of a vec of
18029 aggregate mem_ref_loc.
18030 (free_mem_ref_locs): Inline into ...
18031 (memref_free): ... this and adjust.
18032 (mem_ref_alloc): Adjust.
18033 (mem_ref_locs_alloc): Remove.
18034 (record_mem_ref_loc): Adjust.
18035 (get_all_locs_in_loop): Rewrite into ...
18036 (for_all_locs_in_loop): ... this iterator.
18037 (rewrite_mem_ref_loc): New functor.
18038 (rewrite_mem_refs): Use for_all_locs_in_loop.
18039 (sm_set_flag_if_changed): New functor.
18040 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
18041 (ref_always_accessed): New functor.
18042 (ref_always_accessed_p): Use for_all_locs_in_loop.
18044 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
18046 * tree-pass.h (PROP_gimple_lvec): New.
18047 * passes.c (dump_properties): Handle PROP_gimple_lvec.
18048 (init_optimization_passes): Move pass_lower_vector.
18049 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
18051 (pass_lower_vector): Provide PROP_gimple_lvec.
18052 (pass_lower_vector_ssa): Likewise.
18053 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
18055 2013-03-21 Mark Wielaard <mjw@redhat.com>
18057 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
18059 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
18061 * config/i386/i386.md (*movdi_internal): Disparage slightly
18062 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
18063 conversion alternatives.
18065 2013-03-21 Jakub Jelinek <jakub@redhat.com>
18067 PR middle-end/48087
18068 * diagnostic.def (DK_WERROR): New kind.
18069 * diagnostic.h (werrorcount): Define.
18070 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
18071 promoted to DK_ERROR, increment DK_WERROR counter instead of
18073 * toplev.c (toplev_main): Call print_ignored_options even if
18074 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
18075 even if just werrorcount is non-zero.
18078 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
18080 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
18081 (string_cst_pool_decl): New function.
18082 (optimize_one_addr_into_implicit_ptr): New function.
18083 (resolve_addr_in_expr): Optimize DWARF location expression
18084 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
18085 which doesn't live in memory, but has DW_AT_location or
18086 DW_AT_const_value, or refers to a string literal, into
18087 DW_OP_GNU_implicit_pointer.
18088 (optimize_location_into_implicit_ptr): New function.
18089 (resolve_addr): If removing DW_AT_location of a variable because
18090 it was DW_OP_addr of address of the variable, but the variable doesn't
18091 live in memory, try to emit const value attribute for the initializer.
18093 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
18095 * tree.h (VECTOR_TYPE_P): New macro.
18096 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
18097 TYPE_MODE): Use it.
18098 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
18099 VEC_COND_EXPR cannot be lvalues.
18100 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
18102 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
18104 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
18105 Restrict the transformation to equal modes.
18107 2013-03-21 Richard Biener <rguenther@suse.de>
18109 PR tree-optimization/39326
18110 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
18111 (MEM_ANALYZABLE): Adjust.
18112 (record_mem_ref_loc): Move bitmap ops ...
18113 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
18114 unanalyzable refs, do not record locations for it.
18115 (analyze_memory_references): Allocate ref zero as shared
18117 (refs_independent_p): Do not test for unanalyzed mems here.
18118 (ref_indep_loop_p_1): Special-case disambiguation against
18119 the unanalyzed ref.
18120 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
18122 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
18124 * config/arm/arm-protos.h (tune_params): Add
18125 prefer_neon_for_64bits field.
18126 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
18127 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
18128 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
18129 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
18130 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
18131 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
18132 (arm_option_override): Handle -mneon-for-64bits new option.
18133 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
18134 (prefer_neon_for_64bits): Declare new variable.
18135 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
18136 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
18137 (arch_enabled): Handle new arch types. Remove support for onlya8
18139 (one_cmpldi2): Use new arch names.
18140 (zero_extend<mode>di2, extend<mode>di2): Ditto.
18141 * config/arm/arm.opt (mneon-for-64bits): Add option.
18142 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
18143 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
18144 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
18146 * doc/invoke.texi (-mneon-for-64bits): Document.
18148 2013-03-21 Richard Biener <rguenther@suse.de>
18150 PR tree-optimization/39326
18151 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
18152 (sort_bbs_in_loop_postorder_cmp): New function.
18153 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
18155 2013-03-21 Richard Biener <rguenther@suse.de>
18157 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
18158 (vect_insert_into_interleaving_chain): Likewise.
18159 (vect_drs_dependent_in_basic_block): Inline ...
18160 (vect_slp_analyze_data_ref_dependence): ... here. New function,
18162 (vect_analyze_data_ref_dependence): ... here. Simplify.
18163 (vect_check_interleaving): Simplify.
18164 (vect_analyze_data_ref_dependences): Likewise. Split out ...
18165 (vect_slp_analyze_data_ref_dependences): ... this new function.
18166 (dr_group_sort_cmp): New function.
18167 (vect_analyze_data_ref_accesses): Compute data-reference groups
18168 here instead of in vect_analyze_data_ref_dependence. Use
18169 a more efficient algorithm.
18170 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
18171 vect_slp_analyze_data_ref_dependences. Call
18172 vect_analyze_data_ref_accesses earlier.
18173 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18174 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
18175 (vect_slp_analyze_data_ref_dependences): New prototype.
18177 2013-03-21 Richard Biener <rguenther@suse.de>
18179 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
18180 ref is stored in the loop.
18181 (find_refs_for_sm): Walk only over all stores.
18182 (store_motion_loop): Allocate from lim_bitmap_obstack.
18183 (store_motion): Likewise.
18185 2013-03-21 Richard Biener <rguenther@suse.de>
18187 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
18188 Update virtual SSA form.
18190 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18192 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
18193 * configure: Regenerate.
18194 * config.in: Regenerate.
18195 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
18196 if !HAVE_LD_EH_FRAME_CIEV3.
18198 2013-03-21 Richard Biener <rguenther@suse.de>
18200 * tree-cfg.c (verify_expr_no_block): New function.
18201 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
18202 nor DECL_VALUE_EXPR have locations with associated blocks.
18203 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
18204 (clear_unused_block_pointer): Remove code dealing with
18205 blocks in DECL_DEBUG_EXPR locations.
18207 2013-03-21 Richard Biener <rguenther@suse.de>
18209 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
18210 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
18211 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
18212 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
18213 instead of DECL_DEBUG_EXPR_IS_FROM.
18214 * gimplify.c (gimplify_modify_expr): Likewise.
18215 * tree-cfg.c (verify_expr_location_1): Likewise.
18216 * tree-complex.c (create_one_component_var): Likewise.
18217 * tree-sra.c (create_access_replacement): Likewise.
18218 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
18219 (clear_unused_block_pointer): Likewise.
18220 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
18221 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
18222 * var-tracking.c (var_debug_decl): Likewise.
18223 (track_expr_p): Likewise.
18224 * tree-inline.c (add_local_variables): Likewise. Set
18225 DECL_HAS_DEBUG_EXPR_P after copying it.
18226 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
18227 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
18229 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
18232 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
18233 * configure: Regenerate.
18234 * config.in: Regenerate.
18235 * config/i386/i386.md (*movdf_internal): Use
18236 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
18237 movd instead of movq mnemonic for interunit moves.
18238 (*movdi_internal): Ditto.
18240 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18242 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
18243 (abd<mode>_3): New pattern.
18244 (aba<mode>_3): New pattern.
18245 (fabd<mode>_3): New pattern.
18247 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18249 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
18250 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
18251 occurrence of REGISTER_PREFIX as its empty string.
18253 2013-03-20 Jeff Law <law@redhat.com>
18255 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
18256 addititional equivalences for equality comparisons between an SSA_NAME
18257 and a constant where the SSA_NAME was set from a widening conversion.
18259 2013-03-20 Walter Lee <walt@tilera.com>
18261 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
18263 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
18265 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
18266 depending on type attribute.
18267 (*movti_internal): Ditto.
18268 (*movtf_internal): Ditto.
18269 (*movxf_internal): Ditto.
18270 (*movdf_internal): Ditto.
18271 (*movsf_internal): Ditto.
18273 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
18275 * config/i386/i386.md (*movti_internal): Set prefix attribute to
18276 maybe_vex for sselog1 and ssemov types.
18277 (*movdi_internal): Reorder operand constraints.
18278 (*movsi_internal): Ditto. Set prefix attribute to
18279 maybe_vex for sselog1 and ssemov types.
18280 (*movtf_internal): Set prefix attribute to maybe_vex
18281 for sselog1 and ssemov types.
18282 (*movdf_internal): Ditto. Set prefix_data16 attribute for
18283 DImode ssemov types. Reorder operand constraints.
18284 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
18285 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
18286 attribute for SImode ssemov types. Reorder operand constraints.
18288 2013-03-20 Martin Jambor <mjambor@suse.cz>
18290 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
18291 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
18293 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
18295 * config/rs6000/predicates.md (indexed_address, update_address_mem
18296 update_indexed_address_mem): New predicates.
18297 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
18298 attribute for load/store instructions.
18299 * config/rs6000/dfp.md (movsd_store): Likewise.
18300 (movsd_load): Likewise.
18301 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
18302 (unnamed HI->DI extend define_insn): Likewise.
18303 (unnamed SI->DI extend define_insn): Likewise.
18304 (unnamed QI->SI extend define_insn): Likewise.
18305 (unnamed QI->HI extend define_insn): Likewise.
18306 (unnamed HI->SI extend define_insn): Likewise.
18307 (unnamed HI->SI extend define_insn): Likewise.
18308 (extendsfdf2_fpr): Likewise.
18309 (movsi_internal1): Likewise.
18310 (movsi_internal1_single): Likewise.
18311 (movhi_internal): Likewise.
18312 (movqi_internal): Likewise.
18313 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
18314 attribute for load/store instructions.
18315 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
18317 (mov<mode>_softfloat): Likewise.
18318 (mov<mode>_hardfloat32): Likewise.
18319 (mov<mode>_hardfloat64): Likewise.
18320 (mov<mode>_softfloat64): Likewise.
18321 (movdi_internal32): Likewise.
18322 (movdi_internal64): Likewise.
18323 (probe_stack_<mode>): Likewise.
18325 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18327 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
18328 floating point, and decimal floating point to reload iterator.
18330 * config/rs6000/constraints.md (wl constraint): New constraints to
18331 return FLOAT_REGS if certain options are used to reduce the number
18332 of separate patterns that exist in the file.
18333 (wx constraint): Likewise.
18334 (wz constraint): Likewise.
18336 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18337 -mdebug=reg, print wg, wl, wx, and wz constraints.
18338 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18339 Initialize the reload functions for 64-bit binary/decimal floating
18341 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18342 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18343 create the buffer on the stack to overcome not having a 32-bit
18345 (rs6000_emit_move): Likewise.
18346 (rs6000_secondary_memory_needed_rtx): Likewise.
18347 (rs6000_alloc_sdmode_stack_slot): Likewise.
18348 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18349 via xxlxor, just like DFmode 0.0.
18351 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
18352 define as 1 if we are running on a power7 or newer.
18353 (enum r6000_reg_class_enum): Add new constraints.
18355 * config/rs6000/dfp.md (movsd): Delete, combine with binary
18356 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
18357 with other moves by using conditional constraits (wg). Use LFIWZX
18358 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
18359 (movsd splitter): Likewise.
18360 (movsd_hardfloat): Likewise.
18361 (movsd_softfloat): Likewise.
18363 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
18364 binary and decimal floating point moves.
18365 (fmove_ok): New attributes to combine binary and decimal floating
18366 point moves, and to combine power6x (mfpgpr) moves along normal
18368 (real_value_to_target): Likewise.
18369 (f32_lr): Likewise.
18370 (f32_lm): Likewise.
18371 (f32_li): Likewise.
18372 (f32_sr): Likewise.
18373 (f32_sm): Likewise.
18374 (f32_si): Likewise.
18375 (movsf): Combine binary and decimal floating point moves. Combine
18376 power6x (mfpgpr) moves with other moves by using conditional
18377 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
18378 (mov<mode> for SFmode/SDmode); Likewise.
18379 (SFmode/SDmode splitters): Likewise.
18380 (movsf_hardfloat): Likewise.
18381 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
18382 (movsf_softfloat): Likewise.
18383 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
18385 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
18386 wx and wz constraints.
18388 * config/rs6000/constraints.md (wg constraint): New constraint to
18389 return FLOAT_REGS if -mmfpgpr (power6x) was used.
18391 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
18394 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18395 -mdebug=reg, print wg, wl, wx, and wz constraints.
18396 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18397 Initialize the reload functions for 64-bit binary/decimal floating
18399 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18400 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18401 create the buffer on the stack to overcome not having a 32-bit
18403 (rs6000_emit_move): Likewise.
18404 (rs6000_secondary_memory_needed_rtx): Likewise.
18405 (rs6000_alloc_sdmode_stack_slot): Likewise.
18406 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18407 via xxlxor, just like DFmode 0.0.
18409 * config/rs6000/dfp.md (movdd): Delete, combine with binary
18410 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
18411 with other moves by using conditional constraits (wg). Use LFIWZX
18412 and STFIWX for loading SDmode on power7.
18413 (movdd splitters): Likewise.
18414 (movdd_hardfloat32): Likewise.
18415 (movdd_softfloat32): Likewise.
18416 (movdd_hardfloat64_mfpgpr): Likewise.
18417 (movdd_hardfloat64): Likewise.
18418 (movdd_softfloat64): Likewise.
18420 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
18421 64-bit binary and decimal floating point moves.
18422 (FMOVE64X): Likewise.
18423 (movdf): Combine 64-bit binary and decimal floating point moves.
18424 Combine power6x (mfpgpr) moves with other moves by using
18425 conditional constraits (wg).
18426 (mov<mode> for DFmode/DDmode): Likewise.
18427 (DFmode/DDmode splitters): Likewise.
18428 (movdf_hardfloat32): Likewise.
18429 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
18430 (movdf_softfloat32): Likewise.
18431 (movdf_hardfloat64_mfpgpr): Likewise.
18432 (movdf_hardfloat64): Likewise.
18433 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
18434 (movdf_softfloat64): Likewise.
18435 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
18436 (reload_<mode>_load): Move to later in the file so they aren't in
18437 the middle of the floating point move insns.
18438 (reload_<mode>_store): Likewise.
18440 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
18443 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
18444 constraint if -mdebug=reg.
18445 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
18446 Enable using dd reload support if needed.
18448 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
18449 binary and decimal floating point moves in rs6000.md.
18450 (movtd_internal): Likewise.
18452 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
18453 decimal floating point moves.
18455 (movtf_internal): Likewise.
18456 (mov<mode>_internal, TDmode/TFmode): Likewise.
18457 (movtf_softfloat): Likewise.
18458 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
18460 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
18461 movdi_internal64, using wg constraint for move direct operations.
18462 (movdi_internal64): Likewise.
18464 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
18465 MODES_TIEABLE_P for selected modes. Print the numerical value of
18466 the various virtual registers. Use GPR/FPR first/last values,
18467 instead of hard coding the register numbers. Print which modes
18468 have reload functions registered.
18469 (rs6000_option_override_internal): If -mdebug=reg, trace the options
18470 settings before/after setting cpu, target and subtarget settings.
18471 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
18472 and for secondary reload failures in rs6000_secondary_reload_inner.
18473 (rs6000_secondary_reload_fail): Likewise.
18474 (rs6000_secondary_reload_inner): Likewise.
18476 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
18477 macros for first/last GPR and FPR registers.
18478 (LAST_GPR_REGNO): Likewise.
18479 (FIRST_FPR_REGNO): Likewise.
18480 (LAST_FPR_REGNO): Likewise.
18482 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
18483 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
18484 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
18485 (vcond<mode><mode>): Likewise.
18486 (vcondu<mode><mode>): Likewise.
18487 (vector_gtu<mode>): Likewise.
18488 (vector_gte<mode>): Likewise.
18489 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
18490 to prevent the compiler from converting DImode operations to TImode.
18491 (ior<mode>3): Likewise.
18492 (and<mode>3): Likewise.
18493 (one_cmpl<mode>2): Likewise.
18494 (nor<mode>3): Likewise.
18495 (andc<mode>3): Likewise.
18497 * config/rs6000/constraints.md (wt constraint): New constraint
18498 that returns VSX_REGS if TImode is allowed in VSX registers.
18500 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
18501 constant under VSX.
18503 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
18504 similar to TImode, but it is restricted to being in the GPRs.
18506 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
18507 TImode to occupy a single VSX register.
18509 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
18510 -mvsx-timode for power7/power8.
18511 (power7 cpu): Likewise.
18512 (power8 cpu): Likewise.
18514 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
18515 sure that TFmode/TDmode take up two registers if they are ever
18516 allowed in the upper VSX registers.
18517 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
18519 (rs6000_init_hard_regno_mode_ok): Likewise.
18520 (rs6000_debug_reg_global): Add debugging for PTImode and wt
18521 constraint. Print if LRA is turned on.
18522 (rs6000_option_override_internal): Give an error if -mvsx-timode
18523 and VSX is not enabled.
18524 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
18525 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
18526 to reg+offset addressing. Use PTImode when checking offset
18527 addresses for validity.
18528 (reg_offset_addressing_ok_p): Likewise.
18529 (rs6000_legitimate_offset_address_p): Likewise.
18530 (rs6000_legitimize_address): Likewise.
18531 (rs6000_legitimize_reload_address): Likewise.
18532 (rs6000_legitimate_address_p): Likewise.
18533 (rs6000_eliminate_indexed_memrefs): Likewise.
18534 (rs6000_emit_move): Likewise.
18535 (rs6000_secondary_reload): Likewise.
18536 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
18537 reloads to fpr registers to continue to use reg+offset addressing,
18538 but 64-bit reloads to altivec registers need reg+reg addressing.
18539 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
18540 it. Treat LO_SUM like a PLUS operation.
18541 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
18542 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
18543 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
18544 registers to share a register with a smaller sized type, since VSX
18545 puts scalars in the upper 64-bits.
18546 (print_operand): Add support for PTImode.
18547 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
18548 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
18549 registers, but don't have arithmetic support.
18550 (rs6000_memory_move_cost): Add test for VSX.
18551 (rs6000_opt_masks): Add -mvsx-timode.
18553 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
18556 (VSr): Use wt constraint for TImode.
18557 (VSv): Drop TImode support.
18558 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
18559 (vsx_movti_64bit): Likewise.
18560 (vsx_movti_32bit): Likewise.
18561 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
18562 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
18563 one '?' on the appropriate output constraint. Do not allow TImode
18564 logical operations on 32-bit systems.
18565 (vsx_ior<mode>3): Likewise.
18566 (vsx_xor<mode>3): Likewise.
18567 (vsx_one_cmpl<mode>2): Likewise.
18568 (vsx_nor<mode>3): Likewise.
18569 (vsx_andc<mode>3): Likewise.
18570 (vsx_concat_<mode>): Likewise.
18571 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
18573 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
18574 OPTION_MASK_VSX_TIMODE.
18575 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
18576 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
18578 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
18579 (TI2 iterator): New iterator for TImode, PTImode.
18580 (wd mode attribute): Add values for vector types.
18581 (movti_string): Replace TI move operations with operations for TImode
18582 and PTImode. Add support for TImode being allowed in VSX registers.
18583 (mov<mode>_string, TImode/PTImode): Likewise.
18584 (movti_ppc64): Likewise.
18585 (mov<mode>_ppc64, TImode/PTImode): Likewise.
18586 (TI mode splitters): Likewise.
18588 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
18591 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
18593 PR tree-optimization/56355
18594 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
18595 Also handle integers with undefined overflow.
18597 2013-03-20 Catherine Moore <clm@codesourcery.com>
18598 Maciej W. Rozycki <macro@codesourcery.com>
18599 Tom de Vries <tom@codesourcery.com>
18600 Nathan Sidwell <nathan@codesourcery.com>
18601 Iain Sandoe <iain@codesourcery.com>
18602 Nathan Froyd <froydnj@codesourcery.com>
18603 Chao-ying Fu <fu@mips.com>
18605 * doc/extend.texi: (micromips, nomicromips, nocompression):
18606 Document new function attributes.
18607 * doc/invoke.texi (minterlink-compressed, mmicromips,
18608 m14k, m14ke, m14kec): Document new options.
18609 (minterlink-mips16): Update documentation.
18610 * doc/md.texi (ZC, ZD): Document new constraints.
18611 * configure.ac (gcc_cv_as_micromips): Check if linker
18612 supports the .set micromips directive.
18613 * configure: Regenerate.
18614 * config.in: Regenerate.
18615 * config/mips/mips-tables.opt: Regenerate.
18616 * config/mips/micromips.md: New file.
18617 * constraints.md (ZC, ZD): New constraints.
18618 * config/mips/predicates.md (movep_src_register): New predicate.
18619 (movep_src_operand): New predicate.
18620 (non_volatile_mem_operand): New predicate.
18621 * config/mips/mips.md (multimem): New type.
18622 (length): Differentiate between 17-bit and 18-bit branch offsets.
18623 (MOVEP1, MOVEP2): New mode iterator.
18624 (mov_<load>l): Use ZC constraint.
18625 (mov_<load>r): Likewise.
18626 (mov_<store>l): Likewise.
18627 (mov_<store>r): Likewise.
18628 (*branch_equality<mode>_inverted): Add microMIPS support.
18629 (*branch_equality<mode>): Likewise.
18630 (*jump_absolute): Likewise.
18631 (indirect_jump_<mode>): Likewise.
18632 (tablejump_<mode>): Likewise.
18633 (<optab>_internal): Likewise.
18634 (sibcall_internal): Likewise.
18635 (sibcall_value_internal): Likewise.
18636 (prefetch): Use constraint ZD.
18637 * config/mips/mips.opt (minterlink-compressed): New option.
18638 (minterlink-mips16): Now an alias for minterlink-compressed.
18639 (mmicromips): New option.
18640 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18641 (compare_and_swap_12): Likewise.
18642 (sync_add<mode>): Likewise.
18643 (sync_<optab>_12): Likewise.
18644 (sync_old_<optab>_12): Likewise.
18645 (sync_new_<optab>_12): Likewise.
18646 (sync_nand_12): Likewise.
18647 (sync_old_nand_12): Likewise.
18648 (sync_new_nand_12): Likewise.
18649 (sync_sub<mode>): Likewise.
18650 (sync_old_add<mode>): Likewise.
18651 (sync_old_sub<mode>): Likewise.
18652 (sync_new_add<mode>): Likewise.
18653 (sync_new_sub<mode>): Likewise.
18654 (sync_<optab><mode>): Likewise.
18655 (sync_old_<optab><mode>): Likewise.
18656 (sync_new_<optab><mode>): Likewise.
18657 (sync_nand<mode>): Likewise.
18658 (sync_old_nand<mode>): Likewise.
18659 (sync_new_nand<mode>): Likewise.
18660 (sync_lock_test_and_set<mode>): Likewise.
18661 (test_and_set_12): Likewise.
18662 (atomic_compare_and_swap<mode>): Likewise.
18663 (atomic_exchange<mode>_llsc): Likewise.
18664 (atomic_fetch_add<mode>_llsc): Likewise.
18665 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18666 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18667 (umips_save_restore_pattern_p): Likewise.
18668 (umips_load_store_pair_p): Likewise.
18669 (umips_output_load_store_pair): Likewise.
18670 (umips_movep_target_p): Likewise.
18671 (umips_12bit_offset_address_p): Likewise.
18672 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18673 (mips_base_mips16): Rename this...
18674 (mips_base_compression_flags): ...to this. Update all uses.
18675 (mips_attribute_table): Add micromips, nomicromips and nocompression.
18676 (mips_mips16_decl_p): Delete.
18677 (mips_nomips16_decl_p): Delete.
18678 (mips_get_compress_on_flags): New function.
18679 (mips_get_compress_off_flags): New function.
18680 (mips_get_compress_mode): New function.
18681 (mips_get_compress_on_name): New function.
18682 (mips_get_compress_off_name): New function.
18683 (mips_insert_attributes): Support multiple compression types.
18684 (mips_merge_decl_attributes): Likewise.
18685 (umips_12bit_offset_address_p): New function.
18686 (mips_start_function_definition): Emit .set micromips directive.
18687 (mips_call_may_need_jalx_p): New function.
18688 (mips_function_ok_for_sibcall): Add microMIPS support.
18689 (mips_print_operand_punctuation): Support short delay slots and
18691 (umips_swm_mask, umips_swm_encoding): New.
18692 (umips_build_save_restore): New function.
18693 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18694 (was_mips16_p): Remove.
18695 (old_compression_mode): New.
18696 (mips_set_compression_mode): New function.
18697 (mips_set_current_function): Add microMIPS support.
18698 (mips_option_override): Likewise.
18699 (umips_save_restore_pattern_p): New function.
18700 (umips_output_save_restore): New function.
18701 (umips_load_store_pair_p_1): New function.
18702 (umips_load_store_pair_p): New function.
18703 (umips_output_load_store_pair_1): New function.
18704 (umips_output_load_store_pair): New function.
18705 (umips_movep_target_p) New function.
18706 (mips_prepare_pch_save): Add microMIPS support.
18707 * config/mips/mips.h (TARGET_COMPRESSION): New.
18708 (TARGET_CPU_CPP_BUILTINS): Update macro
18709 to use new compression flags and to support microMIPS.
18710 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18711 (MIPS_ARCH_FLOAT_SPEC): Likewise.
18712 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18713 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18714 (ASM_SPEC): Support mmicromips and mno-micromips.
18715 (M16STORE_REG_P): New macro.
18716 (MIPS_CALL): Support TARGET_MICROMIPS.
18717 (MICROMIPS_J): New macro.
18718 (mips_base_mips16): Rename this...
18719 (mips_base_compression_flags): ...to this.
18720 (UMIPS_12BIT_OFFSET_P): New macro.
18721 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18722 (MULTILIB_DIRNAMES): Likewise.
18723 2013-03-20 Richard Biener <rguenther@suse.de>
18725 PR tree-optimization/56661
18726 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
18727 the result does not have to be distinct.
18729 2013-03-20 Richard Biener <rguenther@suse.de>
18731 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
18734 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18735 Steven Bosscher <steven@gcc.gnu.org>
18737 PR rtl-optimization/56605
18738 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
18740 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
18743 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
18744 that require movd instead of movq.
18746 2013-03-20 Richard Biener <rguenther@suse.de>
18748 * tree-ssa-structalias.c (struct variable_info): Add pointer
18749 to the first field of an aggregate with sub-vars. Make
18750 this and the pointer to the next subfield its ID.
18751 (vi_next): New function.
18752 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
18753 storedanything_id, integer_id): Increment by one.
18754 (new_var_info, get_call_vi, lookup_call_clobber_vi,
18755 get_call_clobber_vi): Adjust.
18756 (solution_set_expand): Simplify and speedup.
18757 (solution_set_add): Inline into ...
18758 (set_union_with_increment): ... this. Adjust accordingly.
18759 (do_sd_constraint): Likewise.
18760 (do_ds_constraint): Likewise.
18761 (do_complex_constraint): Simplify.
18762 (build_pred_graph): Adjust.
18763 (solve_graph): Likewise. Simplify and speedup.
18764 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
18765 get_constraint_for_component_ref, get_constraint_for_1,
18766 first_vi_for_offset, first_or_preceding_vi_for_offset,
18767 create_function_info_for, create_variable_info_for_1,
18768 create_variable_info_for, intra_create_variable_infos): Adjust.
18769 (init_base_vars): Push NULL for ID zero.
18770 (compute_points_to_sets): Adjust.
18772 2013-03-20 Richard Biener <rguenther@suse.de>
18774 * cfgloop.c (verify_loop_structure): Streamline and avoid
18775 ICEing on corrupt loop tree.
18776 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
18779 2013-03-20 Richard Biener <rguenther@suse.de>
18781 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
18782 check whether an SSA update is needed.
18784 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
18786 * config/mips/constraints.md (T): Rename to...
18790 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
18791 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
18793 2013-03-19 Ian Bolton <ian.bolton@arm.com>
18795 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
18796 (*subsi3_carryin_uxtw): Likewise.
18798 2013-03-19 Ian Bolton <ian.bolton@arm.com>
18800 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
18801 (*rorsi3_insn_uxtw): Likewise.
18803 2013-03-19 Ian Bolton <ian.bolton@arm.com>
18805 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
18806 (*extrsi5_insn_uxtw): Likewise.
18808 2013-03-19 Richard Biener <rguenther@suse.de>
18810 PR tree-optimization/56273
18811 * passes.c (init_optimization_passes): Move second VRP after DOM.
18813 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
18815 * config/i386/i386.md (*movti_internal): Merge from
18816 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
18817 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
18818 nox64 isa attributes.
18820 2013-03-18 Richard Biener <rguenther@suse.de>
18822 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
18824 (merge_node_constraints): Likewise.
18825 (build_succ_graph): Likewise.
18826 (valid_graph_edge): Inline into single caller.
18827 (unify_nodes): Likewise. Use bitmap_set_bit return value
18829 (scc_visit): Fix formatting and variable use.
18830 (do_sd_constraint): Use gcc_checking_assert.
18831 (do_ds_constraint): Likewise.
18832 (do_complex_constraint): Likewise.
18833 (condense_visit): Likewise. Cleanup.
18834 (dump_pred_graph): New function.
18835 (perform_var_substitution): Dump the pred-graph before
18836 variable substitution.
18837 (find_equivalent_node): Use gcc_checking_assert.
18838 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
18840 2013-03-18 Richard Biener <rguenther@suse.de>
18842 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
18843 Remove cond_expr_stmt_list argument and do not gimplify the
18845 (vect_loop_versioning): Adjust.
18846 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
18847 Cleanup to use less temporaries.
18848 (vect_create_data_ref_ptr): Cleanup.
18850 2013-03-18 Jakub Jelinek <jakub@redhat.com>
18852 PR tree-optimization/56635
18853 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
18854 require types_compatible_p types.
18856 2013-03-18 Nick Clifton <nickc@redhat.com>
18858 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
18859 spurious backslash.
18861 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
18862 Add missing line to comment describing function.
18864 2013-03-18 Richard Biener <rguenther@suse.de>
18866 PR tree-optimization/56210
18867 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18868 Handle string / character search functions.
18869 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
18871 2013-03-18 Richard Biener <rguenther@suse.de>
18873 PR middle-end/56483
18874 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
18875 and implement properly.
18876 * gimple.h (gimple_cond_single_var_p): Remove.
18878 2013-03-18 Richard Biener <rguenther@suse.de>
18880 * tree-data-ref.h (find_data_references_in_loop): Declare.
18881 * tree-data-ref.c (get_references_in_stmt): Use a stack
18882 vector pre-allocated in the callers.
18883 (find_data_references_in_stmt): Adjust.
18884 (graphite_find_data_references_in_stmt): Likewise.
18885 (create_rdg_vertices): Likewise.
18886 (find_data_references_in_loop): Export.
18887 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
18888 Compute dependences here...
18889 (vect_analyze_data_refs): ...not here. When we encounter
18890 a non-vectorizable data reference in basic-block vectorization
18891 truncate the data reference vector. Do not bother to
18892 fixup data-dependence information for gather loads.
18893 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
18894 of data references, as reported.
18896 2013-03-18 Richard Biener <rguenther@suse.de>
18898 PR tree-optimization/3713
18899 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
18900 has_constants and expr.
18901 (stmt_has_constants): Properly valueize SSA names when deciding
18902 whether the stmt has constants.
18904 2013-03-18 Richard Biener <rguenther@suse.de>
18906 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
18907 whole function when there is nothing to do.
18908 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
18909 * tree-vectorizer.c (vectorize_loops): Update virtual and
18910 loop-closed SSA once.
18911 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
18913 2013-03-18 Richard Biener <rguenther@suse.de>
18915 PR middle-end/56113
18916 * domwalk.c (bb_postorder): New global static.
18917 (cmp_bb_postorder): New function.
18918 (walk_dominator_tree): Replace scheme imposing an order for
18919 visiting dominator sons by one sorting them at the time they
18920 are pushed on the stack.
18922 2013-03-18 Richard Biener <rguenther@suse.de>
18924 PR tree-optimization/39326
18925 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
18926 (struct mem_ref): Replace mem member with ao_ref typed member.
18927 (MEM_ANALYZABLE): Adjust.
18928 (memref_eq): Likewise.
18929 (mem_ref_alloc): Likewise.
18930 (gather_mem_refs_stmt): Likewise.
18931 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
18932 (execute_sm_if_changed_flag_set): Adjust.
18933 (execute_sm): Likewise.
18934 (ref_always_accessed_p): Likewise.
18935 (refs_independent_p): Likewise.
18936 (can_sm_ref_p): Likewise.
18938 2013-03-18 Jakub Jelinek <jakub@redhat.com>
18941 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
18942 return 1 even for !unsignedp.
18944 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
18946 * config/i386/i386.md (isa): Add x64 and nox64.
18947 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
18948 (*pushtf): Enable *roF alternative for x64 isa only.
18949 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
18950 mode attribute of integer alternatives to DImode for TARGET_64BIT.
18951 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
18952 (*movtf_internal): Merge from *movtf_internal_rex64 and
18953 *movtf_internal_sse. Use x64 and nox64 isa attributes.
18954 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
18955 nox64 isa attributes.
18956 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
18957 nox64 isa attributes.
18958 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
18960 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
18962 * config/alpha/alpha.c (TARGET_LRA_P): New define.
18964 2013-03-17 Jakub Jelinek <jakub@redhat.com>
18967 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
18968 class names. Remove trailing comma after "ALL_REGS".
18970 2013-03-16 Jan Hubicka <jh@suse.cz>
18972 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
18973 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
18974 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
18975 of cgraph_get_create_node.
18976 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18978 2013-03-16 Jason Merrill <jason@redhat.com>
18981 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
18982 with DW_AT_default_value.
18984 2013-03-16 Jakub Jelinek <jakub@redhat.com>
18986 * BASE-VER: Set to 4.9.0.
18988 2013-03-14 Andi Kleen <ak@linux.intel.com>
18991 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
18992 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
18993 Document _x* TSX intrinsics.
18995 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18996 David Holsgrove <david.holsgrove@xilinx.com>
18998 * configure.ac: Add MicroBlaze TLS support detection.
18999 * configure: Regenerate.
19000 * config/microblaze/microblaze-protos.h
19001 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
19002 symbol_mentioned_p, label_mentioned_p): Add prototypes.
19003 * config/microblaze/microblaze.c (microblaze_address_type): Add
19004 ADDRESS_TLS and tls_reloc address types.
19005 (microblaze_address_info): Add tls_reloc.
19006 (TARGET_HAVE_TLS): Define.
19007 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
19008 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
19009 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
19010 load_tls_operand, microblaze_call_tls_get_addr,
19011 microblaze_legitimize_tls_address): New functions.
19012 (microblaze_classify_unspec): Handle UNSPEC_TLS.
19013 (get_base_reg): Use microblaze_tls_symbol_p.
19014 (microblaze_classify_address): Handle TLS.
19015 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
19016 label_mentioned_p and microblaze_tls_referenced_p.
19017 (microblaze_legitimize_address): Handle TLS.
19018 (microblaze_address_insns): Handle ADDRESS_TLS.
19019 (pic_address_needs_scratch): Handle TLS.
19020 (print_operand_address): Handle TLS.
19021 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
19022 (microblaze_expand_move): Handle TLS.
19023 (microblaze_legitimate_constant_p): Check
19024 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
19025 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19026 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
19027 (PIC_OFFSET_TABLE_REGNUM): Set.
19028 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
19029 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
19030 (addsi3, movsi_internal2, movdf_internal): Update constraints
19031 * config/microblaze/predicates.md (arith_plus_operand): Define
19032 (move_operand): Redefine as move_src_operand,
19033 check microblaze_tls_referenced_p.
19035 2013-03-14 Ian Bolton <ian.bolton@arm.com>
19037 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
19038 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
19040 2013-03-14 Ian Bolton <ian.bolton@arm.com>
19042 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
19045 2013-03-14 Jakub Jelinek <jakub@redhat.com>
19047 PR tree-optimization/53265
19048 * common.opt (Waggressive-loop-optimizations): New option.
19049 * tree-ssa-loop-niter.c: Include tree-pass.h.
19050 (do_warn_aggressive_loop_optimizations): New function.
19051 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
19052 if number_of_latch_executions returned constant.
19053 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
19054 early. If number_of_latch_executions returned constant, set
19055 nb_iterations_upper_bound back to it.
19056 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
19058 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
19059 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
19061 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
19062 (MULTILIB_OSDIRNAMES): Set.
19063 * genmultilib: If defaultosdirname doesn't start with :: , set
19064 defaultosdirname2 instead, clear it and emit two . multilib_raw
19065 entries instead of just one.
19067 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
19069 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19070 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
19071 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19072 (SUBTARGET_OVERRIDE_OPTIONS): New.
19074 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
19077 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
19078 (musermode): Convert to Var(TARGET_USERMODE).
19079 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
19080 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
19081 * config/sh/sh.c (sh_option_override): Use
19082 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
19083 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
19085 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
19087 (udivsi3_i4_single, divsi3_i4_single): Use
19088 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
19090 2013-03-13 Dave Korn <dave.korn.cygwin@....>
19092 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
19095 2013-03-13 Richard Biener <rguenther@suse.de>
19097 PR tree-optimization/56608
19098 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
19099 calls when vectorizing basic-blocks.
19101 2013-03-13 Jakub Jelinek <jakub@redhat.com>
19104 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
19107 2013-03-12 Jakub Jelinek <jakub@redhat.com>
19109 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
19111 2013-03-11 Jan Hubicka <jh@suse.cz>
19114 * lto-streamer-out.c (output_symbol_p): Skip references from
19115 constructors of external variables.
19117 2013-03-11 Jan Hubicka <jh@suse.cz>
19119 PR middle-end/56571
19120 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
19122 * emit-rtl.c (verify_rtx_sharing): Likewise.
19123 (copy_insn_1): Likewise.
19124 * rtl.c (copy_rtx): Likewise.
19126 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
19129 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
19130 output_operand_lossage message.
19132 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
19135 * arm.c (shift_op): Validate RTL pattern on the fly.
19136 (arm_print_operand, case 'S'): Don't use shift_operator to validate
19139 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19142 * config/pa/pa.md (call_value): Check for calls to powf and direct to
19143 new call patterns that clobber %fr12.
19144 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
19145 split and postreload patterns.
19146 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
19147 registers %fr12 and %fr12R as call used.
19149 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
19151 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
19152 (canon_address, record_store, replace_read, check_mem_read_rtx,
19153 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
19154 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
19155 rest_of_handle_dse): Likewise.
19157 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
19159 PR middle-end/56524
19160 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
19162 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
19163 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
19164 (save_optabs_if_changed): Replace with...
19165 (init_tree_optimization_optabs): ...this.
19166 * optabs.c (save_optabs_if_changed): Rename to...
19167 (init_tree_optimization_optabs): ...this. Take the optimization node
19168 as argument. Do nothing if the base optabs are already correct.
19169 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
19170 to recompute optabs.
19171 * function.h (function): Remove optabs field.
19172 * function.c (invoke_set_current_function_hook): Call
19173 init_tree_optimization_optabs. Use the result to initialize
19176 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
19178 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
19179 if GTMA_HAS_NO_INSTRUMENTATION.
19180 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
19181 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
19182 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
19183 * gimple-pretty-print.c (dump_gimple_transaction): Handle
19184 GTMA_HAS_NO_INSTRUMENTATION.
19186 2013-03-08 Jakub Jelinek <jakub@redhat.com>
19188 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
19191 2013-03-08 Marek Polacek <polacek@redhat.com>
19192 Jakub Jelinek <jakub@redhat.com>
19194 PR tree-optimization/56478
19195 * predict.c (is_comparison_with_loop_invariant_p): Change the
19196 type of loop_step to tree.
19197 (predict_loops): Adjust.
19198 (predict_iv_comparison): Perform the computations on double_ints.
19200 2013-03-08 Richard Biener <rguenther@suse.de>
19202 PR tree-optimization/56570
19203 * tree-cfg.c (verify_expr_location_1): Verify locations for
19205 * tree-sra.c (create_access_replacement): Strip locations
19206 from DECL_DEBUG_EXPRs.
19208 2013-03-08 Richard Biener <rguenther@suse.de>
19210 * tree-inline.c (expand_call_inline): Do not associate
19211 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
19212 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
19214 2013-03-08 Richard Biener <rguenther@suse.de>
19216 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
19217 or block changes with -Og. Fix for location / block encoding
19218 changes and PHI arguments with locations.
19220 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
19222 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
19224 (struct output_info): Likewise.
19225 (register_overhead): Remove bad gcc_assert.
19226 (bitmap_find_bit): If there is only a single bitmap element, do not
19227 count a miss as a search.
19228 (print_statistics): Update for counter type changes.
19229 (dump_bitmap_statistics): Likewise. Print headers such that they
19230 are properly lined up with the printed counters.
19232 2013-03-07 Jakub Jelinek <jakub@redhat.com>
19234 PR tree-optimization/56559
19235 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
19236 check that it has only a single use.
19238 2013-03-07 Richard Biener <rguenther@suse.de>
19240 * doc/invoke.texi (fwhole-program): Discourage use in combination
19243 2013-03-06 Jakub Jelinek <jakub@redhat.com>
19245 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
19247 PR tree-optimization/56539
19248 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
19249 instead of GSI_CONTINUE_LINKING as last argument to
19250 force_gimple_operand_gsi. Adjust function comment.
19252 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
19255 PR middle-end/56548
19256 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
19257 promoted mode, convert the result back to the original mode.
19259 2013-03-06 Richard Biener <rguenther@suse.de>
19261 PR middle-end/56294
19262 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
19263 (insert_updated_phi_nodes_compare_uids): New function.
19264 (update_ssa): Sort symbols_to_rename after UID before
19265 traversing it to insert PHI nodes.
19267 2013-03-06 Richard Biener <rguenther@suse.de>
19269 PR middle-end/50494
19270 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
19271 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
19274 2013-02-13 Richard Biener <rguenther@suse.de>
19277 * varasm.c (output_constant_def_1): Get the decl representing
19278 the constant as argument.
19279 (output_constant_def): Wrap output_constant_def_1.
19280 (make_decl_rtl): Use output_constant_def_1 with the decl
19281 representing the constant.
19282 (build_constant_desc): Optionally re-use a decl already
19283 representing the constant.
19284 (tree_output_constant_def): Adjust.
19286 2013-03-06 Joey Ye <joey.ye@arm.com>
19289 * gcc.c (convert_white_space): New function.
19290 (main): Handles white space in function name.
19292 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
19295 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
19296 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
19297 to SH_DIV_CALL_TABLE for TARGET_SH2.
19298 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
19300 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
19301 call-table options.
19303 2013-03-05 Sterling Augustine <saugustine@google.com>
19304 Cary Coutant <ccoutant@google.com>
19307 * dwarf2out.c (resolve_addr): Don't call
19308 remove_loc_list_addr_table_entries a second time for the same
19311 2013-03-05 Jakub Jelinek <jakub@redhat.com>
19314 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
19315 (avoid_complex_debug_insns): New function.
19316 (expand_debug_locations): Call it.
19318 PR rtl-optimization/56484
19319 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
19320 lifetimes of hard registers on small register class machines.
19322 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
19324 * config/microblaze/microblaze-protos.h: Rename
19325 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
19326 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
19328 (microblaze_fast_interrupt_function_p): New function.
19329 (microblaze_is_interrupt_handler): Rename to
19330 microblaze_is_interrupt_variant and add fast_interrupt check.
19331 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
19332 (save_restore_insns): Likewise.
19333 (compute_frame_size): Likewise.
19334 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
19335 (microblaze_globalize_label): Likewise.
19336 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
19337 * config/microblaze/microblaze.md: Use wrapper
19338 microblaze_is_interrupt_variant.
19340 2013-03-05 Kai Tietz <ktietz@redhat.com>
19342 * sdbout.c (sdbout_one_type): Switch to current function's section
19343 supporting cold/hot.
19345 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
19347 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
19350 2013-03-05 Jakub Jelinek <jakub@redhat.com>
19352 PR middle-end/56461
19353 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
19354 if VALGRIND_GET_VBITS is defined, temporarily make object
19355 memory all defined, and restore previous valgrind addressability
19356 and definability afterwards. Free this_object at the end.
19358 PR middle-end/56461
19359 * lra.c (lra): Call lra_clear_live_ranges if live_p,
19360 right before calling lra_create_live_ranges, also call it
19361 when clearing live_p. Only call lra_clear_live_ranges
19362 at the end if live_p.
19364 PR middle-end/56461
19365 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
19367 2013-03-05 Richard Biener <rguenther@suse.de>
19369 PR tree-optimization/56521
19370 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
19373 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
19376 * except.h (remove_unreachable_eh_regions): New prototype.
19377 * except.c (remove_eh_handler_splicer): New function, split out
19378 of remove_eh_handler.
19379 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
19380 warning about running it on many EH regions one at a time.
19381 (remove_unreachable_eh_regions_worker): New function, walk the
19382 EH tree in depth-first order and remove non-marked regions.
19383 (remove_unreachable_eh_regions): New function.
19384 * tree-eh.c (mark_reachable_handlers): New function, split out
19385 from remove_unreachable_handlers.
19386 (remove_unreachable_handlers): Use mark_reachable_handlers and
19387 remove_unreachable_eh_regions.
19388 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
19389 and remove_unreachable_eh_regions.
19391 2013-03-05 Richard Biener <rguenther@suse.de>
19393 PR middle-end/56525
19394 * loop-init.c (fix_loop_structure): Remove loops in two stages,
19395 not freeing them until the end.
19397 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
19399 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
19401 2013-03-05 Richard Biener <rguenther@suse.de>
19403 PR tree-optimization/56270
19404 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
19405 of loads after scheduling an SLP instance.
19407 2013-03-05 Jakub Jelinek <jakub@redhat.com>
19409 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
19411 (check_gcc_parallelize): Run guality.exp as a separate job from
19412 vect.exp with unsorted.exp and $(dg_target_exps) separately from
19413 struct-layout-1.exp with stackalign.exp.
19415 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
19417 PR middle-end/56461
19418 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
19419 load_index sbitmap even if some bit in it isn't set.
19421 PR middle-end/56461
19422 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
19423 (discover_iteration_bound_by_body_walk): Change queues to
19424 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
19425 spelling in comment. Call safe_push on queues[bound_index] directly.
19426 Release queues[queue_index] in every iteration unconditionally.
19427 Release bounds vector.
19429 PR middle-end/56461
19430 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
19431 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
19432 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
19435 2013-03-05 Richard Biener <rguenther@suse.de>
19438 * tree-inline.c (remap_blocks_to_null): New function.
19439 (expand_call_inline): When expanding a call stmt without
19440 an associated block inline remap all callee blocks to NULL.
19442 2013-03-05 Jakub Jelinek <jakub@redhat.com>
19444 PR rtl-optimization/56494
19445 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
19446 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
19447 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
19449 PR middle-end/56461
19450 * sel-sched-ir.c (free_sched_pools): Release
19451 succs_info_pool.stack[succs_info_pool.max_top] vectors too
19452 if succs_info_pool.max_top isn't -1.
19455 * opts.c (opts_obstack, opts_concat): Moved to...
19456 * opts-common.c (opts_obstack, opts_concat): ... here.
19458 2013-03-04 Jakub Jelinek <jakub@redhat.com>
19460 PR middle-end/56461
19461 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
19463 2013-03-04 Martin Jambor <mjambor@suse.cz>
19465 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
19466 all appropriate places.
19468 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
19470 PR tree-optimization/56424
19471 * ipa-split.c (split_function): Do not set the RSO flag if result is
19472 not by reference and its type is a register type.
19474 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
19476 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
19477 (microblaze_legitimate_pic_operand): Likewise
19478 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
19479 new function microblaze_legitimate_pic_operand
19480 * config/microblaze/microblaze-protos.h
19481 (microblaze_legitimate_pic_operand): Declare.
19483 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19485 * config/microblaze/predicates.md (call_insn_simple_operand):
19486 New predicate for supported rtx code types.
19487 * config/microblaze/microblaze.md (call_internal1): Use
19488 call_insn_simple_operand predicate.
19490 2013-03-04 Jakub Jelinek <jakub@redhat.com>
19492 PR middle-end/56461
19493 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
19494 partitions.ordered_remove.
19496 PR middle-end/56461
19497 * tree-vect-stmts.c (vectorizable_conversion): Don't call
19498 vec_oprnds0.create (1) for modifier == NONE.
19500 PR middle-end/56461
19501 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
19502 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
19503 vec_oprnds1 right before pushing anything to it for
19506 PR middle-end/56461
19507 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
19508 set nbbs to 0 instead of having separate code path.
19509 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
19510 instead of false as last argument if returning NULL.
19512 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
19514 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
19515 the attribute is now called "target" instead of "option".
19516 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
19517 * doc/tm.texi.in (Target Attributes): Likewise document the correct
19518 attribute/pragma name for TARGET_OPTION_VALID_P and
19519 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
19520 * doc/tm.texi: Regenerated.
19522 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
19524 * config/microblaze/microblaze.c:
19525 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
19526 * config/microblaze/microblaze.h: Add -mxl-reorder to
19528 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
19529 instructions emitted if TARGET_REORDER.
19530 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
19531 or 0 for -m/-mno case, but initialises as 2 to detect default use case
19534 2013-03-01 Xinliang David Li <davidxl@google.com>
19536 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
19539 2013-03-01 Jakub Jelinek <jakub@redhat.com>
19541 PR middle-end/56461
19542 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
19543 vector even when returning true. Fix up function comment formatting.
19545 PR middle-end/56461
19546 * ira-build.c (ira_loop_nodes_count): New variable.
19547 (create_loop_tree_nodes): Initialize it.
19548 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
19550 PR middle-end/56461
19551 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
19552 method on dr_chain and result_chain.
19553 * tree-vect-stmts.c (vectorizable_store): Only call
19554 result_chain.create if j == 0.
19556 PR middle-end/56461
19557 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
19558 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
19559 before overwriting it.
19561 2013-03-01 Tobias Burnus <burnus@net-b.de>
19563 * doc/extended.texi (C Extensions): Change order in @menu
19565 (Other MIPS Built-in Functions): Move last MIPS entry before
19566 "picoChip Built-in Functions".
19567 (SH Built-in Functions): Move after RX Built-in Functions.
19568 * doc/gcc.texi (Introduction): Change order in @menu
19570 * doc/md.texi (Constraints): Ditto.
19571 * gty.texi (Type Information): Ditto.
19572 (User-provided marking routines for template types): Make
19574 * doc/invoke.texi (AArch64 Options): Move before
19575 "Adapteva Epiphany Options".
19577 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
19578 Jakub Jelinek <jakub@redhat.com>
19581 * asan.c (gate_asan): Lookup no_sanitize_address instead of
19582 no_address_safety_analysis attribute.
19583 * doc/extend.texi (no_address_safety_attribute): Rename to
19584 no_sanitize_address attribute, mention no_address_safety_analysis
19585 attribute as deprecated alias.
19587 2013-02-28 Jakub Jelinek <jakub@redhat.com>
19589 PR middle-end/56461
19590 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
19591 type to vec<vec<tree> > *.
19592 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
19593 to be vec<tree> instead of vec<tree> *, set vec_defs
19594 to vNULL and call vec_defs.create (number_of_vects), adjust other
19596 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
19597 vectorizable_condition): Adjust vect_get_slp_defs callers.
19599 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
19601 * config/aarch64/aarch64.c
19602 (aarch64_float_const_representable): Remove unused variable.
19604 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
19606 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
19608 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
19610 * config/aarch64/aarch64-builtins.c
19611 (aarch64_init_simd_builtins): Make static.
19613 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
19615 * config/aarch64/aarch64.c
19616 (aarch64_simd_make_constant): Make static.
19618 2013-02-28 Martin Jambor <mjambor@suse.cz>
19620 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
19621 with no initialization to the RHS of debug statements.
19623 2013-02-28 Martin Jambor <mjambor@suse.cz>
19625 PR tree-optimization/56294
19626 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
19628 (get_access_replacement): Do not call create_access_replacement.
19629 Assert a replacement exists.
19630 (get_repl_default_def_ssa_name): Create the replacement declaration
19633 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19635 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
19636 final_end_function.
19638 2013-02-28 Marek Polacek <polacek@redhat.com>
19640 PR rtl-optimization/56466
19641 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
19642 if we're changing a loop.
19643 (peel_loops_completely): Likewise.
19645 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
19648 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
19650 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
19653 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
19654 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
19655 INTX_FTYPE_FX, FX_FTYPE_INTX.
19656 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
19658 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
19660 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
19661 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
19662 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
19663 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
19664 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
19665 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
19666 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
19667 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
19668 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
19669 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
19670 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
19671 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
19672 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
19673 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
19674 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
19675 (avrxmega6): Increase max flash segments from 5 to 6.
19676 * config/avr/t-multilib: Regenerate.
19677 * config/avr/avr-tables.opt: Regenerate.
19678 * doc/avr-mmcu.texi: Regenerate.
19680 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
19682 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
19683 (avr_device_to_arch): Rename to avr_device_to_ld.
19684 (avr_device_to_as): New prototype.
19685 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
19686 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
19687 * config/avr/driver-avr.c (avr_device_to_as): New.
19688 (avr_device_to_arch): Rename to avr_device_to_ld.
19690 2013-02-27 Jakub Jelinek <jakub@redhat.com>
19692 PR middle-end/56461
19693 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
19694 method on dr_chain and result_chain.
19696 PR middle-end/56461
19697 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
19698 pointer_set_destroy on not_executed_last_iteration.
19700 PR middle-end/56461
19701 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
19703 PR middle-end/56461
19704 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
19705 FOR_EACH_DEFINED_FUNCTION when freeing state.
19707 PR middle-end/56461
19708 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
19710 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
19713 PR middle-end/56461
19714 * ipa-cp.c (decide_whether_version_node): Call vec_free on
19715 known_aggs[i].items and release known_aggs vector.
19717 PR middle-end/56461
19718 * ipa-reference.c (propagate): Free node_info even for alias nodes.
19720 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19722 * config/microblaze/microblaze.c (microblaze_emit_compare):
19723 Use xor for EQ/NE comparisions.
19724 * config/microblaze/microblaze.md (cstoresf4): Add constraints
19725 (cbranchsf4): Adjust operator to comparison_operator.
19727 2013-02-27 Jakub Jelinek <jakub@redhat.com>
19729 PR middle-end/56461
19730 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
19732 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
19733 vec_safe_push, always update *slot.
19734 (redirect_edge_var_map_clear): Use vec_free.
19735 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
19736 (free_var_map_entry): Use vec_free.
19737 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
19738 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
19740 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
19742 PR middle-end/45472
19743 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
19744 when the may_trap_p bit of the exprs being merged differs.
19745 Reorder tests for speculativeness in the logical and operator.
19747 2013-02-27 Jakub Jelinek <jakub@redhat.com>
19749 * incpath.c (add_standard_paths): Use reconcat instead of concat
19750 where appropriate and avoid leaking memory.
19752 * opts.h: Include obstack.h.
19753 (opts_concat): New prototype.
19754 (opts_obstack): New declaration.
19755 * opts.c (opts_concat): New function.
19756 (opts_obstack): New variable.
19757 (init_options_struct): Call gcc_init_obstack on opts_obstack.
19758 (finish_options): Use opts_concat instead of concat
19759 and XOBNEWVEC instead of XNEWVEC.
19760 * opts-common.c (generate_canonical_option, decode_cmdline_option,
19761 generate_option): Likewise.
19762 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
19763 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
19766 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
19767 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
19769 2013-02-26 Jakub Jelinek <jakub@redhat.com>
19771 PR middle-end/56461
19772 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
19774 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
19776 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
19777 (arm_block_move_unaligned_straight): Likewise.
19778 (arm_adjust_block_mem): Likewise.
19780 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
19783 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
19784 temp, cond and label.
19785 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
19788 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
19789 * config/c6x/c6x.h (dbx_register_map): Update declaration.
19792 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
19793 of prologue/epilogue functions.
19796 * config/tilegx/tilegx.c (tilegx_expand_prologue):
19797 Remove unused variable cfa_offset.
19798 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
19801 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
19802 type promotion to unsigned.
19805 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
19806 for HOST_WIDE_INT of 32 bit / same size as int.
19807 (arm_block_move_unaligned_straight): Likewise.
19808 (arm_adjust_block_mem): Likewise.
19811 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
19814 2013-02-26 Marek Polacek <polacek@redhat.com>
19816 PR tree-optimization/56426
19817 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
19819 2013-02-26 Richard Biener <rguenther@suse.de>
19822 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
19823 unused variable loops.
19825 2013-02-26 Jakub Jelinek <jakub@redhat.com>
19827 PR tree-optimization/56448
19828 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
19829 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
19830 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
19831 later operands of the references, or even first operand for
19832 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
19834 PR tree-optimization/56443
19835 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
19836 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
19837 to type_for_mode langhook.
19839 2013-02-25 Matt Turner <mattst88@gmail.com>
19841 * doc/invoke.texi: Document r4700.
19843 2013-02-25 Richard Biener <rguenther@suse.de>
19845 PR tree-optimization/56175
19846 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
19848 (simplify_bitwise_binary): ... here. Also guard the conversion
19849 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
19851 2013-02-25 Catherine Moore <clm@codesourcery.com>
19854 2013-02-24 Catherine Moore <clm@codesourcery.com>
19855 Maciej W. Rozycki <macro@codesourcery.com>
19856 Tom de Vries <tom@codesourcery.com>
19857 Nathan Sidwell <nathan@codesourcery.com>
19858 Iain Sandoe <iain@codesourcery.com>
19859 Nathan Froyd <froydnj@codesourcery.com>
19860 Chao-ying Fu <fu@mips.com>
19862 * doc/extend.texi: (micromips, nomicromips, nocompression):
19863 Document new function attributes.
19864 * doc/invoke.texi (minterlink-compressed, mmicromips,
19865 m14k, m14ke, m14kec): Document new options.
19866 (minterlink-mips16): Update documentation.
19867 * doc/md.texi (ZC, ZD): Document new constraints.
19868 * configure.ac (gcc_cv_as_micromips): Check if linker
19869 supports the .set micromips directive.
19870 * configure: Regenerate.
19871 * config.in: Regenerate.
19872 * config/mips/mips-tables.opt: Regenerate.
19873 * config/mips/micromips.md: New file.
19874 * constraints.md (ZC, AD): New constraints.
19875 * config/mips/predicates.md (movep_src_register): New predicate.
19876 (movep_src_operand): New predicate.
19877 (non_volatile_mem_operand): New predicate.
19878 * config/mips/mips.md (multimem): New type.
19879 (length): Differentiate between 17-bit and 18-bit branch offsets.
19880 (MOVEP1, MOVEP2): New mode iterator.
19881 (mov_<load>l): Use ZC constraint.
19882 (mov_<load>r): Likewise.
19883 (mov_<store>l): Likewise.
19884 (mov_<store>r): Likewise.
19885 (*branch_equality<mode>_inverted): Add microMIPS support.
19886 (*branch_equality<mode>): Likewise.
19887 (*jump_absolute): Likewise.
19888 (indirect_jump_<mode>): Likewise.
19889 (tablejump_<mode>): Likewise.
19890 (<optab>_internal): Likewise.
19891 (sibcall_internal): Likewise.
19892 (sibcall_value_internal): Likewise.
19893 (prefetch): Use constraint ZD.
19894 * config/mips/mips.opt (minterlink-compressed): New option.
19895 (minterlink-mips16): Now an alias for minterlink-compressed.
19896 (mmicromips): New option.
19897 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19898 (compare_and_swap_12): Likewise.
19899 (sync_add<mode>): Likewise.
19900 (sync_<optab>_12): Likewise.
19901 (sync_old_<optab>_12): Likewise.
19902 (sync_new_<optab>_12): Likewise.
19903 (sync_nand_12): Likewise.
19904 (sync_old_nand_12): Likewise.
19905 (sync_new_nand_12): Likewise.
19906 (sync_sub<mode>): Likewise.
19907 (sync_old_add<mode>): Likewise.
19908 (sync_old_sub<mode>): Likewise.
19909 (sync_new_add<mode>): Likewise.
19910 (sync_new_sub<mode>): Likewise.
19911 (sync_<optab><mode>): Likewise.
19912 (sync_old_<optab><mode>): Likewise.
19913 (sync_new_<optab><mode>): Likewise.
19914 (sync_nand<mode>): Likewise.
19915 (sync_old_nand<mode>): Likewise.
19916 (sync_new_nand<mode>): Likewise.
19917 (sync_lock_test_and_set<mode>): Likewise.
19918 (test_and_set_12): Likewise.
19919 (atomic_compare_and_swap<mode>): Likewise.
19920 (atomic_exchange<mode>_llsc): Likewise.
19921 (atomic_fetch_add<mode>_llsc): Likewise.
19922 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19923 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19924 (umips_save_restore_pattern_p): Likewise.
19925 (umips_load_store_pair_p): Likewise.
19926 (umips_output_load_store_pair): Likewise.
19927 (umips_movep_target_p): Likewise.
19928 (umips_12bit_offset_address_p): Likewise.
19929 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19930 (mips_base_mips16): Rename this...
19931 (mips_base_compression_flags): ...to this. Update all uses.
19932 (mips_attribute_table): Add micromips, nomicromips and nocompression.
19933 (mips_mips16_decl_p): Delete.
19934 (mips_nomips16_decl_p): Delete.
19935 (mips_get_compress_on_flags): New function.
19936 (mips_get_compress_off_flags): New function.
19937 (mips_get_compress_mode): New function.
19938 (mips_get_compress_on_name): New function.
19939 (mips_get_compress_off_name): New function.
19940 (mips_insert_attributes): Support multiple compression types.
19941 (mips_merge_decl_attributes): Likewise.
19942 (umips_12bit_offset_address_p): New function.
19943 (mips_start_function_definition): Emit .set micromips directive.
19944 (mips_call_may_need_jalx_p): New function.
19945 (mips_function_ok_for_sibcall): Add microMIPS support.
19946 (mips_print_operand_punctuation): Support short delay slots and
19948 (umips_swm_mask, umips_swm_encoding): New.
19949 (umips_build_save_restore): New function.
19950 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19951 (was_mips16_p): Remove.
19952 (old_compression_mode): New.
19953 (mips_set_compression_mode): New function.
19954 (mips_set_current_function): Add microMIPS support.
19955 (mips_option_override): Likewise.
19956 (umips_save_restore_pattern_p): New function.
19957 (umips_output_save_restore): New function.
19958 (umips_load_store_pair_p_1): New function.
19959 (umips_load_store_pair_p): New function.
19960 (umips_output_load_store_pair_1): New function.
19961 (umips_output_load_store_pair): New function.
19962 (umips_movep_target_p) New function.
19963 (mips_prepare_pch_save): Add microMIPS support.
19964 * config/mips/mips.h (TARGET_COMPRESSION): New.
19965 (TARGET_CPU_CPP_BUILTINS): Update macro
19966 to use new compression flags and to support microMIPS.
19967 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19968 (MIPS_ARCH_FLOAT_SPEC): Likewise.
19969 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19970 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19971 (ASM_SPEC): Support mmicromips and mno-micromips.
19972 (M16STORE_REG_P): New macro.
19973 (MIPS_CALL): Support TARGET_MICROMIPS.
19974 (MICROMIPS_J): New macro.
19975 (mips_base_mips16): Rename this...
19976 (mips_base_compression_flags): ...to this.
19977 (UMIPS_12BIT_OFFSET_P): New macro.
19978 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19979 (MULTILIB_DIRNAMES): Likewise.
19981 2013-02-25 Tom de Vries <tom@codesourcery.com>
19983 PR rtl-optimization/56131
19984 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
19985 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
19986 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
19988 2013-02-25 Tobias Burnus <burnus@net-b.de>
19990 * doc/invoke.texi (-fsanitize=): Move from optimization
19991 to debugging options.
19993 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
19995 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
19997 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
19998 Alexander Monakov <amonakov@ispras.ru>
20000 PR middle-end/56077
20001 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
20002 flush pending lists also on non-jumps. Adjust comment.
20004 2013-02-24 Catherine Moore <clm@codesourcery.com>
20005 Maciej W. Rozycki <macro@codesourcery.com>
20006 Tom de Vries <tom@codesourcery.com>
20007 Nathan Sidwell <nathan@codesourcery.com>
20008 Iain Sandoe <iain@codesourcery.com>
20009 Nathan Froyd <froydnj@codesourcery.com>
20010 Chao-ying Fu <fu@mips.com>
20012 * doc/extend.texi: (micromips, nomicromips, nocompression):
20013 Document new function attributes.
20014 * doc/invoke.texi (minterlink-compressed, mmicromips,
20015 m14k, m14ke, m14kec): Document new options.
20016 (minterlink-mips16): Update documentation.
20017 * doc/md.texi (ZC, ZD): Document new constraints.
20018 * configure.ac (gcc_cv_as_micromips): Check if linker
20019 supports the .set micromips directive.
20020 * configure: Regenerate.
20021 * config.in: Regenerate.
20022 * config/mips/mips-tables.opt: Regenerate.
20023 * config/mips/micromips.md: New file.
20024 * constraints.md (ZC, AD): New constraints.
20025 * config/mips/predicates.md (movep_src_register): New predicate.
20026 (movep_src_operand): New predicate.
20027 (non_volatile_mem_operand): New predicate.
20028 * config/mips/mips.md (multimem): New type.
20029 (length): Differentiate between 17-bit and 18-bit branch offsets.
20030 (MOVEP1, MOVEP2): New mode iterator.
20031 (mov_<load>l): Use ZC constraint.
20032 (mov_<load>r): Likewise.
20033 (mov_<store>l): Likewise.
20034 (mov_<store>r): Likewise.
20035 (*branch_equality<mode>_inverted): Add microMIPS support.
20036 (*branch_equality<mode>): Likewise.
20037 (*jump_absolute): Likewise.
20038 (indirect_jump_<mode>): Likewise.
20039 (tablejump_<mode>): Likewise.
20040 (<optab>_internal): Likewise.
20041 (sibcall_internal): Likewise.
20042 (sibcall_value_internal): Likewise.
20043 (prefetch): Use constraint ZD.
20044 * config/mips/mips.opt (minterlink-compressed): New option.
20045 (minterlink-mips16): Now an alias for minterlink-compressed.
20046 (mmicromips): New option.
20047 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
20048 (compare_and_swap_12): Likewise.
20049 (sync_add<mode>): Likewise.
20050 (sync_<optab>_12): Likewise.
20051 (sync_old_<optab>_12): Likewise.
20052 (sync_new_<optab>_12): Likewise.
20053 (sync_nand_12): Likewise.
20054 (sync_old_nand_12): Likewise.
20055 (sync_new_nand_12): Likewise.
20056 (sync_sub<mode>): Likewise.
20057 (sync_old_add<mode>): Likewise.
20058 (sync_old_sub<mode>): Likewise.
20059 (sync_new_add<mode>): Likewise.
20060 (sync_new_sub<mode>): Likewise.
20061 (sync_<optab><mode>): Likewise.
20062 (sync_old_<optab><mode>): Likewise.
20063 (sync_new_<optab><mode>): Likewise.
20064 (sync_nand<mode>): Likewise.
20065 (sync_old_nand<mode>): Likewise.
20066 (sync_new_nand<mode>): Likewise.
20067 (sync_lock_test_and_set<mode>): Likewise.
20068 (test_and_set_12): Likewise.
20069 (atomic_compare_and_swap<mode>): Likewise.
20070 (atomic_exchange<mode>_llsc): Likewise.
20071 (atomic_fetch_add<mode>_llsc): Likewise.
20072 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
20073 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
20074 (umips_save_restore_pattern_p): Likewise.
20075 (umips_load_store_pair_p): Likewise.
20076 (umips_output_load_store_pair): Likewise.
20077 (umips_movep_target_p): Likewise.
20078 (umips_12bit_offset_address_p): Likewise.
20079 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
20080 (mips_base_mips16): Rename this...
20081 (mips_base_compression_flags): ...to this. Update all uses.
20082 (mips_attribute_table): Add micromips, nomicromips and nocompression.
20083 (mips_mips16_decl_p): Delete.
20084 (mips_nomips16_decl_p): Delete.
20085 (mips_get_compress_on_flags): New function.
20086 (mips_get_compress_off_flags): New function.
20087 (mips_get_compress_mode): New function.
20088 (mips_get_compress_on_name): New function.
20089 (mips_get_compress_off_name): New function.
20090 (mips_insert_attributes): Support multiple compression types.
20091 (mips_merge_decl_attributes): Likewise.
20092 (umips_12bit_offset_address_p): New function.
20093 (mips_start_function_definition): Emit .set micromips directive.
20094 (mips_call_may_need_jalx_p): New function.
20095 (mips_function_ok_for_sibcall): Add microMIPS support.
20096 (mips_print_operand_punctuation): Support short delay slots and
20098 (umips_swm_mask, umips_swm_encoding): New.
20099 (umips_build_save_restore): New function.
20100 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
20101 (was_mips16_p): Remove.
20102 (old_compression_mode): New.
20103 (mips_set_compression_mode): New function.
20104 (mips_set_current_function): Add microMIPS support.
20105 (mips_option_override): Likewise.
20106 (umips_save_restore_pattern_p): New function.
20107 (umips_output_save_restore): New function.
20108 (umips_load_store_pair_p_1): New function.
20109 (umips_load_store_pair_p): New function.
20110 (umips_output_load_store_pair_1): New function.
20111 (umips_output_load_store_pair): New function.
20112 (umips_movep_target_p) New function.
20113 (mips_prepare_pch_save): Add microMIPS support.
20114 * config/mips/mips.h (TARGET_COMPRESSION): New.
20115 (TARGET_CPU_CPP_BUILTINS): Update macro
20116 to use new compression flags and to support microMIPS.
20117 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
20118 (MIPS_ARCH_FLOAT_SPEC): Likewise.
20119 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
20120 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
20121 (ASM_SPEC): Support mmicromips and mno-micromips.
20122 (M16STORE_REG_P): New macro.
20123 (MIPS_CALL): Support TARGET_MICROMIPS.
20124 (MICROMIPS_J): New macro.
20125 (mips_base_mips16): Rename this...
20126 (mips_base_compression_flags): ...to this.
20127 (UMIPS_12BIT_OFFSET_P): New macro.
20128 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
20129 (MULTILIB_DIRNAMES): Likewise.
20131 2013-02-24 Jakub Jelinek <jakub@redhat.com>
20134 * target-globals.c (save_target_globals): For init_reg_sets and
20135 target_reinit remporarily set this_fn_optabs to this_target_optabs.
20137 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
20139 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
20140 * config/aarch64/t-aarch64
20141 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
20143 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
20145 PR inline-asm/56148
20146 * lra-constraints.c (process_alt_operands): Reload operand
20147 conflicting with earlier clobber only if no more other conflicting
20150 2013-02-22 Jakub Jelinek <jakub@redhat.com>
20153 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
20154 if not linking a shared library.
20156 2013-02-22 Seth LaForge <sethml@google.com>
20158 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
20160 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
20162 * config/arm/arm.md (split for extendsidi): Update condition.
20163 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
20164 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
20165 (qhs_zextenddi_cstr): Likewise.
20167 2013-02-21 Jakub Jelinek <jakub@redhat.com>
20169 PR middle-end/56420
20170 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
20171 avoid signed wrapping.
20172 (expand_mult): Handle properly multiplication by
20173 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
20174 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
20175 in the compiler if coeff is HOST_WIDE_INT_MIN.
20176 (expand_divmod): Don't make ext_op1 static, change it's type to
20177 uhwi. Avoid undefined behavior in -INTVAL (op1).
20179 PR rtl-optimization/50339
20180 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
20182 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
20183 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
20184 into splitting_ashiftrt field.
20185 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
20187 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
20190 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
20192 PR middle-end/56108
20193 * trans-mem.c (execute_tm_mark): Do not expand transactions that
20194 are sure to go irrevocable.
20196 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
20198 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
20199 scalars are valid operands.
20201 2013-02-21 Martin Jambor <mjambor@suse.cz>
20203 PR tree-optimization/56310
20204 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
20205 only matching indices and non-negative final offsets.
20206 (intersect_aggregates_with_edge): Pass src_idx to
20207 agg_replacements_to_vector. Pass src_idx insstead of index to
20208 intersect_with_agg_replacements.
20210 2013-02-21 Martin Jambor <mjambor@suse.cz>
20212 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
20213 instead of hard-wired defaults.
20215 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
20217 * doc/invoke.texi (MIPS Options): Update documentation of the
20218 floating-point multiply-accumulate instruction restrictions.
20220 2013-02-21 Kostya Serebryany <kcc@google.com>
20222 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
20223 asan_shadow_offset on x86_64 linux.
20225 2013-02-21 Richard Biener <rguenther@suse.de>
20227 PR tree-optimization/56415
20229 2013-02-11 Richard Biener <rguenther@suse.de>
20231 PR tree-optimization/56273
20232 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20235 2013-02-21 Jakub Jelinek <jakub@redhat.com>
20238 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
20241 PR inline-asm/56405
20242 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
20243 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
20245 2013-02-20 Jan Hubicka <jh@suse.cz>
20247 PR tree-optimization/56265
20248 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
20249 when target is referenced for first time.
20251 2013-02-20 Richard Biener <rguenther@suse.de>
20253 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
20254 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
20255 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
20256 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
20257 not return anything.
20258 (rename_ssa_copies): Do not remove unused locals.
20259 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
20260 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
20261 * passes.c (execute_function_todo): Do not schedule unused locals
20262 removal if cleanup_tree_cfg did something.
20263 * tree-ssa-live.c (remove_unused_locals): Dump statistics
20264 about the number of removed locals.
20266 2013-02-20 Richard Biener <rguenther@suse.de>
20268 PR tree-optimization/56398
20269 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
20271 2013-02-20 Martin Jambor <mjambor@suse.cz>
20273 PR tree-optimization/55334
20274 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
20275 restricted pointers to arrays.
20277 2013-02-20 Richard Biener <rguenther@suse.de>
20278 Jakub Jelinek <jakub@redhat.com>
20280 PR tree-optimization/56396
20281 * tree-ssa-ccp.c (n_const_val): New static variable.
20282 (get_value): Return NULL for SSA names we don't have a lattice
20284 (ccp_initialize): Initialize n_const_val.
20285 * tree-ssa-copy.c (n_copy_of): New static variable.
20286 (init_copy_prop): Initialize n_copy_of.
20287 (get_value): Return NULL_TREE for SSA names we don't have a
20290 2013-02-20 Martin Jambor <mjambor@suse.cz>
20292 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
20294 2013-02-20 Richard Biener <rguenther@suse.de>
20296 * genpreds.c (write_lookup_constraint): Do not compare first
20297 letter of the constraint again.
20299 2013-02-20 Richard Biener <rguenther@suse.de>
20301 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
20303 (get_use_iv_cost): Terminate hashtable walk when coming across
20306 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
20308 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
20309 reassociation for avx2 targets.
20311 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20313 * config/microblaze/microblaze.c: microblaze_has_clz = 0
20314 Add version check for v8.10.a to enable microblaze_has_clz
20315 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
20316 version and TARGET_PATTERN_COMPARE check
20317 * config/microblaze/microblaze.md: New clzsi2 instruction
20319 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20321 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
20322 function before branching.
20324 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
20326 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
20328 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
20330 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
20332 PR middle-end/55889
20333 * sel-sched.c: Include ira.h.
20334 (implicit_clobber_conflict_p): New function.
20335 (moveup_expr): Use it.
20336 * Makefile.in (sel-sched.o): Depend on ira.h.
20338 2013-02-19 Richard Biener <rguenther@suse.de>
20340 PR tree-optimization/56384
20341 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
20342 (vn_hash_type): Split out from ...
20343 (vn_hash_constant_with_type): ... here.
20344 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
20345 (vn_phi_eq): Compare types from vn_phi_s structure.
20346 (vn_phi_lookup): Populate vn_phi_s type.
20347 (vn_phi_insert): Likewise.
20349 2013-02-19 Jakub Jelinek <jakub@redhat.com>
20351 PR tree-optimization/56350
20352 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
20353 if haven't found reduction or nested cycle operand, rather than
20354 asserting we must find it.
20356 PR tree-optimization/56381
20357 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
20360 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
20361 Jakub Jelinek <jakub@redhat.com>
20364 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
20365 (swap_optab_enable): Same.
20366 (init_all_optabs): Use argument instead of global.
20367 * tree.h (struct tree_optimization_option): New field target_optabs.
20368 * expr.h (init_all_optabs): Add argument to prototype.
20369 (TREE_OPTIMIZATION_OPTABS): New.
20370 (save_optabs_if_changed): Protoize.
20371 * optabs.h: Declare this_fn_optabs.
20372 * optabs.c (save_optabs_if_changed): New.
20373 Declare this_fn_optabs.
20374 (init_optabs): Add argument to init_all_optabs() call.
20375 * function.c (invoke_set_current_function_hook): Handle per
20377 * function.h (struct function): New field optabs.
20378 * config/mips/mips.c (mips_set_mips16_mode): Handle when
20379 optimization_current_node has changed.
20380 * target-globals.h (save_target_globals_default_opts): Protoize.
20381 * target-globals.c (save_target_globals_default_opts): New.
20383 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
20386 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
20387 registers %fr12 and %fr12R as call used.
20390 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
20391 and HImode, require all displacements to be an integer multiple of
20393 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
20394 only allow QImode and HImode when reload is in progress and strict is
20395 true. Likewise for symbolic addresses. Use base14_operand to check
20396 displacements in REG+BASE addresses.
20398 2013-02-18 Richard Biener <rguenther@suse.de>
20400 PR tree-optimization/56366
20401 * tree-vect-loop.c (get_initial_def_for_induction): Properly
20402 handle sign-conversion of outer-loop initial induction value.
20404 2013-02-18 Richard Biener <rguenther@suse.de>
20406 PR middle-end/56349
20407 * cfghooks.c (merge_blocks): If we merge a latch into another
20408 block adjust references to it.
20409 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
20410 (verify_loop_structure): Verify that a recorded latch is in fact
20413 2013-02-18 Richard Biener <rguenther@suse.de>
20415 PR tree-optimization/56321
20416 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
20417 order SSA name release and virtual operand unlinking.
20419 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20421 * config/microblaze/microblaze.md (save_stack_block): Define.
20422 (restore_stack_block): Likewise.
20424 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20426 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
20427 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
20428 * config/microblaze/microblaze.c (microblaze_option_override):
20429 Bail out early for PIC modes when target does not support PIC.
20431 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20433 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
20434 Replace with a microblaze version.
20435 (microblaze_trampoline_init): Adapt for microblaze.
20436 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
20439 2013-02-16 Jakub Jelinek <jakub@redhat.com>
20440 Dodji Seketeli <dodji@redhat.com>
20443 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
20444 (instrument_mem_region_access): Do not forget to always put
20445 instrumentation of the of 'base' and 'base + len' in a "if (len !=
20446 0) statement, even for cases where either 'base' or 'base + len'
20447 are not instrumented -- because they have been previously
20448 instrumented. Simplify the logic by putting all the statements
20449 instrument 'base + len' inside a sequence, and then insert that
20450 sequence right before the current insertion point. Then, to
20451 instrument 'base + len', just get an iterator on that statement.
20452 And do not forget to update the pointer to iterator the function
20453 received as argument.
20455 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
20457 PR rtl-optimization/56348
20458 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
20460 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
20462 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
20463 (clean_graph_dump_file): Pass base to start_graph_dump.
20465 2013-02-14 Richard Henderson <rth@redhat.com>
20468 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
20470 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
20472 * collect2-aix.h: Define F_LOADONLY.
20474 2013-02-14 Richard Biener <rguenther@suse.de>
20477 * varasm.c (output_constant_def_1): Get the decl representing
20478 the constant as argument.
20479 (output_constant_def): Wrap output_constant_def_1.
20480 (make_decl_rtl): Use output_constant_def_1 with the decl
20481 representing the constant.
20482 (build_constant_desc): Optionally re-use a decl already
20483 representing the constant.
20484 (tree_output_constant_def): Adjust.
20486 2013-02-14 Dodji Seketeli <dodji@redhat.com>
20489 * asan.c (instrument_builtin_call): Really put the length of the
20490 second source argument into src1_len.
20492 2013-02-13 Jakub Jelinek <jakub@redhat.com>
20494 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
20495 argument. If it is false, don't create edge from then_bb to
20497 (insert_if_then_before_iter): Pass true to it.
20498 (build_check_stmt): Pass false to it.
20499 (transform_statements): Flush hash table only on extended basic
20500 block boundaries, rather than at the beginning of every bb.
20501 Don't flush hash table on nonfreeing_call_p calls.
20502 * tree-flow.h (nonfreeing_call_p): New prototype.
20503 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
20505 2013-02-13 David S. Miller <davem@davemloft.net>
20507 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
20509 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
20512 * ira.c (max_regno_before_ira): Move from ...
20514 (fix_reg_equiv_init): Use max_regno_before_ira instead of
20517 2013-02-13 Jakub Jelinek <jakub@redhat.com>
20519 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
20521 2013-02-13 Richard Biener <rguenther@suse.de>
20524 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
20525 globals in MEM_REFs.
20527 2013-02-13 Richard Biener <rguenther@suse.de>
20529 * loop-init.c (loop_optimizer_init): Clear loop state when
20530 re-initializing preserved loops.
20531 * loop-unswitch.c (unswitch_single_loop): Return whether
20532 we unswitched the loop. Do not verify loop state here.
20533 (unswitch_loops): When we unswitched a loop discover new loops.
20535 2013-02-13 Kostya Serebryany <kcc@google.com>
20537 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
20539 * sanitizer.def: Rename __asan_init to __asan_init_v1.
20541 2013-02-12 Dodji Seketeli <dodji@redhat.com>
20543 Avoid instrumenting duplicated memory access in the same basic block
20544 * Makefile.in (asan.o): Add new dependency on hash-table.h
20545 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
20546 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
20547 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
20548 (free_mem_ref_resources, has_mem_ref_been_instrumented)
20549 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
20550 (get_mem_ref_of_assignment): New functions.
20551 (get_mem_refs_of_builtin_call): Extract from
20552 instrument_builtin_call and tweak a little bit to make it fit with
20554 (instrument_builtin_call): Use the new
20555 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
20556 of is_gimple_builtin_call.
20557 (instrument_derefs, instrument_mem_region_access): Insert the
20558 instrumented memory reference into the hash table.
20559 (maybe_instrument_assignment): Renamed instrument_assignment into
20560 this, and change it to advance the iterator when instrumentation
20561 actually happened and return true in that case. This makes it
20562 homogeneous with maybe_instrument_assignment, and thus give a
20563 chance to callers to be more 'regular'.
20564 (transform_statements): Clear the memory reference hash table
20565 whenever we enter a new BB, when we cross a function call, or when
20566 we are done transforming statements. Use
20567 maybe_instrument_assignment instead of instrumentation. No more
20568 need to special case maybe_instrument_assignment and advance the
20569 iterator after calling it; it's now handled just like
20570 maybe_instrument_call. Update comment.
20572 2013-02-13 Richard Biener <rguenther@suse.de>
20574 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
20575 Fix loop discovery code.
20577 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
20579 PR inline-asm/56148
20580 * lra-constraints.c (process_alt_operands): Match early clobber
20581 operand with itself. Check conflicts with earlyclobber only if
20582 the operand is not reloaded. Prefer to reload conflicting operand
20583 if earlyclobber and matching operands are the same.
20585 2013-02-12 Richard Biener <rguenther@suse.de>
20588 * lto-streamer-out.c (write_symbol): Do not output symbols
20589 for hard register variables.
20591 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
20594 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
20595 (umulsidi3_insn, mulsidi3_insn): New insns.
20597 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
20599 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
20600 (struct tune_params): Add vec_costs field.
20601 * config/arm/arm.c (arm_builtin_vectorization_cost)
20602 (arm_add_stmt_cost): New functions.
20603 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
20604 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
20605 (arm_default_vec_cost): New struct of type cpu_vec_costs.
20606 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
20607 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
20608 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
20609 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
20611 2013-02-12 Richard Biener <rguenther@suse.de>
20614 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
20615 decls again if possible.
20617 2013-02-12 Richard Biener <rguenther@suse.de>
20619 PR middle-end/56288
20620 * tree-ssa.c (verify_ssa_name): Fix check, move
20621 SSA_NAME_IN_FREE_LIST check up.
20623 2013-02-12 Jakub Jelinek <jakub@redhat.com>
20624 Steven Bosscher <steven@gcc.gnu.org>
20626 PR rtl-optimization/56151
20627 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
20628 equal to op0 or op1, and last_insn pattern is CODE operation
20629 with MEM dest and one of the operands matches that MEM.
20631 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
20633 * doc/extend.texi: Document Function Multiversioning and "default"
20634 parameter string to target attribute.
20635 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
20636 target attribute parameter is "default".
20637 (ix86_compare_version_priority): Remove checks for target attribute.
20638 (ix86_mangle_function_version_assembler_name): Change error to sorry.
20639 Remove check for target attribute equal to NULL. Add assert.
20640 (ix86_generate_version_dispatcher_body): Change error to sorry.
20642 2013-02-11 Iain Sandoe <iain@codesourcery.com>
20643 Jack Howarth <howarth@bromo.med.uc.edu>
20644 Patrick Marlier <patrick.marlier@gmail.com>
20647 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
20648 define ENDFILE_SPEC as TM_DESTRUCTOR.
20649 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
20651 2013-02-11 Alexander Potapenko <glider@google.com>
20652 Jack Howarth <howarth@bromo.med.uc.edu>
20653 Jakub Jelinek <jakub@redhat.com>
20656 * config/darwin.c (cdtor_record): Rename ctor_record.
20657 (sort_cdtor_records): Rename sort_ctor_records.
20658 (finalize_dtors): New routine to sort destructors by
20659 priority before use in assemble_integer.
20660 (machopic_asm_out_destructor): Use finalize_dtors if needed.
20662 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
20664 PR rtl-optimization/56275
20665 * simplify-rtx.c (avoid_constant_pool_reference): Check that
20666 offset is non-negative and less than cmode size before
20667 calling simplify_subreg.
20669 2013-02-11 Richard Biener <rguenther@suse.de>
20671 PR tree-optimization/56264
20672 * cfgloop.h (fix_loop_structure): Adjust prototype.
20673 * loop-init.c (fix_loop_structure): Return the number of
20674 newly discovered loops.
20675 * tree-cfgcleanup.c (repair_loop_structures): When new loops
20676 are discovered, do a full loop-closed SSA rewrite.
20678 2013-02-11 Richard Biener <rguenther@suse.de>
20680 PR tree-optimization/56273
20681 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20683 (check_array_ref): Fix missing newline in dumps.
20684 (search_for_addr_array): Likewise.
20686 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
20688 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
20690 2013-02-09 Jakub Jelinek <jakub@redhat.com>
20693 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
20695 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
20697 PR rtl-optimization/56246
20698 * lra-constraints.c (simplify_operand_subreg): Try to reuse
20700 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
20701 constraints are satisfied.
20703 2013-02-08 Jeff Law <law@redhat.com>
20706 * emit-rtl.c (reg_is_parm_p): New function.
20707 * regs.h (reg_is_parm_p): New prototype.
20708 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
20709 callee-clobbered registers.
20711 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
20714 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
20715 If there is no implicit builtin declaration, just return NULL.
20717 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
20719 * config/i386/sse.md (FMAMODEM): New mode iterator.
20720 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
20721 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
20723 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
20725 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
20726 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
20727 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
20729 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20731 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
20732 (microblaze*-*-elf): Likewise.
20733 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
20735 * config/microblaze/microblaze-c.c: Add builtin defines for
20736 _LITTLE_ENDIAN and _BIG_ENDIAN.
20737 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
20738 add to TARGET_DEFAULT flags.
20739 Expand ASM_SPEC and LINK_SPEC.
20740 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
20741 * config/microblaze/microblaze.md: Update extendsidi2 and
20742 movdi_internal instructions to use low-order / high-order reg
20744 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
20745 options and inversemask / mask of LITTLE_ENDIAN.
20746 * config/microblaze/t-microblaze: Expand multilib options to
20747 include mlittle-endian (le) and update exceptions patterns.
20749 2013-02-08 Jakub Jelinek <jakub@redhat.com>
20751 PR rtl-optimization/56195
20752 * lra-constraints.c (get_reload_reg): Don't reuse regs
20753 if they have smaller mode than requested, if they have
20754 wider mode than requested, try to return a SUBREG.
20756 PR tree-optimization/56250
20757 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
20758 if type is unsigned and code isn't MULT_EXPR.
20760 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
20762 PR tree-optimization/56064
20763 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
20764 bits according to mode.
20765 * fixed-value.h (fixed_from_double_int)
20766 (const_fixed_from_double_int): Adjust comments.
20768 2013-02-08 Richard Biener <rguenther@suse.de>
20771 * lto-streamer.h (struct data_in): Remove current_file, current_line
20772 and current_col members.
20773 * lto-streamer-out.c (lto_output_location): Stream changed bits
20774 en-block for efficiency.
20775 * lto-streamer-in.c (clear_line_info): Remove.
20776 (lto_input_location): Cache current file, line and column
20777 globally via local statics. Read changed bits en-block.
20778 (input_function): Do not call clear_line_info.
20779 (lto_read_body): Likewise.
20780 (lto_input_toplevel_asms): Likewise.
20782 2013-02-08 Michael Matz <matz@suse.de>
20784 PR tree-optimization/52448
20785 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
20786 (nt_call_phase): New static.
20787 (add_or_mark_expr): Only mark accesses with newer phase than any
20789 (nonfreeing_call_p): New.
20790 (nt_init_block): Update nt_call_phase, mark blocks as visited.
20791 (nt_fini_block): Keep blocks marked as visited.
20792 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
20794 2013-02-08 Richard Biener <rguenther@suse.de>
20796 * ira.c (ira): Free broken dominator information.
20798 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
20800 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
20802 2013-02-08 Marek Polacek <polacek@redhat.com>
20804 * cfgloop.c (verify_loop_structure): Add more checking of headers.
20806 2013-02-08 Richard Biener <rguenther@suse.de>
20808 PR middle-end/56181
20809 * cfgloop.h (flow_loops_find): Adjust.
20810 (bb_loop_header_p): Declare.
20811 * cfgloop.c (bb_loop_header_p): New function split out from ...
20812 (flow_loops_find): ... here. Adjust function signature,
20813 support incremental loop structure update.
20814 (verify_loop_structure): Cleanup. Verify a loop is a loop.
20815 * cfgloopmanip.c (fix_loop_structure): Move ...
20816 * loop-init.c (fix_loop_structure): ... here.
20817 (apply_loop_flags): Split out from ...
20818 (loop_optimizer_init): ... here.
20819 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
20820 in incremental mode, only remove dead loops here.
20822 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
20825 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
20826 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
20827 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
20828 (*round<mode>3.libgcc): New insns for fixed-modes.
20829 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
20830 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
20831 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
20832 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
20833 implementations. Define to __builtin_avr_absFX,
20834 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
20835 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
20836 __builtin_avr_countlsFX, respectively.
20837 * config/avr/avr-c.c (target.h): Include it.
20838 (enum avr_builtin_id): New enum.
20839 (avr_resolve_overloaded_builtin): New static function.
20840 (avr_register_target_pragmas): Use it to set
20841 targetm.resolve_overloaded_builtin.
20842 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
20843 tree nodes used by DEF_BUILTIN.
20844 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
20845 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
20846 <AVR_BUILTIN_xxBITS>: Same.
20848 2013-02-08 Richard Biener <rguenther@suse.de>
20850 * cfgloop.c (verify_loop_structure): Properly handle
20851 a loop exiting to another loop header.
20852 * ira-int.h (ira_loops): Remove.
20853 * ira.c (ira_loops): Remove.
20854 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
20855 (do_reload): Use loop_optimizer_finalize.
20856 * ira-build.c (create_loop_tree_nodes): Use get_loops and
20857 number_of_loops to access the loop tree.
20858 (more_one_region_p): Likewise.
20859 (finish_loop_tree_nodes): Likewise.
20860 (rebuild_regno_allocno_maps): Likewise.
20861 (mark_loops_for_removal): Likewise.
20862 (mark_all_loops_for_removal): Likewise.
20863 (remove_unnecessary_regions): Likewise.
20864 (ira_build): Likewise.
20865 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
20867 2013-02-08 Richard Biener <rguenther@suse.de>
20869 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
20870 * ipa-pure-const.c (analyze_function): Avoid calling
20871 mark_irreducible_loops twice.
20872 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
20874 2013-02-07 David S. Miller <davem@davemloft.net>
20876 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
20878 * var-tracking.c (vt_add_function_parameter): Test the presence of
20879 HAVE_window_save properly and do not remap argument registers when
20880 we have a leaf function.
20882 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
20885 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
20887 * config/i386/i386.c (ix86_print_operand): Ditto.
20889 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
20891 * lra-constraints.c (process_alt_operands): Fix recently added comment.
20893 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
20895 PR rtl-optimization/56225
20896 * lra-constraints.c (process_alt_operands): Check that reload hard
20897 reg can hold value for strict_low_part.
20899 2013-02-07 Jakub Jelinek <jakub@redhat.com>
20902 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
20903 dwarf2out_end_function.
20904 (in_first_function_p, maybe_at_text_label_p,
20905 first_loclabel_num_not_at_text_label): New variables.
20906 (dwarf2out_var_location): In the first function find out
20907 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
20908 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
20911 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
20913 PR rtl-optimization/56178
20914 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
20915 SUBREG of a register. Tidy up related block of code.
20916 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
20917 note if the source is a register or a SUBREG of a register.
20919 2013-02-07 Jakub Jelinek <jakub@redhat.com>
20922 * config/rs6000/rs6000.md (ptrm): New mode attr.
20923 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
20924 call_value_indirect_aix<pttrsize>,
20925 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
20928 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
20930 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
20931 if -bnortl. Convert to strcmp and strncmp.
20933 2013-02-07 Alan Modra <amodra@gmail.com>
20936 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
20937 addresses won't wrap when offsetting.
20938 (rs6000_secondary_reload): Provide secondary reloads needed for
20939 wrapping LO_SUM addresses.
20941 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
20943 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
20944 MACH, just __MACH__.
20946 2013-02-06 Richard Biener <rguenther@suse.de>
20948 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
20949 instead of calling fix_loop_structure.
20951 2013-02-06 Jakub Jelinek <jakub@redhat.com>
20953 PR middle-end/56217
20954 * omp-low.c (use_pointer_for_field): Return false if
20955 lower_send_shared_vars doesn't generate any copy-out code.
20957 2013-02-06 Tom de Vries <tom@codesourcery.com>
20959 PR rtl-optimization/56131
20960 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
20961 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
20962 of the label is NULL. Add comment.
20964 2013-02-05 Jakub Jelinek <jakub@redhat.com>
20966 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
20969 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
20970 (STATIC_LIBTSAN_LIBS): Likewise.
20971 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
20972 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
20973 is defined, don't add anything else beyond that.
20974 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
20975 (LINK_COMMAND_SPEC): Use them.
20977 PR tree-optimization/56205
20978 * tree-stdarg.c (check_all_va_list_escapes): Return true if
20979 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
20980 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
20982 2013-02-05 Richard Biener <rguenther@suse.de>
20984 PR tree-optimization/53342
20985 PR tree-optimization/53185
20986 * tree-vectorizer.h (vect_check_strided_load): Remove.
20987 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
20988 not disallow peeling for vectorized strided loads.
20989 (vect_check_strided_load): Make static and simplify.
20990 (vect_analyze_data_refs): Adjust.
20991 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
20992 correctly when vectorizing strided loads.
20994 2013-02-05 Richard Biener <rguenther@suse.de>
20996 * doc/install.texi: Refer to ISL, not PPL.
20998 2013-02-05 Jan Hubicka <jh@suse.cz>
21000 PR tree-optimization/55789
21001 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
21003 2013-02-05 Jan Hubicka <jh@suse.cz>
21005 PR tree-optimization/55789
21006 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
21007 the dead call anyway.
21009 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
21012 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
21014 2013-02-04 Alexander Potapenko <glider@google.com>
21015 Jack Howarth <howarth@bromo.med.uc.edu>
21016 Jakub Jelinek <jakub@redhat.com>
21019 * config/darwin.c (sort_ctor_records): Stabilized qsort
21020 on constructor priority by using original position.
21021 (finalize_ctors): New routine to sort constructors by
21022 priority before use in assemble_integer.
21023 (machopic_asm_out_constructor): Use finalize_ctors if needed.
21025 2013-02-04 Jakub Jelinek <jakub@redhat.com>
21028 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
21029 about visibility on artificial decls.
21030 * config/sol2.c (solaris_assemble_visibility): Likewise.
21032 2013-02-04 Kai Tietz <ktietz@redhat.com>
21035 * config/i386/i386.c (function_value_ms_64): Add additional valtype
21036 argument and improve checking of return-argument types for 16-byte
21038 (ix86_function_value_1): Add additional valtype argument on call
21039 of function_value_64.
21040 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
21041 handling infunction_value_64 function.
21043 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
21045 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
21047 2013-02-04 Richard Biener <rguenther@suse.de>
21049 PR tree-optimization/56188
21050 * tree-ssa-structalias.c (label_visit): Consider case with
21051 initially non-empty points-to set.
21052 (perform_var_substitution): Dump node mapping and clean up.
21054 2013-02-04 Richard Guenther <rguenther@suse.de>
21057 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
21058 node prevail as last resort.
21059 (lto_symtab_merge_decls): Remove guard on LTRANS here.
21060 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
21062 2013-02-04 Richard Biener <rguenther@suse.de>
21064 PR tree-optimization/56113
21065 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
21067 (equiv_class_lookup_or_add): ... this.
21068 (label_visit): Adjust and fix error in previous patch.
21069 (perform_var_substitution): Adjust.
21071 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
21073 * config/sh/divtab.c: Fix formatting and comments throughout the file.
21074 * config/sh/sh4-300.md: Likewise.
21075 * config/sh/sh4a.md: Likewise.
21076 * config/sh/constraints.md: Likewise.
21077 * config/sh/sh.md: Likewise.
21078 * config/sh/netbsd-elf.h: Likewise.
21079 * config/sh/predicates.md: Likewise.
21080 * config/sh/sh-protos.h: Likewise.
21081 * config/sh/ushmedia.h: Likewise.
21082 * config/sh/linux.h: Likewise.
21083 * config/sh/sh.c: Likewise.
21084 * config/sh/superh.h: Likewise.
21085 * config/sh/elf.h: Likewise.
21086 * config/sh/sh4.md: Likewise.
21087 * config/sh/sh.h: Likewise.
21089 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
21091 * config/pa/constraints.md: Adjust unused letters. Change "T"
21092 constraint to match_test floating_point_store_memory_operand().
21093 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
21094 (base14_operand): New.
21095 (floating_point_store_memory_operand): New.
21096 (integer_store_memory_operand): Revise to use base14_operand and
21097 reg_plus_base_memory_operand.
21098 (move_dest_operand): Allow symbolic_memory_operands.
21099 (symbolic_memory_operand): Check for LO_SOM.
21100 (symbolic_operand): Change default case to break.
21101 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
21102 CONST_DOUBLE values to be reloaded by putting them into memory when
21103 the destination is a floating point register.
21104 (movdf): Remove code to handle CONST_DOUBLE.
21106 (reload_indf_r1): New.
21107 (reload_insf_r1): New.
21108 Consistently use "Q" and "T" constraints with integer and floating
21109 point move instructions, respectively.
21110 (movdi): Remove FAIL.
21111 Change predicate for source operand unamed DImode move from
21112 general_operand to move_src_operand.
21113 (umulsidi3): Change predicate for destination operand to
21115 Likewise for similar unamed patterns.
21116 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
21117 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
21118 (hppa_legitimize_address): Simplify mask calculation.
21119 (pa_emit_move_sequence): Revised handling of secondary reloads from
21120 REG+D addresses for floating point loads and stores. Directly handle
21121 loading CONST0_RTX (mode) to a floating point register.
21122 (pa_secondary_reload): Handle reloading DF and SFmode constant values
21123 to floating point registers. Don't restrict secondary reloads to
21124 floating point registers to integer modes. Revise some comments and
21126 (TARGET_LEGITIMATE_ADDRESS_P): Define.
21127 (pa_legitimate_address_p): New.
21128 (pa_legitimize_reload_address): New.
21129 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
21130 (STRICT_REG_OK_FOR_BASE_P): New.
21131 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
21132 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
21134 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
21135 Andrew Dixie <andrewd@gentrack.com>
21137 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
21140 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
21142 * expmed.c (extract_bit_field_1): Pass the full width of the
21143 structure to get_best_reg_extraction_insn.
21145 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
21148 * configure.ac (use_cxa_atexit): Add AIX.
21149 * configure: Regenerate.
21151 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
21153 2013-02-01 Jakub Jelinek <jakub@redhat.com>
21156 * final.c (need_profile_function): New variable.
21157 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
21158 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
21159 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
21160 notes, targetm.asm_out.function_prologue doesn't emit anything,
21161 HAVE_prologue and profiler should be emitted before prologue,
21162 set need_profile_function instead of emitting it.
21163 (final_scan_insn): If need_profile_function, emit
21164 profile_function on the first NOTE_INSN_BASIC_BLOCK or
21165 NOTE_INSN_FUNCTION_BEG note.
21167 2013-02-01 Richard Henderson <rth@redhat.com>
21169 * config/rs6000/rs6000.md (smulditi3): New.
21172 * config/alpha/alpha.md (umulditi3): New.
21174 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
21176 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
21177 (ASM_OUTPUT_ALIGNED_LOCAL): New.
21179 2013-02-01 Richard Biener <rguenther@suse.de>
21181 PR tree-optimization/56113
21182 * tree-ssa-structalias.c (label_visit): Reduce work for
21183 single-predecessor nodes.
21185 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
21187 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
21188 range isn't testing for zero.
21190 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
21192 PR middle-end/56113
21193 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
21195 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
21196 Nick Clifton <nickc@redhat.com>
21198 * config/v850/constraints.md (Q): Define as a memory constraint.
21199 * config/v850/predicates.md (label_ref_operand): New predicate.
21200 (e3v5_shift_operand): New predicate.
21201 (ior_operator): New predicate.
21202 * config/v850/t-v850: Add e3v5 multilib.
21203 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
21204 (v850_gen_movdi): Prototype.
21205 * config/v850/v850.c: Add support for e3v5 architecture.
21206 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
21208 (construct_save_jarl): Add e3v5 long JARL support.
21209 (v850_adjust_insn_length): New function. Adjust length of call
21210 insns when using e3v5 instructions.
21211 (v850_gen_movdi): New function: Generate instructions to move a
21213 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
21214 (CPP_SPEC): Define __v850e3v5__ as appropriate.
21215 (TARGET_USE_FPU): Enable for e3v5.
21216 (CONST_OK_FOR_W): New macro.
21217 (ADJUST_INSN_LENGTH): Define.
21218 * config/v850/v850.md (UNSPEC_LOOP): Define.
21219 (attr cpu): Add v850e3v5.
21220 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
21221 (movdi): New pattern.
21222 (movdi_internal): New pattern.
21223 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
21224 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
21225 (cstoresf4): Likewise.
21226 (cstoredf4): Likewise.
21227 (insv): New pattern.
21228 (rotlso3_a): New pattern.
21229 (rotlsi3_b): New pattern
21230 (rotlsi3_v850e3v5): New pattern.
21231 (doloop_begin): New pattern.
21232 (fix_loop_counter): New pattern.
21233 (doloop_end): New pattern.
21234 (branch_normal): Add e3v5 long branch support.
21235 (branch_invert): Likewise.
21236 (branch_z_normal): Likewise.
21237 (branch_z_invert): Likewise.
21238 (branch_nz_normal): Likewise.
21239 (branch_nz_invert): Likewise.
21240 (call_internal_short): Add e3v5 register-indirect JARL support.
21241 (call_internal_long): Likewise.
21242 (call_value_internal_short): Likewise.
21243 (call_value_internal_long): Likewise.
21244 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
21245 (mloop): New option.
21246 * config.gcc: Add support for configuring v840e3v5 target.
21247 * doc/invoke.texi: Document new v850 specific command line options.
21249 2013-01-31 Paul Koning <ni1d@arrl.net>
21253 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
21254 children if parent is a class.
21255 (prune_unused_types_prune): Don't add DW_AT_declaration.
21257 2013-01-31 Richard Biener <rguenther@suse.de>
21259 PR tree-optimization/56157
21260 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
21261 match up operand with SLP child.
21263 2013-01-31 Jason Merrill <jason@redhat.com>
21266 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
21267 parameters the first time.
21268 (gen_scheduled_generic_parms_dies): Check completeness here.
21270 2013-01-31 Richard Biener <rguenther@suse.de>
21272 PR middle-end/53073
21273 * common.opt (faggressive-loop-optimizations): New flag,
21274 enabled by default.
21275 * doc/invoke.texi (faggressive-loop-optimizations): Document.
21276 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
21277 infer_loop_bounds_from_undefined by it.
21279 2013-01-31 Richard Biener <rguenther@suse.de>
21281 PR tree-optimization/56150
21282 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
21283 visit virtual operands.
21284 (find_uses_to_rename_bb): Likewise.
21286 2013-01-31 Richard Biener <rguenther@suse.de>
21288 PR tree-optimization/56150
21289 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
21290 mixed store non-store stmts.
21292 2013-01-30 Jakub Jelinek <jakub@redhat.com>
21295 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
21296 LIBASAN_EARLY_SPEC is defined.
21297 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
21298 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
21300 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
21303 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
21304 invalid args instead of ICEing on it.
21305 (ix86_valid_target_attribute_tree): Return error_mark_node if
21306 ix86_valid_target_attribute_inner_p failed.
21307 (ix86_valid_target_attribute_p): Return false only if
21308 ix86_valid_target_attribute_tree returned error_mark_node. Allow
21309 target("default") attribute.
21310 (sorted_attr_string): Change argument from const char * to tree,
21311 merge in all target attribute arguments rather than just one.
21312 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
21313 instead of free. Avoid using strcat.
21314 (ix86_mangle_function_version_assembler_name): Mangle
21315 target("default") as if no target attribute is present. Adjust
21316 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
21317 instead of xmalloc and XDELETEVEC instead of free.
21318 (ix86_function_versions): Don't return true if one of the decls
21319 doesn't have target attribute. If they don't and one of the decls
21320 is DECL_FUNCTION_VERSIONED, report an error. Adjust
21321 sorted_attr_string caller. Use XDELETEVEC instead of free.
21322 (ix86_supports_function_versions): Remove.
21323 (make_name): Fix up formatting.
21324 (make_dispatcher_decl): Remove resolver_name and its initialization.
21325 Avoid leaking memory.
21326 (is_function_default_version): Return true if there is
21327 target("default") attribute rather than no target attribute at all.
21328 (make_resolver_func): Avoid leaking memory.
21329 (ix86_generate_version_dispatcher_body): Likewise.
21330 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
21331 * target.def (supports_function_versions): Remove.
21332 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
21333 * doc/tm.texi: Regenerated.
21335 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
21337 PR rtl-optimization/56144
21338 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
21339 for values with side effects.
21341 2013-01-30 Richard Biener <rguenther@suse.de>
21343 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
21344 (sparseset_pop): Likewise.
21345 * cfganal.c (compute_idf): Likewise. Increase work-stack size
21346 to be able to use quick_push in the worker loop.
21348 2013-01-30 Marek Polacek <polacek@redhat.com>
21350 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
21352 2013-01-30 Richard Biener <rguenther@suse.de>
21355 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
21357 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
21359 PR tree-optimization/56064
21360 * fixed-value.c (fixed_from_double_int): New function.
21361 * fixed-value.h (fixed_from_double_int): New prototype.
21362 (const_fixed_from_double_int): New static inline function.
21363 * fold-const.c (native_interpret_fixed): New static function.
21364 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
21365 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
21366 (native_encode_fixed): New static function.
21367 (native_encode_expr) <FIXED_CST>: Use it.
21368 (native_interpret_int): Move double_int worker code to...
21369 * double-int.c (double_int::from_buffer): ...this new static method.
21370 * double-int.h (double_int::from_buffer): Prototype it.
21372 2013-01-30 Richard Biener <rguenther@suse.de>
21374 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
21375 New pointer-map and obstack.
21376 (init_alias_vars): Allocate pointer-map and obstack.
21377 (delete_points_to_sets): Free them.
21378 (find_what_var_points_to): Cache result.
21379 (find_what_p_points_to): Adjust for changed interface of
21380 find_what_var_points_to.
21381 (compute_points_to_sets): Likewise.
21382 (ipa_pta_execute): Likewise.
21384 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21386 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
21387 * configure: Regenerate.
21388 * config.in: Regenerate.
21389 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
21390 #nobits/#progbits if supported.
21392 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
21395 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
21396 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
21397 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
21399 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
21401 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
21402 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
21404 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
21406 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
21407 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
21409 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
21411 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
21413 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
21414 * config/arm/cortex-a7.md: New bypasses using
21415 arm_mac_accumulator_is_result.
21417 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
21419 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
21420 (cortex_a7_neon_mla): Likewise.
21421 (cortex_a7_fpfmad): New reservation.
21422 (cortex_a7_fpmacs): Use ffmas and update required units.
21423 (cortex_a7_fpmuld): Update required units and latency.
21424 (cortex_a7_fpmacd): Likewise.
21425 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
21426 (cortex_a7_neon). Likewise.
21427 (bypass) Update participating units.
21429 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
21431 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
21432 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
21434 * config/arm/vfp11.md (vfp_farith): Use ffmas.
21435 (vfp_fmul): Use ffmad.
21436 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
21437 (cortex_r4_fmacd): Use ffmad.
21438 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
21439 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
21440 (cortex_a9_fmacd): Use ffmad.
21441 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
21442 (cortex_a8_vfp_macd): Use ffmad.
21443 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
21444 (cortex_a5_fpmacd): Use ffmad.
21445 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
21446 (cortex_a15_vfp_macd): Use ffmad.
21447 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
21449 2013-01-29 Jason Merrill <jason@redhat.com>
21452 * varasm.c (default_assemble_visibility): Don't warn about
21453 visibility on artificial decls.
21455 2013-01-29 Richard Biener <rguenther@suse.de>
21457 PR tree-optimization/56113
21458 * tree-ssa-structalias.c (equiv_class_lookup): Also return
21460 (label_visit): Free duplicate bitmaps and record the leader instead.
21461 (perform_var_substitution): Adjust.
21463 2013-01-29 Richard Biener <rguenther@suse.de>
21465 PR tree-optimization/55270
21466 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
21467 the CFG, schedule loops for fixup.
21469 2013-01-29 Nick Clifton <nickc@redhat.com>
21471 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
21474 2013-01-28 Leif Ekblad <leif@rdos.net>
21476 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
21477 * config/i386/i386.h (TARGET_RDOS): New macro.
21478 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
21479 * config/i386/i386.c (ix86_option_override_internal): For 64bit
21480 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
21481 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
21482 DEFAULT_LARGE_SECTION_THRESHOLD.
21483 * config/i386/i386.md (R14_REG, R15_REG): New constants.
21484 * config/i386/rdos.h: New file.
21485 * config/i386/rdos64.h: New file.
21487 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
21490 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
21493 2013-01-28 Jakub Jelinek <jakub@redhat.com>
21495 PR rtl-optimization/56117
21496 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
21497 call cselib_lookup_from_insn on the MEM before calling
21498 add_insn_mem_dependence.
21500 2013-01-28 Richard Biener <rguenther@suse.de>
21502 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
21503 to a stmt that didn't have one.
21504 (copy_phis_for_bb): Likewise for PHI arguments.
21505 (copy_debug_stmt): Likewise for debug stmts.
21507 2013-01-28 Richard Biener <rguenther@suse.de>
21509 PR tree-optimization/56034
21510 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
21511 (partition_builtin_p): Adjust.
21512 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
21513 it is the last partition.
21514 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
21515 up the vertex for the definition.
21516 (classify_partition): Classify whether a partition is a
21517 PKIND_REDUCTION, thus has uses outside of the loop.
21518 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
21519 Merge all PKIND_REDUCTION partitions into the last partition.
21520 (tree_loop_distribution): Seed partitions from reductions as well.
21522 2013-01-28 Jakub Jelinek <jakub@redhat.com>
21524 PR tree-optimization/56125
21525 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
21526 pow(x,c) into sqrt(x) * powi(x, n/2) or
21527 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
21528 optimizing for size.
21529 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
21530 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
21533 PR tree-optimization/56094
21534 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
21535 to UNKNOWN_LOCATION while gimplifying expr.
21537 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
21540 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
21541 operand 0 in movabs insn template for -masm=intel asm alternative.
21542 (*movabs<mode>_2): Ditto for operand 1.
21544 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
21547 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
21550 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
21552 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
21555 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
21557 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
21558 Undef to avoid warning.
21560 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
21562 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
21563 * configure: Regenerate.
21565 2013-01-25 Jakub Jelinek <jakub@redhat.com>
21567 PR tree-optimization/56098
21568 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
21569 for stmts with volatile ops.
21570 (cond_store_replacement): Don't optimize if assign has volatile ops.
21571 (cond_if_else_store_replacement_1): Don't optimize if either
21572 then_assign or else_assign have volatile ops.
21573 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
21576 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
21578 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
21580 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
21582 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
21583 missing ':' in asm example.
21585 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
21587 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
21588 entries into lane and laneq entries.
21589 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
21590 Remove AdvSIMD scalar modes.
21591 (aarch64_sq<r>dmulh_laneq<mode>): New.
21592 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
21594 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
21595 builtin implementations to relfect changes in RTL in aarch64-simd.md.
21596 * config/aarch64/iterators.md (VCOND): New.
21599 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
21602 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
21603 Add NULL LIBNAME argument to existing definitions.
21604 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
21605 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
21606 * config/avr/avr.c (DEF_BUILTIN): Same.
21607 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
21608 (avr_expand_builtin): Expand to a vanilla call if a libgcc
21609 implementation is available (DECL_ASSEMBLER_NAME is set).
21610 (avr_fold_absfx): New static function.
21611 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
21612 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
21613 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
21614 AVR_BUILTIN_ABSLLK.
21615 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
21616 (abshk, absk, abslk, absllk): Provide as static inline functions.
21618 2013-01-25 Marek Polacek <polacek@redhat.com>
21620 PR tree-optimization/56035
21621 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
21623 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
21625 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
21626 (*movtf_internal_rex64): Add (!o,C) alternative
21627 (*movxf_internal_rex64): Ditto.
21628 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
21630 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
21632 * doc/invoke.texi: fix typo.
21633 * doc/objc.texi: fix typo.
21635 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
21637 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
21638 for the first two alternatives.
21640 2013-01-24 Diego Novillo <dnovillo@google.com>
21642 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
21643 (ggc-zone.o): Remove.
21644 * configure.ac: Remove option --with-gc.
21645 * configure: Re-generate.
21646 * doc/install.texi: Remove documentation for --with-gc.
21647 * gengtype.c (write_enum_defn): Remove. Update all users.
21648 (write_Types_process_field): Remove generation of gt_e_* argument.
21649 (output_type_enum): Remove. Update all users.
21650 (write_enum_defn): Remove. Update all users.
21651 (enum alloc_zone): Remove. Update all users.
21652 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
21653 * ggc-common.c (ggc_splay_alloc): Remove first argument.
21654 Update all callers.
21655 (struct ptr_data): Remove field TYPE. Update all users.
21656 (gt_pch_note_object): Remove argument TYPE. Update all users.
21657 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
21659 * ggc-none.c (ggc_alloc_typed_stat): Remove.
21660 (struct alloc_zone): Remove.
21661 (ggc_internal_alloc_zone_stat): Remove.
21662 (ggc_internal_cleared_alloc_zone_stat): Remove.
21663 * ggc-page.c (ggc_alloc_typed_stat): Remove.
21664 (ggc_pch_count_object): Remove last argument. Update all users.
21665 (ggc_pch_alloc_object): Remove last argument. Update all users.
21666 (struct alloc_zone): Remove.
21667 * ggc-zone.c: Remove.
21668 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
21669 (struct alloc_zone): Remove.
21670 (ggc_alloc_typed_stat): Remove.
21671 (ggc_alloc_typed): Remove.
21672 (ggc_splay_alloc): Remove first argument.
21673 (rtl_zone): Remove. Update all users.
21674 (tree_zone): Remove. Update all users.
21675 (tree_id_zone): Remove. Update all users.
21676 (ggc_internal_zone_alloc_stat): Remove. Update all users.
21677 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
21678 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
21679 * tree-ssanames.c: Remove references to zone allocator in comments.
21681 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
21683 * config/avr/avr.c (avr_out_fract): Make register numbers that
21684 might be outside of source operand signed.
21686 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
21688 * config/i386/constraints.md (Yf): New constraint.
21689 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
21690 of f constraint to conditionaly disable x87 register preferences.
21691 (*movdf_internal): Ditto.
21692 (*movsf_internal): Ditto.
21694 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
21696 PR inline-asm/55934
21697 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
21698 that have operands with impossible constraints.
21699 Add a FIXME for a speed-up opportunity.
21700 * lra-constraints.c (process_alt_operands): Verify that a class
21701 selected from constraints on asms is valid for the operand mode.
21702 (curr_insn_transform): Remove incorrect comment.
21704 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
21706 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
21707 TOC operand is a valid symbol ref in the constant pool.
21709 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
21711 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
21713 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
21716 * config/avr/stdfix.h: New file.
21717 * t-avr (stdfix-gcc.h): New rule to build it.
21718 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
21720 2013-01-23 Kostya Serebryany <kcc@google.com>
21722 * config/darwin.h: remove dependency on
21723 CoreFoundation (asan on Mac OS).
21725 2013-01-23 Jakub Jelinek <jakub@redhat.com>
21728 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
21729 instead of cmpdi_operand for first comparison operand.
21730 Don't assert that comparison operands aren't both constants.
21732 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
21734 * doc/install.texi (Downloading the Source): Update references to
21735 downloading separate components.
21737 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
21739 * doc/extend.texi (__int128): Improve grammar.
21741 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
21744 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
21745 alternative to (o,r).
21746 (*movdi_internal_rex64): Remove (!o,n) alternative.
21747 (DImode immediate->memory splitter): Remove.
21748 (DImode immediate->memory peephole2): Remove.
21749 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
21750 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
21751 alternative to (!o,*r).
21752 (*movtf_internal_sse): New pattern.
21753 (*movxf_internal_rex64): New pattern.
21754 (*movxf_internal): Disable for TARGET_64BIT.
21755 (*movdf_internal_rex64): Remove (!o,F) alternative.
21757 2013-01-22 Jakub Jelinek <jakub@redhat.com>
21759 PR middle-end/56074
21760 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
21761 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
21762 * tree-vect-loop-manip.c (find_loop_location): Also ignore
21763 stmt locations where LOCATION_LOCUS of the stmt location is
21764 UNKNOWN_LOCATION or BUILTINS_LOCATION.
21767 * config/i386/i386.md (UNSPEC_STOS): New.
21768 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
21769 *strsetqi_1): Add UNSPEC_STOS.
21771 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
21774 * doc/invoke.texi: Remove left over -Wsynth example.
21776 2013-01-21 Jakub Jelinek <jakub@redhat.com>
21778 PR tree-optimization/56051
21779 * fold-const.c (fold_binary_loc): Don't fold
21780 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
21781 a narrowing conversion, or widening conversion from signed
21784 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
21786 PR rtl-optimization/56023
21787 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
21788 dependent on debug instruction.
21790 2013-01-21 Martin Jambor <mjambor@suse.cz>
21792 PR middle-end/56022
21793 * function.c (allocate_struct_function): Call
21794 invoke_set_current_function_hook earlier.
21796 2013-01-21 Jakub Jelinek <jakub@redhat.com>
21798 * reload1.c (init_reload): Only initialize reload_obstack
21799 during the first call.
21801 2013-01-21 Marek Polacek <polacek@redhat.com>
21803 * cfgloop.c (verify_loop_structure): Fix up grammar.
21805 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
21807 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
21808 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
21810 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21813 * config/arm/marvell-pj4.md: Update copyright year.
21814 Fix up use of alu to alu_reg and simple_alu_imm.
21816 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
21818 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
21820 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
21823 * lra-constraints.c (curr_insn_transform): Don't reuse original
21824 insn for secondary memory move when memory mode should be different.
21826 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
21828 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
21829 atomic_storedi_1): New patterns.
21831 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
21833 btver2 pipeline descriptions.
21834 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
21836 * config/i386/i386.md (btver2_decode): New type attributes.
21837 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
21839 * config/i386/btver2.md: New file describing btver2 pipelines.
21841 2013-01-19 Andrew Pinski <apinski@cavium.com>
21843 PR tree-optimization/52631
21844 * tree-ssa-sccvn (visit_use): Before looking up the original
21845 statement, try looking up the simplified expression.
21847 2013-01-19 Anthony Green <green@moxielogic.com>
21849 * config/moxie/moxie.c (moxie_expand_prologue): Set
21850 current_function_static_stack_size.
21852 2013-01-18 Jakub Jelinek <jakub@redhat.com>
21854 PR tree-optimization/56029
21855 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
21856 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
21858 2013-01-18 Sharad Singhai <singhai@google.com>
21860 PR tree-optimization/55995
21861 * dumpfile.c (dump_loc): Print location only if available.
21862 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
21864 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
21867 * lra-constraints.c (curr_insn_transform): Reuse original insn for
21868 secondary memory move.
21869 (inherit_reload_reg): Use rclass instead of cl for
21870 check_secondary_memory_needed_p.
21872 2013-01-18 Jakub Jelinek <jakub@redhat.com>
21874 PR middle-end/56015
21875 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
21876 the case where writing real complex part of target modifies op1.
21878 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
21880 * config/aarch64/aarch64-simd.md
21881 (aarch64_vcond_internal<mode>): Handle unordered cases.
21882 * config/aarch64/iterators.md (v_cmp_result): New.
21884 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
21885 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21887 * config/arm/marvell-pj4.md: New file.
21888 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
21889 * config/arm/arm.md (generic_sched): Add marvell_pj4.
21890 (generic_vfp): Likewise.
21891 * config/arm/arm-cores.def: Add marvell-pj4.
21892 * config/arm/arm-tune.md: Regenerate.
21893 * config/arm/arm-tables.opt: Regenerate.
21894 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
21895 * doc/invoke.texi: Document marvell-pj4.
21897 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
21899 * config/aarch64/arm_neon.h: Map scalar types to standard types.
21901 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
21906 * var-tracking.c (negative_power_of_two_p): New.
21907 (global_get_addr_cache, local_get_addr_cache): New.
21908 (get_addr_from_global_cache, get_addr_from_local_cache): New.
21909 (vt_canonicalize_addr): Rewrite using the above. Adjust the
21911 (vt_stack_offset_p): Remove.
21912 (vt_canon_true_dep): Always canonicalize loc's address.
21913 (clobber_overlapping_mems): Make sure we have a MEM.
21914 (local_get_addr_clear_given_value): New.
21915 (val_reset): Clear local cached entries.
21916 (compute_bb_dataflow): Create and release the local cache.
21917 Disable duplicate MEMs clobbering.
21918 (emit_notes_in_bb): Clobber MEMs likewise.
21919 (vt_emit_notes): Create and release the local cache.
21920 (vt_initialize, vt_finalize): Create and release the global
21921 cache, respectively.
21922 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
21924 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
21926 PR libmudflap/53359
21927 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
21928 not found in the symtab.
21930 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
21933 PR rtl-optimization/55547
21934 PR rtl-optimization/53827
21937 * alias.c (offset_overlap_p): New, factored out of...
21938 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
21939 the conservative special case for symbolic constants. Don't
21940 adjust zero sizes on alignment.
21942 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
21944 PR rtl-optimization/52573
21945 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
21946 REG_UNUSED for the same register.
21948 2013-01-17 Richard Biener <rguenther@suse.de>
21949 Marek Polacek <polacek@redhat.com>
21951 PR rtl-optimization/55833
21952 * loop-unswitch.c (unswitch_loops): Move loop verification...
21953 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
21954 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
21955 Set it to true when we're removing a loop from hierarchy tree in
21956 an irreducible region.
21957 (fix_bb_placements): Adjust caller.
21958 (fix_loop_placements): Likewise.
21960 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
21962 * config/avr/builtins.def (DEF_BUILTIN): Factor out
21963 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
21964 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
21965 Remove ID. Adjust comments.
21966 * config/avr/avr-c.c (avr_builtin_name): Remove.
21967 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
21968 * config/avr/avr.c (avr_tolower): New static function.
21969 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
21970 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
21971 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
21974 2013-01-17 Jan Hubicka <jh@suse.cz>
21976 PR tree-optimization/55273
21977 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
21979 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
21982 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
21983 store through atomic_store<mode>_1.
21984 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
21986 2013-01-17 Martin Jambor <mjambor@suse.cz>
21988 PR tree-optimizations/55264
21989 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
21990 for virtual methods.
21991 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
21992 virtual methods before inlining is over.
21993 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
21995 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
21998 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
22000 PR rtl-optimization/56005
22001 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
22002 pending reads for prefetch.
22004 2013-01-16 Ian Bolton <ian.bolton@arm.com>
22006 * config/aarch64/aarch64.md
22007 (*cstoresi_neg_uxtw): New pattern.
22008 (*cmovsi_insn_uxtw): New pattern.
22009 (*<optab>si3_uxtw): New pattern.
22010 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
22011 (*<optab>si3_insn_uxtw): New pattern.
22012 (*bswapsi2_uxtw): New pattern.
22014 2013-01-16 Richard Biener <rguenther@suse.de>
22016 * tree-inline.c (tree_function_versioning): Remove set but
22017 never used variable.
22019 2013-01-16 Richard Biener <rguenther@suse.de>
22021 PR tree-optimization/55964
22022 * tree-flow.h (rename_variables_in_loop): Remove.
22023 (rename_variables_in_bb): Likewise.
22024 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
22025 (copy_loop_before): Adjust and delete update-ssa status.
22026 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
22027 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
22028 (rename_variables_in_loop): Remove.
22029 (slpeel_update_phis_for_duplicate_loop): Likewise.
22030 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
22031 use available cfg machinery instead of duplicating it.
22032 Update PHI nodes and perform poor-mans SSA update here.
22033 (slpeel_tree_peel_loop_to_edge): Adjust.
22035 2013-01-16 Richard Biener <rguenther@suse.de>
22037 PR tree-optimization/54767
22038 PR tree-optimization/53465
22039 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
22040 (vrp_visit_phi_node): For PHI arguments coming via backedges
22041 drop all symbolical range information.
22042 (execute_vrp): Compute backedges.
22044 2013-01-16 Richard Biener <rguenther@suse.de>
22046 * doc/install.texi: Update CLooG and ISL requirements to
22049 2013-01-16 Christian Bruel <christian.bruel@st.com>
22052 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
22053 (broken_move): Handle UNSPECV_SP_SWITCH_B.
22054 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
22056 2013-01-16 DJ Delorie <dj@redhat.com>
22058 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
22059 (UNSPECV_SP_SWITCH_E): New.
22060 (sp_switch_1): Change to an unspec.
22061 (sp_switch_2): Change to an unspec. Don't use post-inc when we
22064 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
22066 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
22067 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
22068 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
22069 (expand_mem_thread_fence): Ditto.
22070 (expand_mem_signal_fence): Ditto.
22071 (expand_atomic_load): Ditto.
22072 (expand_atomic_store): Ditto.
22074 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
22076 PR rtl-optimization/55547
22077 PR rtl-optimization/53827
22080 * alias.c (memrefs_conflict_p): Set sizes to negative after
22083 2013-01-15 Jakub Jelinek <jakub@redhat.com>
22086 * function.c (thread_prologue_and_epilogue_insns): Always
22087 add crtl->drap_reg to set_up_by_prologue.set, even if
22088 stack_realign_drap is false.
22090 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
22092 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
22093 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
22094 *call): Fix indention.
22096 2013-01-15 Tom de Vries <tom@codesourcery.com>
22099 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
22102 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
22104 PR rtl-optimization/55153
22105 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
22107 2013-01-15 Martin Jambor <mjambor@suse.cz>
22109 PR tree-optimization/55920
22110 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
22111 accesses as grp_to_be_debug_replaced.
22113 2013-01-15 Jakub Jelinek <jakub@redhat.com>
22115 PR tree-optimization/55920
22116 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
22117 there is non-useless type conversion needed from debug rhs to lhs,
22118 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
22120 2013-01-15 Joseph Myers <joseph@codesourcery.com>
22121 Mikael Pettersson <mikpe@it.uu.se>
22124 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
22126 (ASM_OUTPUT_CASE_LABEL): Remove.
22127 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
22128 * final.c (shorten_branches): Update alignment of labels before
22129 jump tables if CASE_VECTOR_SHORTEN_MODE.
22131 2013-01-15 Richard Biener <rguenther@suse.de>
22134 * system.h: Do not include gmp.h for building host tools.
22136 2013-01-15 Richard Biener <rguenther@suse.de>
22138 PR middle-end/55882
22139 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
22140 account for bitpos when computing alignment.
22142 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
22144 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
22145 (ix86_target_macros_internal): Likewise.
22147 * config/i386/i386.c (m_CORE2I7): Removed.
22148 (m_CORE_HASWELL): New macro.
22149 (m_CORE_ALL): Likewise.
22150 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
22151 (initial_ix86_arch_features): Likewise.
22152 (processor_target_table): Initializations for Core avx2.
22153 (cpu_names): New names "core-avx2".
22154 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
22155 PROCESSOR_CORE_HASWELL.
22156 (ix86_issue_rate): New case.
22157 (ia32_multipass_dfa_lookahead): Likewise.
22158 (ix86_sched_init_global): Likewise.
22160 * config/i386/i386.h (TARGET_HASWELL): New macro.
22161 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
22162 (processor_type): New PROCESSOR_HASWELL.
22164 2013-01-15 Jakub Jelinek <jakub@redhat.com>
22166 PR tree-optimization/55955
22167 * tree-vect-loop.c (vectorizable_reduction): Give up early on
22168 *SHIFT_EXPR and *ROTATE_EXPR codes.
22170 PR tree-optimization/48766
22171 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
22172 -ftrapv disable -fwrapv.
22174 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
22177 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
22178 etc. to 1 and not to __flash.
22179 Use LL suffix for __INT24_MAX__ with -mint8.
22180 Use ULL suffix for __UINT24_MAX__ with -mint8.
22182 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
22184 * config/avr/avr-arch.h
22185 (struct base_arch_s): Use typedef avr_arch_t instead.
22186 (struct arch_info_s): Use typedef avr_arch_info_t instead.
22187 (struct mcu_type_s): Use typedef avr_mcu_t instead.
22188 * config/avr/avr.c: Same.
22189 * config/avr/avr-devices.c: Same.
22190 * config/avr/driver-avr.c: Same.
22191 * config/avr/gen-avr-mmcu-texi.c: Same.
22192 * config/avr/avr-mcus.def: Adjust comment.
22194 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
22196 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
22197 * config/aarch64/iterators.md (VALLDI): New.
22199 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
22200 Andi Kleen <ak@linux.intel.com>
22203 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
22204 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
22207 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
22209 * config/avr/avr-stdint.h: Remove trailing blanks.
22210 * config/avr/avr-log.h: Same.
22211 * config/avr/avr-arch.h: Same.
22212 * config/avr/avr-devices.c: Same.
22213 * config/avr/avr-dimode.md: Same.
22214 * config/avr/predicates.md: Same.
22215 * config/avr/avr-c.c: Same. And fix typo.
22217 * config/avr/avr-protos.h: Same. And:
22218 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
22219 (init_cumulative_args): Rename to avr_init_cumulative_args.
22220 (expand_prologue): Rename to avr_expand_prologue.
22221 (expand_epilogue): Rename to avr_expand_epilogue.
22222 (adjust_insn_length): Rename to avr_adjust_insn_length.
22223 (notice_update_cc): Rename to avr_notice_update_cc.
22224 (final_prescan_insn): Rename to avr_final_prescan_insn.
22225 * config/avr/avr.c: Same.
22226 * config/avr/avr.h: Same.
22227 * config/avr/avr.md: Remove trailing blanks.
22228 (prologue): Use avr_expand_prologue.
22229 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
22231 2013-01-14 Richard Biener <rguenther@suse.de>
22233 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
22234 verify_location, collect_subblocks): New functions.
22235 (verify_gimple_in_cfg): Verify that locations only reference
22236 BLOCKs in the functions BLOCK tree.
22238 2013-01-14 Richard Biener <rguenther@suse.de>
22240 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
22242 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
22244 (insert_out_of_ssa_copy_on_edge): Likewise.
22245 (rewrite_close_phi_out_of_ssa): Likewise.
22246 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
22248 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
22249 propagated constants.
22250 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
22253 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
22255 * config/avr/avr-modes.def: Add GPL copyright notice.
22257 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
22259 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
22260 MEMMODEL_MASK to determine memory model.
22261 (atomic_store<mode>): Ditto from operands[2].
22262 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
22264 2013-01-13 Jakub Jelinek <jakub@redhat.com>
22267 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
22268 (fold_gimple_assign): Don't call unshare_expr here.
22269 (fold_ctor_reference): Call unshare_expr.
22271 2013-01-13 Terry Guo <terry.guo@arm.com>
22273 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
22274 * doc/fragments.texi: Document MULTILIB_REUSE.
22275 * gcc.c (multilib_reuse): New internal spec.
22276 (set_multilib_dir): Also search multilib from multilib_reuse.
22277 * genmultilib (tmpmultilib3): Refactor code.
22278 (tmpmultilib4): Ditto.
22279 (multilib_reuse): New multilib argument.
22281 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
22283 * Makefile.in: Update copyright.
22285 2013-01-12 Tom de Vries <tom@codesourcery.com>
22287 PR middle-end/55890
22288 * calls.c (expand_call): Check if arg_nr is valid.
22290 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22292 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
22293 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
22294 documentation. Add missing '__' in front of
22295 __builtin_ia32_packssdw256.
22297 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
22300 * config/s390/s390.c (s390_preferred_reload_class): Do not return
22301 NO_REGS for larl operands.
22302 (s390_reload_larl_operand): Use s390_load_address instead of
22305 2013-01-11 Richard Biener <rguenther@suse.de>
22307 * tree-cfg.c (verify_node_sharing_1): Split out from ...
22308 (verify_node_sharing): ... here.
22309 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
22311 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
22313 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
22314 Substitute TREECHECKING.
22315 * configure: Regenerate.
22316 * Makefile.in (TREECHECKING): New.
22318 2013-01-11 Richard Guenther <rguenther@suse.de>
22320 PR tree-optimization/44061
22321 * tree-vrp.c (extract_range_basic): Compute zero as
22322 value-range for __builtin_constant_p of function parameters.
22324 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
22326 Update copyright years.
22328 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
22330 PR rtl-optimization/55672
22331 * lra-eliminations.c (mark_not_eliminable): Permit addition with
22332 const to be eliminable.
22334 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
22336 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
22337 * configure: Regenerate.
22339 2013-01-10 Richard Biener <rguenther@suse.de>
22341 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
22343 2013-01-10 Richard Biener <rguenther@suse.de>
22346 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
22347 locations for virtual PHI arguments.
22348 (rewrite_update_phi_arguments): Likewise.
22350 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
22352 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
22355 2013-01-10 Jakub Jelinek <jakub@redhat.com>
22357 PR tree-optimization/55921
22358 * tree-complex.c (expand_complex_asm): New function.
22359 (expand_complex_operations_1): Call it for GIMPLE_ASM.
22361 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
22364 * config/s390/s390.c (s390_symref_operand_p)
22365 (s390_loadrelative_operand_p): Merge the two functions.
22366 (s390_check_qrst_address, print_operand_address): Add parameters
22367 to s390_loadrelative_operand_p invokation.
22368 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
22369 (s390_reload_larl_operand, s390_secondary_reload): Use
22370 s390_loadrelative_operand_p instead of s390_symref_operand_p.
22371 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
22373 2013-01-09 Mike Stump <mikestump@comcast.net>
22375 * dse.c (record_store): Remove unnecessary assert.
22377 2013-01-09 Jan Hubicka <jh@suse.cz>
22379 PR tree-optimization/55569
22380 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
22381 * cfgloop.h (scale_loop_profile): Likewise.
22383 2013-01-09 Jan Hubicka <jh@suse.cz>
22386 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
22388 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
22390 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
22392 PR middle-end/55114
22393 * expr.h (maybe_emit_group_store): Declare.
22394 * expr.c (maybe_emit_group_store): New function.
22395 * builtins.c (expand_builtin_int_roundingfn): Call it.
22396 (expand_builtin_int_roundingfn_2): Likewise.
22398 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
22400 PR rtl-optimization/55829
22401 * lra-constraints.c (match_reload): Add code for absent output.
22402 (curr_insn_transform): Add code for reloads of matched inputs
22405 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
22407 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
22408 attribute of movddup insn to DF.
22409 (*vec_interleave_lowv2df): Ditto.
22410 (vec_dupv2df): Ditto.
22412 2013-01-09 Jan Hubicka <jh@suse.cz>
22414 PR tree-optimiation/55875
22415 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
22416 EVERY_ITERATION parameter.
22417 (number_of_iterations_exit): Check if exit is executed every iteration.
22418 (idx_infer_loop_bounds): Similarly here.
22419 (n_of_executions_at_most): Simplify
22420 to only test for cases where statement is dominated by the
22421 particular bound; handle correctly the "postdominance" test.
22422 (scev_probably_wraps_p): Use max loop iterations info
22423 as a global bound first.
22425 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
22426 Nick Clifton <nickc@redhat.com>
22428 * config/v850/v850.md (cbranchsf4): New pattern.
22429 (cstoresf4): New pattern.
22430 (cbranchdf4): New pattern.
22431 (cstoredf4): New pattern.
22432 (movsicc): Disallow floating point comparisons.
22433 (cmpsf_le_insn): Fix order of operators.
22434 (cmpsf_lt_insn): Likewise.
22435 (cmpsf_eq_insn): Likewise.
22436 (cmpdf_le_insn): Likewise.
22437 (cmpdf_lt_insn): Likewise.
22438 (cmpdf_eq_insn): Likewise.
22439 (cmpsf_ge_insn): Use LE comparison.
22440 (cmpdf_ge_insn): Likewise.
22441 (cmpsf_gt_insn): Use LT comparison.
22442 (cmpdf_gt_insn): Likewise.
22443 (cmpsf_ne_insn): Delete pattern.
22444 (cmpdf_ne_insn): Delete pattern.
22445 * config/v850/v850.c (v850_gen_float_compare): Use
22446 gen_cmpdf_eq_insn for NE comparison.
22447 (v850_float_z_comparison_operator)
22448 (v850_float_nz_comparison_operator): Move from here ...
22449 * config/v850/predicates.md: ... to here. Move GT and GE
22450 comparisons into v850_float_z_comparison_operator.
22451 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
22453 (v850_float_nz_comparison_operator): Likewise.
22455 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
22457 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
22458 with calls to gen_insvsi/gen_insvdi.
22460 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
22462 * config/i386/i386.c (initial_ix86_tune_features): Set up
22463 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
22465 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
22466 Jakub Jelinek <jakub@redhat.com>
22468 PR tree-optimization/48189
22469 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
22470 If nitercst is 0, don't predict the exit edge.
22472 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
22474 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
22475 in asm_fprintf with reg_names.
22476 (aarch64_print_operand_address): Likewise.
22477 (aarch64_return_addr): Likewise.
22478 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
22480 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
22482 * config/pa/pa.h (VAL_U6_BITS_P): Define.
22483 (INT_U6_BITS): Likewise.
22484 * config/pa/predicates.md (uint6_operand): New predicate.
22485 (shift5_operand, shift6_operand): Likewise.
22486 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
22488 (lshrdi3): Use shift6_operand.
22489 (shrpsi4, shrpdi4): New insn patterns.
22490 (extzv): Delete expander.
22491 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
22492 predicates in unamed zero extract patterns. Tighten common constraint.
22493 (extv): Delete expander.
22494 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
22495 predicates in unamed sign extract patterns. Tighten common constraint.
22496 (insv): Delete expander.
22497 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
22498 predicates in unamed insert patterns. Tighten common constraint.
22499 Change uint32_operand predicate to uint6_operand predicate in unamed
22500 DImode pattern to insert constant values of type 1...1xxxx.
22502 2013-01-04 Jan Hubicka <jh@suse.cz>
22504 PR tree-optimization/55823
22505 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
22508 2013-01-08 Jakub Jelinek <jakub@redhat.com>
22509 Uros Bizjak <ubizjak@gmail.com>
22511 PR rtl-optimization/55845
22512 * df-problems.c (can_move_insns_across): Stop scanning at
22513 volatile_insn_p source instruction or give up if
22514 across_from .. across_to range contains any volatile_insn_p
22517 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
22519 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
22520 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
22522 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
22523 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
22525 2013-01-08 Jakub Jelinek <jakub@redhat.com>
22528 * asan.c (asan_clear_shadow): New function.
22529 (asan_emit_stack_protection): Use it.
22531 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
22533 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
22534 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
22535 with tab instead of space.
22537 2013-01-08 Nick Clifton <nickc@redhat.com>
22539 * config/rl78/rl78.c (rl78_expand_prologue): Always select
22540 register bank 0 at the start of an interrupt handler.
22541 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
22544 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
22546 * config/aarch64/aarch64-simd.md
22547 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
22548 (aarch64_simd_bsl): Likewise.
22549 (aarch64_vcond_internal<mode>): Likewise.
22550 (vcond<mode><mode>): Likewise.
22551 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
22552 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
22554 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
22556 * config/aarch64/aarch64-builtins.c
22557 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
22559 2013-01-08 Martin Jambor <mjambor@suse.cz>
22562 * tree-sra.c (analyze_access_subtree): Return true also after
22563 potentially creating a debug-only replacement.
22565 2013-01-08 Jakub Jelinek <jakub@redhat.com>
22567 PR middle-end/55890
22568 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
22570 PR tree-optimization/54120
22571 * tree-vrp.c (range_fits_type_p): Don't allow
22572 src_precision < precision from signed vr to unsigned_p
22573 if vr->min or vr->max is negative.
22574 (simplify_float_conversion_using_ranges): Test can_float_p
22575 against CODE_FOR_nothing.
22577 2013-01-08 Jakub Jelinek <jakub@redhat.com>
22578 Richard Biener <rguenther@suse.de>
22580 PR middle-end/55851
22581 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
22582 types instead of just INTEGER_TYPE types.
22584 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
22586 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
22589 2013-01-07 Steve Ellcey <sellcey@mips.com>
22592 * config/mips/mips.opt: Change mad to mmad to match documentation.
22594 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
22597 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
22598 .progmemx.data now.
22600 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
22603 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
22604 (avr_addrspace_t): Add .section_name field.
22605 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
22607 (avr_addrspace): Same. Initialize .section_name. Remove last
22608 NULL entry. Put __memx into .progmemx.data.
22609 (progmem_section_prefix): Remove.
22610 (avr_asm_init_sections): No need to initialize progmem_section.
22611 (avr_asm_named_section): Use avr_addrspace[].section_name to get
22612 section name prefix.
22613 (avr_asm_select_section): Ditto. And use get_unnamed_section to
22614 retrieve the progmem section.
22615 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
22616 boundary to run over avr_addrspace[].
22617 (avr_register_target_pragmas): Ditto.
22619 2013-01-06 Jakub Jelinek <jakub@redhat.com>
22621 * varasm.c (output_constant_def_contents): For asan_protect_global
22622 protected strings, adjust DECL_ALIGN if needed, before testing for
22624 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
22625 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
22627 (output_object_block): For asan protected decls, emit asan padding
22628 after their contents.
22629 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
22630 (asan_finish_file): Test it here instead.
22632 2013-01-07 Nick Clifton <nickc@redhat.com>
22633 Matthias Klose <doko@debian.org>
22634 Doug Kwan <dougkwan@google.com>
22635 H.J. Lu <hongjiu.lu@intel.com>
22638 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
22640 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
22642 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
22644 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
22646 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
22648 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
22651 * doc/install.texi (Cross-Compiler-Specific Options): Document
22654 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
22656 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
22657 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
22658 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
22659 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
22660 vqmovun_high_s64): Fix source operand number and update copyright.
22662 2013-01-07 Richard Biener <rguenther@suse.de>
22664 PR middle-end/55890
22665 * gimple.h (gimple_call_builtin_p): New overload.
22666 * gimple.c (validate_call): New function.
22667 (gimple_call_builtin_p): Likewise.
22668 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22669 Use gimple_call_builtin_p.
22670 (find_func_clobbers): Likewise.
22671 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
22672 (strlen_optimize_stmt): Likewise.
22674 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
22676 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
22677 (vld1q_dup_*): Likewise.
22678 (vld1_*): Likewise.
22679 (vld1q_*): Likewise.
22680 (vld1_lane_*): Likewise.
22681 (vld1q_lane_*): Likewise.
22683 2013-01-07 Richard Biener <rguenther@suse.de>
22685 * lto-streamer.h (LTO_minor_version): Bump to 2.
22687 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
22689 * config/aarch64/aarch64-protos.h
22690 (aarch64_const_double_zero_rtx_p): Rename to...
22691 (aarch64_float_const_zero_rtx_p): ...this.
22692 (aarch64_float_const_representable_p): New.
22693 (aarch64_output_simd_mov_immediate): Likewise.
22694 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
22695 move immediate case.
22696 * config/aarch64/aarch64.c
22697 (aarch64_const_double_zero_rtx_p): Rename to...
22698 (aarch64_float_const_zero_rtx_p): ...this.
22699 (aarch64_print_operand): Allow printing of new constants.
22700 (aarch64_valid_floating_const): New.
22701 (aarch64_legitimate_constant_p): Check for valid floating-point
22703 (aarch64_simd_valid_immediate): Likewise.
22704 (aarch64_vect_float_const_representable_p): New.
22705 (aarch64_float_const_representable_p): Likewise.
22706 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
22707 (aarch64_output_simd_mov_immediate): New.
22708 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
22709 (*movdf_aarch64): Likewise.
22710 * config/aarch64/constraints.md (Ufc): New.
22711 (Y): call aarch64_float_const_zero_rtx.
22712 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
22714 2013-01-07 Richard Biener <rguenther@suse.de>
22716 PR tree-optimization/55888
22717 PR tree-optimization/55862
22718 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
22719 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
22720 not if it is contained therein.
22722 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
22724 * config/avr/t-avr: Typo.
22726 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
22729 * config/avr/t-avr: Don't automatically rebuild
22730 $(srcdir)/config/avr/t-multilib
22731 $(srcdir)/config/avr/avr-tables.opt
22732 $(srcdir)/doc/avr-mmcu.texi
22733 (avr-mcus): New phony target to build them on request.
22734 (s-avr-mlib, s-avr-mmcu-texi): Remove.
22735 * avr/avr-mcus.def: Adjust comments.
22737 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
22739 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
22741 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
22743 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
22745 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
22747 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
22749 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
22751 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
22752 to generate profiling.
22753 * config/rs6000/aix64.h (LIB_SPEC): Same.
22755 2013-01-04 Andrew Pinski <apinski@cavium.com>
22757 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
22759 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
22761 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
22763 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
22765 (ix86_expand_move): Ditto.
22766 (ix86_zero_extend_to_Pmode): Ditto.
22767 (ix86_expand_call): Ditto.
22768 (ix86_expand_special_args_builtin): Ditto.
22769 (ix86_expand_builtin): Ditto.
22771 2013-01-04 Richard Biener <rguenther@suse.de>
22773 PR tree-optimization/55862
22774 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
22775 translating them through PHI nodes.
22777 2013-01-04 Martin Jambor <mjambor@suse.cz>
22779 PR tree-optimization/55755
22780 * tree-sra.c (sra_modify_assign): Do not check that an access has no
22781 children when trying to avoid producing a VIEW_CONVERT_EXPR.
22783 2013-01-04 Marek Polacek <polacek@redhat.com>
22785 PR middle-end/55859
22786 * opts.c (default_options_optimization): Clarify error message.
22788 2013-01-04 Richard Biener <rguenther@suse.de>
22790 PR middle-end/55863
22791 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
22794 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
22797 * config/pa/pa.md (movsi): Revert previous change.
22798 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
22801 2013-01-03 Richard Henderson <rth@redhat.com>
22803 * config/i386/i386.c (ix86_expand_move): Always assign to op1
22804 after eliminating TLS symbols.
22806 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
22809 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
22810 * graphite-poly.c (debug_gmp_value): Likewise.
22812 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
22815 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
22816 selected code model, define __code_mode_small__, __code_model_medium__,
22817 __code_model_large__, __code_model_32__ or __code_model_kernel__.
22818 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
22819 xchg temporary register with %k. Declare temporary register as
22821 [__x86_64__]: For medium and large code models, preserve %rbx register.
22823 2013-01-03 Richard Biener <rguenther@suse.de>
22825 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
22826 (dump_subscript): Adjust.
22827 (finalize_ddr_dependent): Do not dump redundant info.
22828 (analyze_siv_subscript): Adjust.
22829 (subscript_dependence_tester): Likewise.
22830 (compute_affine_dependence): Likewise.
22832 2013-01-03 Richard Biener <rguenther@suse.de>
22835 2013-01-03 Richard Biener <rguenther@suse.de>
22837 PR tree-optimization/55857
22838 * tree-vect-stmts.c (vectorizable_load): Do not setup
22839 re-alignment for invariant loads.
22841 2013-01-02 Richard Biener <rguenther@suse.de>
22843 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22844 invariant load do not generate a vector load from the scalar location.
22846 2013-01-03 Richard Biener <rguenther@suse.de>
22848 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
22849 for not vectorizing.
22850 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
22851 not build INDIRECT_REFs, call get_name once only.
22852 (vect_create_data_ref_ptr): Likewise. Dump base object kind
22853 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
22855 2013-01-03 Richard Biener <rguenther@suse.de>
22857 PR tree-optimization/55857
22858 * tree-vect-stmts.c (vectorizable_load): Do not setup
22859 re-alignment for invariant loads.
22861 2013-01-03 Richard Biener <rguenther@suse.de>
22864 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
22865 prefer a built-in decl.
22867 2013-01-03 Jakub Jelinek <jakub@redhat.com>
22869 * gcc.c (process_command): Update copyright notice dates.
22870 * gcov.c (print_version): Likewise.
22871 * gcov-dump.c (print_version): Likewise.
22873 PR rtl-optimization/55838
22874 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
22875 iv0.step, iv1.step and step.
22877 2013-01-03 Jakub Jelinek <jakub@redhat.com>
22878 Marc Glisse <marc.glisse@inria.fr>
22880 PR tree-optimization/55832
22881 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
22882 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
22883 integer_{one,zero}_node.
22885 2013-01-03 Jakub Jelinek <jakub@redhat.com>
22888 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
22889 * var-tracking.c (reverse_op): Don't add reverse ops to
22890 VALUEs that have already
22891 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
22893 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
22895 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
22897 2013-01-02 Teresa Johnson <tejohnson@google.com>
22899 * dumpfile.c (dump_loc): Print filename with location.
22900 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
22901 new location_t parameter to emit complete unroll message with
22902 new dump framework.
22903 (canonicalize_loop_induction_variables): Compute loops location
22904 and pass to try_unroll_loop_completely.
22905 * loop-unroll.c (report_unroll_peel): New function.
22906 (peel_loops_completely): Use new dump format with location
22907 for main dumpfile message, and invoke report_unroll_peel on success.
22908 (decide_unrolling_and_peeling): Ditto.
22909 (decide_peel_once_rolling): Remove old dumpfile message subsumed
22910 by report_unroll_peel.
22911 (decide_peel_completely): Ditto.
22912 (decide_unroll_constant_iterations): Ditto.
22913 (decide_unroll_runtime_iterations): Ditto.
22914 (decide_peel_simple): Ditto.
22915 (decide_unroll_stupid): Ditto.
22916 * cfgloop.c (get_loop_location): New function.
22917 * cfgloop.h (get_loop_location): Declare.
22919 2013-01-02 Sriraman Tallam <tmsriram@google.com>
22921 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
22924 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
22926 PR middle-end/55198
22927 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
22928 BLKmode objects when EXPAND_MEMORY is specified.
22930 2013-01-02 Sriraman Tallam <tmsriram@google.com>
22932 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
22934 (fold_builtin_cpu): Do not share cpu model decls across statements.
22936 2013-01-02 Jason Merrill <jason@redhat.com>
22939 * tree.c (build_array_type_1): Revert earlier change.
22941 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
22943 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
22945 * config/aarch64/aarch64-tune.md: Re-generate.
22947 2013-01-02 Richard Biener <rguenther@suse.de>
22949 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22950 invariant load do not generate a vector load from the scalar location.
22952 2013-01-02 Richard Biener <rguenther@suse.de>
22955 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
22956 * configure: Regenerate.
22958 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
22960 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
22961 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
22962 (expand_builtin_int_roundingfn_2): Keep the original target around
22963 for the fallback case.
22965 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
22967 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
22968 to be clear for sign changes.
22970 2013-01-01 Jan Hubicka <jh@suse.cz>
22972 * ipa-inline-analysis.c: Fix formatting.
22974 2013-01-01 Jakub Jelinek <jakub@redhat.com>
22976 PR tree-optimization/55831
22977 * tree-vect-loop.c (get_initial_def_for_induction): Use
22978 gsi_after_labels instead of gsi_start_bb.
22980 Copyright (C) 2013 Free Software Foundation, Inc.
22982 Copying and distribution of this file, with or without modification,
22983 are permitted in any medium without royalty provided the copyright
22984 notice and this notice are preserved.