1 2013-02-27 Jakub Jelinek <jakub@redhat.com>
3 * opts.h: Include obstack.h.
4 (opts_concat): New prototype.
5 (opts_obstack): New declaration.
6 * opts.c (opts_concat): New function.
7 (opts_obstack): New variable.
8 (init_options_struct): Call gcc_init_obstack on opts_obstack.
9 (finish_options): Use opts_concat instead of concat
10 and XOBNEWVEC instead of XNEWVEC.
11 * opts-common.c (generate_canonical_option, decode_cmdline_option,
12 generate_option): Likewise.
13 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
17 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
20 2013-02-26 Jakub Jelinek <jakub@redhat.com>
23 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
25 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
27 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
28 (arm_block_move_unaligned_straight): Likewise.
29 (arm_adjust_block_mem): Likewise.
31 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
34 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
36 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
39 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
40 * config/c6x/c6x.h (dbx_register_map): Update declaration.
43 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
44 of prologue/epilogue functions.
47 * config/tilegx/tilegx.c (tilegx_expand_prologue):
48 Remove unused variable cfa_offset.
49 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
52 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
53 type promotion to unsigned.
56 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
57 for HOST_WIDE_INT of 32 bit / same size as int.
58 (arm_block_move_unaligned_straight): Likewise.
59 (arm_adjust_block_mem): Likewise.
62 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
65 2013-02-26 Marek Polacek <polacek@redhat.com>
67 PR tree-optimization/56426
68 * tree-ssa-loop.c (tree_ssa_loop_init): Always call
71 2013-02-26 Richard Biener <rguenther@suse.de>
74 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
75 unused variable loops.
77 2013-02-26 Jakub Jelinek <jakub@redhat.com>
79 PR tree-optimization/56448
80 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
81 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
82 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
83 later operands of the references, or even first operand for
84 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
86 PR tree-optimization/56443
87 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
88 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
89 to type_for_mode langhook.
91 2013-02-25 Matt Turner <mattst88@gmail.com>
93 * doc/invoke.texi: Document r4700.
95 2013-02-25 Richard Biener <rguenther@suse.de>
97 PR tree-optimization/56175
98 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
100 (simplify_bitwise_binary): ... here. Also guard the conversion
101 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
103 2013-02-25 Catherine Moore <clm@codesourcery.com>
106 2013-02-24 Catherine Moore <clm@codesourcery.com>
107 Maciej W. Rozycki <macro@codesourcery.com>
108 Tom de Vries <tom@codesourcery.com>
109 Nathan Sidwell <nathan@codesourcery.com>
110 Iain Sandoe <iain@codesourcery.com>
111 Nathan Froyd <froydnj@codesourcery.com>
112 Chao-ying Fu <fu@mips.com>
113 * doc/extend.texi: (micromips, nomicromips, nocompression):
114 Document new function attributes.
115 * doc/invoke.texi (minterlink-compressed, mmicromips,
116 m14k, m14ke, m14kec): Document new options.
117 (minterlink-mips16): Update documentation.
118 * doc/md.texi (ZC, ZD): Document new constraints.
119 * configure.ac (gcc_cv_as_micromips): Check if linker
120 supports the .set micromips directive.
121 * configure: Regenerate.
122 * config.in: Regenerate.
123 * config/mips/mips-tables.opt: Regenerate.
124 * config/mips/micromips.md: New file.
125 * constraints.md (ZC, AD): New constraints.
126 * config/mips/predicates.md (movep_src_register): New predicate.
127 (movep_src_operand): New predicate.
128 (non_volatile_mem_operand): New predicate.
129 * config/mips/mips.md (multimem): New type.
130 (length): Differentiate between 17-bit and 18-bit branch offsets.
131 (MOVEP1, MOVEP2): New mode iterator.
132 (mov_<load>l): Use ZC constraint.
133 (mov_<load>r): Likewise.
134 (mov_<store>l): Likewise.
135 (mov_<store>r): Likewise.
136 (*branch_equality<mode>_inverted): Add microMIPS support.
137 (*branch_equality<mode>): Likewise.
138 (*jump_absolute): Likewise.
139 (indirect_jump_<mode>): Likewise.
140 (tablejump_<mode>): Likewise.
141 (<optab>_internal): Likewise.
142 (sibcall_internal): Likewise.
143 (sibcall_value_internal): Likewise.
144 (prefetch): Use constraint ZD.
145 * config/mips/mips.opt (minterlink-compressed): New option.
146 (minterlink-mips16): Now an alias for minterlink-compressed.
147 (mmicromips): New option.
148 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
149 (compare_and_swap_12): Likewise.
150 (sync_add<mode>): Likewise.
151 (sync_<optab>_12): Likewise.
152 (sync_old_<optab>_12): Likewise.
153 (sync_new_<optab>_12): Likewise.
154 (sync_nand_12): Likewise.
155 (sync_old_nand_12): Likewise.
156 (sync_new_nand_12): Likewise.
157 (sync_sub<mode>): Likewise.
158 (sync_old_add<mode>): Likewise.
159 (sync_old_sub<mode>): Likewise.
160 (sync_new_add<mode>): Likewise.
161 (sync_new_sub<mode>): Likewise.
162 (sync_<optab><mode>): Likewise.
163 (sync_old_<optab><mode>): Likewise.
164 (sync_new_<optab><mode>): Likewise.
165 (sync_nand<mode>): Likewise.
166 (sync_old_nand<mode>): Likewise.
167 (sync_new_nand<mode>): Likewise.
168 (sync_lock_test_and_set<mode>): Likewise.
169 (test_and_set_12): Likewise.
170 (atomic_compare_and_swap<mode>): Likewise.
171 (atomic_exchange<mode>_llsc): Likewise.
172 (atomic_fetch_add<mode>_llsc): Likewise.
173 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
174 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
175 (umips_save_restore_pattern_p): Likewise.
176 (umips_load_store_pair_p): Likewise.
177 (umips_output_load_store_pair): Likewise.
178 (umips_movep_target_p): Likewise.
179 (umips_12bit_offset_address_p): Likewise.
180 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
181 (mips_base_mips16): Rename this...
182 (mips_base_compression_flags): ...to this. Update all uses.
183 (mips_attribute_table): Add micromips, nomicromips and nocompression.
184 (mips_mips16_decl_p): Delete.
185 (mips_nomips16_decl_p): Delete.
186 (mips_get_compress_on_flags): New function.
187 (mips_get_compress_off_flags): New function.
188 (mips_get_compress_mode): New function.
189 (mips_get_compress_on_name): New function.
190 (mips_get_compress_off_name): New function.
191 (mips_insert_attributes): Support multiple compression types.
192 (mips_merge_decl_attributes): Likewise.
193 (umips_12bit_offset_address_p): New function.
194 (mips_start_function_definition): Emit .set micromips directive.
195 (mips_call_may_need_jalx_p): New function.
196 (mips_function_ok_for_sibcall): Add microMIPS support.
197 (mips_print_operand_punctuation): Support short delay slots and
199 (umips_swm_mask, umips_swm_encoding): New.
200 (umips_build_save_restore): New function.
201 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
202 (was_mips16_p): Remove.
203 (old_compression_mode): New.
204 (mips_set_compression_mode): New function.
205 (mips_set_current_function): Add microMIPS support.
206 (mips_option_override): Likewise.
207 (umips_save_restore_pattern_p): New function.
208 (umips_output_save_restore): New function.
209 (umips_load_store_pair_p_1): New function.
210 (umips_load_store_pair_p): New function.
211 (umips_output_load_store_pair_1): New function.
212 (umips_output_load_store_pair): New function.
213 (umips_movep_target_p) New function.
214 (mips_prepare_pch_save): Add microMIPS support.
215 * config/mips/mips.h (TARGET_COMPRESSION): New.
216 (TARGET_CPU_CPP_BUILTINS): Update macro
217 to use new compression flags and to support microMIPS.
218 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
219 (MIPS_ARCH_FLOAT_SPEC): Likewise.
220 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
221 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
222 (ASM_SPEC): Support mmicromips and mno-micromips.
223 (M16STORE_REG_P): New macro.
224 (MIPS_CALL): Support TARGET_MICROMIPS.
225 (MICROMIPS_J): New macro.
226 (mips_base_mips16): Rename this...
227 (mips_base_compression_flags): ...to this.
228 (UMIPS_12BIT_OFFSET_P): New macro.
229 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
230 (MULTILIB_DIRNAMES): Likewise.
232 2013-02-25 Tom de Vries <tom@codesourcery.com>
234 PR rtl-optimization/56131
235 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
236 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
237 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
239 2013-02-25 Tobias Burnus <burnus@net-b.de>
241 * doc/invoke.texi (-fsanitize=): Move from optimization
242 to debugging options.
244 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
246 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
248 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
249 Alexander Monakov <amonakov@ispras.ru>
252 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
253 flush pending lists also on non-jumps. Adjust comment.
255 2013-02-24 Catherine Moore <clm@codesourcery.com>
256 Maciej W. Rozycki <macro@codesourcery.com>
257 Tom de Vries <tom@codesourcery.com>
258 Nathan Sidwell <nathan@codesourcery.com>
259 Iain Sandoe <iain@codesourcery.com>
260 Nathan Froyd <froydnj@codesourcery.com>
261 Chao-ying Fu <fu@mips.com>
263 * doc/extend.texi: (micromips, nomicromips, nocompression):
264 Document new function attributes.
265 * doc/invoke.texi (minterlink-compressed, mmicromips,
266 m14k, m14ke, m14kec): Document new options.
267 (minterlink-mips16): Update documentation.
268 * doc/md.texi (ZC, ZD): Document new constraints.
269 * configure.ac (gcc_cv_as_micromips): Check if linker
270 supports the .set micromips directive.
271 * configure: Regenerate.
272 * config.in: Regenerate.
273 * config/mips/mips-tables.opt: Regenerate.
274 * config/mips/micromips.md: New file.
275 * constraints.md (ZC, AD): New constraints.
276 * config/mips/predicates.md (movep_src_register): New predicate.
277 (movep_src_operand): New predicate.
278 (non_volatile_mem_operand): New predicate.
279 * config/mips/mips.md (multimem): New type.
280 (length): Differentiate between 17-bit and 18-bit branch offsets.
281 (MOVEP1, MOVEP2): New mode iterator.
282 (mov_<load>l): Use ZC constraint.
283 (mov_<load>r): Likewise.
284 (mov_<store>l): Likewise.
285 (mov_<store>r): Likewise.
286 (*branch_equality<mode>_inverted): Add microMIPS support.
287 (*branch_equality<mode>): Likewise.
288 (*jump_absolute): Likewise.
289 (indirect_jump_<mode>): Likewise.
290 (tablejump_<mode>): Likewise.
291 (<optab>_internal): Likewise.
292 (sibcall_internal): Likewise.
293 (sibcall_value_internal): Likewise.
294 (prefetch): Use constraint ZD.
295 * config/mips/mips.opt (minterlink-compressed): New option.
296 (minterlink-mips16): Now an alias for minterlink-compressed.
297 (mmicromips): New option.
298 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
299 (compare_and_swap_12): Likewise.
300 (sync_add<mode>): Likewise.
301 (sync_<optab>_12): Likewise.
302 (sync_old_<optab>_12): Likewise.
303 (sync_new_<optab>_12): Likewise.
304 (sync_nand_12): Likewise.
305 (sync_old_nand_12): Likewise.
306 (sync_new_nand_12): Likewise.
307 (sync_sub<mode>): Likewise.
308 (sync_old_add<mode>): Likewise.
309 (sync_old_sub<mode>): Likewise.
310 (sync_new_add<mode>): Likewise.
311 (sync_new_sub<mode>): Likewise.
312 (sync_<optab><mode>): Likewise.
313 (sync_old_<optab><mode>): Likewise.
314 (sync_new_<optab><mode>): Likewise.
315 (sync_nand<mode>): Likewise.
316 (sync_old_nand<mode>): Likewise.
317 (sync_new_nand<mode>): Likewise.
318 (sync_lock_test_and_set<mode>): Likewise.
319 (test_and_set_12): Likewise.
320 (atomic_compare_and_swap<mode>): Likewise.
321 (atomic_exchange<mode>_llsc): Likewise.
322 (atomic_fetch_add<mode>_llsc): Likewise.
323 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
324 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
325 (umips_save_restore_pattern_p): Likewise.
326 (umips_load_store_pair_p): Likewise.
327 (umips_output_load_store_pair): Likewise.
328 (umips_movep_target_p): Likewise.
329 (umips_12bit_offset_address_p): Likewise.
330 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
331 (mips_base_mips16): Rename this...
332 (mips_base_compression_flags): ...to this. Update all uses.
333 (mips_attribute_table): Add micromips, nomicromips and nocompression.
334 (mips_mips16_decl_p): Delete.
335 (mips_nomips16_decl_p): Delete.
336 (mips_get_compress_on_flags): New function.
337 (mips_get_compress_off_flags): New function.
338 (mips_get_compress_mode): New function.
339 (mips_get_compress_on_name): New function.
340 (mips_get_compress_off_name): New function.
341 (mips_insert_attributes): Support multiple compression types.
342 (mips_merge_decl_attributes): Likewise.
343 (umips_12bit_offset_address_p): New function.
344 (mips_start_function_definition): Emit .set micromips directive.
345 (mips_call_may_need_jalx_p): New function.
346 (mips_function_ok_for_sibcall): Add microMIPS support.
347 (mips_print_operand_punctuation): Support short delay slots and
349 (umips_swm_mask, umips_swm_encoding): New.
350 (umips_build_save_restore): New function.
351 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
352 (was_mips16_p): Remove.
353 (old_compression_mode): New.
354 (mips_set_compression_mode): New function.
355 (mips_set_current_function): Add microMIPS support.
356 (mips_option_override): Likewise.
357 (umips_save_restore_pattern_p): New function.
358 (umips_output_save_restore): New function.
359 (umips_load_store_pair_p_1): New function.
360 (umips_load_store_pair_p): New function.
361 (umips_output_load_store_pair_1): New function.
362 (umips_output_load_store_pair): New function.
363 (umips_movep_target_p) New function.
364 (mips_prepare_pch_save): Add microMIPS support.
365 * config/mips/mips.h (TARGET_COMPRESSION): New.
366 (TARGET_CPU_CPP_BUILTINS): Update macro
367 to use new compression flags and to support microMIPS.
368 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
369 (MIPS_ARCH_FLOAT_SPEC): Likewise.
370 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
371 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
372 (ASM_SPEC): Support mmicromips and mno-micromips.
373 (M16STORE_REG_P): New macro.
374 (MIPS_CALL): Support TARGET_MICROMIPS.
375 (MICROMIPS_J): New macro.
376 (mips_base_mips16): Rename this...
377 (mips_base_compression_flags): ...to this.
378 (UMIPS_12BIT_OFFSET_P): New macro.
379 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
380 (MULTILIB_DIRNAMES): Likewise.
382 2013-02-24 Jakub Jelinek <jakub@redhat.com>
385 * target-globals.c (save_target_globals): For init_reg_sets and
386 target_reinit remporarily set this_fn_optabs to this_target_optabs.
388 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
390 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
391 * config/aarch64/t-aarch64
392 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
394 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
397 * lra-constraints.c (process_alt_operands): Reload operand
398 conflicting with earlier clobber only if no more other conflicting
401 2013-02-22 Jakub Jelinek <jakub@redhat.com>
404 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
405 if not linking a shared library.
407 2013-02-22 Seth LaForge <sethml@google.com>
409 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
411 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
413 * config/arm/arm.md (split for extendsidi): Update condition.
414 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
415 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
416 (qhs_zextenddi_cstr): Likewise.
418 2013-02-21 Jakub Jelinek <jakub@redhat.com>
421 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
422 avoid signed wrapping.
423 (expand_mult): Handle properly multiplication by
424 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
425 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
426 in the compiler if coeff is HOST_WIDE_INT_MIN.
427 (expand_divmod): Don't make ext_op1 static, change it's type to
428 uhwi. Avoid undefined behavior in -INTVAL (op1).
430 PR rtl-optimization/50339
431 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
433 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
434 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
435 into splitting_ashiftrt field.
436 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
438 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
441 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
444 * trans-mem.c (execute_tm_mark): Do not expand transactions that
445 are sure to go irrevocable.
447 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
449 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
450 scalars are valid operands.
452 2013-02-21 Martin Jambor <mjambor@suse.cz>
454 PR tree-optimization/56310
455 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
456 only matching indices and non-negative final offsets.
457 (intersect_aggregates_with_edge): Pass src_idx to
458 agg_replacements_to_vector. Pass src_idx insstead of index to
459 intersect_with_agg_replacements.
461 2013-02-21 Martin Jambor <mjambor@suse.cz>
463 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
464 instead of hard-wired defaults.
466 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
468 * doc/invoke.texi (MIPS Options): Update documentation of the
469 floating-point multiply-accumulate instruction restrictions.
471 2013-02-21 Kostya Serebryany <kcc@google.com>
473 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
474 asan_shadow_offset on x86_64 linux.
476 2013-02-21 Richard Biener <rguenther@suse.de>
478 PR tree-optimization/56415
480 2013-02-11 Richard Biener <rguenther@suse.de>
482 PR tree-optimization/56273
483 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
486 2013-02-21 Jakub Jelinek <jakub@redhat.com>
489 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
493 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
494 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
496 2013-02-20 Jan Hubicka <jh@suse.cz>
498 PR tree-optimization/56265
499 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph when target is
500 referenced for firs ttime.
502 2013-02-20 Richard Biener <rguenther@suse.de>
504 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
505 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
506 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
507 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
509 (rename_ssa_copies): Do not remove unused locals.
510 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
511 * tree-ssanames.c (pass_release_ssa_names): Remove unused
513 * passes.c (execute_function_todo): Do not schedule unused locals
514 removal if cleanup_tree_cfg did something.
515 * tree-ssa-live.c (remove_unused_locals): Dump statistics
516 about the number of removed locals.
518 2013-02-20 Richard Biener <rguenther@suse.de>
520 PR tree-optimization/56398
521 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip
524 2013-02-20 Martin Jambor <mjambor@suse.cz>
526 PR tree-optimization/55334
527 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
528 restricted pointers to arrays.
530 2013-02-20 Richard Biener <rguenther@suse.de>
531 Jakub Jelinek <jakub@redhat.com>
533 PR tree-optimization/56396
534 * tree-ssa-ccp.c (n_const_val): New static variable.
535 (get_value): Return NULL for SSA names we don't have a lattice
537 (ccp_initialize): Initialize n_const_val.
538 * tree-ssa-copy.c (n_copy_of): New static variable.
539 (init_copy_prop): Initialize n_copy_of.
540 (get_value): Return NULL_TREE for SSA names we don't have a
543 2013-02-20 Martin Jambor <mjambor@suse.cz>
545 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
547 2013-02-20 Richard Biener <rguenther@suse.de>
549 * genpreds.c (write_lookup_constraint): Do not compare first
550 letter of the constraint again.
552 2013-02-20 Richard Biener <rguenther@suse.de>
554 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
556 (get_use_iv_cost): Terminate hashtable walk when coming across
559 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
561 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
562 reassociation for avx2 targets.
564 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
566 * config/microblaze/microblaze.c: microblaze_has_clz = 0
567 Add version check for v8.10.a to enable microblaze_has_clz
568 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
569 version and TARGET_PATTERN_COMPARE check
570 * config/microblaze/microblaze.md: New clzsi2 instruction
572 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
574 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
575 function before branching.
577 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
579 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
581 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
583 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
587 * sel-sched.c: Include ira.h.
588 (implicit_clobber_conflict_p): New function.
589 (moveup_expr): Use it.
590 * Makefile.in (sel-sched.o): Depend on ira.h.
592 2013-02-19 Richard Biener <rguenther@suse.de>
594 PR tree-optimization/56384
595 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
596 (vn_hash_type): Split out from ...
597 (vn_hash_constant_with_type): ... here.
598 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
599 (vn_phi_eq): Compare types from vn_phi_s structure.
600 (vn_phi_lookup): Populate vn_phi_s type.
601 (vn_phi_insert): Likewise.
603 2013-02-19 Jakub Jelinek <jakub@redhat.com>
605 PR tree-optimization/56350
606 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
607 if haven't found reduction or nested cycle operand, rather than
608 asserting we must find it.
610 PR tree-optimization/56381
611 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
614 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
615 Jakub Jelinek <jakub@redhat.com>
618 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
619 (swap_optab_enable): Same.
620 (init_all_optabs): Use argument instead of global.
621 * tree.h (struct tree_optimization_option): New field
623 * expr.h (init_all_optabs): Add argument to prototype.
624 (TREE_OPTIMIZATION_OPTABS): New.
625 (save_optabs_if_changed): Protoize.
626 * optabs.h: Declare this_fn_optabs.
627 * optabs.c (save_optabs_if_changed): New.
628 Declare this_fn_optabs.
629 (init_optabs): Add argument to init_all_optabs() call.
630 * function.c (invoke_set_current_function_hook): Handle per
632 * function.h (struct function): New field optabs.
633 * config/mips/mips.c (mips_set_mips16_mode): Handle when
634 optimization_current_node has changed.
635 * target-globals.h (save_target_globals_default_opts): Protoize.
636 * target-globals.c (save_target_globals_default_opts): New.
638 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
641 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
642 registers %fr12 and %fr12R as call used.
645 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
646 and HImode, require all displacements to be an integer multiple of their
648 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
649 only allow QImode and HImode when reload is in progress and strict is
650 true. Likewise for symbolic addresses. Use base14_operand to check
651 displacements in REG+BASE addresses.
653 2013-02-18 Richard Biener <rguenther@suse.de>
655 PR tree-optimization/56366
656 * tree-vect-loop.c (get_initial_def_for_induction): Properly
657 handle sign-conversion of outer-loop initial induction value.
659 2013-02-18 Richard Biener <rguenther@suse.de>
662 * cfghooks.c (merge_blocks): If we merge a latch into another
663 block adjust references to it.
664 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
665 (verify_loop_structure): Verify that a recorded latch is in fact
668 2013-02-18 Richard Biener <rguenther@suse.de>
670 PR tree-optimization/56321
671 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
672 order SSA name release and virtual operand unlinking.
674 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
676 * config/microblaze/microblaze.md (save_stack_block): Define.
677 (restore_stack_block): Likewise.
679 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
681 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
682 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
683 * config/microblaze/microblaze.c (microblaze_option_override):
684 Bail out early for PIC modes when target does not support PIC.
686 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
688 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
689 Replace with a microblaze version.
690 (microblaze_trampoline_init): Adapt for microblaze.
691 * gcc/config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
694 2013-02-16 Jakub Jelinek <jakub@redhat.com>
695 Dodji Seketeli <dodji@redhat.com>
698 * asan.c (get_mem_refs_of_builtin_call): White space and style
700 (instrument_mem_region_access): Do not forget to always put
701 instrumentation of the of 'base' and 'base + len' in a "if (len !=
702 0) statement, even for cases where either 'base' or 'base + len'
703 are not instrumented -- because they have been previously
704 instrumented. Simplify the logic by putting all the statements
705 instrument 'base + len' inside a sequence, and then insert that
706 sequence right before the current insertion point. Then, to
707 instrument 'base + len', just get an iterator on that statement.
708 And do not forget to update the pointer to iterator the function
709 received as argument.
711 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
713 PR rtl-optimization/56348
714 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
716 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
718 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
719 (clean_graph_dump_file): Pass base to start_graph_dump.
721 2013-02-14 Richard Henderson <rth@redhat.com>
724 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
726 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
728 * collect2-aix.h: Define F_LOADONLY.
730 2013-02-14 Richard Biener <rguenther@suse.de>
733 * varasm.c (output_constant_def_1): Get the decl representing
734 the constant as argument.
735 (output_constant_def): Wrap output_constant_def_1.
736 (make_decl_rtl): Use output_constant_def_1 with the decl
737 representing the constant.
738 (build_constant_desc): Optionally re-use a decl already
739 representing the constant.
740 (tree_output_constant_def): Adjust.
742 2013-02-14 Dodji Seketeli <dodji@redhat.com>
745 * asan.c (instrument_builtin_call): Really put the length of the
746 second source argument into src1_len.
748 2013-02-13 Jakub Jelinek <jakub@redhat.com>
750 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
751 argument. If it is false, don't create edge from then_bb to
753 (insert_if_then_before_iter): Pass true to it.
754 (build_check_stmt): Pass false to it.
755 (transform_statements): Flush hash table only on extended basic
756 block boundaries, rather than at the beginning of every bb.
757 Don't flush hash table on nonfreeing_call_p calls.
758 * tree-flow.h (nonfreeing_call_p): New prototype.
759 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
761 2013-02-13 David S. Miller <davem@davemloft.net>
763 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
765 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
768 * ira.c (max_regno_before_ira): Move from ...
770 (fix_reg_equiv_init): Use max_regno_before_ira instead of
773 2013-02-13 Jakub Jelinek <jakub@redhat.com>
775 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
777 2013-02-13 Richard Biener <rguenther@suse.de>
780 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
783 2013-02-13 Richard Biener <rguenther@suse.de>
785 * loop-init.c (loop_optimizer_init): Clear loop state when
786 re-initializing preserved loops.
787 * loop-unswitch.c (unswitch_single_loop): Return whether
788 we unswitched the loop. Do not verify loop state here.
789 (unswitch_loops): When we unswitched a loop discover new
792 2013-02-13 Kostya Serebryany <kcc@google.com>
794 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset on x86_64
796 * sanitizer.def: Rename __asan_init to __asan_init_v1.
798 2013-02-12 Dodji Seketeli <dodji@redhat.com>
800 Avoid instrumenting duplicated memory access in the same basic block
801 * Makefile.in (asan.o): Add new dependency on hash-table.h
802 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
803 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
804 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
805 (free_mem_ref_resources, has_mem_ref_been_instrumented)
806 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
807 (get_mem_ref_of_assignment): New functions.
808 (get_mem_refs_of_builtin_call): Extract from
809 instrument_builtin_call and tweak a little bit to make it fit with
811 (instrument_builtin_call): Use the new
812 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
813 of is_gimple_builtin_call.
814 (instrument_derefs, instrument_mem_region_access): Insert the
815 instrumented memory reference into the hash table.
816 (maybe_instrument_assignment): Renamed instrument_assignment into
817 this, and change it to advance the iterator when instrumentation
818 actually happened and return true in that case. This makes it
819 homogeneous with maybe_instrument_assignment, and thus give a
820 chance to callers to be more 'regular'.
821 (transform_statements): Clear the memory reference hash table
822 whenever we enter a new BB, when we cross a function call, or when
823 we are done transforming statements. Use
824 maybe_instrument_assignment instead of instrumentation. No more
825 need to special case maybe_instrument_assignment and advance the
826 iterator after calling it; it's now handled just like
827 maybe_instrument_call. Update comment.
829 2013-02-13 Richard Biener <rguenther@suse.de>
831 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
832 Fix loop discovery code.
834 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
837 * lra-constraints.c (process_alt_operands): Match early clobber
838 operand with itself. Check conflicts with earlyclobber only if
839 the operand is not reloaded. Prefer to reload conflicting operand
840 if earlyclobber and matching operands are the same.
842 2013-02-12 Richard Biener <rguenther@suse.de>
845 * lto-streamer-out.c (write_symbol): Do not output symbols
846 for hard register variables.
848 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
851 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
852 (umulsidi3_insn, mulsidi3_insn): New insns.
854 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
856 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
857 (struct tune_params): Add vec_costs field.
858 * config/arm/arm.c (arm_builtin_vectorization_cost)
859 (arm_add_stmt_cost): New functions.
860 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
861 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
862 (arm_default_vec_cost): New struct of type cpu_vec_costs.
863 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
864 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
865 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
866 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
868 2013-02-12 Richard Biener <rguenther@suse.de>
871 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
872 decls again if possible.
874 2013-02-12 Richard Biener <rguenther@suse.de>
877 * tree-ssa.c (verify_ssa_name): Fix check, move
878 SSA_NAME_IN_FREE_LIST check up.
880 2013-02-12 Jakub Jelinek <jakub@redhat.com>
881 Steven Bosscher <steven@gcc.gnu.org>
883 PR rtl-optimization/56151
884 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
885 equal to op0 or op1, and last_insn pattern is CODE operation
886 with MEM dest and one of the operands matches that MEM.
888 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
890 * doc/extend.texi: Document Function Multiversioning and "default"
891 parameter string to target attribute.
892 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
893 target attribute parameter is "default".
894 (ix86_compare_version_priority): Remove checks for target attribute.
895 (ix86_mangle_function_version_assembler_name): Change error to sorry.
896 Remove check for target attribute equal to NULL. Add assert.
897 (ix86_generate_version_dispatcher_body): Change error to sorry.
899 2013-02-11 Iain Sandoe <iain@codesourcery.com>
900 Jack Howarth <howarth@bromo.med.uc.edu>
901 Patrick Marlier <patrick.marlier@gmail.com>
904 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
905 define ENDFILE_SPEC as TM_DESTRUCTOR.
906 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
908 2013-02-11 Alexander Potapenko <glider@google.com>
909 Jack Howarth <howarth@bromo.med.uc.edu>
910 Jakub Jelinek <jakub@redhat.com>
913 * config/darwin.c (cdtor_record): Rename ctor_record.
914 (sort_cdtor_records): Rename sort_ctor_records.
915 (finalize_dtors): New routine to sort destructors by
916 priority before use in assemble_integer.
917 (machopic_asm_out_destructor): Use finalize_dtors if needed.
919 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
921 PR rtl-optimization/56275
922 * simplify-rtx.c (avoid_constant_pool_reference): Check that
923 offset is non-negative and less than cmode size before
924 calling simplify_subreg.
926 2013-02-11 Richard Biener <rguenther@suse.de>
928 PR tree-optimization/56264
929 * cfgloop.h (fix_loop_structure): Adjust prototype.
930 * loop-init.c (fix_loop_structure): Return the number of
931 newly discovered loops.
932 * tree-cfgcleanup.c (repair_loop_structures): When new loops
933 are discovered, do a full loop-closed SSA rewrite.
935 2013-02-11 Richard Biener <rguenther@suse.de>
937 PR tree-optimization/56273
938 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
940 (check_array_ref): Fix missing newline in dumps.
941 (search_for_addr_array): Likewise.
943 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
945 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
947 2013-02-09 Jakub Jelinek <jakub@redhat.com>
950 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
952 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
954 PR rtl-optimization/56246
955 * lra-constraints.c (simplify_operand_subreg): Try to reuse
957 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
958 constraints are satisfied.
960 2013-02-08 Jeff Law <law@redhat.com>
963 * emit-rtl.c (reg_is_parm_p): New function.
964 * regs.h (reg_is_parm_p): New prototype.
965 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
966 callee-clobbered registers.
968 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
971 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
972 If there is no implicit builtin declaration, just return NULL.
974 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
976 * config/i386/sse.md (FMAMODEM): New mode iterator.
977 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
978 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
980 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
982 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
983 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
984 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
986 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
988 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
989 (microblaze*-*-elf): Likewise.
990 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
992 * config/microblaze/microblaze-c.c: Add builtin defines for
993 _LITTLE_ENDIAN and _BIG_ENDIAN.
994 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
995 add to TARGET_DEFAULT flags.
996 Expand ASM_SPEC and LINK_SPEC.
997 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
998 * config/microblaze/microblaze.md: Update extendsidi2 and
999 movdi_internal instructions to use low-order / high-order reg
1001 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
1002 options and inversemask / mask of LITTLE_ENDIAN.
1003 * config/microblaze/t-microblaze: Expand multilib options to
1004 include mlittle-endian (le) and update exceptions patterns.
1006 2013-02-08 Jakub Jelinek <jakub@redhat.com>
1008 PR rtl-optimization/56195
1009 * lra-constraints.c (get_reload_reg): Don't reuse regs
1010 if they have smaller mode than requested, if they have
1011 wider mode than requested, try to return a SUBREG.
1013 PR tree-optimization/56250
1014 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
1015 if type is unsigned and code isn't MULT_EXPR.
1017 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
1019 PR tree-optimization/56064
1020 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
1021 bits according to mode.
1022 * fixed-value.h (fixed_from_double_int)
1023 (const_fixed_from_double_int): Adjust comments.
1025 2013-02-08 Richard Biener <rguenther@suse.de>
1028 * lto-streamer.h (struct data_in): Remove current_file, current_line
1029 and current_col members.
1030 * lto-streamer-out.c (lto_output_location): Stream changed bits
1031 en-block for efficiency.
1032 * lto-streamer-in.c (clear_line_info): Remove.
1033 (lto_input_location): Cache current file, line and column
1034 globally via local statics. Read changed bits en-block.
1035 (input_function): Do not call clear_line_info.
1036 (lto_read_body): Likewise.
1037 (lto_input_toplevel_asms): Likewise.
1039 2013-02-08 Michael Matz <matz@suse.de>
1041 PR tree-optimization/52448
1042 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
1043 (nt_call_phase): New static.
1044 (add_or_mark_expr): Only mark accesses with newer phase than any
1046 (nonfreeing_call_p): New.
1047 (nt_init_block): Update nt_call_phase, mark blocks as visited.
1048 (nt_fini_block): Keep blocks marked as visited.
1049 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
1051 2013-02-08 Richard Biener <rguenther@suse.de>
1053 * ira.c (ira): Free broken dominator information.
1055 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
1057 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
1059 2013-02-08 Marek Polacek <polacek@redhat.com>
1061 * cfgloop.c (verify_loop_structure): Add more checking of headers.
1063 2013-02-08 Richard Biener <rguenther@suse.de>
1066 * cfgloop.h (flow_loops_find): Adjust.
1067 (bb_loop_header_p): Declare.
1068 * cfgloop.c (bb_loop_header_p): New function split out from ...
1069 (flow_loops_find): ... here. Adjust function signature,
1070 support incremental loop structure update.
1071 (verify_loop_structure): Cleanup. Verify a loop is a loop.
1072 * cfgloopmanip.c (fix_loop_structure): Move ...
1073 * loop-init.c (fix_loop_structure): ... here.
1074 (apply_loop_flags): Split out from ...
1075 (loop_optimizer_init): ... here.
1076 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
1077 in incremental mode, only remove dead loops here.
1079 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
1082 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
1083 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
1084 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
1085 (*round<mode>3.libgcc): New insns for fixed-modes.
1086 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
1087 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
1088 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
1089 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
1090 implementations. Define to __builtin_avr_absFX,
1091 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
1092 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
1093 __builtin_avr_countlsFX, respectively.
1094 * config/avr/avr-c.c (target.h): Include it.
1095 (enum avr_builtin_id): New enum.
1096 (avr_resolve_overloaded_builtin): New static function.
1097 (avr_register_target_pragmas): Use it to set
1098 targetm.resolve_overloaded_builtin.
1099 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
1100 tree nodes used by DEF_BUILTIN.
1101 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
1102 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
1103 <AVR_BUILTIN_xxBITS>: Same.
1105 2013-02-08 Richard Biener <rguenther@suse.de>
1107 * cfgloop.c (verify_loop_structure): Properly handle
1108 a loop exiting to another loop header.
1109 * ira-int.h (ira_loops): Remove.
1110 * ira.c (ira_loops): Remove.
1111 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
1112 (do_reload): Use loop_optimizer_finalize.
1113 * ira-build.c (create_loop_tree_nodes): Use get_loops and
1114 number_of_loops to access the loop tree.
1115 (more_one_region_p): Likewise.
1116 (finish_loop_tree_nodes): Likewise.
1117 (rebuild_regno_allocno_maps): Likewise.
1118 (mark_loops_for_removal): Likewise.
1119 (mark_all_loops_for_removal): Likewise.
1120 (remove_unnecessary_regions): Likewise.
1121 (ira_build): Likewise.
1122 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
1124 2013-02-08 Richard Biener <rguenther@suse.de>
1126 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
1127 * ipa-pure-const.c (analyze_function): Avoid calling
1128 mark_irreducible_loops twice.
1129 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
1131 2013-02-07 David S. Miller <davem@davemloft.net>
1133 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
1135 * var-tracking.c (vt_add_function_parameter): Test the presence of
1136 HAVE_window_save properly and do not remap argument registers when
1137 we have a leaf function.
1139 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
1142 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
1144 * config/i386/i386.c (ix86_print_operand): Ditto.
1146 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1148 * lra-constraints.c (process_alt_operands): Fix recently added comment.
1150 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1152 PR rtl-optimization/56225
1153 * lra-constraints.c (process_alt_operands): Check that reload hard
1154 reg can hold value for strict_low_part.
1156 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1159 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
1160 dwarf2out_end_function.
1161 (in_first_function_p, maybe_at_text_label_p,
1162 first_loclabel_num_not_at_text_label): New variables.
1163 (dwarf2out_var_location): In the first function find out
1164 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
1165 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
1168 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
1170 PR rtl-optimization/56178
1171 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
1172 SUBREG of a register. Tidy up related block of code.
1173 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
1174 note if the source is a register or a SUBREG of a register.
1176 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1179 * config/rs6000/rs6000.md (ptrm): New mode attr.
1180 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
1181 call_value_indirect_aix<pttrsize>,
1182 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
1185 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1187 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
1188 if -bnortl. Convert to strcmp and strncmp.
1190 2013-02-07 Alan Modra <amodra@gmail.com>
1193 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
1194 addresses won't wrap when offsetting.
1195 (rs6000_secondary_reload): Provide secondary reloads needed for
1196 wrapping LO_SUM addresses.
1198 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
1200 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
1201 MACH, just __MACH__.
1203 2013-02-06 Richard Biener <rguenther@suse.de>
1205 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
1206 instead of calling fix_loop_structure.
1208 2013-02-06 Jakub Jelinek <jakub@redhat.com>
1211 * omp-low.c (use_pointer_for_field): Return false if
1212 lower_send_shared_vars doesn't generate any copy-out code.
1214 2013-02-06 Tom de Vries <tom@codesourcery.com>
1216 PR rtl-optimization/56131
1217 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
1218 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
1219 of the label is NULL. Add comment.
1221 2013-02-05 Jakub Jelinek <jakub@redhat.com>
1223 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
1226 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
1227 (STATIC_LIBTSAN_LIBS): Likewise.
1228 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
1229 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
1230 is defined, don't add anything else beyond that.
1231 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
1232 (LINK_COMMAND_SPEC): Use them.
1234 PR tree-optimization/56205
1235 * tree-stdarg.c (check_all_va_list_escapes): Return true if
1236 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
1237 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
1239 2013-02-05 Richard Biener <rguenther@suse.de>
1241 PR tree-optimization/53342
1242 PR tree-optimization/53185
1243 * tree-vectorizer.h (vect_check_strided_load): Remove.
1244 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
1245 not disallow peeling for vectorized strided loads.
1246 (vect_check_strided_load): Make static and simplify.
1247 (vect_analyze_data_refs): Adjust.
1248 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
1249 correctly when vectorizing strided loads.
1251 2013-02-05 Richard Biener <rguenther@suse.de>
1253 * doc/install.texi: Refer to ISL, not PPL.
1255 2013-02-05 Jan Hubicka <jh@suse.cz>
1257 PR tree-optimization/55789
1258 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
1260 2013-02-05 Jan Hubicka <jh@suse.cz>
1262 PR tree-optimization/55789
1263 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
1264 the dead call anyway.
1266 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
1269 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
1271 2013-02-04 Alexander Potapenko <glider@google.com>
1272 Jack Howarth <howarth@bromo.med.uc.edu>
1273 Jakub Jelinek <jakub@redhat.com>
1276 * config/darwin.c (sort_ctor_records): Stabilized qsort
1277 on constructor priority by using original position.
1278 (finalize_ctors): New routine to sort constructors by
1279 priority before use in assemble_integer.
1280 (machopic_asm_out_constructor): Use finalize_ctors if needed.
1282 2013-02-04 Jakub Jelinek <jakub@redhat.com>
1285 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
1286 about visibility on artificial decls.
1287 * config/sol2.c (solaris_assemble_visibility): Likewise.
1289 2013-02-04 Kai Tietz <ktietz@redhat.com>
1292 * config/i386/i386.c (function_value_ms_64): Add additional valtype
1293 argument and improve checking of return-argument types for 16-byte
1295 (ix86_function_value_1): Add additional valtype argument on call
1296 of function_value_64.
1297 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
1298 handling infunction_value_64 function.
1300 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1302 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
1304 2013-02-04 Richard Biener <rguenther@suse.de>
1306 PR tree-optimization/56188
1307 * tree-ssa-structalias.c (label_visit): Consider case with
1308 initially non-empty points-to set.
1309 (perform_var_substitution): Dump node mapping and clean up.
1311 2013-02-04 Richard Guenther <rguenther@suse.de>
1314 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
1315 node prevail as last resort.
1316 (lto_symtab_merge_decls): Remove guard on LTRANS here.
1317 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
1319 2013-02-04 Richard Biener <rguenther@suse.de>
1321 PR tree-optimization/56113
1322 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
1324 (equiv_class_lookup_or_add): ... this.
1325 (label_visit): Adjust and fix error in previous patch.
1326 (perform_var_substitution): Adjust.
1328 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
1330 * config/sh/divtab.c: Fix formatting and comments throughout the file.
1331 * config/sh/sh4-300.md: Likewise.
1332 * config/sh/sh4a.md: Likewise.
1333 * config/sh/constraints.md: Likewise.
1334 * config/sh/sh.md: Likewise.
1335 * config/sh/netbsd-elf.h: Likewise.
1336 * config/sh/predicates.md: Likewise.
1337 * config/sh/sh-protos.h: Likewise.
1338 * config/sh/ushmedia.h: Likewise.
1339 * config/sh/linux.h: Likewise.
1340 * config/sh/sh.c: Likewise.
1341 * config/sh/superh.h: Likewise.
1342 * config/sh/elf.h: Likewise.
1343 * config/sh/sh4.md: Likewise.
1344 * config/sh/sh.h: Likewise.
1346 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1348 * config/pa/constraints.md: Adjust unused letters. Change "T"
1349 constraint to match_test floating_point_store_memory_operand().
1350 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
1351 (base14_operand): New.
1352 (floating_point_store_memory_operand): New.
1353 (integer_store_memory_operand): Revise to use base14_operand and
1354 reg_plus_base_memory_operand.
1355 (move_dest_operand): Allow symbolic_memory_operands.
1356 (symbolic_memory_operand): Check for LO_SOM.
1357 (symbolic_operand): Change default case to break.
1358 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
1359 CONST_DOUBLE values to be reloaded by putting them into memory when
1360 the destination is a floating point register.
1361 (movdf): Remove code to handle CONST_DOUBLE.
1363 (reload_indf_r1): New.
1364 (reload_insf_r1): New.
1365 Consistently use "Q" and "T" constraints with integer and floating
1366 point move instructions, respectively.
1367 (movdi): Remove FAIL.
1368 Change predicate for source operand unamed DImode move from
1369 general_operand to move_src_operand.
1370 (umulsidi3): Change predicate for destination operand to
1372 Likewise for similar unamed patterns.
1373 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
1374 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
1375 (hppa_legitimize_address): Simplify mask calculation.
1376 (pa_emit_move_sequence): Revised handling of secondary reloads from
1377 REG+D addresses for floating point loads and stores. Directly handle
1378 loading CONST0_RTX (mode) to a floating point register.
1379 (pa_secondary_reload): Handle reloading DF and SFmode constant values
1380 to floating point registers. Don't restrict secondary reloads to
1381 floating point registers to integer modes. Revise some comments and
1383 (TARGET_LEGITIMATE_ADDRESS_P): Define.
1384 (pa_legitimate_address_p): New.
1385 (pa_legitimize_reload_address): New.
1386 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
1387 (STRICT_REG_OK_FOR_BASE_P): New.
1388 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
1389 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
1391 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
1392 Andrew Dixie <andrewd@gentrack.com>
1394 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
1397 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
1399 * expmed.c (extract_bit_field_1): Pass the full width of the
1400 structure to get_best_reg_extraction_insn.
1402 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1405 * configure.ac (use_cxa_atexit): Add AIX.
1406 * configure: Regenerate.
1408 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
1410 2013-02-01 Jakub Jelinek <jakub@redhat.com>
1413 * final.c (need_profile_function): New variable.
1414 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
1415 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
1416 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
1417 notes, targetm.asm_out.function_prologue doesn't emit anything,
1418 HAVE_prologue and profiler should be emitted before prologue,
1419 set need_profile_function instead of emitting it.
1420 (final_scan_insn): If need_profile_function, emit
1421 profile_function on the first NOTE_INSN_BASIC_BLOCK or
1422 NOTE_INSN_FUNCTION_BEG note.
1424 2013-02-01 Richard Henderson <rth@redhat.com>
1426 * config/rs6000/rs6000.md (smulditi3): New.
1429 * config/alpha/alpha.md (umulditi3): New.
1431 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1433 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
1434 (ASM_OUTPUT_ALIGNED_LOCAL): New.
1436 2013-02-01 Richard Biener <rguenther@suse.de>
1438 PR tree-optimization/56113
1439 * tree-ssa-structalias.c (label_visit): Reduce work for
1440 single-predecessor nodes.
1442 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
1444 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
1445 range isn't testing for zero.
1447 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
1450 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
1452 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
1453 Nick Clifton <nickc@redhat.com>
1455 * config/v850/constraints.md (Q): Define as a memory constraint.
1456 * config/v850/predicates.md (label_ref_operand): New predicate.
1457 (e3v5_shift_operand): New predicate.
1458 (ior_operator): New predicate.
1459 * config/v850/t-v850: Add e3v5 multilib.
1460 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
1461 (v850_gen_movdi): Prototype.
1462 * config/v850/v850.c: Add support for e3v5 architecture.
1463 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
1465 (construct_save_jarl): Add e3v5 long JARL support.
1466 (v850_adjust_insn_length): New function. Adjust length of call
1467 insns when using e3v5 instructions.
1468 (v850_gen_movdi): New function: Generate instructions to move a
1470 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
1471 (CPP_SPEC): Define __v850e3v5__ as appropriate.
1472 (TARGET_USE_FPU): Enable for e3v5.
1473 (CONST_OK_FOR_W): New macro.
1474 (ADJUST_INSN_LENGTH): Define.
1475 * config/v850/v850.md (UNSPEC_LOOP): Define.
1476 (attr cpu): Add v850e3v5.
1477 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
1478 (movdi): New pattern.
1479 (movdi_internal): New pattern.
1480 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
1481 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
1482 (cstoresf4): Likewise.
1483 (cstoredf4): Likewise.
1484 (insv): New pattern.
1485 (rotlso3_a): New pattern.
1486 (rotlsi3_b): New pattern
1487 (rotlsi3_v850e3v5): New pattern.
1488 (doloop_begin): New pattern.
1489 (fix_loop_counter): New pattern.
1490 (doloop_end): New pattern.
1491 (branch_normal): Add e3v5 long branch support.
1492 (branch_invert): Likewise.
1493 (branch_z_normal): Likewise.
1494 (branch_z_invert): Likewise.
1495 (branch_nz_normal): Likewise.
1496 (branch_nz_invert): Likewise.
1497 (call_internal_short): Add e3v5 register-indirect JARL support.
1498 (call_internal_long): Likewise.
1499 (call_value_internal_short): Likewise.
1500 (call_value_internal_long): Likewise.
1501 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
1502 (mloop): New option.
1503 * config.gcc: Add support for configuring v840e3v5 target.
1504 * doc/invoke.texi: Document new v850 specific command line options.
1506 2013-01-31 Paul Koning <ni1d@arrl.net>
1510 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
1511 children if parent is a class.
1512 (prune_unused_types_prune): Don't add DW_AT_declaration.
1514 2013-01-31 Richard Biener <rguenther@suse.de>
1516 PR tree-optimization/56157
1517 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
1518 match up operand with SLP child.
1520 2013-01-31 Jason Merrill <jason@redhat.com>
1523 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
1524 parameters the first time.
1525 (gen_scheduled_generic_parms_dies): Check completeness here.
1527 2013-01-31 Richard Biener <rguenther@suse.de>
1530 * common.opt (faggressive-loop-optimizations): New flag,
1532 * doc/invoke.texi (faggressive-loop-optimizations): Document.
1533 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
1534 infer_loop_bounds_from_undefined by it.
1536 2013-01-31 Richard Biener <rguenther@suse.de>
1538 PR tree-optimization/56150
1539 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
1540 visit virtual operands.
1541 (find_uses_to_rename_bb): Likewise.
1543 2013-01-31 Richard Biener <rguenther@suse.de>
1545 PR tree-optimization/56150
1546 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
1547 mixed store non-store stmts.
1549 2013-01-30 Jakub Jelinek <jakub@redhat.com>
1552 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
1553 LIBASAN_EARLY_SPEC is defined.
1554 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
1555 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
1557 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
1560 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
1561 invalid args instead of ICEing on it.
1562 (ix86_valid_target_attribute_tree): Return error_mark_node if
1563 ix86_valid_target_attribute_inner_p failed.
1564 (ix86_valid_target_attribute_p): Return false only if
1565 ix86_valid_target_attribute_tree returned error_mark_node. Allow
1566 target("default") attribute.
1567 (sorted_attr_string): Change argument from const char * to tree,
1568 merge in all target attribute arguments rather than just one.
1569 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
1570 instead of free. Avoid using strcat.
1571 (ix86_mangle_function_version_assembler_name): Mangle
1572 target("default") as if no target attribute is present. Adjust
1573 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
1574 instead of xmalloc and XDELETEVEC instead of free.
1575 (ix86_function_versions): Don't return true if one of the decls
1576 doesn't have target attribute. If they don't and one of the decls
1577 is DECL_FUNCTION_VERSIONED, report an error. Adjust
1578 sorted_attr_string caller. Use XDELETEVEC instead of free.
1579 (ix86_supports_function_versions): Remove.
1580 (make_name): Fix up formatting.
1581 (make_dispatcher_decl): Remove resolver_name and its initialization.
1582 Avoid leaking memory.
1583 (is_function_default_version): Return true if there is
1584 target("default") attribute rather than no target attribute at all.
1585 (make_resolver_func): Avoid leaking memory.
1586 (ix86_generate_version_dispatcher_body): Likewise.
1587 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
1588 * target.def (supports_function_versions): Remove.
1589 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
1590 * doc/tm.texi: Regenerated.
1592 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
1594 PR rtl-optimization/56144
1595 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
1596 for values with side effects.
1598 2013-01-30 Richard Biener <rguenther@suse.de>
1600 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
1601 (sparseset_pop): Likewise.
1602 * cfganal.c (compute_idf): Likewise. Increase work-stack size
1603 to be able to use quick_push in the worker loop.
1605 2013-01-30 Marek Polacek <polacek@redhat.com>
1607 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
1609 2013-01-30 Richard Biener <rguenther@suse.de>
1612 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
1614 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
1616 PR tree-optimization/56064
1617 * fixed-value.c (fixed_from_double_int): New function.
1618 * fixed-value.h (fixed_from_double_int): New prototype.
1619 (const_fixed_from_double_int): New static inline function.
1620 * fold-const.c (native_interpret_fixed): New static function.
1621 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
1622 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
1623 (native_encode_fixed): New static function.
1624 (native_encode_expr) <FIXED_CST>: Use it.
1625 (native_interpret_int): Move double_int worker code to...
1626 * double-int.c (double_int::from_buffer): ...this new static method.
1627 * double-int.h (double_int::from_buffer): Prototype it.
1629 2013-01-30 Richard Biener <rguenther@suse.de>
1631 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
1632 New pointer-map and obstack.
1633 (init_alias_vars): Allocate pointer-map and obstack.
1634 (delete_points_to_sets): Free them.
1635 (find_what_var_points_to): Cache result.
1636 (find_what_p_points_to): Adjust for changed interface of
1637 find_what_var_points_to.
1638 (compute_points_to_sets): Likewise.
1639 (ipa_pta_execute): Likewise.
1641 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1643 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
1644 * configure: Regenerate.
1645 * config.in: Regenerate.
1646 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
1647 #nobits/#progbits if supported.
1649 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
1652 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
1653 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
1654 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
1656 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1658 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
1659 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
1661 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1663 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
1664 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
1666 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1668 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
1670 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
1671 * config/arm/cortex-a7.md: New bypasses using
1672 arm_mac_accumulator_is_result.
1674 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1676 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
1677 (cortex_a7_neon_mla): Likewise.
1678 (cortex_a7_fpfmad): New reservation.
1679 (cortex_a7_fpmacs): Use ffmas and update required units.
1680 (cortex_a7_fpmuld): Update required units and latency.
1681 (cortex_a7_fpmacd): Likewise.
1682 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
1683 (cortex_a7_neon). Likewise.
1684 (bypass) Update participating units.
1686 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1688 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
1689 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
1691 * config/arm/vfp11.md (vfp_farith): Use ffmas.
1692 (vfp_fmul): Use ffmad.
1693 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
1694 (cortex_r4_fmacd): Use ffmad.
1695 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
1696 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
1697 (cortex_a9_fmacd): Use ffmad.
1698 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
1699 (cortex_a8_vfp_macd): Use ffmad.
1700 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
1701 (cortex_a5_fpmacd): Use ffmad.
1702 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
1703 (cortex_a15_vfp_macd): Use ffmad.
1704 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
1706 2013-01-29 Jason Merrill <jason@redhat.com>
1709 * varasm.c (default_assemble_visibility): Don't warn about
1710 visibility on artificial decls.
1712 2013-01-29 Richard Biener <rguenther@suse.de>
1714 PR tree-optimization/56113
1715 * tree-ssa-structalias.c (equiv_class_lookup): Also return
1717 (label_visit): Free duplicate bitmaps and record the leader instead.
1718 (perform_var_substitution): Adjust.
1720 2013-01-29 Richard Biener <rguenther@suse.de>
1722 PR tree-optimization/55270
1723 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
1724 the CFG, schedule loops for fixup.
1726 2013-01-29 Nick Clifton <nickc@redhat.com>
1728 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
1731 2013-01-28 Leif Ekblad <leif@rdos.net>
1733 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
1734 * config/i386/i386.h (TARGET_RDOS): New macro.
1735 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
1736 * config/i386/i386.c (ix86_option_override_internal): For 64bit
1737 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
1738 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
1739 DEFAULT_LARGE_SECTION_THRESHOLD.
1740 * config/i386/i386.md (R14_REG, R15_REG): New constants.
1741 * config/i386/rdos.h: New file.
1742 * config/i386/rdos64.h: New file.
1744 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
1747 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
1750 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1752 PR rtl-optimization/56117
1753 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
1754 call cselib_lookup_from_insn on the MEM before calling
1755 add_insn_mem_dependence.
1757 2013-01-28 Richard Biener <rguenther@suse.de>
1759 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
1760 to a stmt that didn't have one.
1761 (copy_phis_for_bb): Likewise for PHI arguments.
1762 (copy_debug_stmt): Likewise for debug stmts.
1764 2013-01-28 Richard Biener <rguenther@suse.de>
1766 PR tree-optimization/56034
1767 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
1768 (partition_builtin_p): Adjust.
1769 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
1770 it is the last partition.
1771 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
1772 up the vertex for the definition.
1773 (classify_partition): Classify whether a partition is a
1774 PKIND_REDUCTION, thus has uses outside of the loop.
1775 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
1776 Merge all PKIND_REDUCTION partitions into the last partition.
1777 (tree_loop_distribution): Seed partitions from reductions as well.
1779 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1781 PR tree-optimization/56125
1782 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
1783 pow(x,c) into sqrt(x) * powi(x, n/2) or
1784 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
1785 optimizing for size.
1786 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
1787 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
1790 PR tree-optimization/56094
1791 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
1792 to UNKNOWN_LOCATION while gimplifying expr.
1794 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
1797 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
1798 operand 0 in movabs insn template for -masm=intel asm alternative.
1799 (*movabs<mode>_2): Ditto for operand 1.
1801 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
1804 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
1807 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1809 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
1812 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
1814 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
1815 Undef to avoid warning.
1817 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1819 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
1820 * configure: Regenerate.
1822 2013-01-25 Jakub Jelinek <jakub@redhat.com>
1824 PR tree-optimization/56098
1825 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
1826 for stmts with volatile ops.
1827 (cond_store_replacement): Don't optimize if assign has volatile ops.
1828 (cond_if_else_store_replacement_1): Don't optimize if either
1829 then_assign or else_assign have volatile ops.
1830 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
1833 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1835 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
1837 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1839 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
1840 missing ':' in asm example.
1842 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
1844 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
1845 entries into lane and laneq entries.
1846 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
1847 Remove AdvSIMD scalar modes.
1848 (aarch64_sq<r>dmulh_laneq<mode>): New.
1849 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
1851 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
1852 builtin implementations to relfect changes in RTL in aarch64-simd.md.
1853 * config/aarch64/iterators.md (VCOND): New.
1856 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1859 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
1860 Add NULL LIBNAME argument to existing definitions.
1861 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
1862 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
1863 * config/avr/avr.c (DEF_BUILTIN): Same.
1864 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
1865 (avr_expand_builtin): Expand to a vanilla call if a libgcc
1866 implementation is available (DECL_ASSEMBLER_NAME is set).
1867 (avr_fold_absfx): New static function.
1868 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
1869 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
1870 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
1872 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
1873 (abshk, absk, abslk, absllk): Provide as static inline functions.
1875 2013-01-25 Marek Polacek <polacek@redhat.com>
1877 PR tree-optimization/56035
1878 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
1880 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
1882 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
1883 (*movtf_internal_rex64): Add (!o,C) alternative
1884 (*movxf_internal_rex64): Ditto.
1885 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
1887 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
1889 * doc/invoke.texi: fix typo.
1890 * doc/objc.texi: fix typo.
1892 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
1894 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
1895 for the first two alternatives.
1897 2013-01-24 Diego Novillo <dnovillo@google.com>
1899 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
1900 (ggc-zone.o): Remove.
1901 * configure.ac: Remove option --with-gc.
1902 * configure: Re-generate.
1903 * doc/install.texi: Remove documentation for --with-gc.
1904 * gengtype.c (write_enum_defn): Remove. Update all users.
1905 (write_Types_process_field): Remove generation of gt_e_* argument.
1906 (output_type_enum): Remove. Update all users.
1907 (write_enum_defn): Remove. Update all users.
1908 (enum alloc_zone): Remove. Update all users.
1909 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
1910 * ggc-common.c (ggc_splay_alloc): Remove first argument.
1912 (struct ptr_data): Remove field TYPE. Update all users.
1913 (gt_pch_note_object): Remove argument TYPE. Update all users.
1914 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
1916 * ggc-none.c (ggc_alloc_typed_stat): Remove.
1917 (struct alloc_zone): Remove.
1918 (ggc_internal_alloc_zone_stat): Remove.
1919 (ggc_internal_cleared_alloc_zone_stat): Remove.
1920 * ggc-page.c (ggc_alloc_typed_stat): Remove.
1921 (ggc_pch_count_object): Remove last argument. Update all users.
1922 (ggc_pch_alloc_object): Remove last argument. Update all users.
1923 (struct alloc_zone): Remove.
1924 * ggc-zone.c: Remove.
1925 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
1926 (struct alloc_zone): Remove.
1927 (ggc_alloc_typed_stat): Remove.
1928 (ggc_alloc_typed): Remove.
1929 (ggc_splay_alloc): Remove first argument.
1930 (rtl_zone): Remove. Update all users.
1931 (tree_zone): Remove. Update all users.
1932 (tree_id_zone): Remove. Update all users.
1933 (ggc_internal_zone_alloc_stat): Remove. Update all users.
1934 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
1935 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
1936 * tree-ssanames.c: Remove references to zone allocator in comments.
1938 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
1940 * config/avr/avr.c (avr_out_fract): Make register numbers that
1941 might be outside of source operand signed.
1943 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
1945 * config/i386/constraints.md (Yf): New constraint.
1946 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
1947 of f constraint to conditionaly disable x87 register preferences.
1948 (*movdf_internal): Ditto.
1949 (*movsf_internal): Ditto.
1951 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
1954 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
1955 that have operands with impossible constraints.
1956 Add a FIXME for a speed-up opportunity.
1957 * lra-constraints.c (process_alt_operands): Verify that a class
1958 selected from constraints on asms is valid for the operand mode.
1959 (curr_insn_transform): Remove incorrect comment.
1961 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
1963 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
1964 TOC operand is a valid symbol ref in the constant pool.
1966 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1968 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
1970 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
1973 * config/avr/stdfix.h: New file.
1974 * t-avr (stdfix-gcc.h): New rule to build it.
1975 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
1977 2013-01-23 Kostya Serebryany <kcc@google.com>
1979 * config/darwin.h: remove dependency on
1980 CoreFoundation (asan on Mac OS).
1982 2013-01-23 Jakub Jelinek <jakub@redhat.com>
1985 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
1986 instead of cmpdi_operand for first comparison operand.
1987 Don't assert that comparison operands aren't both constants.
1989 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
1991 * doc/install.texi (Downloading the Source): Update references to
1992 downloading separate components.
1994 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
1996 * doc/extend.texi (__int128): Improve grammar.
1998 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
2001 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
2002 alternative to (o,r).
2003 (*movdi_internal_rex64): Remove (!o,n) alternative.
2004 (DImode immediate->memory splitter): Remove.
2005 (DImode immediate->memory peephole2): Remove.
2006 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
2007 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
2008 alternative to (!o,*r).
2009 (*movtf_internal_sse): New pattern.
2010 (*movxf_internal_rex64): New pattern.
2011 (*movxf_internal): Disable for TARGET_64BIT.
2012 (*movdf_internal_rex64): Remove (!o,F) alternative.
2014 2013-01-22 Jakub Jelinek <jakub@redhat.com>
2017 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
2018 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
2019 * tree-vect-loop-manip.c (find_loop_location): Also ignore
2020 stmt locations where LOCATION_LOCUS of the stmt location is
2021 UNKNOWN_LOCATION or BUILTINS_LOCATION.
2024 * config/i386/i386.md (UNSPEC_STOS): New.
2025 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
2026 *strsetqi_1): Add UNSPEC_STOS.
2028 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
2031 * doc/invoke.texi: Remove left over -Wsynth example.
2033 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2035 PR tree-optimization/56051
2036 * fold-const.c (fold_binary_loc): Don't fold
2037 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
2038 a narrowing conversion, or widening conversion from signed
2041 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2043 PR rtl-optimization/56023
2044 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
2045 dependent on debug instruction.
2047 2013-01-21 Martin Jambor <mjambor@suse.cz>
2050 * function.c (allocate_struct_function): Call
2051 invoke_set_current_function_hook earlier.
2053 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2055 * reload1.c (init_reload): Only initialize reload_obstack
2056 during the first call.
2058 2013-01-21 Marek Polacek <polacek@redhat.com>
2060 * cfgloop.c (verify_loop_structure): Fix up grammar.
2062 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
2064 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
2065 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
2067 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2070 * config/arm/marvell-pj4.md: Update copyright year.
2071 Fix up use of alu to alu_reg and simple_alu_imm.
2073 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2075 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
2077 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
2080 * lra-constraints.c (curr_insn_transform): Don't reuse original
2081 insn for secondary memory move when memory mode should be different.
2083 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2085 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
2086 atomic_storedi_1): New patterns.
2088 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2090 btver2 pipeline descriptions.
2091 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
2093 * config/i386/i386.md (btver2_decode): New type attributes.
2094 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
2096 * config/i386/btver2.md: New file describing btver2 pipelines.
2098 2013-01-19 Andrew Pinski <apinski@cavium.com>
2100 PR tree-optimization/52631
2101 * tree-ssa-sccvn (visit_use): Before looking up the original
2102 statement, try looking up the simplified expression.
2104 2013-01-19 Anthony Green <green@moxielogic.com>
2106 * config/moxie/moxie.c (moxie_expand_prologue): Set
2107 current_function_static_stack_size.
2109 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2111 PR tree-optimization/56029
2112 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
2113 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
2115 2013-01-18 Sharad Singhai <singhai@google.com>
2117 PR tree-optimization/55995
2118 * dumpfile.c (dump_loc): Print location only if available.
2119 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
2121 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
2124 * lra-constraints.c (curr_insn_transform): Reuse original insn for
2125 secondary memory move.
2126 (inherit_reload_reg): Use rclass instead of cl for
2127 check_secondary_memory_needed_p.
2129 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2132 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
2133 the case where writing real complex part of target modifies op1.
2135 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
2137 * config/aarch64/aarch64-simd.md
2138 (aarch64_vcond_internal<mode>): Handle unordered cases.
2139 * config/aarch64/iterators.md (v_cmp_result): New.
2141 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
2142 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2144 * config/arm/marvell-pj4.md: New file.
2145 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
2146 * config/arm/arm.md (generic_sched): Add marvell_pj4.
2147 (generic_vfp): Likewise.
2148 * config/arm/arm-cores.def: Add marvell-pj4.
2149 * config/arm/arm-tune.md: Regenerate.
2150 * config/arm/arm-tables.opt: Regenerate.
2151 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
2152 * doc/invoke.texi: Document marvell-pj4.
2154 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
2156 * config/aarch64/arm_neon.h: Map scalar types to standard types.
2158 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2163 * var-tracking.c (negative_power_of_two_p): New.
2164 (global_get_addr_cache, local_get_addr_cache): New.
2165 (get_addr_from_global_cache, get_addr_from_local_cache): New.
2166 (vt_canonicalize_addr): Rewrite using the above. Adjust the
2168 (vt_stack_offset_p): Remove.
2169 (vt_canon_true_dep): Always canonicalize loc's address.
2170 (clobber_overlapping_mems): Make sure we have a MEM.
2171 (local_get_addr_clear_given_value): New.
2172 (val_reset): Clear local cached entries.
2173 (compute_bb_dataflow): Create and release the local cache.
2174 Disable duplicate MEMs clobbering.
2175 (emit_notes_in_bb): Clobber MEMs likewise.
2176 (vt_emit_notes): Create and release the local cache.
2177 (vt_initialize, vt_finalize): Create and release the global
2178 cache, respectively.
2179 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
2181 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2184 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
2185 not found in the symtab.
2187 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2190 PR rtl-optimization/55547
2191 PR rtl-optimization/53827
2194 * alias.c (offset_overlap_p): New, factored out of...
2195 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
2196 the conservative special case for symbolic constants. Don't
2197 adjust zero sizes on alignment.
2199 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
2201 PR rtl-optimization/52573
2202 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
2203 REG_UNUSED for the same register.
2205 2013-01-17 Richard Biener <rguenther@suse.de>
2206 Marek Polacek <polacek@redhat.com>
2208 PR rtl-optimization/55833
2209 * loop-unswitch.c (unswitch_loops): Move loop verification...
2210 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
2211 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
2212 Set it to true when we're removing a loop from hierarchy tree in
2213 an irreducible region.
2214 (fix_bb_placements): Adjust caller.
2215 (fix_loop_placements): Likewise.
2217 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
2219 * config/avr/builtins.def (DEF_BUILTIN): Factor out
2220 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
2221 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
2222 Remove ID. Adjust comments.
2223 * config/avr/avr-c.c (avr_builtin_name): Remove.
2224 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
2225 * config/avr/avr.c (avr_tolower): New static function.
2226 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
2227 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
2228 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
2231 2013-01-17 Jan Hubicka <jh@suse.cz>
2233 PR tree-optimization/55273
2234 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
2236 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
2239 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
2240 store through atomic_store<mode>_1.
2241 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
2243 2013-01-17 Martin Jambor <mjambor@suse.cz>
2245 PR tree-optimizations/55264
2246 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
2247 for virtual methods.
2248 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
2249 virtual methods before inlining is over.
2250 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
2252 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
2255 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
2257 PR rtl-optimization/56005
2258 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
2259 pending reads for prefetch.
2261 2013-01-16 Ian Bolton <ian.bolton@arm.com>
2263 * config/aarch64/aarch64.md
2264 (*cstoresi_neg_uxtw): New pattern.
2265 (*cmovsi_insn_uxtw): New pattern.
2266 (*<optab>si3_uxtw): New pattern.
2267 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
2268 (*<optab>si3_insn_uxtw): New pattern.
2269 (*bswapsi2_uxtw): New pattern.
2271 2013-01-16 Richard Biener <rguenther@suse.de>
2273 * tree-inline.c (tree_function_versioning): Remove set but
2274 never used variable.
2276 2013-01-16 Richard Biener <rguenther@suse.de>
2278 PR tree-optimization/55964
2279 * tree-flow.h (rename_variables_in_loop): Remove.
2280 (rename_variables_in_bb): Likewise.
2281 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
2282 (copy_loop_before): Adjust and delete update-ssa status.
2283 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
2284 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
2285 (rename_variables_in_loop): Remove.
2286 (slpeel_update_phis_for_duplicate_loop): Likewise.
2287 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
2288 use available cfg machinery instead of duplicating it.
2289 Update PHI nodes and perform poor-mans SSA update here.
2290 (slpeel_tree_peel_loop_to_edge): Adjust.
2292 2013-01-16 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/54767
2295 PR tree-optimization/53465
2296 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
2297 (vrp_visit_phi_node): For PHI arguments coming via backedges
2298 drop all symbolical range information.
2299 (execute_vrp): Compute backedges.
2301 2013-01-16 Richard Biener <rguenther@suse.de>
2303 * doc/install.texi: Update CLooG and ISL requirements to
2306 2013-01-16 Christian Bruel <christian.bruel@st.com>
2309 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
2310 (broken_move): Handle UNSPECV_SP_SWITCH_B.
2311 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
2313 2013-01-16 DJ Delorie <dj@redhat.com>
2315 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
2316 (UNSPECV_SP_SWITCH_E): New.
2317 (sp_switch_1): Change to an unspec.
2318 (sp_switch_2): Change to an unspec. Don't use post-inc when we
2321 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
2323 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
2324 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
2325 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
2326 (expand_mem_thread_fence): Ditto.
2327 (expand_mem_signal_fence): Ditto.
2328 (expand_atomic_load): Ditto.
2329 (expand_atomic_store): Ditto.
2331 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
2333 PR rtl-optimization/55547
2334 PR rtl-optimization/53827
2337 * alias.c (memrefs_conflict_p): Set sizes to negative after
2340 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2343 * function.c (thread_prologue_and_epilogue_insns): Always
2344 add crtl->drap_reg to set_up_by_prologue.set, even if
2345 stack_realign_drap is false.
2347 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2349 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
2350 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
2351 *call): Fix indention.
2353 2013-01-15 Tom de Vries <tom@codesourcery.com>
2356 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
2359 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
2361 PR rtl-optimization/55153
2362 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
2364 2013-01-15 Martin Jambor <mjambor@suse.cz>
2366 PR tree-optimization/55920
2367 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
2368 accesses as grp_to_be_debug_replaced.
2370 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2372 PR tree-optimization/55920
2373 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
2374 there is non-useless type conversion needed from debug rhs to lhs,
2375 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
2377 2013-01-15 Joseph Myers <joseph@codesourcery.com>
2378 Mikael Pettersson <mikpe@it.uu.se>
2381 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
2383 (ASM_OUTPUT_CASE_LABEL): Remove.
2384 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
2385 * final.c (shorten_branches): Update alignment of labels before
2386 jump tables if CASE_VECTOR_SHORTEN_MODE.
2388 2013-01-15 Richard Biener <rguenther@suse.de>
2391 * system.h: Do not include gmp.h for building host tools.
2393 2013-01-15 Richard Biener <rguenther@suse.de>
2396 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
2397 account for bitpos when computing alignment.
2399 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
2401 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
2402 (ix86_target_macros_internal): Likewise.
2404 * config/i386/i386.c (m_CORE2I7): Removed.
2405 (m_CORE_HASWELL): New macro.
2406 (m_CORE_ALL): Likewise.
2407 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
2408 (initial_ix86_arch_features): Likewise.
2409 (processor_target_table): Initializations for Core avx2.
2410 (cpu_names): New names "core-avx2".
2411 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
2412 PROCESSOR_CORE_HASWELL.
2413 (ix86_issue_rate): New case.
2414 (ia32_multipass_dfa_lookahead): Likewise.
2415 (ix86_sched_init_global): Likewise.
2417 * config/i386/i386.h (TARGET_HASWELL): New macro.
2418 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
2419 (processor_type): New PROCESSOR_HASWELL.
2421 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2423 PR tree-optimization/55955
2424 * tree-vect-loop.c (vectorizable_reduction): Give up early on
2425 *SHIFT_EXPR and *ROTATE_EXPR codes.
2427 PR tree-optimization/48766
2428 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
2429 -ftrapv disable -fwrapv.
2431 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2434 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
2435 etc. to 1 and not to __flash.
2436 Use LL suffix for __INT24_MAX__ with -mint8.
2437 Use ULL suffix for __UINT24_MAX__ with -mint8.
2439 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2441 * config/avr/avr-arch.h
2442 (struct base_arch_s): Use typedef avr_arch_t instead.
2443 (struct arch_info_s): Use typedef avr_arch_info_t instead.
2444 (struct mcu_type_s): Use typedef avr_mcu_t instead.
2445 * config/avr/avr.c: Same.
2446 * config/avr/avr-devices.c: Same.
2447 * config/avr/driver-avr.c: Same.
2448 * config/avr/gen-avr-mmcu-texi.c: Same.
2449 * config/avr/avr-mcus.def: Adjust comment.
2451 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
2453 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
2454 * config/aarch64/iterators.md (VALLDI): New.
2456 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
2457 Andi Kleen <ak@linux.intel.com>
2460 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
2461 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
2464 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2466 * config/avr/avr-stdint.h: Remove trailing blanks.
2467 * config/avr/avr-log.h: Same.
2468 * config/avr/avr-arch.h: Same.
2469 * config/avr/avr-devices.c: Same.
2470 * config/avr/avr-dimode.md: Same.
2471 * config/avr/predicates.md: Same.
2472 * config/avr/avr-c.c: Same. And fix typo.
2474 * config/avr/avr-protos.h: Same. And:
2475 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
2476 (init_cumulative_args): Rename to avr_init_cumulative_args.
2477 (expand_prologue): Rename to avr_expand_prologue.
2478 (expand_epilogue): Rename to avr_expand_epilogue.
2479 (adjust_insn_length): Rename to avr_adjust_insn_length.
2480 (notice_update_cc): Rename to avr_notice_update_cc.
2481 (final_prescan_insn): Rename to avr_final_prescan_insn.
2482 * config/avr/avr.c: Same.
2483 * config/avr/avr.h: Same.
2484 * config/avr/avr.md: Remove trailing blanks.
2485 (prologue): Use avr_expand_prologue.
2486 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
2488 2013-01-14 Richard Biener <rguenther@suse.de>
2490 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
2491 verify_location, collect_subblocks): New functions.
2492 (verify_gimple_in_cfg): Verify that locations only reference
2493 BLOCKs in the functions BLOCK tree.
2495 2013-01-14 Richard Biener <rguenther@suse.de>
2497 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
2499 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
2501 (insert_out_of_ssa_copy_on_edge): Likewise.
2502 (rewrite_close_phi_out_of_ssa): Likewise.
2503 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
2505 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
2506 propagated constants.
2507 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
2510 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2512 * config/avr/avr-modes.def: Add GPL copyright notice.
2514 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
2516 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
2517 MEMMODEL_MASK to determine memory model.
2518 (atomic_store<mode>): Ditto from operands[2].
2519 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
2521 2013-01-13 Jakub Jelinek <jakub@redhat.com>
2524 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
2525 (fold_gimple_assign): Don't call unshare_expr here.
2526 (fold_ctor_reference): Call unshare_expr.
2528 2013-01-13 Terry Guo <terry.guo@arm.com>
2530 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
2531 * doc/fragments.texi: Document MULTILIB_REUSE.
2532 * gcc.c (multilib_reuse): New internal spec.
2533 (set_multilib_dir): Also search multilib from multilib_reuse.
2534 * genmultilib (tmpmultilib3): Refactor code.
2535 (tmpmultilib4): Ditto.
2536 (multilib_reuse): New multilib argument.
2538 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
2540 * Makefile.in: Update copyright.
2542 2013-01-12 Tom de Vries <tom@codesourcery.com>
2545 * calls.c (expand_call): Check if arg_nr is valid.
2547 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2549 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
2550 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
2551 documentation. Add missing '__' in front of
2552 __builtin_ia32_packssdw256.
2554 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2557 * config/s390/s390.c (s390_preferred_reload_class): Do not return
2558 NO_REGS for larl operands.
2559 (s390_reload_larl_operand): Use s390_load_address instead of
2562 2013-01-11 Richard Biener <rguenther@suse.de>
2564 * tree-cfg.c (verify_node_sharing_1): Split out from ...
2565 (verify_node_sharing): ... here.
2566 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
2568 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
2570 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
2571 Substitute TREECHECKING.
2572 * configure: Regenerate.
2573 * Makefile.in (TREECHECKING): New.
2575 2013-01-11 Richard Guenther <rguenther@suse.de>
2577 PR tree-optimization/44061
2578 * tree-vrp.c (extract_range_basic): Compute zero as
2579 value-range for __builtin_constant_p of function parameters.
2581 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
2583 Update copyright years.
2585 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
2587 PR rtl-optimization/55672
2588 * lra-eliminations.c (mark_not_eliminable): Permit addition with
2589 const to be eliminable.
2591 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
2593 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
2594 * configure: Regenerate.
2596 2013-01-10 Richard Biener <rguenther@suse.de>
2598 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
2600 2013-01-10 Richard Biener <rguenther@suse.de>
2603 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
2604 locations for virtual PHI arguments.
2605 (rewrite_update_phi_arguments): Likewise.
2607 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
2609 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
2612 2013-01-10 Jakub Jelinek <jakub@redhat.com>
2614 PR tree-optimization/55921
2615 * tree-complex.c (expand_complex_asm): New function.
2616 (expand_complex_operations_1): Call it for GIMPLE_ASM.
2618 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2621 * config/s390/s390.c (s390_symref_operand_p)
2622 (s390_loadrelative_operand_p): Merge the two functions.
2623 (s390_check_qrst_address, print_operand_address): Add parameters
2624 to s390_loadrelative_operand_p invokation.
2625 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
2626 (s390_reload_larl_operand, s390_secondary_reload): Use
2627 s390_loadrelative_operand_p instead of s390_symref_operand_p.
2628 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
2630 2013-01-09 Mike Stump <mikestump@comcast.net>
2632 * dse.c (record_store): Remove unnecessary assert.
2634 2013-01-09 Jan Hubicka <jh@suse.cz>
2636 PR tree-optimization/55569
2637 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
2638 * cfgloop.h (scale_loop_profile): Likewise.
2640 2013-01-09 Jan Hubicka <jh@suse.cz>
2643 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
2645 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
2647 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
2650 * expr.h (maybe_emit_group_store): Declare.
2651 * expr.c (maybe_emit_group_store): New function.
2652 * builtins.c (expand_builtin_int_roundingfn): Call it.
2653 (expand_builtin_int_roundingfn_2): Likewise.
2655 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
2657 PR rtl-optimization/55829
2658 * lra-constraints.c (match_reload): Add code for absent output.
2659 (curr_insn_transform): Add code for reloads of matched inputs
2662 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
2664 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
2665 attribute of movddup insn to DF.
2666 (*vec_interleave_lowv2df): Ditto.
2667 (vec_dupv2df): Ditto.
2669 2013-01-09 Jan Hubicka <jh@suse.cz>
2671 PR tree-optimiation/55875
2672 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
2673 EVERY_ITERATION parameter.
2674 (number_of_iterations_exit): Check if exit is executed every iteration.
2675 (idx_infer_loop_bounds): Similarly here.
2676 (n_of_executions_at_most): Simplify
2677 to only test for cases where statement is dominated by the
2678 particular bound; handle correctly the "postdominance" test.
2679 (scev_probably_wraps_p): Use max loop iterations info
2680 as a global bound first.
2682 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
2683 Nick Clifton <nickc@redhat.com>
2685 * config/v850/v850.md (cbranchsf4): New pattern.
2686 (cstoresf4): New pattern.
2687 (cbranchdf4): New pattern.
2688 (cstoredf4): New pattern.
2689 (movsicc): Disallow floating point comparisons.
2690 (cmpsf_le_insn): Fix order of operators.
2691 (cmpsf_lt_insn): Likewise.
2692 (cmpsf_eq_insn): Likewise.
2693 (cmpdf_le_insn): Likewise.
2694 (cmpdf_lt_insn): Likewise.
2695 (cmpdf_eq_insn): Likewise.
2696 (cmpsf_ge_insn): Use LE comparison.
2697 (cmpdf_ge_insn): Likewise.
2698 (cmpsf_gt_insn): Use LT comparison.
2699 (cmpdf_gt_insn): Likewise.
2700 (cmpsf_ne_insn): Delete pattern.
2701 (cmpdf_ne_insn): Delete pattern.
2702 * config/v850/v850.c (v850_gen_float_compare): Use
2703 gen_cmpdf_eq_insn for NE comparison.
2704 (v850_float_z_comparison_operator)
2705 (v850_float_nz_comparison_operator): Move from here ...
2706 * config/v850/predicates.md: ... to here. Move GT and GE
2707 comparisons into v850_float_z_comparison_operator.
2708 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
2710 (v850_float_nz_comparison_operator): Likewise.
2712 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2714 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
2715 with calls to gen_insvsi/gen_insvdi.
2717 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2719 * config/i386/i386.c (initial_ix86_tune_features): Set up
2720 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
2722 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
2723 Jakub Jelinek <jakub@redhat.com>
2725 PR tree-optimization/48189
2726 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
2727 If nitercst is 0, don't predict the exit edge.
2729 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
2731 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
2732 in asm_fprintf with reg_names.
2733 (aarch64_print_operand_address): Likewise.
2734 (aarch64_return_addr): Likewise.
2735 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
2737 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2739 * config/pa/pa.h (VAL_U6_BITS_P): Define.
2740 (INT_U6_BITS): Likewise.
2741 * config/pa/predicates.md (uint6_operand): New predicate.
2742 (shift5_operand, shift6_operand): Likewise.
2743 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
2745 (lshrdi3): Use shift6_operand.
2746 (shrpsi4, shrpdi4): New insn patterns.
2747 (extzv): Delete expander.
2748 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
2749 predicates in unamed zero extract patterns. Tighten common constraint.
2750 (extv): Delete expander.
2751 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
2752 predicates in unamed sign extract patterns. Tighten common constraint.
2753 (insv): Delete expander.
2754 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
2755 predicates in unamed insert patterns. Tighten common constraint.
2756 Change uint32_operand predicate to uint6_operand predicate in unamed
2757 DImode pattern to insert constant values of type 1...1xxxx.
2759 2013-01-04 Jan Hubicka <jh@suse.cz>
2761 PR tree-optimization/55823
2762 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
2765 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2766 Uros Bizjak <ubizjak@gmail.com>
2768 PR rtl-optimization/55845
2769 * df-problems.c (can_move_insns_across): Stop scanning at
2770 volatile_insn_p source instruction or give up if
2771 across_from .. across_to range contains any volatile_insn_p
2774 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2776 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
2777 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
2779 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
2780 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
2782 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2785 * asan.c (asan_clear_shadow): New function.
2786 (asan_emit_stack_protection): Use it.
2788 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2790 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
2791 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
2792 with tab instead of space.
2794 2013-01-08 Nick Clifton <nickc@redhat.com>
2796 * config/rl78/rl78.c (rl78_expand_prologue): Always select
2797 register bank 0 at the start of an interrupt handler.
2798 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
2801 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2803 * config/aarch64/aarch64-simd.md
2804 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
2805 (aarch64_simd_bsl): Likewise.
2806 (aarch64_vcond_internal<mode>): Likewise.
2807 (vcond<mode><mode>): Likewise.
2808 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
2809 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
2811 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2813 * config/aarch64/aarch64-builtins.c
2814 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
2816 2013-01-08 Martin Jambor <mjambor@suse.cz>
2819 * tree-sra.c (analyze_access_subtree): Return true also after
2820 potentially creating a debug-only replacement.
2822 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2825 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
2827 PR tree-optimization/54120
2828 * tree-vrp.c (range_fits_type_p): Don't allow
2829 src_precision < precision from signed vr to unsigned_p
2830 if vr->min or vr->max is negative.
2831 (simplify_float_conversion_using_ranges): Test can_float_p
2832 against CODE_FOR_nothing.
2834 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2835 Richard Biener <rguenther@suse.de>
2838 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
2839 types instead of just INTEGER_TYPE types.
2841 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
2843 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
2846 2013-01-07 Steve Ellcey <sellcey@mips.com>
2849 * config/mips/mips.opt: Change mad to mmad to match documentation.
2851 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2854 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
2857 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2860 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
2861 (avr_addrspace_t): Add .section_name field.
2862 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
2864 (avr_addrspace): Same. Initialize .section_name. Remove last
2865 NULL entry. Put __memx into .progmemx.data.
2866 (progmem_section_prefix): Remove.
2867 (avr_asm_init_sections): No need to initialize progmem_section.
2868 (avr_asm_named_section): Use avr_addrspace[].section_name to get
2869 section name prefix.
2870 (avr_asm_select_section): Ditto. And use get_unnamed_section to
2871 retrieve the progmem section.
2872 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
2873 boundary to run over avr_addrspace[].
2874 (avr_register_target_pragmas): Ditto.
2876 2013-01-06 Jakub Jelinek <jakub@redhat.com>
2878 * varasm.c (output_constant_def_contents): For asan_protect_global
2879 protected strings, adjust DECL_ALIGN if needed, before testing for
2881 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
2882 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
2884 (output_object_block): For asan protected decls, emit asan padding
2885 after their contents.
2886 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
2887 (asan_finish_file): Test it here instead.
2889 2013-01-07 Nick Clifton <nickc@redhat.com>
2890 Matthias Klose <doko@debian.org>
2891 Doug Kwan <dougkwan@google.com>
2892 H.J. Lu <hongjiu.lu@intel.com>
2895 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
2897 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
2899 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
2901 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
2903 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
2905 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2908 * doc/install.texi (Cross-Compiler-Specific Options): Document
2911 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
2913 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
2914 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
2915 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
2916 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
2917 vqmovun_high_s64): Fix source operand number and update copyright.
2919 2013-01-07 Richard Biener <rguenther@suse.de>
2922 * gimple.h (gimple_call_builtin_p): New overload.
2923 * gimple.c (validate_call): New function.
2924 (gimple_call_builtin_p): Likewise.
2925 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2926 Use gimple_call_builtin_p.
2927 (find_func_clobbers): Likewise.
2928 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
2929 (strlen_optimize_stmt): Likewise.
2931 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2933 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
2934 (vld1q_dup_*): Likewise.
2936 (vld1q_*): Likewise.
2937 (vld1_lane_*): Likewise.
2938 (vld1q_lane_*): Likewise.
2940 2013-01-07 Richard Biener <rguenther@suse.de>
2942 * lto-streamer.h (LTO_minor_version): Bump to 2.
2944 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2946 * config/aarch64/aarch64-protos.h
2947 (aarch64_const_double_zero_rtx_p): Rename to...
2948 (aarch64_float_const_zero_rtx_p): ...this.
2949 (aarch64_float_const_representable_p): New.
2950 (aarch64_output_simd_mov_immediate): Likewise.
2951 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
2952 move immediate case.
2953 * config/aarch64/aarch64.c
2954 (aarch64_const_double_zero_rtx_p): Rename to...
2955 (aarch64_float_const_zero_rtx_p): ...this.
2956 (aarch64_print_operand): Allow printing of new constants.
2957 (aarch64_valid_floating_const): New.
2958 (aarch64_legitimate_constant_p): Check for valid floating-point
2960 (aarch64_simd_valid_immediate): Likewise.
2961 (aarch64_vect_float_const_representable_p): New.
2962 (aarch64_float_const_representable_p): Likewise.
2963 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
2964 (aarch64_output_simd_mov_immediate): New.
2965 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
2966 (*movdf_aarch64): Likewise.
2967 * config/aarch64/constraints.md (Ufc): New.
2968 (Y): call aarch64_float_const_zero_rtx.
2969 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
2971 2013-01-07 Richard Biener <rguenther@suse.de>
2973 PR tree-optimization/55888
2974 PR tree-optimization/55862
2975 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
2976 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
2977 not if it is contained therein.
2979 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2981 * config/avr/t-avr: Typo.
2983 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2986 * config/avr/t-avr: Don't automatically rebuild
2987 $(srcdir)/config/avr/t-multilib
2988 $(srcdir)/config/avr/avr-tables.opt
2989 $(srcdir)/doc/avr-mmcu.texi
2990 (avr-mcus): New phony target to build them on request.
2991 (s-avr-mlib, s-avr-mmcu-texi): Remove.
2992 * avr/avr-mcus.def: Adjust comments.
2994 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
2996 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
2998 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
3000 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
3002 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
3004 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
3006 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
3008 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
3009 to generate profiling.
3010 * config/rs6000/aix64.h (LIB_SPEC): Same.
3012 2013-01-04 Andrew Pinski <apinski@cavium.com>
3014 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
3016 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
3018 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
3020 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
3022 (ix86_expand_move): Ditto.
3023 (ix86_zero_extend_to_Pmode): Ditto.
3024 (ix86_expand_call): Ditto.
3025 (ix86_expand_special_args_builtin): Ditto.
3026 (ix86_expand_builtin): Ditto.
3028 2013-01-04 Richard Biener <rguenther@suse.de>
3030 PR tree-optimization/55862
3031 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
3032 translating them through PHI nodes.
3034 2013-01-04 Martin Jambor <mjambor@suse.cz>
3036 PR tree-optimization/55755
3037 * tree-sra.c (sra_modify_assign): Do not check that an access has no
3038 children when trying to avoid producing a VIEW_CONVERT_EXPR.
3040 2013-01-04 Marek Polacek <polacek@redhat.com>
3043 * opts.c (default_options_optimization): Clarify error message.
3045 2013-01-04 Richard Biener <rguenther@suse.de>
3048 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
3051 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3054 * config/pa/pa.md (movsi): Revert previous change.
3055 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
3058 2013-01-03 Richard Henderson <rth@redhat.com>
3060 * config/i386/i386.c (ix86_expand_move): Always assign to op1
3061 after eliminating TLS symbols.
3063 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
3066 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
3067 * graphite-poly.c (debug_gmp_value): Likewise.
3069 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
3072 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
3073 selected code model, define __code_mode_small__, __code_model_medium__,
3074 __code_model_large__, __code_model_32__ or __code_model_kernel__.
3075 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
3076 xchg temporary register with %k. Declare temporary register as
3078 [__x86_64__]: For medium and large code models, preserve %rbx register.
3080 2013-01-03 Richard Biener <rguenther@suse.de>
3082 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
3083 (dump_subscript): Adjust.
3084 (finalize_ddr_dependent): Do not dump redundant info.
3085 (analyze_siv_subscript): Adjust.
3086 (subscript_dependence_tester): Likewise.
3087 (compute_affine_dependence): Likewise.
3089 2013-01-03 Richard Biener <rguenther@suse.de>
3092 2013-01-03 Richard Biener <rguenther@suse.de>
3094 PR tree-optimization/55857
3095 * tree-vect-stmts.c (vectorizable_load): Do not setup
3096 re-alignment for invariant loads.
3098 2013-01-02 Richard Biener <rguenther@suse.de>
3100 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3101 invariant load do not generate a vector load from the scalar location.
3103 2013-01-03 Richard Biener <rguenther@suse.de>
3105 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
3106 for not vectorizing.
3107 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
3108 not build INDIRECT_REFs, call get_name once only.
3109 (vect_create_data_ref_ptr): Likewise. Dump base object kind
3110 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
3112 2013-01-03 Richard Biener <rguenther@suse.de>
3114 PR tree-optimization/55857
3115 * tree-vect-stmts.c (vectorizable_load): Do not setup
3116 re-alignment for invariant loads.
3118 2013-01-03 Richard Biener <rguenther@suse.de>
3121 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
3122 prefer a built-in decl.
3124 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3126 * gcc.c (process_command): Update copyright notice dates.
3127 * gcov.c (print_version): Likewise.
3128 * gcov-dump.c (print_version): Likewise.
3130 PR rtl-optimization/55838
3131 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
3132 iv0.step, iv1.step and step.
3134 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3135 Marc Glisse <marc.glisse@inria.fr>
3137 PR tree-optimization/55832
3138 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
3139 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
3140 integer_{one,zero}_node.
3142 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3145 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
3146 * var-tracking.c (reverse_op): Don't add reverse ops to
3147 VALUEs that have already
3148 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
3150 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
3152 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
3154 2013-01-02 Teresa Johnson <tejohnson@google.com>
3156 * dumpfile.c (dump_loc): Print filename with location.
3157 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
3158 new location_t parameter to emit complete unroll message with
3160 (canonicalize_loop_induction_variables): Compute loops location
3161 and pass to try_unroll_loop_completely.
3162 * loop-unroll.c (report_unroll_peel): New function.
3163 (peel_loops_completely): Use new dump format with location
3164 for main dumpfile message, and invoke report_unroll_peel on success.
3165 (decide_unrolling_and_peeling): Ditto.
3166 (decide_peel_once_rolling): Remove old dumpfile message subsumed
3167 by report_unroll_peel.
3168 (decide_peel_completely): Ditto.
3169 (decide_unroll_constant_iterations): Ditto.
3170 (decide_unroll_runtime_iterations): Ditto.
3171 (decide_peel_simple): Ditto.
3172 (decide_unroll_stupid): Ditto.
3173 * cfgloop.c (get_loop_location): New function.
3174 * cfgloop.h (get_loop_location): Declare.
3176 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3178 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
3181 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3184 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
3185 BLKmode objects when EXPAND_MEMORY is specified.
3187 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3189 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
3191 (fold_builtin_cpu): Do not share cpu model decls across statements.
3193 2013-01-02 Jason Merrill <jason@redhat.com>
3196 * tree.c (build_array_type_1): Revert earlier change.
3198 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
3200 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
3202 * config/aarch64/aarch64-tune.md: Re-generate.
3204 2013-01-02 Richard Biener <rguenther@suse.de>
3206 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3207 invariant load do not generate a vector load from the scalar location.
3209 2013-01-02 Richard Biener <rguenther@suse.de>
3212 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
3213 * configure: Regenerate.
3215 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3217 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
3218 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
3219 (expand_builtin_int_roundingfn_2): Keep the original target around
3220 for the fallback case.
3222 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3224 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
3225 to be clear for sign changes.
3227 2013-01-01 Jan Hubicka <jh@suse.cz>
3229 * ipa-inline-analysis.c: Fix formatting.
3231 2013-01-01 Jakub Jelinek <jakub@redhat.com>
3233 PR tree-optimization/55831
3234 * tree-vect-loop.c (get_initial_def_for_induction): Use
3235 gsi_after_labels instead of gsi_start_bb.
3237 Copyright (C) 2013 Free Software Foundation, Inc.
3239 Copying and distribution of this file, with or without modification,
3240 are permitted in any medium without royalty provided the copyright
3241 notice and this notice are preserved.