1 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
3 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
4 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
5 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
6 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
7 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
8 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
9 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
10 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
11 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
12 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
13 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
17 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18 (avrxmega6): Increase max flash segments from 5 to 6.
19 * config/avr/t-multilib: Regenerate.
20 * config/avr/avr-tables.opt: Regenerate.
21 * doc/avr-mmcu.texi: Regenerate.
23 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
25 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
26 (avr_device_to_arch): Rename to avr_device_to_ld.
27 (avr_device_to_as): New prototype.
28 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
29 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
30 * config/avr/driver-avr.c (avr_device_to_as): New.
31 (avr_device_to_arch): Rename to avr_device_to_ld.
33 2013-02-27 Jakub Jelinek <jakub@redhat.com>
36 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
37 method on dr_chain and result_chain.
40 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
41 pointer_set_destroy on not_executed_last_iteration.
44 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs
48 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
49 FOR_EACH_DEFINED_FUNCTION when freeing state.
52 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
54 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
58 * ipa-cp.c (decide_whether_version_node): Call vec_free on
59 known_aggs[i].items and release known_aggs vector.
62 * ipa-reference.c (propagate): Free node_info even for alias nodes.
64 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
66 * config/microblaze/microblaze.c (microblaze_emit_compare):
67 Use xor for EQ/NE comparisions.
68 * config/microblaze/microblaze.md (cstoresf4): Add constraints
69 (cbranchsf4): Adjust operator to comparison_operator.
71 2013-02-27 Jakub Jelinek <jakub@redhat.com>
74 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
76 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
77 vec_safe_push, always update *slot.
78 (redirect_edge_var_map_clear): Use vec_free.
79 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
80 (free_var_map_entry): Use vec_free.
81 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
82 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
84 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
87 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
88 when the may_trap_p bit of the exprs being merged differs.
89 Reorder tests for speculativeness in the logical and operator.
91 2013-02-27 Jakub Jelinek <jakub@redhat.com>
93 * incpath.c (add_standard_paths): Use reconcat instead of concat
94 where appropriate and avoid leaking memory.
96 * opts.h: Include obstack.h.
97 (opts_concat): New prototype.
98 (opts_obstack): New declaration.
99 * opts.c (opts_concat): New function.
100 (opts_obstack): New variable.
101 (init_options_struct): Call gcc_init_obstack on opts_obstack.
102 (finish_options): Use opts_concat instead of concat
103 and XOBNEWVEC instead of XNEWVEC.
104 * opts-common.c (generate_canonical_option, decode_cmdline_option,
105 generate_option): Likewise.
106 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
107 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
110 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
111 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
113 2013-02-26 Jakub Jelinek <jakub@redhat.com>
116 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
118 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
120 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
121 (arm_block_move_unaligned_straight): Likewise.
122 (arm_adjust_block_mem): Likewise.
124 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
127 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
128 temp, cond and label.
129 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
132 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
133 * config/c6x/c6x.h (dbx_register_map): Update declaration.
136 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
137 of prologue/epilogue functions.
140 * config/tilegx/tilegx.c (tilegx_expand_prologue):
141 Remove unused variable cfa_offset.
142 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
145 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
146 type promotion to unsigned.
149 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
150 for HOST_WIDE_INT of 32 bit / same size as int.
151 (arm_block_move_unaligned_straight): Likewise.
152 (arm_adjust_block_mem): Likewise.
155 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
158 2013-02-26 Marek Polacek <polacek@redhat.com>
160 PR tree-optimization/56426
161 * tree-ssa-loop.c (tree_ssa_loop_init): Always call
164 2013-02-26 Richard Biener <rguenther@suse.de>
167 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
168 unused variable loops.
170 2013-02-26 Jakub Jelinek <jakub@redhat.com>
172 PR tree-optimization/56448
173 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
174 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
175 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
176 later operands of the references, or even first operand for
177 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
179 PR tree-optimization/56443
180 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
181 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
182 to type_for_mode langhook.
184 2013-02-25 Matt Turner <mattst88@gmail.com>
186 * doc/invoke.texi: Document r4700.
188 2013-02-25 Richard Biener <rguenther@suse.de>
190 PR tree-optimization/56175
191 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
193 (simplify_bitwise_binary): ... here. Also guard the conversion
194 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
196 2013-02-25 Catherine Moore <clm@codesourcery.com>
199 2013-02-24 Catherine Moore <clm@codesourcery.com>
200 Maciej W. Rozycki <macro@codesourcery.com>
201 Tom de Vries <tom@codesourcery.com>
202 Nathan Sidwell <nathan@codesourcery.com>
203 Iain Sandoe <iain@codesourcery.com>
204 Nathan Froyd <froydnj@codesourcery.com>
205 Chao-ying Fu <fu@mips.com>
206 * doc/extend.texi: (micromips, nomicromips, nocompression):
207 Document new function attributes.
208 * doc/invoke.texi (minterlink-compressed, mmicromips,
209 m14k, m14ke, m14kec): Document new options.
210 (minterlink-mips16): Update documentation.
211 * doc/md.texi (ZC, ZD): Document new constraints.
212 * configure.ac (gcc_cv_as_micromips): Check if linker
213 supports the .set micromips directive.
214 * configure: Regenerate.
215 * config.in: Regenerate.
216 * config/mips/mips-tables.opt: Regenerate.
217 * config/mips/micromips.md: New file.
218 * constraints.md (ZC, AD): New constraints.
219 * config/mips/predicates.md (movep_src_register): New predicate.
220 (movep_src_operand): New predicate.
221 (non_volatile_mem_operand): New predicate.
222 * config/mips/mips.md (multimem): New type.
223 (length): Differentiate between 17-bit and 18-bit branch offsets.
224 (MOVEP1, MOVEP2): New mode iterator.
225 (mov_<load>l): Use ZC constraint.
226 (mov_<load>r): Likewise.
227 (mov_<store>l): Likewise.
228 (mov_<store>r): Likewise.
229 (*branch_equality<mode>_inverted): Add microMIPS support.
230 (*branch_equality<mode>): Likewise.
231 (*jump_absolute): Likewise.
232 (indirect_jump_<mode>): Likewise.
233 (tablejump_<mode>): Likewise.
234 (<optab>_internal): Likewise.
235 (sibcall_internal): Likewise.
236 (sibcall_value_internal): Likewise.
237 (prefetch): Use constraint ZD.
238 * config/mips/mips.opt (minterlink-compressed): New option.
239 (minterlink-mips16): Now an alias for minterlink-compressed.
240 (mmicromips): New option.
241 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
242 (compare_and_swap_12): Likewise.
243 (sync_add<mode>): Likewise.
244 (sync_<optab>_12): Likewise.
245 (sync_old_<optab>_12): Likewise.
246 (sync_new_<optab>_12): Likewise.
247 (sync_nand_12): Likewise.
248 (sync_old_nand_12): Likewise.
249 (sync_new_nand_12): Likewise.
250 (sync_sub<mode>): Likewise.
251 (sync_old_add<mode>): Likewise.
252 (sync_old_sub<mode>): Likewise.
253 (sync_new_add<mode>): Likewise.
254 (sync_new_sub<mode>): Likewise.
255 (sync_<optab><mode>): Likewise.
256 (sync_old_<optab><mode>): Likewise.
257 (sync_new_<optab><mode>): Likewise.
258 (sync_nand<mode>): Likewise.
259 (sync_old_nand<mode>): Likewise.
260 (sync_new_nand<mode>): Likewise.
261 (sync_lock_test_and_set<mode>): Likewise.
262 (test_and_set_12): Likewise.
263 (atomic_compare_and_swap<mode>): Likewise.
264 (atomic_exchange<mode>_llsc): Likewise.
265 (atomic_fetch_add<mode>_llsc): Likewise.
266 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
267 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
268 (umips_save_restore_pattern_p): Likewise.
269 (umips_load_store_pair_p): Likewise.
270 (umips_output_load_store_pair): Likewise.
271 (umips_movep_target_p): Likewise.
272 (umips_12bit_offset_address_p): Likewise.
273 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
274 (mips_base_mips16): Rename this...
275 (mips_base_compression_flags): ...to this. Update all uses.
276 (mips_attribute_table): Add micromips, nomicromips and nocompression.
277 (mips_mips16_decl_p): Delete.
278 (mips_nomips16_decl_p): Delete.
279 (mips_get_compress_on_flags): New function.
280 (mips_get_compress_off_flags): New function.
281 (mips_get_compress_mode): New function.
282 (mips_get_compress_on_name): New function.
283 (mips_get_compress_off_name): New function.
284 (mips_insert_attributes): Support multiple compression types.
285 (mips_merge_decl_attributes): Likewise.
286 (umips_12bit_offset_address_p): New function.
287 (mips_start_function_definition): Emit .set micromips directive.
288 (mips_call_may_need_jalx_p): New function.
289 (mips_function_ok_for_sibcall): Add microMIPS support.
290 (mips_print_operand_punctuation): Support short delay slots and
292 (umips_swm_mask, umips_swm_encoding): New.
293 (umips_build_save_restore): New function.
294 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
295 (was_mips16_p): Remove.
296 (old_compression_mode): New.
297 (mips_set_compression_mode): New function.
298 (mips_set_current_function): Add microMIPS support.
299 (mips_option_override): Likewise.
300 (umips_save_restore_pattern_p): New function.
301 (umips_output_save_restore): New function.
302 (umips_load_store_pair_p_1): New function.
303 (umips_load_store_pair_p): New function.
304 (umips_output_load_store_pair_1): New function.
305 (umips_output_load_store_pair): New function.
306 (umips_movep_target_p) New function.
307 (mips_prepare_pch_save): Add microMIPS support.
308 * config/mips/mips.h (TARGET_COMPRESSION): New.
309 (TARGET_CPU_CPP_BUILTINS): Update macro
310 to use new compression flags and to support microMIPS.
311 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
312 (MIPS_ARCH_FLOAT_SPEC): Likewise.
313 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
314 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
315 (ASM_SPEC): Support mmicromips and mno-micromips.
316 (M16STORE_REG_P): New macro.
317 (MIPS_CALL): Support TARGET_MICROMIPS.
318 (MICROMIPS_J): New macro.
319 (mips_base_mips16): Rename this...
320 (mips_base_compression_flags): ...to this.
321 (UMIPS_12BIT_OFFSET_P): New macro.
322 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
323 (MULTILIB_DIRNAMES): Likewise.
325 2013-02-25 Tom de Vries <tom@codesourcery.com>
327 PR rtl-optimization/56131
328 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
329 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
330 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
332 2013-02-25 Tobias Burnus <burnus@net-b.de>
334 * doc/invoke.texi (-fsanitize=): Move from optimization
335 to debugging options.
337 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
339 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
341 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
342 Alexander Monakov <amonakov@ispras.ru>
345 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
346 flush pending lists also on non-jumps. Adjust comment.
348 2013-02-24 Catherine Moore <clm@codesourcery.com>
349 Maciej W. Rozycki <macro@codesourcery.com>
350 Tom de Vries <tom@codesourcery.com>
351 Nathan Sidwell <nathan@codesourcery.com>
352 Iain Sandoe <iain@codesourcery.com>
353 Nathan Froyd <froydnj@codesourcery.com>
354 Chao-ying Fu <fu@mips.com>
356 * doc/extend.texi: (micromips, nomicromips, nocompression):
357 Document new function attributes.
358 * doc/invoke.texi (minterlink-compressed, mmicromips,
359 m14k, m14ke, m14kec): Document new options.
360 (minterlink-mips16): Update documentation.
361 * doc/md.texi (ZC, ZD): Document new constraints.
362 * configure.ac (gcc_cv_as_micromips): Check if linker
363 supports the .set micromips directive.
364 * configure: Regenerate.
365 * config.in: Regenerate.
366 * config/mips/mips-tables.opt: Regenerate.
367 * config/mips/micromips.md: New file.
368 * constraints.md (ZC, AD): New constraints.
369 * config/mips/predicates.md (movep_src_register): New predicate.
370 (movep_src_operand): New predicate.
371 (non_volatile_mem_operand): New predicate.
372 * config/mips/mips.md (multimem): New type.
373 (length): Differentiate between 17-bit and 18-bit branch offsets.
374 (MOVEP1, MOVEP2): New mode iterator.
375 (mov_<load>l): Use ZC constraint.
376 (mov_<load>r): Likewise.
377 (mov_<store>l): Likewise.
378 (mov_<store>r): Likewise.
379 (*branch_equality<mode>_inverted): Add microMIPS support.
380 (*branch_equality<mode>): Likewise.
381 (*jump_absolute): Likewise.
382 (indirect_jump_<mode>): Likewise.
383 (tablejump_<mode>): Likewise.
384 (<optab>_internal): Likewise.
385 (sibcall_internal): Likewise.
386 (sibcall_value_internal): Likewise.
387 (prefetch): Use constraint ZD.
388 * config/mips/mips.opt (minterlink-compressed): New option.
389 (minterlink-mips16): Now an alias for minterlink-compressed.
390 (mmicromips): New option.
391 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
392 (compare_and_swap_12): Likewise.
393 (sync_add<mode>): Likewise.
394 (sync_<optab>_12): Likewise.
395 (sync_old_<optab>_12): Likewise.
396 (sync_new_<optab>_12): Likewise.
397 (sync_nand_12): Likewise.
398 (sync_old_nand_12): Likewise.
399 (sync_new_nand_12): Likewise.
400 (sync_sub<mode>): Likewise.
401 (sync_old_add<mode>): Likewise.
402 (sync_old_sub<mode>): Likewise.
403 (sync_new_add<mode>): Likewise.
404 (sync_new_sub<mode>): Likewise.
405 (sync_<optab><mode>): Likewise.
406 (sync_old_<optab><mode>): Likewise.
407 (sync_new_<optab><mode>): Likewise.
408 (sync_nand<mode>): Likewise.
409 (sync_old_nand<mode>): Likewise.
410 (sync_new_nand<mode>): Likewise.
411 (sync_lock_test_and_set<mode>): Likewise.
412 (test_and_set_12): Likewise.
413 (atomic_compare_and_swap<mode>): Likewise.
414 (atomic_exchange<mode>_llsc): Likewise.
415 (atomic_fetch_add<mode>_llsc): Likewise.
416 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
417 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
418 (umips_save_restore_pattern_p): Likewise.
419 (umips_load_store_pair_p): Likewise.
420 (umips_output_load_store_pair): Likewise.
421 (umips_movep_target_p): Likewise.
422 (umips_12bit_offset_address_p): Likewise.
423 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
424 (mips_base_mips16): Rename this...
425 (mips_base_compression_flags): ...to this. Update all uses.
426 (mips_attribute_table): Add micromips, nomicromips and nocompression.
427 (mips_mips16_decl_p): Delete.
428 (mips_nomips16_decl_p): Delete.
429 (mips_get_compress_on_flags): New function.
430 (mips_get_compress_off_flags): New function.
431 (mips_get_compress_mode): New function.
432 (mips_get_compress_on_name): New function.
433 (mips_get_compress_off_name): New function.
434 (mips_insert_attributes): Support multiple compression types.
435 (mips_merge_decl_attributes): Likewise.
436 (umips_12bit_offset_address_p): New function.
437 (mips_start_function_definition): Emit .set micromips directive.
438 (mips_call_may_need_jalx_p): New function.
439 (mips_function_ok_for_sibcall): Add microMIPS support.
440 (mips_print_operand_punctuation): Support short delay slots and
442 (umips_swm_mask, umips_swm_encoding): New.
443 (umips_build_save_restore): New function.
444 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
445 (was_mips16_p): Remove.
446 (old_compression_mode): New.
447 (mips_set_compression_mode): New function.
448 (mips_set_current_function): Add microMIPS support.
449 (mips_option_override): Likewise.
450 (umips_save_restore_pattern_p): New function.
451 (umips_output_save_restore): New function.
452 (umips_load_store_pair_p_1): New function.
453 (umips_load_store_pair_p): New function.
454 (umips_output_load_store_pair_1): New function.
455 (umips_output_load_store_pair): New function.
456 (umips_movep_target_p) New function.
457 (mips_prepare_pch_save): Add microMIPS support.
458 * config/mips/mips.h (TARGET_COMPRESSION): New.
459 (TARGET_CPU_CPP_BUILTINS): Update macro
460 to use new compression flags and to support microMIPS.
461 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
462 (MIPS_ARCH_FLOAT_SPEC): Likewise.
463 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
464 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
465 (ASM_SPEC): Support mmicromips and mno-micromips.
466 (M16STORE_REG_P): New macro.
467 (MIPS_CALL): Support TARGET_MICROMIPS.
468 (MICROMIPS_J): New macro.
469 (mips_base_mips16): Rename this...
470 (mips_base_compression_flags): ...to this.
471 (UMIPS_12BIT_OFFSET_P): New macro.
472 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
473 (MULTILIB_DIRNAMES): Likewise.
475 2013-02-24 Jakub Jelinek <jakub@redhat.com>
478 * target-globals.c (save_target_globals): For init_reg_sets and
479 target_reinit remporarily set this_fn_optabs to this_target_optabs.
481 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
483 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
484 * config/aarch64/t-aarch64
485 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
487 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
490 * lra-constraints.c (process_alt_operands): Reload operand
491 conflicting with earlier clobber only if no more other conflicting
494 2013-02-22 Jakub Jelinek <jakub@redhat.com>
497 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
498 if not linking a shared library.
500 2013-02-22 Seth LaForge <sethml@google.com>
502 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
504 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
506 * config/arm/arm.md (split for extendsidi): Update condition.
507 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
508 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
509 (qhs_zextenddi_cstr): Likewise.
511 2013-02-21 Jakub Jelinek <jakub@redhat.com>
514 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
515 avoid signed wrapping.
516 (expand_mult): Handle properly multiplication by
517 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
518 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
519 in the compiler if coeff is HOST_WIDE_INT_MIN.
520 (expand_divmod): Don't make ext_op1 static, change it's type to
521 uhwi. Avoid undefined behavior in -INTVAL (op1).
523 PR rtl-optimization/50339
524 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
526 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
527 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
528 into splitting_ashiftrt field.
529 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
531 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
534 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
537 * trans-mem.c (execute_tm_mark): Do not expand transactions that
538 are sure to go irrevocable.
540 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
542 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
543 scalars are valid operands.
545 2013-02-21 Martin Jambor <mjambor@suse.cz>
547 PR tree-optimization/56310
548 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
549 only matching indices and non-negative final offsets.
550 (intersect_aggregates_with_edge): Pass src_idx to
551 agg_replacements_to_vector. Pass src_idx insstead of index to
552 intersect_with_agg_replacements.
554 2013-02-21 Martin Jambor <mjambor@suse.cz>
556 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
557 instead of hard-wired defaults.
559 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
561 * doc/invoke.texi (MIPS Options): Update documentation of the
562 floating-point multiply-accumulate instruction restrictions.
564 2013-02-21 Kostya Serebryany <kcc@google.com>
566 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
567 asan_shadow_offset on x86_64 linux.
569 2013-02-21 Richard Biener <rguenther@suse.de>
571 PR tree-optimization/56415
573 2013-02-11 Richard Biener <rguenther@suse.de>
575 PR tree-optimization/56273
576 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
579 2013-02-21 Jakub Jelinek <jakub@redhat.com>
582 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
586 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
587 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
589 2013-02-20 Jan Hubicka <jh@suse.cz>
591 PR tree-optimization/56265
592 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph when target is
593 referenced for firs ttime.
595 2013-02-20 Richard Biener <rguenther@suse.de>
597 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
598 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
599 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
600 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
602 (rename_ssa_copies): Do not remove unused locals.
603 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
604 * tree-ssanames.c (pass_release_ssa_names): Remove unused
606 * passes.c (execute_function_todo): Do not schedule unused locals
607 removal if cleanup_tree_cfg did something.
608 * tree-ssa-live.c (remove_unused_locals): Dump statistics
609 about the number of removed locals.
611 2013-02-20 Richard Biener <rguenther@suse.de>
613 PR tree-optimization/56398
614 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip
617 2013-02-20 Martin Jambor <mjambor@suse.cz>
619 PR tree-optimization/55334
620 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
621 restricted pointers to arrays.
623 2013-02-20 Richard Biener <rguenther@suse.de>
624 Jakub Jelinek <jakub@redhat.com>
626 PR tree-optimization/56396
627 * tree-ssa-ccp.c (n_const_val): New static variable.
628 (get_value): Return NULL for SSA names we don't have a lattice
630 (ccp_initialize): Initialize n_const_val.
631 * tree-ssa-copy.c (n_copy_of): New static variable.
632 (init_copy_prop): Initialize n_copy_of.
633 (get_value): Return NULL_TREE for SSA names we don't have a
636 2013-02-20 Martin Jambor <mjambor@suse.cz>
638 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
640 2013-02-20 Richard Biener <rguenther@suse.de>
642 * genpreds.c (write_lookup_constraint): Do not compare first
643 letter of the constraint again.
645 2013-02-20 Richard Biener <rguenther@suse.de>
647 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
649 (get_use_iv_cost): Terminate hashtable walk when coming across
652 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
654 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
655 reassociation for avx2 targets.
657 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
659 * config/microblaze/microblaze.c: microblaze_has_clz = 0
660 Add version check for v8.10.a to enable microblaze_has_clz
661 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
662 version and TARGET_PATTERN_COMPARE check
663 * config/microblaze/microblaze.md: New clzsi2 instruction
665 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
667 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
668 function before branching.
670 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
672 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
674 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
676 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
680 * sel-sched.c: Include ira.h.
681 (implicit_clobber_conflict_p): New function.
682 (moveup_expr): Use it.
683 * Makefile.in (sel-sched.o): Depend on ira.h.
685 2013-02-19 Richard Biener <rguenther@suse.de>
687 PR tree-optimization/56384
688 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
689 (vn_hash_type): Split out from ...
690 (vn_hash_constant_with_type): ... here.
691 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
692 (vn_phi_eq): Compare types from vn_phi_s structure.
693 (vn_phi_lookup): Populate vn_phi_s type.
694 (vn_phi_insert): Likewise.
696 2013-02-19 Jakub Jelinek <jakub@redhat.com>
698 PR tree-optimization/56350
699 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
700 if haven't found reduction or nested cycle operand, rather than
701 asserting we must find it.
703 PR tree-optimization/56381
704 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
707 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
708 Jakub Jelinek <jakub@redhat.com>
711 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
712 (swap_optab_enable): Same.
713 (init_all_optabs): Use argument instead of global.
714 * tree.h (struct tree_optimization_option): New field
716 * expr.h (init_all_optabs): Add argument to prototype.
717 (TREE_OPTIMIZATION_OPTABS): New.
718 (save_optabs_if_changed): Protoize.
719 * optabs.h: Declare this_fn_optabs.
720 * optabs.c (save_optabs_if_changed): New.
721 Declare this_fn_optabs.
722 (init_optabs): Add argument to init_all_optabs() call.
723 * function.c (invoke_set_current_function_hook): Handle per
725 * function.h (struct function): New field optabs.
726 * config/mips/mips.c (mips_set_mips16_mode): Handle when
727 optimization_current_node has changed.
728 * target-globals.h (save_target_globals_default_opts): Protoize.
729 * target-globals.c (save_target_globals_default_opts): New.
731 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
734 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
735 registers %fr12 and %fr12R as call used.
738 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
739 and HImode, require all displacements to be an integer multiple of their
741 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
742 only allow QImode and HImode when reload is in progress and strict is
743 true. Likewise for symbolic addresses. Use base14_operand to check
744 displacements in REG+BASE addresses.
746 2013-02-18 Richard Biener <rguenther@suse.de>
748 PR tree-optimization/56366
749 * tree-vect-loop.c (get_initial_def_for_induction): Properly
750 handle sign-conversion of outer-loop initial induction value.
752 2013-02-18 Richard Biener <rguenther@suse.de>
755 * cfghooks.c (merge_blocks): If we merge a latch into another
756 block adjust references to it.
757 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
758 (verify_loop_structure): Verify that a recorded latch is in fact
761 2013-02-18 Richard Biener <rguenther@suse.de>
763 PR tree-optimization/56321
764 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
765 order SSA name release and virtual operand unlinking.
767 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
769 * config/microblaze/microblaze.md (save_stack_block): Define.
770 (restore_stack_block): Likewise.
772 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
774 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
775 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
776 * config/microblaze/microblaze.c (microblaze_option_override):
777 Bail out early for PIC modes when target does not support PIC.
779 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
781 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
782 Replace with a microblaze version.
783 (microblaze_trampoline_init): Adapt for microblaze.
784 * gcc/config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
787 2013-02-16 Jakub Jelinek <jakub@redhat.com>
788 Dodji Seketeli <dodji@redhat.com>
791 * asan.c (get_mem_refs_of_builtin_call): White space and style
793 (instrument_mem_region_access): Do not forget to always put
794 instrumentation of the of 'base' and 'base + len' in a "if (len !=
795 0) statement, even for cases where either 'base' or 'base + len'
796 are not instrumented -- because they have been previously
797 instrumented. Simplify the logic by putting all the statements
798 instrument 'base + len' inside a sequence, and then insert that
799 sequence right before the current insertion point. Then, to
800 instrument 'base + len', just get an iterator on that statement.
801 And do not forget to update the pointer to iterator the function
802 received as argument.
804 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
806 PR rtl-optimization/56348
807 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
809 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
811 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
812 (clean_graph_dump_file): Pass base to start_graph_dump.
814 2013-02-14 Richard Henderson <rth@redhat.com>
817 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
819 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
821 * collect2-aix.h: Define F_LOADONLY.
823 2013-02-14 Richard Biener <rguenther@suse.de>
826 * varasm.c (output_constant_def_1): Get the decl representing
827 the constant as argument.
828 (output_constant_def): Wrap output_constant_def_1.
829 (make_decl_rtl): Use output_constant_def_1 with the decl
830 representing the constant.
831 (build_constant_desc): Optionally re-use a decl already
832 representing the constant.
833 (tree_output_constant_def): Adjust.
835 2013-02-14 Dodji Seketeli <dodji@redhat.com>
838 * asan.c (instrument_builtin_call): Really put the length of the
839 second source argument into src1_len.
841 2013-02-13 Jakub Jelinek <jakub@redhat.com>
843 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
844 argument. If it is false, don't create edge from then_bb to
846 (insert_if_then_before_iter): Pass true to it.
847 (build_check_stmt): Pass false to it.
848 (transform_statements): Flush hash table only on extended basic
849 block boundaries, rather than at the beginning of every bb.
850 Don't flush hash table on nonfreeing_call_p calls.
851 * tree-flow.h (nonfreeing_call_p): New prototype.
852 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
854 2013-02-13 David S. Miller <davem@davemloft.net>
856 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
858 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
861 * ira.c (max_regno_before_ira): Move from ...
863 (fix_reg_equiv_init): Use max_regno_before_ira instead of
866 2013-02-13 Jakub Jelinek <jakub@redhat.com>
868 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
870 2013-02-13 Richard Biener <rguenther@suse.de>
873 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
876 2013-02-13 Richard Biener <rguenther@suse.de>
878 * loop-init.c (loop_optimizer_init): Clear loop state when
879 re-initializing preserved loops.
880 * loop-unswitch.c (unswitch_single_loop): Return whether
881 we unswitched the loop. Do not verify loop state here.
882 (unswitch_loops): When we unswitched a loop discover new
885 2013-02-13 Kostya Serebryany <kcc@google.com>
887 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset on x86_64
889 * sanitizer.def: Rename __asan_init to __asan_init_v1.
891 2013-02-12 Dodji Seketeli <dodji@redhat.com>
893 Avoid instrumenting duplicated memory access in the same basic block
894 * Makefile.in (asan.o): Add new dependency on hash-table.h
895 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
896 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
897 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
898 (free_mem_ref_resources, has_mem_ref_been_instrumented)
899 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
900 (get_mem_ref_of_assignment): New functions.
901 (get_mem_refs_of_builtin_call): Extract from
902 instrument_builtin_call and tweak a little bit to make it fit with
904 (instrument_builtin_call): Use the new
905 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
906 of is_gimple_builtin_call.
907 (instrument_derefs, instrument_mem_region_access): Insert the
908 instrumented memory reference into the hash table.
909 (maybe_instrument_assignment): Renamed instrument_assignment into
910 this, and change it to advance the iterator when instrumentation
911 actually happened and return true in that case. This makes it
912 homogeneous with maybe_instrument_assignment, and thus give a
913 chance to callers to be more 'regular'.
914 (transform_statements): Clear the memory reference hash table
915 whenever we enter a new BB, when we cross a function call, or when
916 we are done transforming statements. Use
917 maybe_instrument_assignment instead of instrumentation. No more
918 need to special case maybe_instrument_assignment and advance the
919 iterator after calling it; it's now handled just like
920 maybe_instrument_call. Update comment.
922 2013-02-13 Richard Biener <rguenther@suse.de>
924 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
925 Fix loop discovery code.
927 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
930 * lra-constraints.c (process_alt_operands): Match early clobber
931 operand with itself. Check conflicts with earlyclobber only if
932 the operand is not reloaded. Prefer to reload conflicting operand
933 if earlyclobber and matching operands are the same.
935 2013-02-12 Richard Biener <rguenther@suse.de>
938 * lto-streamer-out.c (write_symbol): Do not output symbols
939 for hard register variables.
941 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
944 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
945 (umulsidi3_insn, mulsidi3_insn): New insns.
947 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
949 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
950 (struct tune_params): Add vec_costs field.
951 * config/arm/arm.c (arm_builtin_vectorization_cost)
952 (arm_add_stmt_cost): New functions.
953 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
954 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
955 (arm_default_vec_cost): New struct of type cpu_vec_costs.
956 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
957 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
958 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
959 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
961 2013-02-12 Richard Biener <rguenther@suse.de>
964 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
965 decls again if possible.
967 2013-02-12 Richard Biener <rguenther@suse.de>
970 * tree-ssa.c (verify_ssa_name): Fix check, move
971 SSA_NAME_IN_FREE_LIST check up.
973 2013-02-12 Jakub Jelinek <jakub@redhat.com>
974 Steven Bosscher <steven@gcc.gnu.org>
976 PR rtl-optimization/56151
977 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
978 equal to op0 or op1, and last_insn pattern is CODE operation
979 with MEM dest and one of the operands matches that MEM.
981 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
983 * doc/extend.texi: Document Function Multiversioning and "default"
984 parameter string to target attribute.
985 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
986 target attribute parameter is "default".
987 (ix86_compare_version_priority): Remove checks for target attribute.
988 (ix86_mangle_function_version_assembler_name): Change error to sorry.
989 Remove check for target attribute equal to NULL. Add assert.
990 (ix86_generate_version_dispatcher_body): Change error to sorry.
992 2013-02-11 Iain Sandoe <iain@codesourcery.com>
993 Jack Howarth <howarth@bromo.med.uc.edu>
994 Patrick Marlier <patrick.marlier@gmail.com>
997 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
998 define ENDFILE_SPEC as TM_DESTRUCTOR.
999 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
1001 2013-02-11 Alexander Potapenko <glider@google.com>
1002 Jack Howarth <howarth@bromo.med.uc.edu>
1003 Jakub Jelinek <jakub@redhat.com>
1006 * config/darwin.c (cdtor_record): Rename ctor_record.
1007 (sort_cdtor_records): Rename sort_ctor_records.
1008 (finalize_dtors): New routine to sort destructors by
1009 priority before use in assemble_integer.
1010 (machopic_asm_out_destructor): Use finalize_dtors if needed.
1012 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
1014 PR rtl-optimization/56275
1015 * simplify-rtx.c (avoid_constant_pool_reference): Check that
1016 offset is non-negative and less than cmode size before
1017 calling simplify_subreg.
1019 2013-02-11 Richard Biener <rguenther@suse.de>
1021 PR tree-optimization/56264
1022 * cfgloop.h (fix_loop_structure): Adjust prototype.
1023 * loop-init.c (fix_loop_structure): Return the number of
1024 newly discovered loops.
1025 * tree-cfgcleanup.c (repair_loop_structures): When new loops
1026 are discovered, do a full loop-closed SSA rewrite.
1028 2013-02-11 Richard Biener <rguenther@suse.de>
1030 PR tree-optimization/56273
1031 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
1033 (check_array_ref): Fix missing newline in dumps.
1034 (search_for_addr_array): Likewise.
1036 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
1038 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
1040 2013-02-09 Jakub Jelinek <jakub@redhat.com>
1043 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
1045 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
1047 PR rtl-optimization/56246
1048 * lra-constraints.c (simplify_operand_subreg): Try to reuse
1050 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
1051 constraints are satisfied.
1053 2013-02-08 Jeff Law <law@redhat.com>
1056 * emit-rtl.c (reg_is_parm_p): New function.
1057 * regs.h (reg_is_parm_p): New prototype.
1058 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
1059 callee-clobbered registers.
1061 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
1064 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
1065 If there is no implicit builtin declaration, just return NULL.
1067 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
1069 * config/i386/sse.md (FMAMODEM): New mode iterator.
1070 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
1071 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
1073 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
1075 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
1076 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
1077 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
1079 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1081 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
1082 (microblaze*-*-elf): Likewise.
1083 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
1085 * config/microblaze/microblaze-c.c: Add builtin defines for
1086 _LITTLE_ENDIAN and _BIG_ENDIAN.
1087 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
1088 add to TARGET_DEFAULT flags.
1089 Expand ASM_SPEC and LINK_SPEC.
1090 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
1091 * config/microblaze/microblaze.md: Update extendsidi2 and
1092 movdi_internal instructions to use low-order / high-order reg
1094 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
1095 options and inversemask / mask of LITTLE_ENDIAN.
1096 * config/microblaze/t-microblaze: Expand multilib options to
1097 include mlittle-endian (le) and update exceptions patterns.
1099 2013-02-08 Jakub Jelinek <jakub@redhat.com>
1101 PR rtl-optimization/56195
1102 * lra-constraints.c (get_reload_reg): Don't reuse regs
1103 if they have smaller mode than requested, if they have
1104 wider mode than requested, try to return a SUBREG.
1106 PR tree-optimization/56250
1107 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
1108 if type is unsigned and code isn't MULT_EXPR.
1110 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
1112 PR tree-optimization/56064
1113 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
1114 bits according to mode.
1115 * fixed-value.h (fixed_from_double_int)
1116 (const_fixed_from_double_int): Adjust comments.
1118 2013-02-08 Richard Biener <rguenther@suse.de>
1121 * lto-streamer.h (struct data_in): Remove current_file, current_line
1122 and current_col members.
1123 * lto-streamer-out.c (lto_output_location): Stream changed bits
1124 en-block for efficiency.
1125 * lto-streamer-in.c (clear_line_info): Remove.
1126 (lto_input_location): Cache current file, line and column
1127 globally via local statics. Read changed bits en-block.
1128 (input_function): Do not call clear_line_info.
1129 (lto_read_body): Likewise.
1130 (lto_input_toplevel_asms): Likewise.
1132 2013-02-08 Michael Matz <matz@suse.de>
1134 PR tree-optimization/52448
1135 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
1136 (nt_call_phase): New static.
1137 (add_or_mark_expr): Only mark accesses with newer phase than any
1139 (nonfreeing_call_p): New.
1140 (nt_init_block): Update nt_call_phase, mark blocks as visited.
1141 (nt_fini_block): Keep blocks marked as visited.
1142 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
1144 2013-02-08 Richard Biener <rguenther@suse.de>
1146 * ira.c (ira): Free broken dominator information.
1148 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
1150 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
1152 2013-02-08 Marek Polacek <polacek@redhat.com>
1154 * cfgloop.c (verify_loop_structure): Add more checking of headers.
1156 2013-02-08 Richard Biener <rguenther@suse.de>
1159 * cfgloop.h (flow_loops_find): Adjust.
1160 (bb_loop_header_p): Declare.
1161 * cfgloop.c (bb_loop_header_p): New function split out from ...
1162 (flow_loops_find): ... here. Adjust function signature,
1163 support incremental loop structure update.
1164 (verify_loop_structure): Cleanup. Verify a loop is a loop.
1165 * cfgloopmanip.c (fix_loop_structure): Move ...
1166 * loop-init.c (fix_loop_structure): ... here.
1167 (apply_loop_flags): Split out from ...
1168 (loop_optimizer_init): ... here.
1169 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
1170 in incremental mode, only remove dead loops here.
1172 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
1175 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
1176 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
1177 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
1178 (*round<mode>3.libgcc): New insns for fixed-modes.
1179 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
1180 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
1181 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
1182 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
1183 implementations. Define to __builtin_avr_absFX,
1184 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
1185 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
1186 __builtin_avr_countlsFX, respectively.
1187 * config/avr/avr-c.c (target.h): Include it.
1188 (enum avr_builtin_id): New enum.
1189 (avr_resolve_overloaded_builtin): New static function.
1190 (avr_register_target_pragmas): Use it to set
1191 targetm.resolve_overloaded_builtin.
1192 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
1193 tree nodes used by DEF_BUILTIN.
1194 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
1195 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
1196 <AVR_BUILTIN_xxBITS>: Same.
1198 2013-02-08 Richard Biener <rguenther@suse.de>
1200 * cfgloop.c (verify_loop_structure): Properly handle
1201 a loop exiting to another loop header.
1202 * ira-int.h (ira_loops): Remove.
1203 * ira.c (ira_loops): Remove.
1204 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
1205 (do_reload): Use loop_optimizer_finalize.
1206 * ira-build.c (create_loop_tree_nodes): Use get_loops and
1207 number_of_loops to access the loop tree.
1208 (more_one_region_p): Likewise.
1209 (finish_loop_tree_nodes): Likewise.
1210 (rebuild_regno_allocno_maps): Likewise.
1211 (mark_loops_for_removal): Likewise.
1212 (mark_all_loops_for_removal): Likewise.
1213 (remove_unnecessary_regions): Likewise.
1214 (ira_build): Likewise.
1215 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
1217 2013-02-08 Richard Biener <rguenther@suse.de>
1219 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
1220 * ipa-pure-const.c (analyze_function): Avoid calling
1221 mark_irreducible_loops twice.
1222 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
1224 2013-02-07 David S. Miller <davem@davemloft.net>
1226 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
1228 * var-tracking.c (vt_add_function_parameter): Test the presence of
1229 HAVE_window_save properly and do not remap argument registers when
1230 we have a leaf function.
1232 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
1235 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
1237 * config/i386/i386.c (ix86_print_operand): Ditto.
1239 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1241 * lra-constraints.c (process_alt_operands): Fix recently added comment.
1243 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1245 PR rtl-optimization/56225
1246 * lra-constraints.c (process_alt_operands): Check that reload hard
1247 reg can hold value for strict_low_part.
1249 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1252 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
1253 dwarf2out_end_function.
1254 (in_first_function_p, maybe_at_text_label_p,
1255 first_loclabel_num_not_at_text_label): New variables.
1256 (dwarf2out_var_location): In the first function find out
1257 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
1258 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
1261 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
1263 PR rtl-optimization/56178
1264 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
1265 SUBREG of a register. Tidy up related block of code.
1266 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
1267 note if the source is a register or a SUBREG of a register.
1269 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1272 * config/rs6000/rs6000.md (ptrm): New mode attr.
1273 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
1274 call_value_indirect_aix<pttrsize>,
1275 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
1278 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1280 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
1281 if -bnortl. Convert to strcmp and strncmp.
1283 2013-02-07 Alan Modra <amodra@gmail.com>
1286 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
1287 addresses won't wrap when offsetting.
1288 (rs6000_secondary_reload): Provide secondary reloads needed for
1289 wrapping LO_SUM addresses.
1291 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
1293 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
1294 MACH, just __MACH__.
1296 2013-02-06 Richard Biener <rguenther@suse.de>
1298 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
1299 instead of calling fix_loop_structure.
1301 2013-02-06 Jakub Jelinek <jakub@redhat.com>
1304 * omp-low.c (use_pointer_for_field): Return false if
1305 lower_send_shared_vars doesn't generate any copy-out code.
1307 2013-02-06 Tom de Vries <tom@codesourcery.com>
1309 PR rtl-optimization/56131
1310 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
1311 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
1312 of the label is NULL. Add comment.
1314 2013-02-05 Jakub Jelinek <jakub@redhat.com>
1316 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
1319 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
1320 (STATIC_LIBTSAN_LIBS): Likewise.
1321 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
1322 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
1323 is defined, don't add anything else beyond that.
1324 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
1325 (LINK_COMMAND_SPEC): Use them.
1327 PR tree-optimization/56205
1328 * tree-stdarg.c (check_all_va_list_escapes): Return true if
1329 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
1330 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
1332 2013-02-05 Richard Biener <rguenther@suse.de>
1334 PR tree-optimization/53342
1335 PR tree-optimization/53185
1336 * tree-vectorizer.h (vect_check_strided_load): Remove.
1337 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
1338 not disallow peeling for vectorized strided loads.
1339 (vect_check_strided_load): Make static and simplify.
1340 (vect_analyze_data_refs): Adjust.
1341 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
1342 correctly when vectorizing strided loads.
1344 2013-02-05 Richard Biener <rguenther@suse.de>
1346 * doc/install.texi: Refer to ISL, not PPL.
1348 2013-02-05 Jan Hubicka <jh@suse.cz>
1350 PR tree-optimization/55789
1351 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
1353 2013-02-05 Jan Hubicka <jh@suse.cz>
1355 PR tree-optimization/55789
1356 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
1357 the dead call anyway.
1359 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
1362 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
1364 2013-02-04 Alexander Potapenko <glider@google.com>
1365 Jack Howarth <howarth@bromo.med.uc.edu>
1366 Jakub Jelinek <jakub@redhat.com>
1369 * config/darwin.c (sort_ctor_records): Stabilized qsort
1370 on constructor priority by using original position.
1371 (finalize_ctors): New routine to sort constructors by
1372 priority before use in assemble_integer.
1373 (machopic_asm_out_constructor): Use finalize_ctors if needed.
1375 2013-02-04 Jakub Jelinek <jakub@redhat.com>
1378 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
1379 about visibility on artificial decls.
1380 * config/sol2.c (solaris_assemble_visibility): Likewise.
1382 2013-02-04 Kai Tietz <ktietz@redhat.com>
1385 * config/i386/i386.c (function_value_ms_64): Add additional valtype
1386 argument and improve checking of return-argument types for 16-byte
1388 (ix86_function_value_1): Add additional valtype argument on call
1389 of function_value_64.
1390 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
1391 handling infunction_value_64 function.
1393 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1395 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
1397 2013-02-04 Richard Biener <rguenther@suse.de>
1399 PR tree-optimization/56188
1400 * tree-ssa-structalias.c (label_visit): Consider case with
1401 initially non-empty points-to set.
1402 (perform_var_substitution): Dump node mapping and clean up.
1404 2013-02-04 Richard Guenther <rguenther@suse.de>
1407 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
1408 node prevail as last resort.
1409 (lto_symtab_merge_decls): Remove guard on LTRANS here.
1410 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
1412 2013-02-04 Richard Biener <rguenther@suse.de>
1414 PR tree-optimization/56113
1415 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
1417 (equiv_class_lookup_or_add): ... this.
1418 (label_visit): Adjust and fix error in previous patch.
1419 (perform_var_substitution): Adjust.
1421 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
1423 * config/sh/divtab.c: Fix formatting and comments throughout the file.
1424 * config/sh/sh4-300.md: Likewise.
1425 * config/sh/sh4a.md: Likewise.
1426 * config/sh/constraints.md: Likewise.
1427 * config/sh/sh.md: Likewise.
1428 * config/sh/netbsd-elf.h: Likewise.
1429 * config/sh/predicates.md: Likewise.
1430 * config/sh/sh-protos.h: Likewise.
1431 * config/sh/ushmedia.h: Likewise.
1432 * config/sh/linux.h: Likewise.
1433 * config/sh/sh.c: Likewise.
1434 * config/sh/superh.h: Likewise.
1435 * config/sh/elf.h: Likewise.
1436 * config/sh/sh4.md: Likewise.
1437 * config/sh/sh.h: Likewise.
1439 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1441 * config/pa/constraints.md: Adjust unused letters. Change "T"
1442 constraint to match_test floating_point_store_memory_operand().
1443 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
1444 (base14_operand): New.
1445 (floating_point_store_memory_operand): New.
1446 (integer_store_memory_operand): Revise to use base14_operand and
1447 reg_plus_base_memory_operand.
1448 (move_dest_operand): Allow symbolic_memory_operands.
1449 (symbolic_memory_operand): Check for LO_SOM.
1450 (symbolic_operand): Change default case to break.
1451 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
1452 CONST_DOUBLE values to be reloaded by putting them into memory when
1453 the destination is a floating point register.
1454 (movdf): Remove code to handle CONST_DOUBLE.
1456 (reload_indf_r1): New.
1457 (reload_insf_r1): New.
1458 Consistently use "Q" and "T" constraints with integer and floating
1459 point move instructions, respectively.
1460 (movdi): Remove FAIL.
1461 Change predicate for source operand unamed DImode move from
1462 general_operand to move_src_operand.
1463 (umulsidi3): Change predicate for destination operand to
1465 Likewise for similar unamed patterns.
1466 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
1467 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
1468 (hppa_legitimize_address): Simplify mask calculation.
1469 (pa_emit_move_sequence): Revised handling of secondary reloads from
1470 REG+D addresses for floating point loads and stores. Directly handle
1471 loading CONST0_RTX (mode) to a floating point register.
1472 (pa_secondary_reload): Handle reloading DF and SFmode constant values
1473 to floating point registers. Don't restrict secondary reloads to
1474 floating point registers to integer modes. Revise some comments and
1476 (TARGET_LEGITIMATE_ADDRESS_P): Define.
1477 (pa_legitimate_address_p): New.
1478 (pa_legitimize_reload_address): New.
1479 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
1480 (STRICT_REG_OK_FOR_BASE_P): New.
1481 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
1482 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
1484 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
1485 Andrew Dixie <andrewd@gentrack.com>
1487 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
1490 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
1492 * expmed.c (extract_bit_field_1): Pass the full width of the
1493 structure to get_best_reg_extraction_insn.
1495 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1498 * configure.ac (use_cxa_atexit): Add AIX.
1499 * configure: Regenerate.
1501 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
1503 2013-02-01 Jakub Jelinek <jakub@redhat.com>
1506 * final.c (need_profile_function): New variable.
1507 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
1508 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
1509 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
1510 notes, targetm.asm_out.function_prologue doesn't emit anything,
1511 HAVE_prologue and profiler should be emitted before prologue,
1512 set need_profile_function instead of emitting it.
1513 (final_scan_insn): If need_profile_function, emit
1514 profile_function on the first NOTE_INSN_BASIC_BLOCK or
1515 NOTE_INSN_FUNCTION_BEG note.
1517 2013-02-01 Richard Henderson <rth@redhat.com>
1519 * config/rs6000/rs6000.md (smulditi3): New.
1522 * config/alpha/alpha.md (umulditi3): New.
1524 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1526 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
1527 (ASM_OUTPUT_ALIGNED_LOCAL): New.
1529 2013-02-01 Richard Biener <rguenther@suse.de>
1531 PR tree-optimization/56113
1532 * tree-ssa-structalias.c (label_visit): Reduce work for
1533 single-predecessor nodes.
1535 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
1537 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
1538 range isn't testing for zero.
1540 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
1543 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
1545 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
1546 Nick Clifton <nickc@redhat.com>
1548 * config/v850/constraints.md (Q): Define as a memory constraint.
1549 * config/v850/predicates.md (label_ref_operand): New predicate.
1550 (e3v5_shift_operand): New predicate.
1551 (ior_operator): New predicate.
1552 * config/v850/t-v850: Add e3v5 multilib.
1553 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
1554 (v850_gen_movdi): Prototype.
1555 * config/v850/v850.c: Add support for e3v5 architecture.
1556 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
1558 (construct_save_jarl): Add e3v5 long JARL support.
1559 (v850_adjust_insn_length): New function. Adjust length of call
1560 insns when using e3v5 instructions.
1561 (v850_gen_movdi): New function: Generate instructions to move a
1563 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
1564 (CPP_SPEC): Define __v850e3v5__ as appropriate.
1565 (TARGET_USE_FPU): Enable for e3v5.
1566 (CONST_OK_FOR_W): New macro.
1567 (ADJUST_INSN_LENGTH): Define.
1568 * config/v850/v850.md (UNSPEC_LOOP): Define.
1569 (attr cpu): Add v850e3v5.
1570 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
1571 (movdi): New pattern.
1572 (movdi_internal): New pattern.
1573 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
1574 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
1575 (cstoresf4): Likewise.
1576 (cstoredf4): Likewise.
1577 (insv): New pattern.
1578 (rotlso3_a): New pattern.
1579 (rotlsi3_b): New pattern
1580 (rotlsi3_v850e3v5): New pattern.
1581 (doloop_begin): New pattern.
1582 (fix_loop_counter): New pattern.
1583 (doloop_end): New pattern.
1584 (branch_normal): Add e3v5 long branch support.
1585 (branch_invert): Likewise.
1586 (branch_z_normal): Likewise.
1587 (branch_z_invert): Likewise.
1588 (branch_nz_normal): Likewise.
1589 (branch_nz_invert): Likewise.
1590 (call_internal_short): Add e3v5 register-indirect JARL support.
1591 (call_internal_long): Likewise.
1592 (call_value_internal_short): Likewise.
1593 (call_value_internal_long): Likewise.
1594 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
1595 (mloop): New option.
1596 * config.gcc: Add support for configuring v840e3v5 target.
1597 * doc/invoke.texi: Document new v850 specific command line options.
1599 2013-01-31 Paul Koning <ni1d@arrl.net>
1603 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
1604 children if parent is a class.
1605 (prune_unused_types_prune): Don't add DW_AT_declaration.
1607 2013-01-31 Richard Biener <rguenther@suse.de>
1609 PR tree-optimization/56157
1610 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
1611 match up operand with SLP child.
1613 2013-01-31 Jason Merrill <jason@redhat.com>
1616 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
1617 parameters the first time.
1618 (gen_scheduled_generic_parms_dies): Check completeness here.
1620 2013-01-31 Richard Biener <rguenther@suse.de>
1623 * common.opt (faggressive-loop-optimizations): New flag,
1625 * doc/invoke.texi (faggressive-loop-optimizations): Document.
1626 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
1627 infer_loop_bounds_from_undefined by it.
1629 2013-01-31 Richard Biener <rguenther@suse.de>
1631 PR tree-optimization/56150
1632 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
1633 visit virtual operands.
1634 (find_uses_to_rename_bb): Likewise.
1636 2013-01-31 Richard Biener <rguenther@suse.de>
1638 PR tree-optimization/56150
1639 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
1640 mixed store non-store stmts.
1642 2013-01-30 Jakub Jelinek <jakub@redhat.com>
1645 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
1646 LIBASAN_EARLY_SPEC is defined.
1647 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
1648 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
1650 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
1653 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
1654 invalid args instead of ICEing on it.
1655 (ix86_valid_target_attribute_tree): Return error_mark_node if
1656 ix86_valid_target_attribute_inner_p failed.
1657 (ix86_valid_target_attribute_p): Return false only if
1658 ix86_valid_target_attribute_tree returned error_mark_node. Allow
1659 target("default") attribute.
1660 (sorted_attr_string): Change argument from const char * to tree,
1661 merge in all target attribute arguments rather than just one.
1662 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
1663 instead of free. Avoid using strcat.
1664 (ix86_mangle_function_version_assembler_name): Mangle
1665 target("default") as if no target attribute is present. Adjust
1666 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
1667 instead of xmalloc and XDELETEVEC instead of free.
1668 (ix86_function_versions): Don't return true if one of the decls
1669 doesn't have target attribute. If they don't and one of the decls
1670 is DECL_FUNCTION_VERSIONED, report an error. Adjust
1671 sorted_attr_string caller. Use XDELETEVEC instead of free.
1672 (ix86_supports_function_versions): Remove.
1673 (make_name): Fix up formatting.
1674 (make_dispatcher_decl): Remove resolver_name and its initialization.
1675 Avoid leaking memory.
1676 (is_function_default_version): Return true if there is
1677 target("default") attribute rather than no target attribute at all.
1678 (make_resolver_func): Avoid leaking memory.
1679 (ix86_generate_version_dispatcher_body): Likewise.
1680 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
1681 * target.def (supports_function_versions): Remove.
1682 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
1683 * doc/tm.texi: Regenerated.
1685 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
1687 PR rtl-optimization/56144
1688 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
1689 for values with side effects.
1691 2013-01-30 Richard Biener <rguenther@suse.de>
1693 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
1694 (sparseset_pop): Likewise.
1695 * cfganal.c (compute_idf): Likewise. Increase work-stack size
1696 to be able to use quick_push in the worker loop.
1698 2013-01-30 Marek Polacek <polacek@redhat.com>
1700 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
1702 2013-01-30 Richard Biener <rguenther@suse.de>
1705 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
1707 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
1709 PR tree-optimization/56064
1710 * fixed-value.c (fixed_from_double_int): New function.
1711 * fixed-value.h (fixed_from_double_int): New prototype.
1712 (const_fixed_from_double_int): New static inline function.
1713 * fold-const.c (native_interpret_fixed): New static function.
1714 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
1715 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
1716 (native_encode_fixed): New static function.
1717 (native_encode_expr) <FIXED_CST>: Use it.
1718 (native_interpret_int): Move double_int worker code to...
1719 * double-int.c (double_int::from_buffer): ...this new static method.
1720 * double-int.h (double_int::from_buffer): Prototype it.
1722 2013-01-30 Richard Biener <rguenther@suse.de>
1724 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
1725 New pointer-map and obstack.
1726 (init_alias_vars): Allocate pointer-map and obstack.
1727 (delete_points_to_sets): Free them.
1728 (find_what_var_points_to): Cache result.
1729 (find_what_p_points_to): Adjust for changed interface of
1730 find_what_var_points_to.
1731 (compute_points_to_sets): Likewise.
1732 (ipa_pta_execute): Likewise.
1734 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1736 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
1737 * configure: Regenerate.
1738 * config.in: Regenerate.
1739 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
1740 #nobits/#progbits if supported.
1742 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
1745 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
1746 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
1747 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
1749 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1751 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
1752 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
1754 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1756 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
1757 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
1759 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1761 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
1763 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
1764 * config/arm/cortex-a7.md: New bypasses using
1765 arm_mac_accumulator_is_result.
1767 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1769 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
1770 (cortex_a7_neon_mla): Likewise.
1771 (cortex_a7_fpfmad): New reservation.
1772 (cortex_a7_fpmacs): Use ffmas and update required units.
1773 (cortex_a7_fpmuld): Update required units and latency.
1774 (cortex_a7_fpmacd): Likewise.
1775 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
1776 (cortex_a7_neon). Likewise.
1777 (bypass) Update participating units.
1779 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1781 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
1782 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
1784 * config/arm/vfp11.md (vfp_farith): Use ffmas.
1785 (vfp_fmul): Use ffmad.
1786 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
1787 (cortex_r4_fmacd): Use ffmad.
1788 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
1789 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
1790 (cortex_a9_fmacd): Use ffmad.
1791 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
1792 (cortex_a8_vfp_macd): Use ffmad.
1793 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
1794 (cortex_a5_fpmacd): Use ffmad.
1795 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
1796 (cortex_a15_vfp_macd): Use ffmad.
1797 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
1799 2013-01-29 Jason Merrill <jason@redhat.com>
1802 * varasm.c (default_assemble_visibility): Don't warn about
1803 visibility on artificial decls.
1805 2013-01-29 Richard Biener <rguenther@suse.de>
1807 PR tree-optimization/56113
1808 * tree-ssa-structalias.c (equiv_class_lookup): Also return
1810 (label_visit): Free duplicate bitmaps and record the leader instead.
1811 (perform_var_substitution): Adjust.
1813 2013-01-29 Richard Biener <rguenther@suse.de>
1815 PR tree-optimization/55270
1816 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
1817 the CFG, schedule loops for fixup.
1819 2013-01-29 Nick Clifton <nickc@redhat.com>
1821 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
1824 2013-01-28 Leif Ekblad <leif@rdos.net>
1826 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
1827 * config/i386/i386.h (TARGET_RDOS): New macro.
1828 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
1829 * config/i386/i386.c (ix86_option_override_internal): For 64bit
1830 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
1831 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
1832 DEFAULT_LARGE_SECTION_THRESHOLD.
1833 * config/i386/i386.md (R14_REG, R15_REG): New constants.
1834 * config/i386/rdos.h: New file.
1835 * config/i386/rdos64.h: New file.
1837 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
1840 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
1843 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1845 PR rtl-optimization/56117
1846 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
1847 call cselib_lookup_from_insn on the MEM before calling
1848 add_insn_mem_dependence.
1850 2013-01-28 Richard Biener <rguenther@suse.de>
1852 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
1853 to a stmt that didn't have one.
1854 (copy_phis_for_bb): Likewise for PHI arguments.
1855 (copy_debug_stmt): Likewise for debug stmts.
1857 2013-01-28 Richard Biener <rguenther@suse.de>
1859 PR tree-optimization/56034
1860 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
1861 (partition_builtin_p): Adjust.
1862 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
1863 it is the last partition.
1864 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
1865 up the vertex for the definition.
1866 (classify_partition): Classify whether a partition is a
1867 PKIND_REDUCTION, thus has uses outside of the loop.
1868 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
1869 Merge all PKIND_REDUCTION partitions into the last partition.
1870 (tree_loop_distribution): Seed partitions from reductions as well.
1872 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1874 PR tree-optimization/56125
1875 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
1876 pow(x,c) into sqrt(x) * powi(x, n/2) or
1877 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
1878 optimizing for size.
1879 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
1880 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
1883 PR tree-optimization/56094
1884 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
1885 to UNKNOWN_LOCATION while gimplifying expr.
1887 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
1890 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
1891 operand 0 in movabs insn template for -masm=intel asm alternative.
1892 (*movabs<mode>_2): Ditto for operand 1.
1894 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
1897 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
1900 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1902 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
1905 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
1907 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
1908 Undef to avoid warning.
1910 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1912 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
1913 * configure: Regenerate.
1915 2013-01-25 Jakub Jelinek <jakub@redhat.com>
1917 PR tree-optimization/56098
1918 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
1919 for stmts with volatile ops.
1920 (cond_store_replacement): Don't optimize if assign has volatile ops.
1921 (cond_if_else_store_replacement_1): Don't optimize if either
1922 then_assign or else_assign have volatile ops.
1923 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
1926 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1928 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
1930 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1932 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
1933 missing ':' in asm example.
1935 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
1937 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
1938 entries into lane and laneq entries.
1939 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
1940 Remove AdvSIMD scalar modes.
1941 (aarch64_sq<r>dmulh_laneq<mode>): New.
1942 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
1944 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
1945 builtin implementations to relfect changes in RTL in aarch64-simd.md.
1946 * config/aarch64/iterators.md (VCOND): New.
1949 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1952 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
1953 Add NULL LIBNAME argument to existing definitions.
1954 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
1955 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
1956 * config/avr/avr.c (DEF_BUILTIN): Same.
1957 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
1958 (avr_expand_builtin): Expand to a vanilla call if a libgcc
1959 implementation is available (DECL_ASSEMBLER_NAME is set).
1960 (avr_fold_absfx): New static function.
1961 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
1962 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
1963 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
1965 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
1966 (abshk, absk, abslk, absllk): Provide as static inline functions.
1968 2013-01-25 Marek Polacek <polacek@redhat.com>
1970 PR tree-optimization/56035
1971 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
1973 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
1975 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
1976 (*movtf_internal_rex64): Add (!o,C) alternative
1977 (*movxf_internal_rex64): Ditto.
1978 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
1980 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
1982 * doc/invoke.texi: fix typo.
1983 * doc/objc.texi: fix typo.
1985 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
1987 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
1988 for the first two alternatives.
1990 2013-01-24 Diego Novillo <dnovillo@google.com>
1992 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
1993 (ggc-zone.o): Remove.
1994 * configure.ac: Remove option --with-gc.
1995 * configure: Re-generate.
1996 * doc/install.texi: Remove documentation for --with-gc.
1997 * gengtype.c (write_enum_defn): Remove. Update all users.
1998 (write_Types_process_field): Remove generation of gt_e_* argument.
1999 (output_type_enum): Remove. Update all users.
2000 (write_enum_defn): Remove. Update all users.
2001 (enum alloc_zone): Remove. Update all users.
2002 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
2003 * ggc-common.c (ggc_splay_alloc): Remove first argument.
2005 (struct ptr_data): Remove field TYPE. Update all users.
2006 (gt_pch_note_object): Remove argument TYPE. Update all users.
2007 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
2009 * ggc-none.c (ggc_alloc_typed_stat): Remove.
2010 (struct alloc_zone): Remove.
2011 (ggc_internal_alloc_zone_stat): Remove.
2012 (ggc_internal_cleared_alloc_zone_stat): Remove.
2013 * ggc-page.c (ggc_alloc_typed_stat): Remove.
2014 (ggc_pch_count_object): Remove last argument. Update all users.
2015 (ggc_pch_alloc_object): Remove last argument. Update all users.
2016 (struct alloc_zone): Remove.
2017 * ggc-zone.c: Remove.
2018 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
2019 (struct alloc_zone): Remove.
2020 (ggc_alloc_typed_stat): Remove.
2021 (ggc_alloc_typed): Remove.
2022 (ggc_splay_alloc): Remove first argument.
2023 (rtl_zone): Remove. Update all users.
2024 (tree_zone): Remove. Update all users.
2025 (tree_id_zone): Remove. Update all users.
2026 (ggc_internal_zone_alloc_stat): Remove. Update all users.
2027 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
2028 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
2029 * tree-ssanames.c: Remove references to zone allocator in comments.
2031 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
2033 * config/avr/avr.c (avr_out_fract): Make register numbers that
2034 might be outside of source operand signed.
2036 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
2038 * config/i386/constraints.md (Yf): New constraint.
2039 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
2040 of f constraint to conditionaly disable x87 register preferences.
2041 (*movdf_internal): Ditto.
2042 (*movsf_internal): Ditto.
2044 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
2047 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
2048 that have operands with impossible constraints.
2049 Add a FIXME for a speed-up opportunity.
2050 * lra-constraints.c (process_alt_operands): Verify that a class
2051 selected from constraints on asms is valid for the operand mode.
2052 (curr_insn_transform): Remove incorrect comment.
2054 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
2056 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
2057 TOC operand is a valid symbol ref in the constant pool.
2059 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2061 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
2063 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
2066 * config/avr/stdfix.h: New file.
2067 * t-avr (stdfix-gcc.h): New rule to build it.
2068 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
2070 2013-01-23 Kostya Serebryany <kcc@google.com>
2072 * config/darwin.h: remove dependency on
2073 CoreFoundation (asan on Mac OS).
2075 2013-01-23 Jakub Jelinek <jakub@redhat.com>
2078 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
2079 instead of cmpdi_operand for first comparison operand.
2080 Don't assert that comparison operands aren't both constants.
2082 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
2084 * doc/install.texi (Downloading the Source): Update references to
2085 downloading separate components.
2087 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
2089 * doc/extend.texi (__int128): Improve grammar.
2091 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
2094 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
2095 alternative to (o,r).
2096 (*movdi_internal_rex64): Remove (!o,n) alternative.
2097 (DImode immediate->memory splitter): Remove.
2098 (DImode immediate->memory peephole2): Remove.
2099 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
2100 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
2101 alternative to (!o,*r).
2102 (*movtf_internal_sse): New pattern.
2103 (*movxf_internal_rex64): New pattern.
2104 (*movxf_internal): Disable for TARGET_64BIT.
2105 (*movdf_internal_rex64): Remove (!o,F) alternative.
2107 2013-01-22 Jakub Jelinek <jakub@redhat.com>
2110 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
2111 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
2112 * tree-vect-loop-manip.c (find_loop_location): Also ignore
2113 stmt locations where LOCATION_LOCUS of the stmt location is
2114 UNKNOWN_LOCATION or BUILTINS_LOCATION.
2117 * config/i386/i386.md (UNSPEC_STOS): New.
2118 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
2119 *strsetqi_1): Add UNSPEC_STOS.
2121 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
2124 * doc/invoke.texi: Remove left over -Wsynth example.
2126 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2128 PR tree-optimization/56051
2129 * fold-const.c (fold_binary_loc): Don't fold
2130 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
2131 a narrowing conversion, or widening conversion from signed
2134 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2136 PR rtl-optimization/56023
2137 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
2138 dependent on debug instruction.
2140 2013-01-21 Martin Jambor <mjambor@suse.cz>
2143 * function.c (allocate_struct_function): Call
2144 invoke_set_current_function_hook earlier.
2146 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2148 * reload1.c (init_reload): Only initialize reload_obstack
2149 during the first call.
2151 2013-01-21 Marek Polacek <polacek@redhat.com>
2153 * cfgloop.c (verify_loop_structure): Fix up grammar.
2155 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
2157 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
2158 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
2160 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2163 * config/arm/marvell-pj4.md: Update copyright year.
2164 Fix up use of alu to alu_reg and simple_alu_imm.
2166 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2168 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
2170 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
2173 * lra-constraints.c (curr_insn_transform): Don't reuse original
2174 insn for secondary memory move when memory mode should be different.
2176 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2178 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
2179 atomic_storedi_1): New patterns.
2181 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2183 btver2 pipeline descriptions.
2184 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
2186 * config/i386/i386.md (btver2_decode): New type attributes.
2187 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
2189 * config/i386/btver2.md: New file describing btver2 pipelines.
2191 2013-01-19 Andrew Pinski <apinski@cavium.com>
2193 PR tree-optimization/52631
2194 * tree-ssa-sccvn (visit_use): Before looking up the original
2195 statement, try looking up the simplified expression.
2197 2013-01-19 Anthony Green <green@moxielogic.com>
2199 * config/moxie/moxie.c (moxie_expand_prologue): Set
2200 current_function_static_stack_size.
2202 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2204 PR tree-optimization/56029
2205 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
2206 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
2208 2013-01-18 Sharad Singhai <singhai@google.com>
2210 PR tree-optimization/55995
2211 * dumpfile.c (dump_loc): Print location only if available.
2212 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
2214 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
2217 * lra-constraints.c (curr_insn_transform): Reuse original insn for
2218 secondary memory move.
2219 (inherit_reload_reg): Use rclass instead of cl for
2220 check_secondary_memory_needed_p.
2222 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2225 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
2226 the case where writing real complex part of target modifies op1.
2228 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
2230 * config/aarch64/aarch64-simd.md
2231 (aarch64_vcond_internal<mode>): Handle unordered cases.
2232 * config/aarch64/iterators.md (v_cmp_result): New.
2234 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
2235 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2237 * config/arm/marvell-pj4.md: New file.
2238 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
2239 * config/arm/arm.md (generic_sched): Add marvell_pj4.
2240 (generic_vfp): Likewise.
2241 * config/arm/arm-cores.def: Add marvell-pj4.
2242 * config/arm/arm-tune.md: Regenerate.
2243 * config/arm/arm-tables.opt: Regenerate.
2244 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
2245 * doc/invoke.texi: Document marvell-pj4.
2247 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
2249 * config/aarch64/arm_neon.h: Map scalar types to standard types.
2251 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2256 * var-tracking.c (negative_power_of_two_p): New.
2257 (global_get_addr_cache, local_get_addr_cache): New.
2258 (get_addr_from_global_cache, get_addr_from_local_cache): New.
2259 (vt_canonicalize_addr): Rewrite using the above. Adjust the
2261 (vt_stack_offset_p): Remove.
2262 (vt_canon_true_dep): Always canonicalize loc's address.
2263 (clobber_overlapping_mems): Make sure we have a MEM.
2264 (local_get_addr_clear_given_value): New.
2265 (val_reset): Clear local cached entries.
2266 (compute_bb_dataflow): Create and release the local cache.
2267 Disable duplicate MEMs clobbering.
2268 (emit_notes_in_bb): Clobber MEMs likewise.
2269 (vt_emit_notes): Create and release the local cache.
2270 (vt_initialize, vt_finalize): Create and release the global
2271 cache, respectively.
2272 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
2274 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2277 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
2278 not found in the symtab.
2280 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2283 PR rtl-optimization/55547
2284 PR rtl-optimization/53827
2287 * alias.c (offset_overlap_p): New, factored out of...
2288 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
2289 the conservative special case for symbolic constants. Don't
2290 adjust zero sizes on alignment.
2292 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
2294 PR rtl-optimization/52573
2295 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
2296 REG_UNUSED for the same register.
2298 2013-01-17 Richard Biener <rguenther@suse.de>
2299 Marek Polacek <polacek@redhat.com>
2301 PR rtl-optimization/55833
2302 * loop-unswitch.c (unswitch_loops): Move loop verification...
2303 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
2304 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
2305 Set it to true when we're removing a loop from hierarchy tree in
2306 an irreducible region.
2307 (fix_bb_placements): Adjust caller.
2308 (fix_loop_placements): Likewise.
2310 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
2312 * config/avr/builtins.def (DEF_BUILTIN): Factor out
2313 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
2314 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
2315 Remove ID. Adjust comments.
2316 * config/avr/avr-c.c (avr_builtin_name): Remove.
2317 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
2318 * config/avr/avr.c (avr_tolower): New static function.
2319 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
2320 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
2321 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
2324 2013-01-17 Jan Hubicka <jh@suse.cz>
2326 PR tree-optimization/55273
2327 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
2329 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
2332 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
2333 store through atomic_store<mode>_1.
2334 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
2336 2013-01-17 Martin Jambor <mjambor@suse.cz>
2338 PR tree-optimizations/55264
2339 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
2340 for virtual methods.
2341 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
2342 virtual methods before inlining is over.
2343 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
2345 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
2348 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
2350 PR rtl-optimization/56005
2351 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
2352 pending reads for prefetch.
2354 2013-01-16 Ian Bolton <ian.bolton@arm.com>
2356 * config/aarch64/aarch64.md
2357 (*cstoresi_neg_uxtw): New pattern.
2358 (*cmovsi_insn_uxtw): New pattern.
2359 (*<optab>si3_uxtw): New pattern.
2360 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
2361 (*<optab>si3_insn_uxtw): New pattern.
2362 (*bswapsi2_uxtw): New pattern.
2364 2013-01-16 Richard Biener <rguenther@suse.de>
2366 * tree-inline.c (tree_function_versioning): Remove set but
2367 never used variable.
2369 2013-01-16 Richard Biener <rguenther@suse.de>
2371 PR tree-optimization/55964
2372 * tree-flow.h (rename_variables_in_loop): Remove.
2373 (rename_variables_in_bb): Likewise.
2374 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
2375 (copy_loop_before): Adjust and delete update-ssa status.
2376 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
2377 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
2378 (rename_variables_in_loop): Remove.
2379 (slpeel_update_phis_for_duplicate_loop): Likewise.
2380 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
2381 use available cfg machinery instead of duplicating it.
2382 Update PHI nodes and perform poor-mans SSA update here.
2383 (slpeel_tree_peel_loop_to_edge): Adjust.
2385 2013-01-16 Richard Biener <rguenther@suse.de>
2387 PR tree-optimization/54767
2388 PR tree-optimization/53465
2389 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
2390 (vrp_visit_phi_node): For PHI arguments coming via backedges
2391 drop all symbolical range information.
2392 (execute_vrp): Compute backedges.
2394 2013-01-16 Richard Biener <rguenther@suse.de>
2396 * doc/install.texi: Update CLooG and ISL requirements to
2399 2013-01-16 Christian Bruel <christian.bruel@st.com>
2402 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
2403 (broken_move): Handle UNSPECV_SP_SWITCH_B.
2404 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
2406 2013-01-16 DJ Delorie <dj@redhat.com>
2408 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
2409 (UNSPECV_SP_SWITCH_E): New.
2410 (sp_switch_1): Change to an unspec.
2411 (sp_switch_2): Change to an unspec. Don't use post-inc when we
2414 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
2416 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
2417 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
2418 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
2419 (expand_mem_thread_fence): Ditto.
2420 (expand_mem_signal_fence): Ditto.
2421 (expand_atomic_load): Ditto.
2422 (expand_atomic_store): Ditto.
2424 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
2426 PR rtl-optimization/55547
2427 PR rtl-optimization/53827
2430 * alias.c (memrefs_conflict_p): Set sizes to negative after
2433 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2436 * function.c (thread_prologue_and_epilogue_insns): Always
2437 add crtl->drap_reg to set_up_by_prologue.set, even if
2438 stack_realign_drap is false.
2440 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2442 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
2443 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
2444 *call): Fix indention.
2446 2013-01-15 Tom de Vries <tom@codesourcery.com>
2449 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
2452 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
2454 PR rtl-optimization/55153
2455 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
2457 2013-01-15 Martin Jambor <mjambor@suse.cz>
2459 PR tree-optimization/55920
2460 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
2461 accesses as grp_to_be_debug_replaced.
2463 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2465 PR tree-optimization/55920
2466 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
2467 there is non-useless type conversion needed from debug rhs to lhs,
2468 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
2470 2013-01-15 Joseph Myers <joseph@codesourcery.com>
2471 Mikael Pettersson <mikpe@it.uu.se>
2474 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
2476 (ASM_OUTPUT_CASE_LABEL): Remove.
2477 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
2478 * final.c (shorten_branches): Update alignment of labels before
2479 jump tables if CASE_VECTOR_SHORTEN_MODE.
2481 2013-01-15 Richard Biener <rguenther@suse.de>
2484 * system.h: Do not include gmp.h for building host tools.
2486 2013-01-15 Richard Biener <rguenther@suse.de>
2489 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
2490 account for bitpos when computing alignment.
2492 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
2494 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
2495 (ix86_target_macros_internal): Likewise.
2497 * config/i386/i386.c (m_CORE2I7): Removed.
2498 (m_CORE_HASWELL): New macro.
2499 (m_CORE_ALL): Likewise.
2500 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
2501 (initial_ix86_arch_features): Likewise.
2502 (processor_target_table): Initializations for Core avx2.
2503 (cpu_names): New names "core-avx2".
2504 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
2505 PROCESSOR_CORE_HASWELL.
2506 (ix86_issue_rate): New case.
2507 (ia32_multipass_dfa_lookahead): Likewise.
2508 (ix86_sched_init_global): Likewise.
2510 * config/i386/i386.h (TARGET_HASWELL): New macro.
2511 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
2512 (processor_type): New PROCESSOR_HASWELL.
2514 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2516 PR tree-optimization/55955
2517 * tree-vect-loop.c (vectorizable_reduction): Give up early on
2518 *SHIFT_EXPR and *ROTATE_EXPR codes.
2520 PR tree-optimization/48766
2521 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
2522 -ftrapv disable -fwrapv.
2524 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2527 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
2528 etc. to 1 and not to __flash.
2529 Use LL suffix for __INT24_MAX__ with -mint8.
2530 Use ULL suffix for __UINT24_MAX__ with -mint8.
2532 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2534 * config/avr/avr-arch.h
2535 (struct base_arch_s): Use typedef avr_arch_t instead.
2536 (struct arch_info_s): Use typedef avr_arch_info_t instead.
2537 (struct mcu_type_s): Use typedef avr_mcu_t instead.
2538 * config/avr/avr.c: Same.
2539 * config/avr/avr-devices.c: Same.
2540 * config/avr/driver-avr.c: Same.
2541 * config/avr/gen-avr-mmcu-texi.c: Same.
2542 * config/avr/avr-mcus.def: Adjust comment.
2544 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
2546 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
2547 * config/aarch64/iterators.md (VALLDI): New.
2549 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
2550 Andi Kleen <ak@linux.intel.com>
2553 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
2554 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
2557 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2559 * config/avr/avr-stdint.h: Remove trailing blanks.
2560 * config/avr/avr-log.h: Same.
2561 * config/avr/avr-arch.h: Same.
2562 * config/avr/avr-devices.c: Same.
2563 * config/avr/avr-dimode.md: Same.
2564 * config/avr/predicates.md: Same.
2565 * config/avr/avr-c.c: Same. And fix typo.
2567 * config/avr/avr-protos.h: Same. And:
2568 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
2569 (init_cumulative_args): Rename to avr_init_cumulative_args.
2570 (expand_prologue): Rename to avr_expand_prologue.
2571 (expand_epilogue): Rename to avr_expand_epilogue.
2572 (adjust_insn_length): Rename to avr_adjust_insn_length.
2573 (notice_update_cc): Rename to avr_notice_update_cc.
2574 (final_prescan_insn): Rename to avr_final_prescan_insn.
2575 * config/avr/avr.c: Same.
2576 * config/avr/avr.h: Same.
2577 * config/avr/avr.md: Remove trailing blanks.
2578 (prologue): Use avr_expand_prologue.
2579 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
2581 2013-01-14 Richard Biener <rguenther@suse.de>
2583 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
2584 verify_location, collect_subblocks): New functions.
2585 (verify_gimple_in_cfg): Verify that locations only reference
2586 BLOCKs in the functions BLOCK tree.
2588 2013-01-14 Richard Biener <rguenther@suse.de>
2590 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
2592 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
2594 (insert_out_of_ssa_copy_on_edge): Likewise.
2595 (rewrite_close_phi_out_of_ssa): Likewise.
2596 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
2598 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
2599 propagated constants.
2600 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
2603 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2605 * config/avr/avr-modes.def: Add GPL copyright notice.
2607 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
2609 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
2610 MEMMODEL_MASK to determine memory model.
2611 (atomic_store<mode>): Ditto from operands[2].
2612 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
2614 2013-01-13 Jakub Jelinek <jakub@redhat.com>
2617 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
2618 (fold_gimple_assign): Don't call unshare_expr here.
2619 (fold_ctor_reference): Call unshare_expr.
2621 2013-01-13 Terry Guo <terry.guo@arm.com>
2623 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
2624 * doc/fragments.texi: Document MULTILIB_REUSE.
2625 * gcc.c (multilib_reuse): New internal spec.
2626 (set_multilib_dir): Also search multilib from multilib_reuse.
2627 * genmultilib (tmpmultilib3): Refactor code.
2628 (tmpmultilib4): Ditto.
2629 (multilib_reuse): New multilib argument.
2631 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
2633 * Makefile.in: Update copyright.
2635 2013-01-12 Tom de Vries <tom@codesourcery.com>
2638 * calls.c (expand_call): Check if arg_nr is valid.
2640 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2642 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
2643 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
2644 documentation. Add missing '__' in front of
2645 __builtin_ia32_packssdw256.
2647 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2650 * config/s390/s390.c (s390_preferred_reload_class): Do not return
2651 NO_REGS for larl operands.
2652 (s390_reload_larl_operand): Use s390_load_address instead of
2655 2013-01-11 Richard Biener <rguenther@suse.de>
2657 * tree-cfg.c (verify_node_sharing_1): Split out from ...
2658 (verify_node_sharing): ... here.
2659 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
2661 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
2663 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
2664 Substitute TREECHECKING.
2665 * configure: Regenerate.
2666 * Makefile.in (TREECHECKING): New.
2668 2013-01-11 Richard Guenther <rguenther@suse.de>
2670 PR tree-optimization/44061
2671 * tree-vrp.c (extract_range_basic): Compute zero as
2672 value-range for __builtin_constant_p of function parameters.
2674 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
2676 Update copyright years.
2678 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
2680 PR rtl-optimization/55672
2681 * lra-eliminations.c (mark_not_eliminable): Permit addition with
2682 const to be eliminable.
2684 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
2686 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
2687 * configure: Regenerate.
2689 2013-01-10 Richard Biener <rguenther@suse.de>
2691 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
2693 2013-01-10 Richard Biener <rguenther@suse.de>
2696 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
2697 locations for virtual PHI arguments.
2698 (rewrite_update_phi_arguments): Likewise.
2700 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
2702 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
2705 2013-01-10 Jakub Jelinek <jakub@redhat.com>
2707 PR tree-optimization/55921
2708 * tree-complex.c (expand_complex_asm): New function.
2709 (expand_complex_operations_1): Call it for GIMPLE_ASM.
2711 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2714 * config/s390/s390.c (s390_symref_operand_p)
2715 (s390_loadrelative_operand_p): Merge the two functions.
2716 (s390_check_qrst_address, print_operand_address): Add parameters
2717 to s390_loadrelative_operand_p invokation.
2718 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
2719 (s390_reload_larl_operand, s390_secondary_reload): Use
2720 s390_loadrelative_operand_p instead of s390_symref_operand_p.
2721 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
2723 2013-01-09 Mike Stump <mikestump@comcast.net>
2725 * dse.c (record_store): Remove unnecessary assert.
2727 2013-01-09 Jan Hubicka <jh@suse.cz>
2729 PR tree-optimization/55569
2730 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
2731 * cfgloop.h (scale_loop_profile): Likewise.
2733 2013-01-09 Jan Hubicka <jh@suse.cz>
2736 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
2738 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
2740 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
2743 * expr.h (maybe_emit_group_store): Declare.
2744 * expr.c (maybe_emit_group_store): New function.
2745 * builtins.c (expand_builtin_int_roundingfn): Call it.
2746 (expand_builtin_int_roundingfn_2): Likewise.
2748 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
2750 PR rtl-optimization/55829
2751 * lra-constraints.c (match_reload): Add code for absent output.
2752 (curr_insn_transform): Add code for reloads of matched inputs
2755 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
2757 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
2758 attribute of movddup insn to DF.
2759 (*vec_interleave_lowv2df): Ditto.
2760 (vec_dupv2df): Ditto.
2762 2013-01-09 Jan Hubicka <jh@suse.cz>
2764 PR tree-optimiation/55875
2765 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
2766 EVERY_ITERATION parameter.
2767 (number_of_iterations_exit): Check if exit is executed every iteration.
2768 (idx_infer_loop_bounds): Similarly here.
2769 (n_of_executions_at_most): Simplify
2770 to only test for cases where statement is dominated by the
2771 particular bound; handle correctly the "postdominance" test.
2772 (scev_probably_wraps_p): Use max loop iterations info
2773 as a global bound first.
2775 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
2776 Nick Clifton <nickc@redhat.com>
2778 * config/v850/v850.md (cbranchsf4): New pattern.
2779 (cstoresf4): New pattern.
2780 (cbranchdf4): New pattern.
2781 (cstoredf4): New pattern.
2782 (movsicc): Disallow floating point comparisons.
2783 (cmpsf_le_insn): Fix order of operators.
2784 (cmpsf_lt_insn): Likewise.
2785 (cmpsf_eq_insn): Likewise.
2786 (cmpdf_le_insn): Likewise.
2787 (cmpdf_lt_insn): Likewise.
2788 (cmpdf_eq_insn): Likewise.
2789 (cmpsf_ge_insn): Use LE comparison.
2790 (cmpdf_ge_insn): Likewise.
2791 (cmpsf_gt_insn): Use LT comparison.
2792 (cmpdf_gt_insn): Likewise.
2793 (cmpsf_ne_insn): Delete pattern.
2794 (cmpdf_ne_insn): Delete pattern.
2795 * config/v850/v850.c (v850_gen_float_compare): Use
2796 gen_cmpdf_eq_insn for NE comparison.
2797 (v850_float_z_comparison_operator)
2798 (v850_float_nz_comparison_operator): Move from here ...
2799 * config/v850/predicates.md: ... to here. Move GT and GE
2800 comparisons into v850_float_z_comparison_operator.
2801 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
2803 (v850_float_nz_comparison_operator): Likewise.
2805 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2807 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
2808 with calls to gen_insvsi/gen_insvdi.
2810 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2812 * config/i386/i386.c (initial_ix86_tune_features): Set up
2813 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
2815 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
2816 Jakub Jelinek <jakub@redhat.com>
2818 PR tree-optimization/48189
2819 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
2820 If nitercst is 0, don't predict the exit edge.
2822 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
2824 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
2825 in asm_fprintf with reg_names.
2826 (aarch64_print_operand_address): Likewise.
2827 (aarch64_return_addr): Likewise.
2828 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
2830 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2832 * config/pa/pa.h (VAL_U6_BITS_P): Define.
2833 (INT_U6_BITS): Likewise.
2834 * config/pa/predicates.md (uint6_operand): New predicate.
2835 (shift5_operand, shift6_operand): Likewise.
2836 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
2838 (lshrdi3): Use shift6_operand.
2839 (shrpsi4, shrpdi4): New insn patterns.
2840 (extzv): Delete expander.
2841 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
2842 predicates in unamed zero extract patterns. Tighten common constraint.
2843 (extv): Delete expander.
2844 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
2845 predicates in unamed sign extract patterns. Tighten common constraint.
2846 (insv): Delete expander.
2847 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
2848 predicates in unamed insert patterns. Tighten common constraint.
2849 Change uint32_operand predicate to uint6_operand predicate in unamed
2850 DImode pattern to insert constant values of type 1...1xxxx.
2852 2013-01-04 Jan Hubicka <jh@suse.cz>
2854 PR tree-optimization/55823
2855 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
2858 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2859 Uros Bizjak <ubizjak@gmail.com>
2861 PR rtl-optimization/55845
2862 * df-problems.c (can_move_insns_across): Stop scanning at
2863 volatile_insn_p source instruction or give up if
2864 across_from .. across_to range contains any volatile_insn_p
2867 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2869 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
2870 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
2872 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
2873 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
2875 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2878 * asan.c (asan_clear_shadow): New function.
2879 (asan_emit_stack_protection): Use it.
2881 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2883 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
2884 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
2885 with tab instead of space.
2887 2013-01-08 Nick Clifton <nickc@redhat.com>
2889 * config/rl78/rl78.c (rl78_expand_prologue): Always select
2890 register bank 0 at the start of an interrupt handler.
2891 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
2894 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2896 * config/aarch64/aarch64-simd.md
2897 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
2898 (aarch64_simd_bsl): Likewise.
2899 (aarch64_vcond_internal<mode>): Likewise.
2900 (vcond<mode><mode>): Likewise.
2901 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
2902 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
2904 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2906 * config/aarch64/aarch64-builtins.c
2907 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
2909 2013-01-08 Martin Jambor <mjambor@suse.cz>
2912 * tree-sra.c (analyze_access_subtree): Return true also after
2913 potentially creating a debug-only replacement.
2915 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2918 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
2920 PR tree-optimization/54120
2921 * tree-vrp.c (range_fits_type_p): Don't allow
2922 src_precision < precision from signed vr to unsigned_p
2923 if vr->min or vr->max is negative.
2924 (simplify_float_conversion_using_ranges): Test can_float_p
2925 against CODE_FOR_nothing.
2927 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2928 Richard Biener <rguenther@suse.de>
2931 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
2932 types instead of just INTEGER_TYPE types.
2934 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
2936 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
2939 2013-01-07 Steve Ellcey <sellcey@mips.com>
2942 * config/mips/mips.opt: Change mad to mmad to match documentation.
2944 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2947 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
2950 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2953 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
2954 (avr_addrspace_t): Add .section_name field.
2955 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
2957 (avr_addrspace): Same. Initialize .section_name. Remove last
2958 NULL entry. Put __memx into .progmemx.data.
2959 (progmem_section_prefix): Remove.
2960 (avr_asm_init_sections): No need to initialize progmem_section.
2961 (avr_asm_named_section): Use avr_addrspace[].section_name to get
2962 section name prefix.
2963 (avr_asm_select_section): Ditto. And use get_unnamed_section to
2964 retrieve the progmem section.
2965 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
2966 boundary to run over avr_addrspace[].
2967 (avr_register_target_pragmas): Ditto.
2969 2013-01-06 Jakub Jelinek <jakub@redhat.com>
2971 * varasm.c (output_constant_def_contents): For asan_protect_global
2972 protected strings, adjust DECL_ALIGN if needed, before testing for
2974 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
2975 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
2977 (output_object_block): For asan protected decls, emit asan padding
2978 after their contents.
2979 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
2980 (asan_finish_file): Test it here instead.
2982 2013-01-07 Nick Clifton <nickc@redhat.com>
2983 Matthias Klose <doko@debian.org>
2984 Doug Kwan <dougkwan@google.com>
2985 H.J. Lu <hongjiu.lu@intel.com>
2988 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
2990 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
2992 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
2994 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
2996 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
2998 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
3001 * doc/install.texi (Cross-Compiler-Specific Options): Document
3004 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
3006 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
3007 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
3008 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
3009 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
3010 vqmovun_high_s64): Fix source operand number and update copyright.
3012 2013-01-07 Richard Biener <rguenther@suse.de>
3015 * gimple.h (gimple_call_builtin_p): New overload.
3016 * gimple.c (validate_call): New function.
3017 (gimple_call_builtin_p): Likewise.
3018 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3019 Use gimple_call_builtin_p.
3020 (find_func_clobbers): Likewise.
3021 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
3022 (strlen_optimize_stmt): Likewise.
3024 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3026 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
3027 (vld1q_dup_*): Likewise.
3029 (vld1q_*): Likewise.
3030 (vld1_lane_*): Likewise.
3031 (vld1q_lane_*): Likewise.
3033 2013-01-07 Richard Biener <rguenther@suse.de>
3035 * lto-streamer.h (LTO_minor_version): Bump to 2.
3037 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3039 * config/aarch64/aarch64-protos.h
3040 (aarch64_const_double_zero_rtx_p): Rename to...
3041 (aarch64_float_const_zero_rtx_p): ...this.
3042 (aarch64_float_const_representable_p): New.
3043 (aarch64_output_simd_mov_immediate): Likewise.
3044 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
3045 move immediate case.
3046 * config/aarch64/aarch64.c
3047 (aarch64_const_double_zero_rtx_p): Rename to...
3048 (aarch64_float_const_zero_rtx_p): ...this.
3049 (aarch64_print_operand): Allow printing of new constants.
3050 (aarch64_valid_floating_const): New.
3051 (aarch64_legitimate_constant_p): Check for valid floating-point
3053 (aarch64_simd_valid_immediate): Likewise.
3054 (aarch64_vect_float_const_representable_p): New.
3055 (aarch64_float_const_representable_p): Likewise.
3056 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
3057 (aarch64_output_simd_mov_immediate): New.
3058 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
3059 (*movdf_aarch64): Likewise.
3060 * config/aarch64/constraints.md (Ufc): New.
3061 (Y): call aarch64_float_const_zero_rtx.
3062 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
3064 2013-01-07 Richard Biener <rguenther@suse.de>
3066 PR tree-optimization/55888
3067 PR tree-optimization/55862
3068 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
3069 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
3070 not if it is contained therein.
3072 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
3074 * config/avr/t-avr: Typo.
3076 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
3079 * config/avr/t-avr: Don't automatically rebuild
3080 $(srcdir)/config/avr/t-multilib
3081 $(srcdir)/config/avr/avr-tables.opt
3082 $(srcdir)/doc/avr-mmcu.texi
3083 (avr-mcus): New phony target to build them on request.
3084 (s-avr-mlib, s-avr-mmcu-texi): Remove.
3085 * avr/avr-mcus.def: Adjust comments.
3087 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
3089 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
3091 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
3093 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
3095 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
3097 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
3099 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
3101 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
3102 to generate profiling.
3103 * config/rs6000/aix64.h (LIB_SPEC): Same.
3105 2013-01-04 Andrew Pinski <apinski@cavium.com>
3107 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
3109 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
3111 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
3113 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
3115 (ix86_expand_move): Ditto.
3116 (ix86_zero_extend_to_Pmode): Ditto.
3117 (ix86_expand_call): Ditto.
3118 (ix86_expand_special_args_builtin): Ditto.
3119 (ix86_expand_builtin): Ditto.
3121 2013-01-04 Richard Biener <rguenther@suse.de>
3123 PR tree-optimization/55862
3124 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
3125 translating them through PHI nodes.
3127 2013-01-04 Martin Jambor <mjambor@suse.cz>
3129 PR tree-optimization/55755
3130 * tree-sra.c (sra_modify_assign): Do not check that an access has no
3131 children when trying to avoid producing a VIEW_CONVERT_EXPR.
3133 2013-01-04 Marek Polacek <polacek@redhat.com>
3136 * opts.c (default_options_optimization): Clarify error message.
3138 2013-01-04 Richard Biener <rguenther@suse.de>
3141 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
3144 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3147 * config/pa/pa.md (movsi): Revert previous change.
3148 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
3151 2013-01-03 Richard Henderson <rth@redhat.com>
3153 * config/i386/i386.c (ix86_expand_move): Always assign to op1
3154 after eliminating TLS symbols.
3156 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
3159 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
3160 * graphite-poly.c (debug_gmp_value): Likewise.
3162 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
3165 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
3166 selected code model, define __code_mode_small__, __code_model_medium__,
3167 __code_model_large__, __code_model_32__ or __code_model_kernel__.
3168 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
3169 xchg temporary register with %k. Declare temporary register as
3171 [__x86_64__]: For medium and large code models, preserve %rbx register.
3173 2013-01-03 Richard Biener <rguenther@suse.de>
3175 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
3176 (dump_subscript): Adjust.
3177 (finalize_ddr_dependent): Do not dump redundant info.
3178 (analyze_siv_subscript): Adjust.
3179 (subscript_dependence_tester): Likewise.
3180 (compute_affine_dependence): Likewise.
3182 2013-01-03 Richard Biener <rguenther@suse.de>
3185 2013-01-03 Richard Biener <rguenther@suse.de>
3187 PR tree-optimization/55857
3188 * tree-vect-stmts.c (vectorizable_load): Do not setup
3189 re-alignment for invariant loads.
3191 2013-01-02 Richard Biener <rguenther@suse.de>
3193 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3194 invariant load do not generate a vector load from the scalar location.
3196 2013-01-03 Richard Biener <rguenther@suse.de>
3198 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
3199 for not vectorizing.
3200 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
3201 not build INDIRECT_REFs, call get_name once only.
3202 (vect_create_data_ref_ptr): Likewise. Dump base object kind
3203 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
3205 2013-01-03 Richard Biener <rguenther@suse.de>
3207 PR tree-optimization/55857
3208 * tree-vect-stmts.c (vectorizable_load): Do not setup
3209 re-alignment for invariant loads.
3211 2013-01-03 Richard Biener <rguenther@suse.de>
3214 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
3215 prefer a built-in decl.
3217 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3219 * gcc.c (process_command): Update copyright notice dates.
3220 * gcov.c (print_version): Likewise.
3221 * gcov-dump.c (print_version): Likewise.
3223 PR rtl-optimization/55838
3224 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
3225 iv0.step, iv1.step and step.
3227 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3228 Marc Glisse <marc.glisse@inria.fr>
3230 PR tree-optimization/55832
3231 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
3232 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
3233 integer_{one,zero}_node.
3235 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3238 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
3239 * var-tracking.c (reverse_op): Don't add reverse ops to
3240 VALUEs that have already
3241 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
3243 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
3245 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
3247 2013-01-02 Teresa Johnson <tejohnson@google.com>
3249 * dumpfile.c (dump_loc): Print filename with location.
3250 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
3251 new location_t parameter to emit complete unroll message with
3253 (canonicalize_loop_induction_variables): Compute loops location
3254 and pass to try_unroll_loop_completely.
3255 * loop-unroll.c (report_unroll_peel): New function.
3256 (peel_loops_completely): Use new dump format with location
3257 for main dumpfile message, and invoke report_unroll_peel on success.
3258 (decide_unrolling_and_peeling): Ditto.
3259 (decide_peel_once_rolling): Remove old dumpfile message subsumed
3260 by report_unroll_peel.
3261 (decide_peel_completely): Ditto.
3262 (decide_unroll_constant_iterations): Ditto.
3263 (decide_unroll_runtime_iterations): Ditto.
3264 (decide_peel_simple): Ditto.
3265 (decide_unroll_stupid): Ditto.
3266 * cfgloop.c (get_loop_location): New function.
3267 * cfgloop.h (get_loop_location): Declare.
3269 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3271 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
3274 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3277 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
3278 BLKmode objects when EXPAND_MEMORY is specified.
3280 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3282 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
3284 (fold_builtin_cpu): Do not share cpu model decls across statements.
3286 2013-01-02 Jason Merrill <jason@redhat.com>
3289 * tree.c (build_array_type_1): Revert earlier change.
3291 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
3293 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
3295 * config/aarch64/aarch64-tune.md: Re-generate.
3297 2013-01-02 Richard Biener <rguenther@suse.de>
3299 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3300 invariant load do not generate a vector load from the scalar location.
3302 2013-01-02 Richard Biener <rguenther@suse.de>
3305 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
3306 * configure: Regenerate.
3308 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3310 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
3311 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
3312 (expand_builtin_int_roundingfn_2): Keep the original target around
3313 for the fallback case.
3315 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3317 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
3318 to be clear for sign changes.
3320 2013-01-01 Jan Hubicka <jh@suse.cz>
3322 * ipa-inline-analysis.c: Fix formatting.
3324 2013-01-01 Jakub Jelinek <jakub@redhat.com>
3326 PR tree-optimization/55831
3327 * tree-vect-loop.c (get_initial_def_for_induction): Use
3328 gsi_after_labels instead of gsi_start_bb.
3330 Copyright (C) 2013 Free Software Foundation, Inc.
3332 Copying and distribution of this file, with or without modification,
3333 are permitted in any medium without royalty provided the copyright
3334 notice and this notice are preserved.