1 2013-03-25 Catherine Moore <clm@codesourcery.com>
3 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
4 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
5 * config/mip/predicates.md (lwsp_swsp_operand,
6 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
7 sb16_operand, db4_operand, db7_operand, ib3_operand,
8 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
9 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
10 andi16_operand): New predicates.
11 * config/mips/mips.md (compression): New attribute.
12 (enabled): New attribute.
13 (length): Consider compression in computing length.
14 (shift_compression): New code attribute.
15 (*add<mode>3): New operands. Record compression.
16 (sub<mode>3): Likewise.
17 (one_cmpl<mode>2): Likewise.
18 (*and<mode>3): Likewise.
19 (*ior<mode>3): Likewise.
20 (unnamed pattern for xor): Likewise.
21 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22 (*<optab><mode>3): Likewise.
23 (*mov<mode>_internal: Likewise.
24 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
25 (mips_unsigned_immediate_p): New.
26 (umips_lwsp_swsp_address_p): New.
27 (m16_based_address_p): New.
28 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
29 (mips_unsigned_immediate_p): New prototype.
30 (lwsp_swsp_address_p): New prototype.
31 (m16_based_address_p): New prototype.
32 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
33 (mips_signed_immediate_p): New function.
34 (m16_based_address_p): New function.
35 (lwsp_swsp_address_p): New function.
36 (mips_print_operand_punctuation): Recognize short delay slot insns
37 for microMIPS.add<mode>3"
39 2013-03-13 Catherine Moore <clm@codesourcery.com>
42 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
45 2013-03-05 Richard Sandiford <rdsandiford@googlemail.com>
47 * config/mips/constraints.md (T): Rename to...
51 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
52 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
54 2013-02-25 Catherine Moore <clm@codesourcery.com>
57 2013-02-24 Catherine Moore <clm@codesourcery.com>
58 Maciej W. Rozycki <macro@codesourcery.com>
59 Tom de Vries <tom@codesourcery.com>
60 Nathan Sidwell <nathan@codesourcery.com>
61 Iain Sandoe <iain@codesourcery.com>
62 Nathan Froyd <froydnj@codesourcery.com>
63 Chao-ying Fu <fu@mips.com>
64 Initial microMIPS support.
66 2013-02-25 Richard Biener <rguenther@suse.de>
68 PR tree-optimization/56175
69 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
71 (simplify_bitwise_binary): ... here. Also guard the conversion
72 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
76 2013-02-24 Catherine Moore <clm@codesourcery.com>
77 Maciej W. Rozycki <macro@codesourcery.com>
78 Tom de Vries <tom@codesourcery.com>
79 Nathan Sidwell <nathan@codesourcery.com>
80 Iain Sandoe <iain@codesourcery.com>
81 Nathan Froyd <froydnj@codesourcery.com>
82 Chao-ying Fu <fu@mips.com>
83 * doc/extend.texi: (micromips, nomicromips, nocompression):
84 Document new function attributes.
85 * doc/invoke.texi (minterlink-compressed, mmicromips,
86 m14k, m14ke, m14kec): Document new options.
87 (minterlink-mips16): Update documentation.
88 * doc/md.texi (ZC, ZD): Document new constraints.
89 * configure.ac (gcc_cv_as_micromips): Check if linker
90 supports the .set micromips directive.
91 * configure: Regenerate.
92 * config.in: Regenerate.
93 * config/mips/mips-tables.opt: Regenerate.
94 * config/mips/micromips.md: New file.
95 * constraints.md (ZC, AD): New constraints.
96 * config/mips/predicates.md (movep_src_register): New predicate.
97 (movep_src_operand): New predicate.
98 (non_volatile_mem_operand): New predicate.
99 * config/mips/mips.md (multimem): New type.
100 (length): Differentiate between 17-bit and 18-bit branch offsets.
101 (MOVEP1, MOVEP2): New mode iterator.
102 (mov_<load>l): Use ZC constraint.
103 (mov_<load>r): Likewise.
104 (mov_<store>l): Likewise.
105 (mov_<store>r): Likewise.
106 (*branch_equality<mode>_inverted): Add microMIPS support.
107 (*branch_equality<mode>): Likewise.
108 (*jump_absolute): Likewise.
109 (indirect_jump_<mode>): Likewise.
110 (tablejump_<mode>): Likewise.
111 (<optab>_internal): Likewise.
112 (sibcall_internal): Likewise.
113 (sibcall_value_internal): Likewise.
114 (prefetch): Use constraint ZD.
115 * config/mips/mips.opt (minterlink-compressed): New option.
116 (minterlink-mips16): Now an alias for minterlink-compressed.
117 (mmicromips): New option.
118 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
119 (compare_and_swap_12): Likewise.
120 (sync_add<mode>): Likewise.
121 (sync_<optab>_12): Likewise.
122 (sync_old_<optab>_12): Likewise.
123 (sync_new_<optab>_12): Likewise.
124 (sync_nand_12): Likewise.
125 (sync_old_nand_12): Likewise.
126 (sync_new_nand_12): Likewise.
127 (sync_sub<mode>): Likewise.
128 (sync_old_add<mode>): Likewise.
129 (sync_old_sub<mode>): Likewise.
130 (sync_new_add<mode>): Likewise.
131 (sync_new_sub<mode>): Likewise.
132 (sync_<optab><mode>): Likewise.
133 (sync_old_<optab><mode>): Likewise.
134 (sync_new_<optab><mode>): Likewise.
135 (sync_nand<mode>): Likewise.
136 (sync_old_nand<mode>): Likewise.
137 (sync_new_nand<mode>): Likewise.
138 (sync_lock_test_and_set<mode>): Likewise.
139 (test_and_set_12): Likewise.
140 (atomic_compare_and_swap<mode>): Likewise.
141 (atomic_exchange<mode>_llsc): Likewise.
142 (atomic_fetch_add<mode>_llsc): Likewise.
143 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
144 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
145 (umips_save_restore_pattern_p): Likewise.
146 (umips_load_store_pair_p): Likewise.
147 (umips_output_load_store_pair): Likewise.
148 (umips_movep_target_p): Likewise.
149 (umips_12bit_offset_address_p): Likewise.
150 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
151 (mips_base_mips16): Rename this...
152 (mips_base_compression_flags): ...to this. Update all uses.
153 (mips_attribute_table): Add micromips, nomicromips and nocompression.
154 (mips_mips16_decl_p): Delete.
155 (mips_nomips16_decl_p): Delete.
156 (mips_get_compress_on_flags): New function.
157 (mips_get_compress_off_flags): New function.
158 (mips_get_compress_mode): New function.
159 (mips_get_compress_on_name): New function.
160 (mips_get_compress_off_name): New function.
161 (mips_insert_attributes): Support multiple compression types.
162 (mips_merge_decl_attributes): Likewise.
163 (umips_12bit_offset_address_p): New function.
164 (mips_start_function_definition): Emit .set micromips directive.
165 (mips_call_may_need_jalx_p): New function.
166 (mips_function_ok_for_sibcall): Add microMIPS support.
167 (mips_print_operand_punctuation): Support short delay slots and
169 (umips_swm_mask, umips_swm_encoding): New.
170 (umips_build_save_restore): New function.
171 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
172 (was_mips16_p): Remove.
173 (old_compression_mode): New.
174 (mips_set_compression_mode): New function.
175 (mips_set_current_function): Add microMIPS support.
176 (mips_option_override): Likewise.
177 (umips_save_restore_pattern_p): New function.
178 (umips_output_save_restore): New function.
179 (umips_load_store_pair_p_1): New function.
180 (umips_load_store_pair_p): New function.
181 (umips_output_load_store_pair_1): New function.
182 (umips_output_load_store_pair): New function.
183 (umips_movep_target_p) New function.
184 (mips_prepare_pch_save): Add microMIPS support.
185 * config/mips/mips.h (TARGET_COMPRESSION): New.
186 (TARGET_CPU_CPP_BUILTINS): Update macro
187 to use new compression flags and to support microMIPS.
188 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
189 (MIPS_ARCH_FLOAT_SPEC): Likewise.
190 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
191 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
192 (ASM_SPEC): Support mmicromips and mno-micromips.
193 (M16STORE_REG_P): New macro.
194 (MIPS_CALL): Support TARGET_MICROMIPS.
195 (MICROMIPS_J): New macro.
196 (mips_base_mips16): Rename this...
197 (mips_base_compression_flags): ...to this.
198 (UMIPS_12BIT_OFFSET_P): New macro.
199 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
200 (MULTILIB_DIRNAMES): Likewise.
202 2013-02-25 Tom de Vries <tom@codesourcery.com>
204 PR rtl-optimization/56131
205 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
206 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
207 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
209 2013-02-25 Tobias Burnus <burnus@net-b.de>
211 * doc/invoke.texi (-fsanitize=): Move from optimization
212 to debugging options.
214 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
216 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
218 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
219 Alexander Monakov <amonakov@ispras.ru>
222 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
223 flush pending lists also on non-jumps. Adjust comment.
225 2013-02-24 Catherine Moore <clm@codesourcery.com>
226 Maciej W. Rozycki <macro@codesourcery.com>
227 Tom de Vries <tom@codesourcery.com>
228 Nathan Sidwell <nathan@codesourcery.com>
229 Iain Sandoe <iain@codesourcery.com>
230 Nathan Froyd <froydnj@codesourcery.com>
231 Chao-ying Fu <fu@mips.com>
233 * doc/extend.texi: (micromips, nomicromips, nocompression):
234 Document new function attributes.
235 * doc/invoke.texi (minterlink-compressed, mmicromips,
236 m14k, m14ke, m14kec): Document new options.
237 (minterlink-mips16): Update documentation.
238 * doc/md.texi (ZC, ZD): Document new constraints.
239 * configure.ac (gcc_cv_as_micromips): Check if linker
240 supports the .set micromips directive.
241 * configure: Regenerate.
242 * config.in: Regenerate.
243 * config/mips/mips-tables.opt: Regenerate.
244 * config/mips/micromips.md: New file.
245 * constraints.md (ZC, AD): New constraints.
246 * config/mips/predicates.md (movep_src_register): New predicate.
247 (movep_src_operand): New predicate.
248 (non_volatile_mem_operand): New predicate.
249 * config/mips/mips.md (multimem): New type.
250 (length): Differentiate between 17-bit and 18-bit branch offsets.
251 (MOVEP1, MOVEP2): New mode iterator.
252 (mov_<load>l): Use ZC constraint.
253 (mov_<load>r): Likewise.
254 (mov_<store>l): Likewise.
255 (mov_<store>r): Likewise.
256 (*branch_equality<mode>_inverted): Add microMIPS support.
257 (*branch_equality<mode>): Likewise.
258 (*jump_absolute): Likewise.
259 (indirect_jump_<mode>): Likewise.
260 (tablejump_<mode>): Likewise.
261 (<optab>_internal): Likewise.
262 (sibcall_internal): Likewise.
263 (sibcall_value_internal): Likewise.
264 (prefetch): Use constraint ZD.
265 * config/mips/mips.opt (minterlink-compressed): New option.
266 (minterlink-mips16): Now an alias for minterlink-compressed.
267 (mmicromips): New option.
268 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
269 (compare_and_swap_12): Likewise.
270 (sync_add<mode>): Likewise.
271 (sync_<optab>_12): Likewise.
272 (sync_old_<optab>_12): Likewise.
273 (sync_new_<optab>_12): Likewise.
274 (sync_nand_12): Likewise.
275 (sync_old_nand_12): Likewise.
276 (sync_new_nand_12): Likewise.
277 (sync_sub<mode>): Likewise.
278 (sync_old_add<mode>): Likewise.
279 (sync_old_sub<mode>): Likewise.
280 (sync_new_add<mode>): Likewise.
281 (sync_new_sub<mode>): Likewise.
282 (sync_<optab><mode>): Likewise.
283 (sync_old_<optab><mode>): Likewise.
284 (sync_new_<optab><mode>): Likewise.
285 (sync_nand<mode>): Likewise.
286 (sync_old_nand<mode>): Likewise.
287 (sync_new_nand<mode>): Likewise.
288 (sync_lock_test_and_set<mode>): Likewise.
289 (test_and_set_12): Likewise.
290 (atomic_compare_and_swap<mode>): Likewise.
291 (atomic_exchange<mode>_llsc): Likewise.
292 (atomic_fetch_add<mode>_llsc): Likewise.
293 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
294 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
295 (umips_save_restore_pattern_p): Likewise.
296 (umips_load_store_pair_p): Likewise.
297 (umips_output_load_store_pair): Likewise.
298 (umips_movep_target_p): Likewise.
299 (umips_12bit_offset_address_p): Likewise.
300 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
301 (mips_base_mips16): Rename this...
302 (mips_base_compression_flags): ...to this. Update all uses.
303 (mips_attribute_table): Add micromips, nomicromips and nocompression.
304 (mips_mips16_decl_p): Delete.
305 (mips_nomips16_decl_p): Delete.
306 (mips_get_compress_on_flags): New function.
307 (mips_get_compress_off_flags): New function.
308 (mips_get_compress_mode): New function.
309 (mips_get_compress_on_name): New function.
310 (mips_get_compress_off_name): New function.
311 (mips_insert_attributes): Support multiple compression types.
312 (mips_merge_decl_attributes): Likewise.
313 (umips_12bit_offset_address_p): New function.
314 (mips_start_function_definition): Emit .set micromips directive.
315 (mips_call_may_need_jalx_p): New function.
316 (mips_function_ok_for_sibcall): Add microMIPS support.
317 (mips_print_operand_punctuation): Support short delay slots and
319 (umips_swm_mask, umips_swm_encoding): New.
320 (umips_build_save_restore): New function.
321 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
322 (was_mips16_p): Remove.
323 (old_compression_mode): New.
324 (mips_set_compression_mode): New function.
325 (mips_set_current_function): Add microMIPS support.
326 (mips_option_override): Likewise.
327 (umips_save_restore_pattern_p): New function.
328 (umips_output_save_restore): New function.
329 (umips_load_store_pair_p_1): New function.
330 (umips_load_store_pair_p): New function.
331 (umips_output_load_store_pair_1): New function.
332 (umips_output_load_store_pair): New function.
333 (umips_movep_target_p) New function.
334 (mips_prepare_pch_save): Add microMIPS support.
335 * config/mips/mips.h (TARGET_COMPRESSION): New.
336 (TARGET_CPU_CPP_BUILTINS): Update macro
337 to use new compression flags and to support microMIPS.
338 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
339 (MIPS_ARCH_FLOAT_SPEC): Likewise.
340 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
341 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
342 (ASM_SPEC): Support mmicromips and mno-micromips.
343 (M16STORE_REG_P): New macro.
344 (MIPS_CALL): Support TARGET_MICROMIPS.
345 (MICROMIPS_J): New macro.
346 (mips_base_mips16): Rename this...
347 (mips_base_compression_flags): ...to this.
348 (UMIPS_12BIT_OFFSET_P): New macro.
349 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
350 (MULTILIB_DIRNAMES): Likewise.
352 2013-02-24 Jakub Jelinek <jakub@redhat.com>
355 * target-globals.c (save_target_globals): For init_reg_sets and
356 target_reinit remporarily set this_fn_optabs to this_target_optabs.
358 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
360 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
361 * config/aarch64/t-aarch64
362 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
364 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
367 * lra-constraints.c (process_alt_operands): Reload operand
368 conflicting with earlier clobber only if no more other conflicting
371 2013-02-22 Jakub Jelinek <jakub@redhat.com>
374 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
375 if not linking a shared library.
377 2013-02-22 Seth LaForge <sethml@google.com>
379 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
381 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
383 * config/arm/arm.md (split for extendsidi): Update condition.
384 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
385 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
386 (qhs_zextenddi_cstr): Likewise.
388 2013-02-21 Jakub Jelinek <jakub@redhat.com>
391 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
392 avoid signed wrapping.
393 (expand_mult): Handle properly multiplication by
394 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
395 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
396 in the compiler if coeff is HOST_WIDE_INT_MIN.
397 (expand_divmod): Don't make ext_op1 static, change it's type to
398 uhwi. Avoid undefined behavior in -INTVAL (op1).
400 PR rtl-optimization/50339
401 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
403 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
404 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
405 into splitting_ashiftrt field.
406 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
408 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
411 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
414 * trans-mem.c (execute_tm_mark): Do not expand transactions that
415 are sure to go irrevocable.
417 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
419 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
420 scalars are valid operands.
422 2013-02-21 Martin Jambor <mjambor@suse.cz>
424 PR tree-optimization/56310
425 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
426 only matching indices and non-negative final offsets.
427 (intersect_aggregates_with_edge): Pass src_idx to
428 agg_replacements_to_vector. Pass src_idx insstead of index to
429 intersect_with_agg_replacements.
431 2013-02-21 Martin Jambor <mjambor@suse.cz>
433 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
434 instead of hard-wired defaults.
436 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
438 * doc/invoke.texi (MIPS Options): Update documentation of the
439 floating-point multiply-accumulate instruction restrictions.
441 2013-02-21 Kostya Serebryany <kcc@google.com>
443 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
444 asan_shadow_offset on x86_64 linux.
446 2013-02-21 Richard Biener <rguenther@suse.de>
448 PR tree-optimization/56415
450 2013-02-11 Richard Biener <rguenther@suse.de>
452 PR tree-optimization/56273
453 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
456 2013-02-21 Jakub Jelinek <jakub@redhat.com>
459 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
463 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
464 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
466 2013-02-20 Jan Hubicka <jh@suse.cz>
468 PR tree-optimization/56265
469 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph when target is
470 referenced for firs ttime.
472 2013-02-20 Richard Biener <rguenther@suse.de>
474 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
475 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
476 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
477 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
479 (rename_ssa_copies): Do not remove unused locals.
480 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
481 * tree-ssanames.c (pass_release_ssa_names): Remove unused
483 * passes.c (execute_function_todo): Do not schedule unused locals
484 removal if cleanup_tree_cfg did something.
485 * tree-ssa-live.c (remove_unused_locals): Dump statistics
486 about the number of removed locals.
488 2013-02-20 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/56398
491 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip
494 2013-02-20 Martin Jambor <mjambor@suse.cz>
496 PR tree-optimization/55334
497 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
498 restricted pointers to arrays.
500 2013-02-20 Richard Biener <rguenther@suse.de>
501 Jakub Jelinek <jakub@redhat.com>
503 PR tree-optimization/56396
504 * tree-ssa-ccp.c (n_const_val): New static variable.
505 (get_value): Return NULL for SSA names we don't have a lattice
507 (ccp_initialize): Initialize n_const_val.
508 * tree-ssa-copy.c (n_copy_of): New static variable.
509 (init_copy_prop): Initialize n_copy_of.
510 (get_value): Return NULL_TREE for SSA names we don't have a
513 2013-02-20 Martin Jambor <mjambor@suse.cz>
515 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
517 2013-02-20 Richard Biener <rguenther@suse.de>
519 * genpreds.c (write_lookup_constraint): Do not compare first
520 letter of the constraint again.
522 2013-02-20 Richard Biener <rguenther@suse.de>
524 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
526 (get_use_iv_cost): Terminate hashtable walk when coming across
529 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
531 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
532 reassociation for avx2 targets.
534 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
536 * config/microblaze/microblaze.c: microblaze_has_clz = 0
537 Add version check for v8.10.a to enable microblaze_has_clz
538 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
539 version and TARGET_PATTERN_COMPARE check
540 * config/microblaze/microblaze.md: New clzsi2 instruction
542 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
544 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
545 function before branching.
547 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
549 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
551 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
553 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
557 * sel-sched.c: Include ira.h.
558 (implicit_clobber_conflict_p): New function.
559 (moveup_expr): Use it.
560 * Makefile.in (sel-sched.o): Depend on ira.h.
562 2013-02-19 Richard Biener <rguenther@suse.de>
564 PR tree-optimization/56384
565 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
566 (vn_hash_type): Split out from ...
567 (vn_hash_constant_with_type): ... here.
568 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
569 (vn_phi_eq): Compare types from vn_phi_s structure.
570 (vn_phi_lookup): Populate vn_phi_s type.
571 (vn_phi_insert): Likewise.
573 2013-02-19 Jakub Jelinek <jakub@redhat.com>
575 PR tree-optimization/56350
576 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
577 if haven't found reduction or nested cycle operand, rather than
578 asserting we must find it.
580 PR tree-optimization/56381
581 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
584 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
585 Jakub Jelinek <jakub@redhat.com>
588 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
589 (swap_optab_enable): Same.
590 (init_all_optabs): Use argument instead of global.
591 * tree.h (struct tree_optimization_option): New field
593 * expr.h (init_all_optabs): Add argument to prototype.
594 (TREE_OPTIMIZATION_OPTABS): New.
595 (save_optabs_if_changed): Protoize.
596 * optabs.h: Declare this_fn_optabs.
597 * optabs.c (save_optabs_if_changed): New.
598 Declare this_fn_optabs.
599 (init_optabs): Add argument to init_all_optabs() call.
600 * function.c (invoke_set_current_function_hook): Handle per
602 * function.h (struct function): New field optabs.
603 * config/mips/mips.c (mips_set_mips16_mode): Handle when
604 optimization_current_node has changed.
605 * target-globals.h (save_target_globals_default_opts): Protoize.
606 * target-globals.c (save_target_globals_default_opts): New.
608 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
611 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
612 registers %fr12 and %fr12R as call used.
615 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
616 and HImode, require all displacements to be an integer multiple of their
618 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
619 only allow QImode and HImode when reload is in progress and strict is
620 true. Likewise for symbolic addresses. Use base14_operand to check
621 displacements in REG+BASE addresses.
623 2013-02-18 Richard Biener <rguenther@suse.de>
625 PR tree-optimization/56366
626 * tree-vect-loop.c (get_initial_def_for_induction): Properly
627 handle sign-conversion of outer-loop initial induction value.
629 2013-02-18 Richard Biener <rguenther@suse.de>
632 * cfghooks.c (merge_blocks): If we merge a latch into another
633 block adjust references to it.
634 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
635 (verify_loop_structure): Verify that a recorded latch is in fact
638 2013-02-18 Richard Biener <rguenther@suse.de>
640 PR tree-optimization/56321
641 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
642 order SSA name release and virtual operand unlinking.
644 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
646 * config/microblaze/microblaze.md (save_stack_block): Define.
647 (restore_stack_block): Likewise.
649 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
651 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
652 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
653 * config/microblaze/microblaze.c (microblaze_option_override):
654 Bail out early for PIC modes when target does not support PIC.
656 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
658 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
659 Replace with a microblaze version.
660 (microblaze_trampoline_init): Adapt for microblaze.
661 * gcc/config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
664 2013-02-16 Jakub Jelinek <jakub@redhat.com>
665 Dodji Seketeli <dodji@redhat.com>
668 * asan.c (get_mem_refs_of_builtin_call): White space and style
670 (instrument_mem_region_access): Do not forget to always put
671 instrumentation of the of 'base' and 'base + len' in a "if (len !=
672 0) statement, even for cases where either 'base' or 'base + len'
673 are not instrumented -- because they have been previously
674 instrumented. Simplify the logic by putting all the statements
675 instrument 'base + len' inside a sequence, and then insert that
676 sequence right before the current insertion point. Then, to
677 instrument 'base + len', just get an iterator on that statement.
678 And do not forget to update the pointer to iterator the function
679 received as argument.
681 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
683 PR rtl-optimization/56348
684 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
686 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
688 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
689 (clean_graph_dump_file): Pass base to start_graph_dump.
691 2013-02-14 Richard Henderson <rth@redhat.com>
694 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
696 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
698 * collect2-aix.h: Define F_LOADONLY.
700 2013-02-14 Richard Biener <rguenther@suse.de>
703 * varasm.c (output_constant_def_1): Get the decl representing
704 the constant as argument.
705 (output_constant_def): Wrap output_constant_def_1.
706 (make_decl_rtl): Use output_constant_def_1 with the decl
707 representing the constant.
708 (build_constant_desc): Optionally re-use a decl already
709 representing the constant.
710 (tree_output_constant_def): Adjust.
712 2013-02-14 Dodji Seketeli <dodji@redhat.com>
715 * asan.c (instrument_builtin_call): Really put the length of the
716 second source argument into src1_len.
718 2013-02-13 Jakub Jelinek <jakub@redhat.com>
720 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
721 argument. If it is false, don't create edge from then_bb to
723 (insert_if_then_before_iter): Pass true to it.
724 (build_check_stmt): Pass false to it.
725 (transform_statements): Flush hash table only on extended basic
726 block boundaries, rather than at the beginning of every bb.
727 Don't flush hash table on nonfreeing_call_p calls.
728 * tree-flow.h (nonfreeing_call_p): New prototype.
729 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
731 2013-02-13 David S. Miller <davem@davemloft.net>
733 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
735 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
738 * ira.c (max_regno_before_ira): Move from ...
740 (fix_reg_equiv_init): Use max_regno_before_ira instead of
743 2013-02-13 Jakub Jelinek <jakub@redhat.com>
745 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
747 2013-02-13 Richard Biener <rguenther@suse.de>
750 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
753 2013-02-13 Richard Biener <rguenther@suse.de>
755 * loop-init.c (loop_optimizer_init): Clear loop state when
756 re-initializing preserved loops.
757 * loop-unswitch.c (unswitch_single_loop): Return whether
758 we unswitched the loop. Do not verify loop state here.
759 (unswitch_loops): When we unswitched a loop discover new
762 2013-02-13 Kostya Serebryany <kcc@google.com>
764 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset on x86_64
766 * sanitizer.def: Rename __asan_init to __asan_init_v1.
768 2013-02-12 Dodji Seketeli <dodji@redhat.com>
770 Avoid instrumenting duplicated memory access in the same basic block
771 * Makefile.in (asan.o): Add new dependency on hash-table.h
772 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
773 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
774 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
775 (free_mem_ref_resources, has_mem_ref_been_instrumented)
776 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
777 (get_mem_ref_of_assignment): New functions.
778 (get_mem_refs_of_builtin_call): Extract from
779 instrument_builtin_call and tweak a little bit to make it fit with
781 (instrument_builtin_call): Use the new
782 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
783 of is_gimple_builtin_call.
784 (instrument_derefs, instrument_mem_region_access): Insert the
785 instrumented memory reference into the hash table.
786 (maybe_instrument_assignment): Renamed instrument_assignment into
787 this, and change it to advance the iterator when instrumentation
788 actually happened and return true in that case. This makes it
789 homogeneous with maybe_instrument_assignment, and thus give a
790 chance to callers to be more 'regular'.
791 (transform_statements): Clear the memory reference hash table
792 whenever we enter a new BB, when we cross a function call, or when
793 we are done transforming statements. Use
794 maybe_instrument_assignment instead of instrumentation. No more
795 need to special case maybe_instrument_assignment and advance the
796 iterator after calling it; it's now handled just like
797 maybe_instrument_call. Update comment.
799 2013-02-13 Richard Biener <rguenther@suse.de>
801 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
802 Fix loop discovery code.
804 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
807 * lra-constraints.c (process_alt_operands): Match early clobber
808 operand with itself. Check conflicts with earlyclobber only if
809 the operand is not reloaded. Prefer to reload conflicting operand
810 if earlyclobber and matching operands are the same.
812 2013-02-12 Richard Biener <rguenther@suse.de>
815 * lto-streamer-out.c (write_symbol): Do not output symbols
816 for hard register variables.
818 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
821 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
822 (umulsidi3_insn, mulsidi3_insn): New insns.
824 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
826 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
827 (struct tune_params): Add vec_costs field.
828 * config/arm/arm.c (arm_builtin_vectorization_cost)
829 (arm_add_stmt_cost): New functions.
830 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
831 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
832 (arm_default_vec_cost): New struct of type cpu_vec_costs.
833 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
834 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
835 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
836 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
838 2013-02-12 Richard Biener <rguenther@suse.de>
841 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
842 decls again if possible.
844 2013-02-12 Richard Biener <rguenther@suse.de>
847 * tree-ssa.c (verify_ssa_name): Fix check, move
848 SSA_NAME_IN_FREE_LIST check up.
850 2013-02-12 Jakub Jelinek <jakub@redhat.com>
851 Steven Bosscher <steven@gcc.gnu.org>
853 PR rtl-optimization/56151
854 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
855 equal to op0 or op1, and last_insn pattern is CODE operation
856 with MEM dest and one of the operands matches that MEM.
858 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
860 * doc/extend.texi: Document Function Multiversioning and "default"
861 parameter string to target attribute.
862 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
863 target attribute parameter is "default".
864 (ix86_compare_version_priority): Remove checks for target attribute.
865 (ix86_mangle_function_version_assembler_name): Change error to sorry.
866 Remove check for target attribute equal to NULL. Add assert.
867 (ix86_generate_version_dispatcher_body): Change error to sorry.
869 2013-02-11 Iain Sandoe <iain@codesourcery.com>
870 Jack Howarth <howarth@bromo.med.uc.edu>
871 Patrick Marlier <patrick.marlier@gmail.com>
874 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
875 define ENDFILE_SPEC as TM_DESTRUCTOR.
876 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
878 2013-02-11 Alexander Potapenko <glider@google.com>
879 Jack Howarth <howarth@bromo.med.uc.edu>
880 Jakub Jelinek <jakub@redhat.com>
883 * config/darwin.c (cdtor_record): Rename ctor_record.
884 (sort_cdtor_records): Rename sort_ctor_records.
885 (finalize_dtors): New routine to sort destructors by
886 priority before use in assemble_integer.
887 (machopic_asm_out_destructor): Use finalize_dtors if needed.
889 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
891 PR rtl-optimization/56275
892 * simplify-rtx.c (avoid_constant_pool_reference): Check that
893 offset is non-negative and less than cmode size before
894 calling simplify_subreg.
896 2013-02-11 Richard Biener <rguenther@suse.de>
898 PR tree-optimization/56264
899 * cfgloop.h (fix_loop_structure): Adjust prototype.
900 * loop-init.c (fix_loop_structure): Return the number of
901 newly discovered loops.
902 * tree-cfgcleanup.c (repair_loop_structures): When new loops
903 are discovered, do a full loop-closed SSA rewrite.
905 2013-02-11 Richard Biener <rguenther@suse.de>
907 PR tree-optimization/56273
908 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
910 (check_array_ref): Fix missing newline in dumps.
911 (search_for_addr_array): Likewise.
913 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
915 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
917 2013-02-09 Jakub Jelinek <jakub@redhat.com>
920 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
922 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
924 PR rtl-optimization/56246
925 * lra-constraints.c (simplify_operand_subreg): Try to reuse
927 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
928 constraints are satisfied.
930 2013-02-08 Jeff Law <law@redhat.com>
933 * emit-rtl.c (reg_is_parm_p): New function.
934 * regs.h (reg_is_parm_p): New prototype.
935 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
936 callee-clobbered registers.
938 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
941 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
942 If there is no implicit builtin declaration, just return NULL.
944 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
946 * config/i386/sse.md (FMAMODEM): New mode iterator.
947 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
948 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
950 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
952 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
953 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
954 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
956 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
958 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
959 (microblaze*-*-elf): Likewise.
960 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
962 * config/microblaze/microblaze-c.c: Add builtin defines for
963 _LITTLE_ENDIAN and _BIG_ENDIAN.
964 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
965 add to TARGET_DEFAULT flags.
966 Expand ASM_SPEC and LINK_SPEC.
967 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
968 * config/microblaze/microblaze.md: Update extendsidi2 and
969 movdi_internal instructions to use low-order / high-order reg
971 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
972 options and inversemask / mask of LITTLE_ENDIAN.
973 * config/microblaze/t-microblaze: Expand multilib options to
974 include mlittle-endian (le) and update exceptions patterns.
976 2013-02-08 Jakub Jelinek <jakub@redhat.com>
978 PR rtl-optimization/56195
979 * lra-constraints.c (get_reload_reg): Don't reuse regs
980 if they have smaller mode than requested, if they have
981 wider mode than requested, try to return a SUBREG.
983 PR tree-optimization/56250
984 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
985 if type is unsigned and code isn't MULT_EXPR.
987 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
989 PR tree-optimization/56064
990 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
991 bits according to mode.
992 * fixed-value.h (fixed_from_double_int)
993 (const_fixed_from_double_int): Adjust comments.
995 2013-02-08 Richard Biener <rguenther@suse.de>
998 * lto-streamer.h (struct data_in): Remove current_file, current_line
999 and current_col members.
1000 * lto-streamer-out.c (lto_output_location): Stream changed bits
1001 en-block for efficiency.
1002 * lto-streamer-in.c (clear_line_info): Remove.
1003 (lto_input_location): Cache current file, line and column
1004 globally via local statics. Read changed bits en-block.
1005 (input_function): Do not call clear_line_info.
1006 (lto_read_body): Likewise.
1007 (lto_input_toplevel_asms): Likewise.
1009 2013-02-08 Michael Matz <matz@suse.de>
1011 PR tree-optimization/52448
1012 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
1013 (nt_call_phase): New static.
1014 (add_or_mark_expr): Only mark accesses with newer phase than any
1016 (nonfreeing_call_p): New.
1017 (nt_init_block): Update nt_call_phase, mark blocks as visited.
1018 (nt_fini_block): Keep blocks marked as visited.
1019 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
1021 2013-02-08 Richard Biener <rguenther@suse.de>
1023 * ira.c (ira): Free broken dominator information.
1025 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
1027 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
1029 2013-02-08 Marek Polacek <polacek@redhat.com>
1031 * cfgloop.c (verify_loop_structure): Add more checking of headers.
1033 2013-02-08 Richard Biener <rguenther@suse.de>
1036 * cfgloop.h (flow_loops_find): Adjust.
1037 (bb_loop_header_p): Declare.
1038 * cfgloop.c (bb_loop_header_p): New function split out from ...
1039 (flow_loops_find): ... here. Adjust function signature,
1040 support incremental loop structure update.
1041 (verify_loop_structure): Cleanup. Verify a loop is a loop.
1042 * cfgloopmanip.c (fix_loop_structure): Move ...
1043 * loop-init.c (fix_loop_structure): ... here.
1044 (apply_loop_flags): Split out from ...
1045 (loop_optimizer_init): ... here.
1046 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
1047 in incremental mode, only remove dead loops here.
1049 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
1052 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
1053 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
1054 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
1055 (*round<mode>3.libgcc): New insns for fixed-modes.
1056 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
1057 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
1058 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
1059 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
1060 implementations. Define to __builtin_avr_absFX,
1061 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
1062 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
1063 __builtin_avr_countlsFX, respectively.
1064 * config/avr/avr-c.c (target.h): Include it.
1065 (enum avr_builtin_id): New enum.
1066 (avr_resolve_overloaded_builtin): New static function.
1067 (avr_register_target_pragmas): Use it to set
1068 targetm.resolve_overloaded_builtin.
1069 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
1070 tree nodes used by DEF_BUILTIN.
1071 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
1072 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
1073 <AVR_BUILTIN_xxBITS>: Same.
1075 2013-02-08 Richard Biener <rguenther@suse.de>
1077 * cfgloop.c (verify_loop_structure): Properly handle
1078 a loop exiting to another loop header.
1079 * ira-int.h (ira_loops): Remove.
1080 * ira.c (ira_loops): Remove.
1081 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
1082 (do_reload): Use loop_optimizer_finalize.
1083 * ira-build.c (create_loop_tree_nodes): Use get_loops and
1084 number_of_loops to access the loop tree.
1085 (more_one_region_p): Likewise.
1086 (finish_loop_tree_nodes): Likewise.
1087 (rebuild_regno_allocno_maps): Likewise.
1088 (mark_loops_for_removal): Likewise.
1089 (mark_all_loops_for_removal): Likewise.
1090 (remove_unnecessary_regions): Likewise.
1091 (ira_build): Likewise.
1092 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
1094 2013-02-08 Richard Biener <rguenther@suse.de>
1096 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
1097 * ipa-pure-const.c (analyze_function): Avoid calling
1098 mark_irreducible_loops twice.
1099 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
1101 2013-02-07 David S. Miller <davem@davemloft.net>
1103 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
1105 * var-tracking.c (vt_add_function_parameter): Test the presence of
1106 HAVE_window_save properly and do not remap argument registers when
1107 we have a leaf function.
1109 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
1112 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
1114 * config/i386/i386.c (ix86_print_operand): Ditto.
1116 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1118 * lra-constraints.c (process_alt_operands): Fix recently added comment.
1120 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
1122 PR rtl-optimization/56225
1123 * lra-constraints.c (process_alt_operands): Check that reload hard
1124 reg can hold value for strict_low_part.
1126 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1129 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
1130 dwarf2out_end_function.
1131 (in_first_function_p, maybe_at_text_label_p,
1132 first_loclabel_num_not_at_text_label): New variables.
1133 (dwarf2out_var_location): In the first function find out
1134 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
1135 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
1138 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
1140 PR rtl-optimization/56178
1141 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
1142 SUBREG of a register. Tidy up related block of code.
1143 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
1144 note if the source is a register or a SUBREG of a register.
1146 2013-02-07 Jakub Jelinek <jakub@redhat.com>
1149 * config/rs6000/rs6000.md (ptrm): New mode attr.
1150 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
1151 call_value_indirect_aix<pttrsize>,
1152 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
1155 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1157 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
1158 if -bnortl. Convert to strcmp and strncmp.
1160 2013-02-07 Alan Modra <amodra@gmail.com>
1163 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
1164 addresses won't wrap when offsetting.
1165 (rs6000_secondary_reload): Provide secondary reloads needed for
1166 wrapping LO_SUM addresses.
1168 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
1170 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
1171 MACH, just __MACH__.
1173 2013-02-06 Richard Biener <rguenther@suse.de>
1175 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
1176 instead of calling fix_loop_structure.
1178 2013-02-06 Jakub Jelinek <jakub@redhat.com>
1181 * omp-low.c (use_pointer_for_field): Return false if
1182 lower_send_shared_vars doesn't generate any copy-out code.
1184 2013-02-06 Tom de Vries <tom@codesourcery.com>
1186 PR rtl-optimization/56131
1187 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
1188 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
1189 of the label is NULL. Add comment.
1191 2013-02-05 Jakub Jelinek <jakub@redhat.com>
1193 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
1196 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
1197 (STATIC_LIBTSAN_LIBS): Likewise.
1198 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
1199 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
1200 is defined, don't add anything else beyond that.
1201 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
1202 (LINK_COMMAND_SPEC): Use them.
1204 PR tree-optimization/56205
1205 * tree-stdarg.c (check_all_va_list_escapes): Return true if
1206 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
1207 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
1209 2013-02-05 Richard Biener <rguenther@suse.de>
1211 PR tree-optimization/53342
1212 PR tree-optimization/53185
1213 * tree-vectorizer.h (vect_check_strided_load): Remove.
1214 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
1215 not disallow peeling for vectorized strided loads.
1216 (vect_check_strided_load): Make static and simplify.
1217 (vect_analyze_data_refs): Adjust.
1218 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
1219 correctly when vectorizing strided loads.
1221 2013-02-05 Richard Biener <rguenther@suse.de>
1223 * doc/install.texi: Refer to ISL, not PPL.
1225 2013-02-05 Jan Hubicka <jh@suse.cz>
1227 PR tree-optimization/55789
1228 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
1230 2013-02-05 Jan Hubicka <jh@suse.cz>
1232 PR tree-optimization/55789
1233 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
1234 the dead call anyway.
1236 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
1239 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
1241 2013-02-04 Alexander Potapenko <glider@google.com>
1242 Jack Howarth <howarth@bromo.med.uc.edu>
1243 Jakub Jelinek <jakub@redhat.com>
1246 * config/darwin.c (sort_ctor_records): Stabilized qsort
1247 on constructor priority by using original position.
1248 (finalize_ctors): New routine to sort constructors by
1249 priority before use in assemble_integer.
1250 (machopic_asm_out_constructor): Use finalize_ctors if needed.
1252 2013-02-04 Jakub Jelinek <jakub@redhat.com>
1255 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
1256 about visibility on artificial decls.
1257 * config/sol2.c (solaris_assemble_visibility): Likewise.
1259 2013-02-04 Kai Tietz <ktietz@redhat.com>
1262 * config/i386/i386.c (function_value_ms_64): Add additional valtype
1263 argument and improve checking of return-argument types for 16-byte
1265 (ix86_function_value_1): Add additional valtype argument on call
1266 of function_value_64.
1267 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
1268 handling infunction_value_64 function.
1270 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1272 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
1274 2013-02-04 Richard Biener <rguenther@suse.de>
1276 PR tree-optimization/56188
1277 * tree-ssa-structalias.c (label_visit): Consider case with
1278 initially non-empty points-to set.
1279 (perform_var_substitution): Dump node mapping and clean up.
1281 2013-02-04 Richard Guenther <rguenther@suse.de>
1284 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
1285 node prevail as last resort.
1286 (lto_symtab_merge_decls): Remove guard on LTRANS here.
1287 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
1289 2013-02-04 Richard Biener <rguenther@suse.de>
1291 PR tree-optimization/56113
1292 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
1294 (equiv_class_lookup_or_add): ... this.
1295 (label_visit): Adjust and fix error in previous patch.
1296 (perform_var_substitution): Adjust.
1298 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
1300 * config/sh/divtab.c: Fix formatting and comments throughout the file.
1301 * config/sh/sh4-300.md: Likewise.
1302 * config/sh/sh4a.md: Likewise.
1303 * config/sh/constraints.md: Likewise.
1304 * config/sh/sh.md: Likewise.
1305 * config/sh/netbsd-elf.h: Likewise.
1306 * config/sh/predicates.md: Likewise.
1307 * config/sh/sh-protos.h: Likewise.
1308 * config/sh/ushmedia.h: Likewise.
1309 * config/sh/linux.h: Likewise.
1310 * config/sh/sh.c: Likewise.
1311 * config/sh/superh.h: Likewise.
1312 * config/sh/elf.h: Likewise.
1313 * config/sh/sh4.md: Likewise.
1314 * config/sh/sh.h: Likewise.
1316 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1318 * config/pa/constraints.md: Adjust unused letters. Change "T"
1319 constraint to match_test floating_point_store_memory_operand().
1320 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
1321 (base14_operand): New.
1322 (floating_point_store_memory_operand): New.
1323 (integer_store_memory_operand): Revise to use base14_operand and
1324 reg_plus_base_memory_operand.
1325 (move_dest_operand): Allow symbolic_memory_operands.
1326 (symbolic_memory_operand): Check for LO_SOM.
1327 (symbolic_operand): Change default case to break.
1328 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
1329 CONST_DOUBLE values to be reloaded by putting them into memory when
1330 the destination is a floating point register.
1331 (movdf): Remove code to handle CONST_DOUBLE.
1333 (reload_indf_r1): New.
1334 (reload_insf_r1): New.
1335 Consistently use "Q" and "T" constraints with integer and floating
1336 point move instructions, respectively.
1337 (movdi): Remove FAIL.
1338 Change predicate for source operand unamed DImode move from
1339 general_operand to move_src_operand.
1340 (umulsidi3): Change predicate for destination operand to
1342 Likewise for similar unamed patterns.
1343 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
1344 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
1345 (hppa_legitimize_address): Simplify mask calculation.
1346 (pa_emit_move_sequence): Revised handling of secondary reloads from
1347 REG+D addresses for floating point loads and stores. Directly handle
1348 loading CONST0_RTX (mode) to a floating point register.
1349 (pa_secondary_reload): Handle reloading DF and SFmode constant values
1350 to floating point registers. Don't restrict secondary reloads to
1351 floating point registers to integer modes. Revise some comments and
1353 (TARGET_LEGITIMATE_ADDRESS_P): Define.
1354 (pa_legitimate_address_p): New.
1355 (pa_legitimize_reload_address): New.
1356 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
1357 (STRICT_REG_OK_FOR_BASE_P): New.
1358 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
1359 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
1361 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
1362 Andrew Dixie <andrewd@gentrack.com>
1364 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
1367 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
1369 * expmed.c (extract_bit_field_1): Pass the full width of the
1370 structure to get_best_reg_extraction_insn.
1372 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1375 * configure.ac (use_cxa_atexit): Add AIX.
1376 * configure: Regenerate.
1378 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
1380 2013-02-01 Jakub Jelinek <jakub@redhat.com>
1383 * final.c (need_profile_function): New variable.
1384 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
1385 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
1386 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
1387 notes, targetm.asm_out.function_prologue doesn't emit anything,
1388 HAVE_prologue and profiler should be emitted before prologue,
1389 set need_profile_function instead of emitting it.
1390 (final_scan_insn): If need_profile_function, emit
1391 profile_function on the first NOTE_INSN_BASIC_BLOCK or
1392 NOTE_INSN_FUNCTION_BEG note.
1394 2013-02-01 Richard Henderson <rth@redhat.com>
1396 * config/rs6000/rs6000.md (smulditi3): New.
1399 * config/alpha/alpha.md (umulditi3): New.
1401 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
1403 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
1404 (ASM_OUTPUT_ALIGNED_LOCAL): New.
1406 2013-02-01 Richard Biener <rguenther@suse.de>
1408 PR tree-optimization/56113
1409 * tree-ssa-structalias.c (label_visit): Reduce work for
1410 single-predecessor nodes.
1412 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
1414 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
1415 range isn't testing for zero.
1417 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
1420 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
1422 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
1423 Nick Clifton <nickc@redhat.com>
1425 * config/v850/constraints.md (Q): Define as a memory constraint.
1426 * config/v850/predicates.md (label_ref_operand): New predicate.
1427 (e3v5_shift_operand): New predicate.
1428 (ior_operator): New predicate.
1429 * config/v850/t-v850: Add e3v5 multilib.
1430 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
1431 (v850_gen_movdi): Prototype.
1432 * config/v850/v850.c: Add support for e3v5 architecture.
1433 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
1435 (construct_save_jarl): Add e3v5 long JARL support.
1436 (v850_adjust_insn_length): New function. Adjust length of call
1437 insns when using e3v5 instructions.
1438 (v850_gen_movdi): New function: Generate instructions to move a
1440 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
1441 (CPP_SPEC): Define __v850e3v5__ as appropriate.
1442 (TARGET_USE_FPU): Enable for e3v5.
1443 (CONST_OK_FOR_W): New macro.
1444 (ADJUST_INSN_LENGTH): Define.
1445 * config/v850/v850.md (UNSPEC_LOOP): Define.
1446 (attr cpu): Add v850e3v5.
1447 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
1448 (movdi): New pattern.
1449 (movdi_internal): New pattern.
1450 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
1451 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
1452 (cstoresf4): Likewise.
1453 (cstoredf4): Likewise.
1454 (insv): New pattern.
1455 (rotlso3_a): New pattern.
1456 (rotlsi3_b): New pattern
1457 (rotlsi3_v850e3v5): New pattern.
1458 (doloop_begin): New pattern.
1459 (fix_loop_counter): New pattern.
1460 (doloop_end): New pattern.
1461 (branch_normal): Add e3v5 long branch support.
1462 (branch_invert): Likewise.
1463 (branch_z_normal): Likewise.
1464 (branch_z_invert): Likewise.
1465 (branch_nz_normal): Likewise.
1466 (branch_nz_invert): Likewise.
1467 (call_internal_short): Add e3v5 register-indirect JARL support.
1468 (call_internal_long): Likewise.
1469 (call_value_internal_short): Likewise.
1470 (call_value_internal_long): Likewise.
1471 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
1472 (mloop): New option.
1473 * config.gcc: Add support for configuring v840e3v5 target.
1474 * doc/invoke.texi: Document new v850 specific command line options.
1476 2013-01-31 Paul Koning <ni1d@arrl.net>
1480 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
1481 children if parent is a class.
1482 (prune_unused_types_prune): Don't add DW_AT_declaration.
1484 2013-01-31 Richard Biener <rguenther@suse.de>
1486 PR tree-optimization/56157
1487 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
1488 match up operand with SLP child.
1490 2013-01-31 Jason Merrill <jason@redhat.com>
1493 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
1494 parameters the first time.
1495 (gen_scheduled_generic_parms_dies): Check completeness here.
1497 2013-01-31 Richard Biener <rguenther@suse.de>
1500 * common.opt (faggressive-loop-optimizations): New flag,
1502 * doc/invoke.texi (faggressive-loop-optimizations): Document.
1503 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
1504 infer_loop_bounds_from_undefined by it.
1506 2013-01-31 Richard Biener <rguenther@suse.de>
1508 PR tree-optimization/56150
1509 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
1510 visit virtual operands.
1511 (find_uses_to_rename_bb): Likewise.
1513 2013-01-31 Richard Biener <rguenther@suse.de>
1515 PR tree-optimization/56150
1516 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
1517 mixed store non-store stmts.
1519 2013-01-30 Jakub Jelinek <jakub@redhat.com>
1522 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
1523 LIBASAN_EARLY_SPEC is defined.
1524 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
1525 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
1527 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
1530 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
1531 invalid args instead of ICEing on it.
1532 (ix86_valid_target_attribute_tree): Return error_mark_node if
1533 ix86_valid_target_attribute_inner_p failed.
1534 (ix86_valid_target_attribute_p): Return false only if
1535 ix86_valid_target_attribute_tree returned error_mark_node. Allow
1536 target("default") attribute.
1537 (sorted_attr_string): Change argument from const char * to tree,
1538 merge in all target attribute arguments rather than just one.
1539 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
1540 instead of free. Avoid using strcat.
1541 (ix86_mangle_function_version_assembler_name): Mangle
1542 target("default") as if no target attribute is present. Adjust
1543 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
1544 instead of xmalloc and XDELETEVEC instead of free.
1545 (ix86_function_versions): Don't return true if one of the decls
1546 doesn't have target attribute. If they don't and one of the decls
1547 is DECL_FUNCTION_VERSIONED, report an error. Adjust
1548 sorted_attr_string caller. Use XDELETEVEC instead of free.
1549 (ix86_supports_function_versions): Remove.
1550 (make_name): Fix up formatting.
1551 (make_dispatcher_decl): Remove resolver_name and its initialization.
1552 Avoid leaking memory.
1553 (is_function_default_version): Return true if there is
1554 target("default") attribute rather than no target attribute at all.
1555 (make_resolver_func): Avoid leaking memory.
1556 (ix86_generate_version_dispatcher_body): Likewise.
1557 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
1558 * target.def (supports_function_versions): Remove.
1559 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
1560 * doc/tm.texi: Regenerated.
1562 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
1564 PR rtl-optimization/56144
1565 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
1566 for values with side effects.
1568 2013-01-30 Richard Biener <rguenther@suse.de>
1570 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
1571 (sparseset_pop): Likewise.
1572 * cfganal.c (compute_idf): Likewise. Increase work-stack size
1573 to be able to use quick_push in the worker loop.
1575 2013-01-30 Marek Polacek <polacek@redhat.com>
1577 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
1579 2013-01-30 Richard Biener <rguenther@suse.de>
1582 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
1584 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
1586 PR tree-optimization/56064
1587 * fixed-value.c (fixed_from_double_int): New function.
1588 * fixed-value.h (fixed_from_double_int): New prototype.
1589 (const_fixed_from_double_int): New static inline function.
1590 * fold-const.c (native_interpret_fixed): New static function.
1591 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
1592 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
1593 (native_encode_fixed): New static function.
1594 (native_encode_expr) <FIXED_CST>: Use it.
1595 (native_interpret_int): Move double_int worker code to...
1596 * double-int.c (double_int::from_buffer): ...this new static method.
1597 * double-int.h (double_int::from_buffer): Prototype it.
1599 2013-01-30 Richard Biener <rguenther@suse.de>
1601 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
1602 New pointer-map and obstack.
1603 (init_alias_vars): Allocate pointer-map and obstack.
1604 (delete_points_to_sets): Free them.
1605 (find_what_var_points_to): Cache result.
1606 (find_what_p_points_to): Adjust for changed interface of
1607 find_what_var_points_to.
1608 (compute_points_to_sets): Likewise.
1609 (ipa_pta_execute): Likewise.
1611 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1613 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
1614 * configure: Regenerate.
1615 * config.in: Regenerate.
1616 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
1617 #nobits/#progbits if supported.
1619 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
1622 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
1623 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
1624 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
1626 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1628 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
1629 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
1631 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1633 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
1634 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
1636 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1638 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
1640 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
1641 * config/arm/cortex-a7.md: New bypasses using
1642 arm_mac_accumulator_is_result.
1644 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1646 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
1647 (cortex_a7_neon_mla): Likewise.
1648 (cortex_a7_fpfmad): New reservation.
1649 (cortex_a7_fpmacs): Use ffmas and update required units.
1650 (cortex_a7_fpmuld): Update required units and latency.
1651 (cortex_a7_fpmacd): Likewise.
1652 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
1653 (cortex_a7_neon). Likewise.
1654 (bypass) Update participating units.
1656 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
1658 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
1659 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
1661 * config/arm/vfp11.md (vfp_farith): Use ffmas.
1662 (vfp_fmul): Use ffmad.
1663 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
1664 (cortex_r4_fmacd): Use ffmad.
1665 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
1666 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
1667 (cortex_a9_fmacd): Use ffmad.
1668 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
1669 (cortex_a8_vfp_macd): Use ffmad.
1670 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
1671 (cortex_a5_fpmacd): Use ffmad.
1672 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
1673 (cortex_a15_vfp_macd): Use ffmad.
1674 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
1676 2013-01-29 Jason Merrill <jason@redhat.com>
1679 * varasm.c (default_assemble_visibility): Don't warn about
1680 visibility on artificial decls.
1682 2013-01-29 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/56113
1685 * tree-ssa-structalias.c (equiv_class_lookup): Also return
1687 (label_visit): Free duplicate bitmaps and record the leader instead.
1688 (perform_var_substitution): Adjust.
1690 2013-01-29 Richard Biener <rguenther@suse.de>
1692 PR tree-optimization/55270
1693 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
1694 the CFG, schedule loops for fixup.
1696 2013-01-29 Nick Clifton <nickc@redhat.com>
1698 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
1701 2013-01-28 Leif Ekblad <leif@rdos.net>
1703 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
1704 * config/i386/i386.h (TARGET_RDOS): New macro.
1705 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
1706 * config/i386/i386.c (ix86_option_override_internal): For 64bit
1707 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
1708 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
1709 DEFAULT_LARGE_SECTION_THRESHOLD.
1710 * config/i386/i386.md (R14_REG, R15_REG): New constants.
1711 * config/i386/rdos.h: New file.
1712 * config/i386/rdos64.h: New file.
1714 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
1717 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
1720 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1722 PR rtl-optimization/56117
1723 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
1724 call cselib_lookup_from_insn on the MEM before calling
1725 add_insn_mem_dependence.
1727 2013-01-28 Richard Biener <rguenther@suse.de>
1729 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
1730 to a stmt that didn't have one.
1731 (copy_phis_for_bb): Likewise for PHI arguments.
1732 (copy_debug_stmt): Likewise for debug stmts.
1734 2013-01-28 Richard Biener <rguenther@suse.de>
1736 PR tree-optimization/56034
1737 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
1738 (partition_builtin_p): Adjust.
1739 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
1740 it is the last partition.
1741 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
1742 up the vertex for the definition.
1743 (classify_partition): Classify whether a partition is a
1744 PKIND_REDUCTION, thus has uses outside of the loop.
1745 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
1746 Merge all PKIND_REDUCTION partitions into the last partition.
1747 (tree_loop_distribution): Seed partitions from reductions as well.
1749 2013-01-28 Jakub Jelinek <jakub@redhat.com>
1751 PR tree-optimization/56125
1752 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
1753 pow(x,c) into sqrt(x) * powi(x, n/2) or
1754 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
1755 optimizing for size.
1756 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
1757 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
1760 PR tree-optimization/56094
1761 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
1762 to UNKNOWN_LOCATION while gimplifying expr.
1764 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
1767 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
1768 operand 0 in movabs insn template for -masm=intel asm alternative.
1769 (*movabs<mode>_2): Ditto for operand 1.
1771 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
1774 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
1777 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1779 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
1782 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
1784 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
1785 Undef to avoid warning.
1787 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
1789 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
1790 * configure: Regenerate.
1792 2013-01-25 Jakub Jelinek <jakub@redhat.com>
1794 PR tree-optimization/56098
1795 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
1796 for stmts with volatile ops.
1797 (cond_store_replacement): Don't optimize if assign has volatile ops.
1798 (cond_if_else_store_replacement_1): Don't optimize if either
1799 then_assign or else_assign have volatile ops.
1800 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
1803 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1805 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
1807 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1809 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
1810 missing ':' in asm example.
1812 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
1814 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
1815 entries into lane and laneq entries.
1816 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
1817 Remove AdvSIMD scalar modes.
1818 (aarch64_sq<r>dmulh_laneq<mode>): New.
1819 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
1821 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
1822 builtin implementations to relfect changes in RTL in aarch64-simd.md.
1823 * config/aarch64/iterators.md (VCOND): New.
1826 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
1829 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
1830 Add NULL LIBNAME argument to existing definitions.
1831 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
1832 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
1833 * config/avr/avr.c (DEF_BUILTIN): Same.
1834 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
1835 (avr_expand_builtin): Expand to a vanilla call if a libgcc
1836 implementation is available (DECL_ASSEMBLER_NAME is set).
1837 (avr_fold_absfx): New static function.
1838 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
1839 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
1840 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
1842 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
1843 (abshk, absk, abslk, absllk): Provide as static inline functions.
1845 2013-01-25 Marek Polacek <polacek@redhat.com>
1847 PR tree-optimization/56035
1848 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
1850 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
1852 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
1853 (*movtf_internal_rex64): Add (!o,C) alternative
1854 (*movxf_internal_rex64): Ditto.
1855 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
1857 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
1859 * doc/invoke.texi: fix typo.
1860 * doc/objc.texi: fix typo.
1862 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
1864 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
1865 for the first two alternatives.
1867 2013-01-24 Diego Novillo <dnovillo@google.com>
1869 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
1870 (ggc-zone.o): Remove.
1871 * configure.ac: Remove option --with-gc.
1872 * configure: Re-generate.
1873 * doc/install.texi: Remove documentation for --with-gc.
1874 * gengtype.c (write_enum_defn): Remove. Update all users.
1875 (write_Types_process_field): Remove generation of gt_e_* argument.
1876 (output_type_enum): Remove. Update all users.
1877 (write_enum_defn): Remove. Update all users.
1878 (enum alloc_zone): Remove. Update all users.
1879 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
1880 * ggc-common.c (ggc_splay_alloc): Remove first argument.
1882 (struct ptr_data): Remove field TYPE. Update all users.
1883 (gt_pch_note_object): Remove argument TYPE. Update all users.
1884 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
1886 * ggc-none.c (ggc_alloc_typed_stat): Remove.
1887 (struct alloc_zone): Remove.
1888 (ggc_internal_alloc_zone_stat): Remove.
1889 (ggc_internal_cleared_alloc_zone_stat): Remove.
1890 * ggc-page.c (ggc_alloc_typed_stat): Remove.
1891 (ggc_pch_count_object): Remove last argument. Update all users.
1892 (ggc_pch_alloc_object): Remove last argument. Update all users.
1893 (struct alloc_zone): Remove.
1894 * ggc-zone.c: Remove.
1895 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
1896 (struct alloc_zone): Remove.
1897 (ggc_alloc_typed_stat): Remove.
1898 (ggc_alloc_typed): Remove.
1899 (ggc_splay_alloc): Remove first argument.
1900 (rtl_zone): Remove. Update all users.
1901 (tree_zone): Remove. Update all users.
1902 (tree_id_zone): Remove. Update all users.
1903 (ggc_internal_zone_alloc_stat): Remove. Update all users.
1904 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
1905 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
1906 * tree-ssanames.c: Remove references to zone allocator in comments.
1908 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
1910 * config/avr/avr.c (avr_out_fract): Make register numbers that
1911 might be outside of source operand signed.
1913 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
1915 * config/i386/constraints.md (Yf): New constraint.
1916 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
1917 of f constraint to conditionaly disable x87 register preferences.
1918 (*movdf_internal): Ditto.
1919 (*movsf_internal): Ditto.
1921 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
1924 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
1925 that have operands with impossible constraints.
1926 Add a FIXME for a speed-up opportunity.
1927 * lra-constraints.c (process_alt_operands): Verify that a class
1928 selected from constraints on asms is valid for the operand mode.
1929 (curr_insn_transform): Remove incorrect comment.
1931 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
1933 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
1934 TOC operand is a valid symbol ref in the constant pool.
1936 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1938 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
1940 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
1943 * config/avr/stdfix.h: New file.
1944 * t-avr (stdfix-gcc.h): New rule to build it.
1945 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
1947 2013-01-23 Kostya Serebryany <kcc@google.com>
1949 * config/darwin.h: remove dependency on
1950 CoreFoundation (asan on Mac OS).
1952 2013-01-23 Jakub Jelinek <jakub@redhat.com>
1955 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
1956 instead of cmpdi_operand for first comparison operand.
1957 Don't assert that comparison operands aren't both constants.
1959 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
1961 * doc/install.texi (Downloading the Source): Update references to
1962 downloading separate components.
1964 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
1966 * doc/extend.texi (__int128): Improve grammar.
1968 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
1971 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
1972 alternative to (o,r).
1973 (*movdi_internal_rex64): Remove (!o,n) alternative.
1974 (DImode immediate->memory splitter): Remove.
1975 (DImode immediate->memory peephole2): Remove.
1976 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
1977 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
1978 alternative to (!o,*r).
1979 (*movtf_internal_sse): New pattern.
1980 (*movxf_internal_rex64): New pattern.
1981 (*movxf_internal): Disable for TARGET_64BIT.
1982 (*movdf_internal_rex64): Remove (!o,F) alternative.
1984 2013-01-22 Jakub Jelinek <jakub@redhat.com>
1987 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
1988 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
1989 * tree-vect-loop-manip.c (find_loop_location): Also ignore
1990 stmt locations where LOCATION_LOCUS of the stmt location is
1991 UNKNOWN_LOCATION or BUILTINS_LOCATION.
1994 * config/i386/i386.md (UNSPEC_STOS): New.
1995 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
1996 *strsetqi_1): Add UNSPEC_STOS.
1998 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
2001 * doc/invoke.texi: Remove left over -Wsynth example.
2003 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2005 PR tree-optimization/56051
2006 * fold-const.c (fold_binary_loc): Don't fold
2007 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
2008 a narrowing conversion, or widening conversion from signed
2011 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2013 PR rtl-optimization/56023
2014 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
2015 dependent on debug instruction.
2017 2013-01-21 Martin Jambor <mjambor@suse.cz>
2020 * function.c (allocate_struct_function): Call
2021 invoke_set_current_function_hook earlier.
2023 2013-01-21 Jakub Jelinek <jakub@redhat.com>
2025 * reload1.c (init_reload): Only initialize reload_obstack
2026 during the first call.
2028 2013-01-21 Marek Polacek <polacek@redhat.com>
2030 * cfgloop.c (verify_loop_structure): Fix up grammar.
2032 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
2034 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
2035 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
2037 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2040 * config/arm/marvell-pj4.md: Update copyright year.
2041 Fix up use of alu to alu_reg and simple_alu_imm.
2043 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
2045 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
2047 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
2050 * lra-constraints.c (curr_insn_transform): Don't reuse original
2051 insn for secondary memory move when memory mode should be different.
2053 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2055 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
2056 atomic_storedi_1): New patterns.
2058 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2060 btver2 pipeline descriptions.
2061 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
2063 * config/i386/i386.md (btver2_decode): New type attributes.
2064 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
2066 * config/i386/btver2.md: New file describing btver2 pipelines.
2068 2013-01-19 Andrew Pinski <apinski@cavium.com>
2070 PR tree-optimization/52631
2071 * tree-ssa-sccvn (visit_use): Before looking up the original
2072 statement, try looking up the simplified expression.
2074 2013-01-19 Anthony Green <green@moxielogic.com>
2076 * config/moxie/moxie.c (moxie_expand_prologue): Set
2077 current_function_static_stack_size.
2079 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2081 PR tree-optimization/56029
2082 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
2083 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
2085 2013-01-18 Sharad Singhai <singhai@google.com>
2087 PR tree-optimization/55995
2088 * dumpfile.c (dump_loc): Print location only if available.
2089 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
2091 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
2094 * lra-constraints.c (curr_insn_transform): Reuse original insn for
2095 secondary memory move.
2096 (inherit_reload_reg): Use rclass instead of cl for
2097 check_secondary_memory_needed_p.
2099 2013-01-18 Jakub Jelinek <jakub@redhat.com>
2102 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
2103 the case where writing real complex part of target modifies op1.
2105 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
2107 * config/aarch64/aarch64-simd.md
2108 (aarch64_vcond_internal<mode>): Handle unordered cases.
2109 * config/aarch64/iterators.md (v_cmp_result): New.
2111 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
2112 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2114 * config/arm/marvell-pj4.md: New file.
2115 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
2116 * config/arm/arm.md (generic_sched): Add marvell_pj4.
2117 (generic_vfp): Likewise.
2118 * config/arm/arm-cores.def: Add marvell-pj4.
2119 * config/arm/arm-tune.md: Regenerate.
2120 * config/arm/arm-tables.opt: Regenerate.
2121 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
2122 * doc/invoke.texi: Document marvell-pj4.
2124 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
2126 * config/aarch64/arm_neon.h: Map scalar types to standard types.
2128 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2133 * var-tracking.c (negative_power_of_two_p): New.
2134 (global_get_addr_cache, local_get_addr_cache): New.
2135 (get_addr_from_global_cache, get_addr_from_local_cache): New.
2136 (vt_canonicalize_addr): Rewrite using the above. Adjust the
2138 (vt_stack_offset_p): Remove.
2139 (vt_canon_true_dep): Always canonicalize loc's address.
2140 (clobber_overlapping_mems): Make sure we have a MEM.
2141 (local_get_addr_clear_given_value): New.
2142 (val_reset): Clear local cached entries.
2143 (compute_bb_dataflow): Create and release the local cache.
2144 Disable duplicate MEMs clobbering.
2145 (emit_notes_in_bb): Clobber MEMs likewise.
2146 (vt_emit_notes): Create and release the local cache.
2147 (vt_initialize, vt_finalize): Create and release the global
2148 cache, respectively.
2149 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
2151 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2154 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
2155 not found in the symtab.
2157 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
2160 PR rtl-optimization/55547
2161 PR rtl-optimization/53827
2164 * alias.c (offset_overlap_p): New, factored out of...
2165 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
2166 the conservative special case for symbolic constants. Don't
2167 adjust zero sizes on alignment.
2169 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
2171 PR rtl-optimization/52573
2172 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
2173 REG_UNUSED for the same register.
2175 2013-01-17 Richard Biener <rguenther@suse.de>
2176 Marek Polacek <polacek@redhat.com>
2178 PR rtl-optimization/55833
2179 * loop-unswitch.c (unswitch_loops): Move loop verification...
2180 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
2181 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
2182 Set it to true when we're removing a loop from hierarchy tree in
2183 an irreducible region.
2184 (fix_bb_placements): Adjust caller.
2185 (fix_loop_placements): Likewise.
2187 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
2189 * config/avr/builtins.def (DEF_BUILTIN): Factor out
2190 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
2191 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
2192 Remove ID. Adjust comments.
2193 * config/avr/avr-c.c (avr_builtin_name): Remove.
2194 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
2195 * config/avr/avr.c (avr_tolower): New static function.
2196 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
2197 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
2198 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
2201 2013-01-17 Jan Hubicka <jh@suse.cz>
2203 PR tree-optimization/55273
2204 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
2206 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
2209 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
2210 store through atomic_store<mode>_1.
2211 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
2213 2013-01-17 Martin Jambor <mjambor@suse.cz>
2215 PR tree-optimizations/55264
2216 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
2217 for virtual methods.
2218 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
2219 virtual methods before inlining is over.
2220 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
2222 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
2225 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
2227 PR rtl-optimization/56005
2228 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
2229 pending reads for prefetch.
2231 2013-01-16 Ian Bolton <ian.bolton@arm.com>
2233 * config/aarch64/aarch64.md
2234 (*cstoresi_neg_uxtw): New pattern.
2235 (*cmovsi_insn_uxtw): New pattern.
2236 (*<optab>si3_uxtw): New pattern.
2237 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
2238 (*<optab>si3_insn_uxtw): New pattern.
2239 (*bswapsi2_uxtw): New pattern.
2241 2013-01-16 Richard Biener <rguenther@suse.de>
2243 * tree-inline.c (tree_function_versioning): Remove set but
2244 never used variable.
2246 2013-01-16 Richard Biener <rguenther@suse.de>
2248 PR tree-optimization/55964
2249 * tree-flow.h (rename_variables_in_loop): Remove.
2250 (rename_variables_in_bb): Likewise.
2251 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
2252 (copy_loop_before): Adjust and delete update-ssa status.
2253 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
2254 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
2255 (rename_variables_in_loop): Remove.
2256 (slpeel_update_phis_for_duplicate_loop): Likewise.
2257 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
2258 use available cfg machinery instead of duplicating it.
2259 Update PHI nodes and perform poor-mans SSA update here.
2260 (slpeel_tree_peel_loop_to_edge): Adjust.
2262 2013-01-16 Richard Biener <rguenther@suse.de>
2264 PR tree-optimization/54767
2265 PR tree-optimization/53465
2266 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
2267 (vrp_visit_phi_node): For PHI arguments coming via backedges
2268 drop all symbolical range information.
2269 (execute_vrp): Compute backedges.
2271 2013-01-16 Richard Biener <rguenther@suse.de>
2273 * doc/install.texi: Update CLooG and ISL requirements to
2276 2013-01-16 Christian Bruel <christian.bruel@st.com>
2279 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
2280 (broken_move): Handle UNSPECV_SP_SWITCH_B.
2281 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
2283 2013-01-16 DJ Delorie <dj@redhat.com>
2285 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
2286 (UNSPECV_SP_SWITCH_E): New.
2287 (sp_switch_1): Change to an unspec.
2288 (sp_switch_2): Change to an unspec. Don't use post-inc when we
2291 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
2293 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
2294 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
2295 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
2296 (expand_mem_thread_fence): Ditto.
2297 (expand_mem_signal_fence): Ditto.
2298 (expand_atomic_load): Ditto.
2299 (expand_atomic_store): Ditto.
2301 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
2303 PR rtl-optimization/55547
2304 PR rtl-optimization/53827
2307 * alias.c (memrefs_conflict_p): Set sizes to negative after
2310 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2313 * function.c (thread_prologue_and_epilogue_insns): Always
2314 add crtl->drap_reg to set_up_by_prologue.set, even if
2315 stack_realign_drap is false.
2317 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2319 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
2320 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
2321 *call): Fix indention.
2323 2013-01-15 Tom de Vries <tom@codesourcery.com>
2326 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
2329 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
2331 PR rtl-optimization/55153
2332 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
2334 2013-01-15 Martin Jambor <mjambor@suse.cz>
2336 PR tree-optimization/55920
2337 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
2338 accesses as grp_to_be_debug_replaced.
2340 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2342 PR tree-optimization/55920
2343 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
2344 there is non-useless type conversion needed from debug rhs to lhs,
2345 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
2347 2013-01-15 Joseph Myers <joseph@codesourcery.com>
2348 Mikael Pettersson <mikpe@it.uu.se>
2351 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
2353 (ASM_OUTPUT_CASE_LABEL): Remove.
2354 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
2355 * final.c (shorten_branches): Update alignment of labels before
2356 jump tables if CASE_VECTOR_SHORTEN_MODE.
2358 2013-01-15 Richard Biener <rguenther@suse.de>
2361 * system.h: Do not include gmp.h for building host tools.
2363 2013-01-15 Richard Biener <rguenther@suse.de>
2366 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
2367 account for bitpos when computing alignment.
2369 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
2371 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
2372 (ix86_target_macros_internal): Likewise.
2374 * config/i386/i386.c (m_CORE2I7): Removed.
2375 (m_CORE_HASWELL): New macro.
2376 (m_CORE_ALL): Likewise.
2377 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
2378 (initial_ix86_arch_features): Likewise.
2379 (processor_target_table): Initializations for Core avx2.
2380 (cpu_names): New names "core-avx2".
2381 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
2382 PROCESSOR_CORE_HASWELL.
2383 (ix86_issue_rate): New case.
2384 (ia32_multipass_dfa_lookahead): Likewise.
2385 (ix86_sched_init_global): Likewise.
2387 * config/i386/i386.h (TARGET_HASWELL): New macro.
2388 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
2389 (processor_type): New PROCESSOR_HASWELL.
2391 2013-01-15 Jakub Jelinek <jakub@redhat.com>
2393 PR tree-optimization/55955
2394 * tree-vect-loop.c (vectorizable_reduction): Give up early on
2395 *SHIFT_EXPR and *ROTATE_EXPR codes.
2397 PR tree-optimization/48766
2398 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
2399 -ftrapv disable -fwrapv.
2401 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2404 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
2405 etc. to 1 and not to __flash.
2406 Use LL suffix for __INT24_MAX__ with -mint8.
2407 Use ULL suffix for __UINT24_MAX__ with -mint8.
2409 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2411 * config/avr/avr-arch.h
2412 (struct base_arch_s): Use typedef avr_arch_t instead.
2413 (struct arch_info_s): Use typedef avr_arch_info_t instead.
2414 (struct mcu_type_s): Use typedef avr_mcu_t instead.
2415 * config/avr/avr.c: Same.
2416 * config/avr/avr-devices.c: Same.
2417 * config/avr/driver-avr.c: Same.
2418 * config/avr/gen-avr-mmcu-texi.c: Same.
2419 * config/avr/avr-mcus.def: Adjust comment.
2421 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
2423 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
2424 * config/aarch64/iterators.md (VALLDI): New.
2426 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
2427 Andi Kleen <ak@linux.intel.com>
2430 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
2431 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
2434 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2436 * config/avr/avr-stdint.h: Remove trailing blanks.
2437 * config/avr/avr-log.h: Same.
2438 * config/avr/avr-arch.h: Same.
2439 * config/avr/avr-devices.c: Same.
2440 * config/avr/avr-dimode.md: Same.
2441 * config/avr/predicates.md: Same.
2442 * config/avr/avr-c.c: Same. And fix typo.
2444 * config/avr/avr-protos.h: Same. And:
2445 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
2446 (init_cumulative_args): Rename to avr_init_cumulative_args.
2447 (expand_prologue): Rename to avr_expand_prologue.
2448 (expand_epilogue): Rename to avr_expand_epilogue.
2449 (adjust_insn_length): Rename to avr_adjust_insn_length.
2450 (notice_update_cc): Rename to avr_notice_update_cc.
2451 (final_prescan_insn): Rename to avr_final_prescan_insn.
2452 * config/avr/avr.c: Same.
2453 * config/avr/avr.h: Same.
2454 * config/avr/avr.md: Remove trailing blanks.
2455 (prologue): Use avr_expand_prologue.
2456 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
2458 2013-01-14 Richard Biener <rguenther@suse.de>
2460 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
2461 verify_location, collect_subblocks): New functions.
2462 (verify_gimple_in_cfg): Verify that locations only reference
2463 BLOCKs in the functions BLOCK tree.
2465 2013-01-14 Richard Biener <rguenther@suse.de>
2467 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
2469 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
2471 (insert_out_of_ssa_copy_on_edge): Likewise.
2472 (rewrite_close_phi_out_of_ssa): Likewise.
2473 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
2475 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
2476 propagated constants.
2477 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
2480 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
2482 * config/avr/avr-modes.def: Add GPL copyright notice.
2484 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
2486 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
2487 MEMMODEL_MASK to determine memory model.
2488 (atomic_store<mode>): Ditto from operands[2].
2489 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
2491 2013-01-13 Jakub Jelinek <jakub@redhat.com>
2494 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
2495 (fold_gimple_assign): Don't call unshare_expr here.
2496 (fold_ctor_reference): Call unshare_expr.
2498 2013-01-13 Terry Guo <terry.guo@arm.com>
2500 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
2501 * doc/fragments.texi: Document MULTILIB_REUSE.
2502 * gcc.c (multilib_reuse): New internal spec.
2503 (set_multilib_dir): Also search multilib from multilib_reuse.
2504 * genmultilib (tmpmultilib3): Refactor code.
2505 (tmpmultilib4): Ditto.
2506 (multilib_reuse): New multilib argument.
2508 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
2510 * Makefile.in: Update copyright.
2512 2013-01-12 Tom de Vries <tom@codesourcery.com>
2515 * calls.c (expand_call): Check if arg_nr is valid.
2517 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2519 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
2520 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
2521 documentation. Add missing '__' in front of
2522 __builtin_ia32_packssdw256.
2524 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2527 * config/s390/s390.c (s390_preferred_reload_class): Do not return
2528 NO_REGS for larl operands.
2529 (s390_reload_larl_operand): Use s390_load_address instead of
2532 2013-01-11 Richard Biener <rguenther@suse.de>
2534 * tree-cfg.c (verify_node_sharing_1): Split out from ...
2535 (verify_node_sharing): ... here.
2536 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
2538 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
2540 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
2541 Substitute TREECHECKING.
2542 * configure: Regenerate.
2543 * Makefile.in (TREECHECKING): New.
2545 2013-01-11 Richard Guenther <rguenther@suse.de>
2547 PR tree-optimization/44061
2548 * tree-vrp.c (extract_range_basic): Compute zero as
2549 value-range for __builtin_constant_p of function parameters.
2551 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
2553 Update copyright years.
2555 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
2557 PR rtl-optimization/55672
2558 * lra-eliminations.c (mark_not_eliminable): Permit addition with
2559 const to be eliminable.
2561 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
2563 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
2564 * configure: Regenerate.
2566 2013-01-10 Richard Biener <rguenther@suse.de>
2568 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
2570 2013-01-10 Richard Biener <rguenther@suse.de>
2573 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
2574 locations for virtual PHI arguments.
2575 (rewrite_update_phi_arguments): Likewise.
2577 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
2579 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
2582 2013-01-10 Jakub Jelinek <jakub@redhat.com>
2584 PR tree-optimization/55921
2585 * tree-complex.c (expand_complex_asm): New function.
2586 (expand_complex_operations_1): Call it for GIMPLE_ASM.
2588 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2591 * config/s390/s390.c (s390_symref_operand_p)
2592 (s390_loadrelative_operand_p): Merge the two functions.
2593 (s390_check_qrst_address, print_operand_address): Add parameters
2594 to s390_loadrelative_operand_p invokation.
2595 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
2596 (s390_reload_larl_operand, s390_secondary_reload): Use
2597 s390_loadrelative_operand_p instead of s390_symref_operand_p.
2598 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
2600 2013-01-09 Mike Stump <mikestump@comcast.net>
2602 * dse.c (record_store): Remove unnecessary assert.
2604 2013-01-09 Jan Hubicka <jh@suse.cz>
2606 PR tree-optimization/55569
2607 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
2608 * cfgloop.h (scale_loop_profile): Likewise.
2610 2013-01-09 Jan Hubicka <jh@suse.cz>
2613 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
2615 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
2617 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
2620 * expr.h (maybe_emit_group_store): Declare.
2621 * expr.c (maybe_emit_group_store): New function.
2622 * builtins.c (expand_builtin_int_roundingfn): Call it.
2623 (expand_builtin_int_roundingfn_2): Likewise.
2625 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
2627 PR rtl-optimization/55829
2628 * lra-constraints.c (match_reload): Add code for absent output.
2629 (curr_insn_transform): Add code for reloads of matched inputs
2632 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
2634 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
2635 attribute of movddup insn to DF.
2636 (*vec_interleave_lowv2df): Ditto.
2637 (vec_dupv2df): Ditto.
2639 2013-01-09 Jan Hubicka <jh@suse.cz>
2641 PR tree-optimiation/55875
2642 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
2643 EVERY_ITERATION parameter.
2644 (number_of_iterations_exit): Check if exit is executed every iteration.
2645 (idx_infer_loop_bounds): Similarly here.
2646 (n_of_executions_at_most): Simplify
2647 to only test for cases where statement is dominated by the
2648 particular bound; handle correctly the "postdominance" test.
2649 (scev_probably_wraps_p): Use max loop iterations info
2650 as a global bound first.
2652 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
2653 Nick Clifton <nickc@redhat.com>
2655 * config/v850/v850.md (cbranchsf4): New pattern.
2656 (cstoresf4): New pattern.
2657 (cbranchdf4): New pattern.
2658 (cstoredf4): New pattern.
2659 (movsicc): Disallow floating point comparisons.
2660 (cmpsf_le_insn): Fix order of operators.
2661 (cmpsf_lt_insn): Likewise.
2662 (cmpsf_eq_insn): Likewise.
2663 (cmpdf_le_insn): Likewise.
2664 (cmpdf_lt_insn): Likewise.
2665 (cmpdf_eq_insn): Likewise.
2666 (cmpsf_ge_insn): Use LE comparison.
2667 (cmpdf_ge_insn): Likewise.
2668 (cmpsf_gt_insn): Use LT comparison.
2669 (cmpdf_gt_insn): Likewise.
2670 (cmpsf_ne_insn): Delete pattern.
2671 (cmpdf_ne_insn): Delete pattern.
2672 * config/v850/v850.c (v850_gen_float_compare): Use
2673 gen_cmpdf_eq_insn for NE comparison.
2674 (v850_float_z_comparison_operator)
2675 (v850_float_nz_comparison_operator): Move from here ...
2676 * config/v850/predicates.md: ... to here. Move GT and GE
2677 comparisons into v850_float_z_comparison_operator.
2678 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
2680 (v850_float_nz_comparison_operator): Likewise.
2682 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2684 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
2685 with calls to gen_insvsi/gen_insvdi.
2687 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2689 * config/i386/i386.c (initial_ix86_tune_features): Set up
2690 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
2692 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
2693 Jakub Jelinek <jakub@redhat.com>
2695 PR tree-optimization/48189
2696 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
2697 If nitercst is 0, don't predict the exit edge.
2699 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
2701 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
2702 in asm_fprintf with reg_names.
2703 (aarch64_print_operand_address): Likewise.
2704 (aarch64_return_addr): Likewise.
2705 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
2707 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2709 * config/pa/pa.h (VAL_U6_BITS_P): Define.
2710 (INT_U6_BITS): Likewise.
2711 * config/pa/predicates.md (uint6_operand): New predicate.
2712 (shift5_operand, shift6_operand): Likewise.
2713 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
2715 (lshrdi3): Use shift6_operand.
2716 (shrpsi4, shrpdi4): New insn patterns.
2717 (extzv): Delete expander.
2718 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
2719 predicates in unamed zero extract patterns. Tighten common constraint.
2720 (extv): Delete expander.
2721 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
2722 predicates in unamed sign extract patterns. Tighten common constraint.
2723 (insv): Delete expander.
2724 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
2725 predicates in unamed insert patterns. Tighten common constraint.
2726 Change uint32_operand predicate to uint6_operand predicate in unamed
2727 DImode pattern to insert constant values of type 1...1xxxx.
2729 2013-01-04 Jan Hubicka <jh@suse.cz>
2731 PR tree-optimization/55823
2732 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
2735 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2736 Uros Bizjak <ubizjak@gmail.com>
2738 PR rtl-optimization/55845
2739 * df-problems.c (can_move_insns_across): Stop scanning at
2740 volatile_insn_p source instruction or give up if
2741 across_from .. across_to range contains any volatile_insn_p
2744 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2746 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
2747 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
2749 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
2750 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
2752 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2755 * asan.c (asan_clear_shadow): New function.
2756 (asan_emit_stack_protection): Use it.
2758 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
2760 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
2761 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
2762 with tab instead of space.
2764 2013-01-08 Nick Clifton <nickc@redhat.com>
2766 * config/rl78/rl78.c (rl78_expand_prologue): Always select
2767 register bank 0 at the start of an interrupt handler.
2768 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
2771 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2773 * config/aarch64/aarch64-simd.md
2774 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
2775 (aarch64_simd_bsl): Likewise.
2776 (aarch64_vcond_internal<mode>): Likewise.
2777 (vcond<mode><mode>): Likewise.
2778 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
2779 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
2781 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
2783 * config/aarch64/aarch64-builtins.c
2784 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
2786 2013-01-08 Martin Jambor <mjambor@suse.cz>
2789 * tree-sra.c (analyze_access_subtree): Return true also after
2790 potentially creating a debug-only replacement.
2792 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2795 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
2797 PR tree-optimization/54120
2798 * tree-vrp.c (range_fits_type_p): Don't allow
2799 src_precision < precision from signed vr to unsigned_p
2800 if vr->min or vr->max is negative.
2801 (simplify_float_conversion_using_ranges): Test can_float_p
2802 against CODE_FOR_nothing.
2804 2013-01-08 Jakub Jelinek <jakub@redhat.com>
2805 Richard Biener <rguenther@suse.de>
2808 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
2809 types instead of just INTEGER_TYPE types.
2811 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
2813 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
2816 2013-01-07 Steve Ellcey <sellcey@mips.com>
2819 * config/mips/mips.opt: Change mad to mmad to match documentation.
2821 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2824 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
2827 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2830 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
2831 (avr_addrspace_t): Add .section_name field.
2832 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
2834 (avr_addrspace): Same. Initialize .section_name. Remove last
2835 NULL entry. Put __memx into .progmemx.data.
2836 (progmem_section_prefix): Remove.
2837 (avr_asm_init_sections): No need to initialize progmem_section.
2838 (avr_asm_named_section): Use avr_addrspace[].section_name to get
2839 section name prefix.
2840 (avr_asm_select_section): Ditto. And use get_unnamed_section to
2841 retrieve the progmem section.
2842 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
2843 boundary to run over avr_addrspace[].
2844 (avr_register_target_pragmas): Ditto.
2846 2013-01-06 Jakub Jelinek <jakub@redhat.com>
2848 * varasm.c (output_constant_def_contents): For asan_protect_global
2849 protected strings, adjust DECL_ALIGN if needed, before testing for
2851 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
2852 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
2854 (output_object_block): For asan protected decls, emit asan padding
2855 after their contents.
2856 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
2857 (asan_finish_file): Test it here instead.
2859 2013-01-07 Nick Clifton <nickc@redhat.com>
2860 Matthias Klose <doko@debian.org>
2861 Doug Kwan <dougkwan@google.com>
2862 H.J. Lu <hongjiu.lu@intel.com>
2865 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
2867 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
2869 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
2871 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
2873 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
2875 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2878 * doc/install.texi (Cross-Compiler-Specific Options): Document
2881 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
2883 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
2884 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
2885 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
2886 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
2887 vqmovun_high_s64): Fix source operand number and update copyright.
2889 2013-01-07 Richard Biener <rguenther@suse.de>
2892 * gimple.h (gimple_call_builtin_p): New overload.
2893 * gimple.c (validate_call): New function.
2894 (gimple_call_builtin_p): Likewise.
2895 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2896 Use gimple_call_builtin_p.
2897 (find_func_clobbers): Likewise.
2898 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
2899 (strlen_optimize_stmt): Likewise.
2901 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2903 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
2904 (vld1q_dup_*): Likewise.
2906 (vld1q_*): Likewise.
2907 (vld1_lane_*): Likewise.
2908 (vld1q_lane_*): Likewise.
2910 2013-01-07 Richard Biener <rguenther@suse.de>
2912 * lto-streamer.h (LTO_minor_version): Bump to 2.
2914 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2916 * config/aarch64/aarch64-protos.h
2917 (aarch64_const_double_zero_rtx_p): Rename to...
2918 (aarch64_float_const_zero_rtx_p): ...this.
2919 (aarch64_float_const_representable_p): New.
2920 (aarch64_output_simd_mov_immediate): Likewise.
2921 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
2922 move immediate case.
2923 * config/aarch64/aarch64.c
2924 (aarch64_const_double_zero_rtx_p): Rename to...
2925 (aarch64_float_const_zero_rtx_p): ...this.
2926 (aarch64_print_operand): Allow printing of new constants.
2927 (aarch64_valid_floating_const): New.
2928 (aarch64_legitimate_constant_p): Check for valid floating-point
2930 (aarch64_simd_valid_immediate): Likewise.
2931 (aarch64_vect_float_const_representable_p): New.
2932 (aarch64_float_const_representable_p): Likewise.
2933 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
2934 (aarch64_output_simd_mov_immediate): New.
2935 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
2936 (*movdf_aarch64): Likewise.
2937 * config/aarch64/constraints.md (Ufc): New.
2938 (Y): call aarch64_float_const_zero_rtx.
2939 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
2941 2013-01-07 Richard Biener <rguenther@suse.de>
2943 PR tree-optimization/55888
2944 PR tree-optimization/55862
2945 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
2946 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
2947 not if it is contained therein.
2949 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2951 * config/avr/t-avr: Typo.
2953 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
2956 * config/avr/t-avr: Don't automatically rebuild
2957 $(srcdir)/config/avr/t-multilib
2958 $(srcdir)/config/avr/avr-tables.opt
2959 $(srcdir)/doc/avr-mmcu.texi
2960 (avr-mcus): New phony target to build them on request.
2961 (s-avr-mlib, s-avr-mmcu-texi): Remove.
2962 * avr/avr-mcus.def: Adjust comments.
2964 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
2966 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
2968 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
2970 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
2972 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
2974 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
2976 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
2978 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
2979 to generate profiling.
2980 * config/rs6000/aix64.h (LIB_SPEC): Same.
2982 2013-01-04 Andrew Pinski <apinski@cavium.com>
2984 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
2986 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
2988 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
2990 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
2992 (ix86_expand_move): Ditto.
2993 (ix86_zero_extend_to_Pmode): Ditto.
2994 (ix86_expand_call): Ditto.
2995 (ix86_expand_special_args_builtin): Ditto.
2996 (ix86_expand_builtin): Ditto.
2998 2013-01-04 Richard Biener <rguenther@suse.de>
3000 PR tree-optimization/55862
3001 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
3002 translating them through PHI nodes.
3004 2013-01-04 Martin Jambor <mjambor@suse.cz>
3006 PR tree-optimization/55755
3007 * tree-sra.c (sra_modify_assign): Do not check that an access has no
3008 children when trying to avoid producing a VIEW_CONVERT_EXPR.
3010 2013-01-04 Marek Polacek <polacek@redhat.com>
3013 * opts.c (default_options_optimization): Clarify error message.
3015 2013-01-04 Richard Biener <rguenther@suse.de>
3018 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
3021 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3024 * config/pa/pa.md (movsi): Revert previous change.
3025 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
3028 2013-01-03 Richard Henderson <rth@redhat.com>
3030 * config/i386/i386.c (ix86_expand_move): Always assign to op1
3031 after eliminating TLS symbols.
3033 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
3036 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
3037 * graphite-poly.c (debug_gmp_value): Likewise.
3039 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
3042 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
3043 selected code model, define __code_mode_small__, __code_model_medium__,
3044 __code_model_large__, __code_model_32__ or __code_model_kernel__.
3045 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
3046 xchg temporary register with %k. Declare temporary register as
3048 [__x86_64__]: For medium and large code models, preserve %rbx register.
3050 2013-01-03 Richard Biener <rguenther@suse.de>
3052 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
3053 (dump_subscript): Adjust.
3054 (finalize_ddr_dependent): Do not dump redundant info.
3055 (analyze_siv_subscript): Adjust.
3056 (subscript_dependence_tester): Likewise.
3057 (compute_affine_dependence): Likewise.
3059 2013-01-03 Richard Biener <rguenther@suse.de>
3062 2013-01-03 Richard Biener <rguenther@suse.de>
3064 PR tree-optimization/55857
3065 * tree-vect-stmts.c (vectorizable_load): Do not setup
3066 re-alignment for invariant loads.
3068 2013-01-02 Richard Biener <rguenther@suse.de>
3070 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3071 invariant load do not generate a vector load from the scalar location.
3073 2013-01-03 Richard Biener <rguenther@suse.de>
3075 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
3076 for not vectorizing.
3077 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
3078 not build INDIRECT_REFs, call get_name once only.
3079 (vect_create_data_ref_ptr): Likewise. Dump base object kind
3080 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
3082 2013-01-03 Richard Biener <rguenther@suse.de>
3084 PR tree-optimization/55857
3085 * tree-vect-stmts.c (vectorizable_load): Do not setup
3086 re-alignment for invariant loads.
3088 2013-01-03 Richard Biener <rguenther@suse.de>
3091 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
3092 prefer a built-in decl.
3094 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3096 * gcc.c (process_command): Update copyright notice dates.
3097 * gcov.c (print_version): Likewise.
3098 * gcov-dump.c (print_version): Likewise.
3100 PR rtl-optimization/55838
3101 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
3102 iv0.step, iv1.step and step.
3104 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3105 Marc Glisse <marc.glisse@inria.fr>
3107 PR tree-optimization/55832
3108 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
3109 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
3110 integer_{one,zero}_node.
3112 2013-01-03 Jakub Jelinek <jakub@redhat.com>
3115 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
3116 * var-tracking.c (reverse_op): Don't add reverse ops to
3117 VALUEs that have already
3118 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
3120 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
3122 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
3124 2013-01-02 Teresa Johnson <tejohnson@google.com>
3126 * dumpfile.c (dump_loc): Print filename with location.
3127 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
3128 new location_t parameter to emit complete unroll message with
3130 (canonicalize_loop_induction_variables): Compute loops location
3131 and pass to try_unroll_loop_completely.
3132 * loop-unroll.c (report_unroll_peel): New function.
3133 (peel_loops_completely): Use new dump format with location
3134 for main dumpfile message, and invoke report_unroll_peel on success.
3135 (decide_unrolling_and_peeling): Ditto.
3136 (decide_peel_once_rolling): Remove old dumpfile message subsumed
3137 by report_unroll_peel.
3138 (decide_peel_completely): Ditto.
3139 (decide_unroll_constant_iterations): Ditto.
3140 (decide_unroll_runtime_iterations): Ditto.
3141 (decide_peel_simple): Ditto.
3142 (decide_unroll_stupid): Ditto.
3143 * cfgloop.c (get_loop_location): New function.
3144 * cfgloop.h (get_loop_location): Declare.
3146 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3148 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
3151 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3154 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
3155 BLKmode objects when EXPAND_MEMORY is specified.
3157 2013-01-02 Sriraman Tallam <tmsriram@google.com>
3159 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
3161 (fold_builtin_cpu): Do not share cpu model decls across statements.
3163 2013-01-02 Jason Merrill <jason@redhat.com>
3166 * tree.c (build_array_type_1): Revert earlier change.
3168 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
3170 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
3172 * config/aarch64/aarch64-tune.md: Re-generate.
3174 2013-01-02 Richard Biener <rguenther@suse.de>
3176 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
3177 invariant load do not generate a vector load from the scalar location.
3179 2013-01-02 Richard Biener <rguenther@suse.de>
3182 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
3183 * configure: Regenerate.
3185 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3187 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
3188 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
3189 (expand_builtin_int_roundingfn_2): Keep the original target around
3190 for the fallback case.
3192 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3194 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
3195 to be clear for sign changes.
3197 2013-01-01 Jan Hubicka <jh@suse.cz>
3199 * ipa-inline-analysis.c: Fix formatting.
3201 2013-01-01 Jakub Jelinek <jakub@redhat.com>
3203 PR tree-optimization/55831
3204 * tree-vect-loop.c (get_initial_def_for_induction): Use
3205 gsi_after_labels instead of gsi_start_bb.
3207 Copyright (C) 2013 Free Software Foundation, Inc.
3209 Copying and distribution of this file, with or without modification,
3210 are permitted in any medium without royalty provided the copyright
3211 notice and this notice are preserved.