1 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
3 * config/arm/arm-protos.h (tune_params): Add
4 prefer_neon_for_64bits field.
5 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
6 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
7 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
8 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
9 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
10 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
11 (arm_option_override): Handle -mneon-for-64bits new option.
12 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13 (prefer_neon_for_64bits): Declare new variable.
14 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
15 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and
17 (arch_enabled): Handle new arch types. Remove support for onlya8
19 (one_cmpldi2): Use new arch names.
20 * config/arm/arm.opt (mneon-for-64bits): Add option.
21 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
23 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
25 * doc/invoke.texi (-mneon-for-64bits): Document.
27 2013-03-21 Richard Biener <rguenther@suse.de>
29 PR tree-optimization/39326
30 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
31 (sort_bbs_in_loop_postorder_cmp): New function.
32 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop
35 2013-03-21 Richard Biener <rguenther@suse.de>
37 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
38 (vect_insert_into_interleaving_chain): Likewise.
39 (vect_drs_dependent_in_basic_block): Inline ...
40 (vect_slp_analyze_data_ref_dependence): ... here. New function,
42 (vect_analyze_data_ref_dependence): ... here. Simplify.
43 (vect_check_interleaving): Simplify.
44 (vect_analyze_data_ref_dependences): Likewise. Split out ...
45 (vect_slp_analyze_data_ref_dependences): ... this new function.
46 (dr_group_sort_cmp): New function.
47 (vect_analyze_data_ref_accesses): Compute data-reference groups
48 here instead of in vect_analyze_data_ref_dependence. Use
49 a more efficient algorithm.
50 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
51 vect_slp_analyze_data_ref_dependences. Call
52 vect_analyze_data_ref_accesses earlier.
53 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
54 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
55 (vect_slp_analyze_data_ref_dependences): New prototype.
57 2013-03-21 Richard Biener <rguenther@suse.de>
59 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
60 ref is stored in the loop.
61 (find_refs_for_sm): Walk only over all stores.
62 (store_motion_loop): Allocate from lim_bitmap_obstack.
63 (store_motion): Likewise.
65 2013-03-21 Richard Biener <rguenther@suse.de>
67 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
68 Update virtual SSA form.
70 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
72 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
73 * configure: Regenerate.
74 * config.in: Regenerate.
75 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
76 if !HAVE_LD_EH_FRAME_CIEV3.
78 2013-03-21 Richard Biener <rguenther@suse.de>
80 * tree-cfg.c (verify_expr_no_block): New function.
81 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
82 nor DECL_VALUE_EXPR have locations with associated blocks.
83 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
84 (clear_unused_block_pointer): Remove code dealing with
85 blocks in DECL_DEBUG_EXPR locations.
87 2013-03-21 Richard Biener <rguenther@suse.de>
89 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
90 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
91 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
92 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
93 instead of DECL_DEBUG_EXPR_IS_FROM.
94 * gimplify.c (gimplify_modify_expr): Likewise.
95 * tree-cfg.c (verify_expr_location_1): Likewise.
96 * tree-complex.c (create_one_component_var): Likewise.
97 * tree-sra.c (create_access_replacement): Likewise.
98 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
99 (clear_unused_block_pointer): Likewise.
100 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
101 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
102 * var-tracking.c (var_debug_decl): Likewise.
103 (track_expr_p): Likewise.
104 * tree-inline.c (add_local_variables): Likewise. Set
105 DECL_HAS_DEBUG_EXPR_P after copying it.
106 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
107 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
109 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
112 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
113 * configure: Regenerate.
114 * config.in: Regenerate.
115 * config/i386/i386.md (*movdf_internal): Use
116 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
117 movd instead of movq mnemonic for interunit moves.
118 (*movdi_internal): Ditto.
120 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
122 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
123 (abd<mode>_3): New pattern.
124 (aba<mode>_3): New pattern.
125 (fabd<mode>_3): New pattern.
127 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
129 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
130 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
131 occurrence of REGISTER_PREFIX as its empty string.
133 2013-03-20 Jeff Law <law@redhat.com>
135 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
136 addititional equivalences for equality comparisons between an SSA_NAME
137 and a constant where the SSA_NAME was set from a widening conversion.
139 2013-03-20 Walter Lee <walt@tilera.com>
141 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
143 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
145 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
146 depending on type attribute.
147 (*movti_internal): Ditto.
148 (*movtf_internal): Ditto.
149 (*movxf_internal): Ditto.
150 (*movdf_internal): Ditto.
151 (*movsf_internal): Ditto.
153 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
155 * config/i386/i386.md (*movti_internal): Set prefix attribute to
156 maybe_vex for sselog1 and ssemov types.
157 (*movdi_internal): Reorder operand constraints.
158 (*movsi_internal): Ditto. Set prefix attribute to
159 maybe_vex for sselog1 and ssemov types.
160 (*movtf_internal): Set prefix attribute to maybe_vex
161 for sselog1 and ssemov types.
162 (*movdf_internal): Ditto. Set prefix_data16 attribute for
163 DImode ssemov types. Reorder operand constraints.
164 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
165 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
166 attribute for SImode ssemov types. Reorder operand constraints.
168 2013-03-20 Martin Jambor <mjambor@suse.cz>
170 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
171 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
173 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
175 * config/rs6000/predicates.md (indexed_address, update_address_mem
176 update_indexed_address_mem): New predicates.
177 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
178 attribute for load/store instructions.
179 * config/rs6000/dfp.md (movsd_store): Likewise.
180 (movsd_load): Likewise.
181 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
182 (unnamed HI->DI extend define_insn): Likewise.
183 (unnamed SI->DI extend define_insn): Likewise.
184 (unnamed QI->SI extend define_insn): Likewise.
185 (unnamed QI->HI extend define_insn): Likewise.
186 (unnamed HI->SI extend define_insn): Likewise.
187 (unnamed HI->SI extend define_insn): Likewise.
188 (extendsfdf2_fpr): Likewise.
189 (movsi_internal1): Likewise.
190 (movsi_internal1_single): Likewise.
191 (movhi_internal): Likewise.
192 (movqi_internal): Likewise.
193 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
194 attribute for load/store instructions.
195 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
197 (mov<mode>_softfloat): Likewise.
198 (mov<mode>_hardfloat32): Likewise.
199 (mov<mode>_hardfloat64): Likewise.
200 (mov<mode>_softfloat64): Likewise.
201 (movdi_internal32): Likewise.
202 (movdi_internal64): Likewise.
203 (probe_stack_<mode>): Likewise.
205 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
207 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
208 floating point, and decimal floating point to reload iterator.
210 * config/rs6000/constraints.md (wl constraint): New constraints to
211 return FLOAT_REGS if certain options are used to reduce the number
212 of separate patterns that exist in the file.
213 (wx constraint): Likewise.
214 (wz constraint): Likewise.
216 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
217 -mdebug=reg, print wg, wl, wx, and wz constraints.
218 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
219 Initialize the reload functions for 64-bit binary/decimal floating
221 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
222 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
223 create the buffer on the stack to overcome not having a 32-bit
225 (rs6000_emit_move): Likewise.
226 (rs6000_secondary_memory_needed_rtx): Likewise.
227 (rs6000_alloc_sdmode_stack_slot): Likewise.
228 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
229 via xxlxor, just like DFmode 0.0.
231 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
232 define as 1 if we are running on a power7 or newer.
233 (enum r6000_reg_class_enum): Add new constraints.
235 * config/rs6000/dfp.md (movsd): Delete, combine with binary
236 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
237 with other moves by using conditional constraits (wg). Use LFIWZX
238 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
239 (movsd splitter): Likewise.
240 (movsd_hardfloat): Likewise.
241 (movsd_softfloat): Likewise.
243 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
244 binary and decimal floating point moves.
245 (fmove_ok): New attributes to combine binary and decimal floating
246 point moves, and to combine power6x (mfpgpr) moves along normal
248 (real_value_to_target): Likewise.
255 (movsf): Combine binary and decimal floating point moves. Combine
256 power6x (mfpgpr) moves with other moves by using conditional
257 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
258 (mov<mode> for SFmode/SDmode); Likewise.
259 (SFmode/SDmode splitters): Likewise.
260 (movsf_hardfloat): Likewise.
261 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
262 (movsf_softfloat): Likewise.
263 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
265 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
266 wx and wz constraints.
268 * config/rs6000/constraints.md (wg constraint): New constraint to
269 return FLOAT_REGS if -mmfpgpr (power6x) was used.
271 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
274 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
275 -mdebug=reg, print wg, wl, wx, and wz constraints.
276 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
277 Initialize the reload functions for 64-bit binary/decimal floating
279 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
280 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
281 create the buffer on the stack to overcome not having a 32-bit
283 (rs6000_emit_move): Likewise.
284 (rs6000_secondary_memory_needed_rtx): Likewise.
285 (rs6000_alloc_sdmode_stack_slot): Likewise.
286 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
287 via xxlxor, just like DFmode 0.0.
289 * config/rs6000/dfp.md (movdd): Delete, combine with binary
290 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
291 with other moves by using conditional constraits (wg). Use LFIWZX
292 and STFIWX for loading SDmode on power7.
293 (movdd splitters): Likewise.
294 (movdd_hardfloat32): Likewise.
295 (movdd_softfloat32): Likewise.
296 (movdd_hardfloat64_mfpgpr): Likewise.
297 (movdd_hardfloat64): Likewise.
298 (movdd_softfloat64): Likewise.
300 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
301 64-bit binary and decimal floating point moves.
302 (FMOVE64X): Likewise.
303 (movdf): Combine 64-bit binary and decimal floating point moves.
304 Combine power6x (mfpgpr) moves with other moves by using
305 conditional constraits (wg).
306 (mov<mode> for DFmode/DDmode): Likewise.
307 (DFmode/DDmode splitters): Likewise.
308 (movdf_hardfloat32): Likewise.
309 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
310 (movdf_softfloat32): Likewise.
311 (movdf_hardfloat64_mfpgpr): Likewise.
312 (movdf_hardfloat64): Likewise.
313 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
314 (movdf_softfloat64): Likewise.
315 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
316 (reload_<mode>_load): Move to later in the file so they aren't in
317 the middle of the floating point move insns.
318 (reload_<mode>_store): Likewise.
320 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
323 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
324 constraint if -mdebug=reg.
325 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
326 Enable using dd reload support if needed.
328 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
329 binary and decimal floating point moves in rs6000.md.
330 (movtd_internal): Likewise.
332 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
333 decimal floating point moves.
335 (movtf_internal): Likewise.
336 (mov<mode>_internal, TDmode/TFmode): Likewise.
337 (movtf_softfloat): Likewise.
338 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
340 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
341 movdi_internal64, using wg constraint for move direct operations.
342 (movdi_internal64): Likewise.
344 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
345 MODES_TIEABLE_P for selected modes. Print the numerical value of
346 the various virtual registers. Use GPR/FPR first/last values,
347 instead of hard coding the register numbers. Print which modes
348 have reload functions registered.
349 (rs6000_option_override_internal): If -mdebug=reg, trace the options
350 settings before/after setting cpu, target and subtarget settings.
351 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
352 and for secondary reload failures in rs6000_secondary_reload_inner.
353 (rs6000_secondary_reload_fail): Likewise.
354 (rs6000_secondary_reload_inner): Likewise.
356 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
357 macros for first/last GPR and FPR registers.
358 (LAST_GPR_REGNO): Likewise.
359 (FIRST_FPR_REGNO): Likewise.
360 (LAST_FPR_REGNO): Likewise.
362 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
363 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
364 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
365 (vcond<mode><mode>): Likewise.
366 (vcondu<mode><mode>): Likewise.
367 (vector_gtu<mode>): Likewise.
368 (vector_gte<mode>): Likewise.
369 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
370 to prevent the compiler from converting DImode operations to TImode.
371 (ior<mode>3): Likewise.
372 (and<mode>3): Likewise.
373 (one_cmpl<mode>2): Likewise.
374 (nor<mode>3): Likewise.
375 (andc<mode>3): Likewise.
377 * config/rs6000/constraints.md (wt constraint): New constraint
378 that returns VSX_REGS if TImode is allowed in VSX registers.
380 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
383 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
384 similar to TImode, but it is restricted to being in the GPRs.
386 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
387 TImode to occupy a single VSX register.
389 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
390 -mvsx-timode for power7/power8.
391 (power7 cpu): Likewise.
392 (power8 cpu): Likewise.
394 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
395 sure that TFmode/TDmode take up two registers if they are ever
396 allowed in the upper VSX registers.
397 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
399 (rs6000_init_hard_regno_mode_ok): Likewise.
400 (rs6000_debug_reg_global): Add debugging for PTImode and wt
401 constraint. Print if LRA is turned on.
402 (rs6000_option_override_internal): Give an error if -mvsx-timode
403 and VSX is not enabled.
404 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
405 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
406 to reg+offset addressing. Use PTImode when checking offset
407 addresses for validity.
408 (reg_offset_addressing_ok_p): Likewise.
409 (rs6000_legitimate_offset_address_p): Likewise.
410 (rs6000_legitimize_address): Likewise.
411 (rs6000_legitimize_reload_address): Likewise.
412 (rs6000_legitimate_address_p): Likewise.
413 (rs6000_eliminate_indexed_memrefs): Likewise.
414 (rs6000_emit_move): Likewise.
415 (rs6000_secondary_reload): Likewise.
416 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
417 reloads to fpr registers to continue to use reg+offset addressing,
418 but 64-bit reloads to altivec registers need reg+reg addressing.
419 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
420 it. Treat LO_SUM like a PLUS operation.
421 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
422 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
423 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
424 registers to share a register with a smaller sized type, since VSX
425 puts scalars in the upper 64-bits.
426 (print_operand): Add support for PTImode.
427 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
428 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
429 registers, but don't have arithmetic support.
430 (rs6000_memory_move_cost): Add test for VSX.
431 (rs6000_opt_masks): Add -mvsx-timode.
433 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
436 (VSr): Use wt constraint for TImode.
437 (VSv): Drop TImode support.
438 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
439 (vsx_movti_64bit): Likewise.
440 (vsx_movti_32bit): Likewise.
441 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
442 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
443 one '?' on the appropriate output constraint. Do not allow TImode
444 logical operations on 32-bit systems.
445 (vsx_ior<mode>3): Likewise.
446 (vsx_xor<mode>3): Likewise.
447 (vsx_one_cmpl<mode>2): Likewise.
448 (vsx_nor<mode>3): Likewise.
449 (vsx_andc<mode>3): Likewise.
450 (vsx_concat_<mode>): Likewise.
451 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
453 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
454 OPTION_MASK_VSX_TIMODE.
455 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
456 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
458 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
459 (TI2 iterator): New iterator for TImode, PTImode.
460 (wd mode attribute): Add values for vector types.
461 (movti_string): Replace TI move operations with operations for TImode
462 and PTImode. Add support for TImode being allowed in VSX registers.
463 (mov<mode>_string, TImode/PTImode): Likewise.
464 (movti_ppc64): Likewise.
465 (mov<mode>_ppc64, TImode/PTImode): Likewise.
466 (TI mode splitters): Likewise.
468 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
471 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
473 PR tree-optimization/56355
474 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
475 Also handle integers with undefined overflow.
477 2013-03-20 Catherine Moore <clm@codesourcery.com>
478 Maciej W. Rozycki <macro@codesourcery.com>
479 Tom de Vries <tom@codesourcery.com>
480 Nathan Sidwell <nathan@codesourcery.com>
481 Iain Sandoe <iain@codesourcery.com>
482 Nathan Froyd <froydnj@codesourcery.com>
483 Chao-ying Fu <fu@mips.com>
485 * doc/extend.texi: (micromips, nomicromips, nocompression):
486 Document new function attributes.
487 * doc/invoke.texi (minterlink-compressed, mmicromips,
488 m14k, m14ke, m14kec): Document new options.
489 (minterlink-mips16): Update documentation.
490 * doc/md.texi (ZC, ZD): Document new constraints.
491 * configure.ac (gcc_cv_as_micromips): Check if linker
492 supports the .set micromips directive.
493 * configure: Regenerate.
494 * config.in: Regenerate.
495 * config/mips/mips-tables.opt: Regenerate.
496 * config/mips/micromips.md: New file.
497 * constraints.md (ZC, ZD): New constraints.
498 * config/mips/predicates.md (movep_src_register): New predicate.
499 (movep_src_operand): New predicate.
500 (non_volatile_mem_operand): New predicate.
501 * config/mips/mips.md (multimem): New type.
502 (length): Differentiate between 17-bit and 18-bit branch offsets.
503 (MOVEP1, MOVEP2): New mode iterator.
504 (mov_<load>l): Use ZC constraint.
505 (mov_<load>r): Likewise.
506 (mov_<store>l): Likewise.
507 (mov_<store>r): Likewise.
508 (*branch_equality<mode>_inverted): Add microMIPS support.
509 (*branch_equality<mode>): Likewise.
510 (*jump_absolute): Likewise.
511 (indirect_jump_<mode>): Likewise.
512 (tablejump_<mode>): Likewise.
513 (<optab>_internal): Likewise.
514 (sibcall_internal): Likewise.
515 (sibcall_value_internal): Likewise.
516 (prefetch): Use constraint ZD.
517 * config/mips/mips.opt (minterlink-compressed): New option.
518 (minterlink-mips16): Now an alias for minterlink-compressed.
519 (mmicromips): New option.
520 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
521 (compare_and_swap_12): Likewise.
522 (sync_add<mode>): Likewise.
523 (sync_<optab>_12): Likewise.
524 (sync_old_<optab>_12): Likewise.
525 (sync_new_<optab>_12): Likewise.
526 (sync_nand_12): Likewise.
527 (sync_old_nand_12): Likewise.
528 (sync_new_nand_12): Likewise.
529 (sync_sub<mode>): Likewise.
530 (sync_old_add<mode>): Likewise.
531 (sync_old_sub<mode>): Likewise.
532 (sync_new_add<mode>): Likewise.
533 (sync_new_sub<mode>): Likewise.
534 (sync_<optab><mode>): Likewise.
535 (sync_old_<optab><mode>): Likewise.
536 (sync_new_<optab><mode>): Likewise.
537 (sync_nand<mode>): Likewise.
538 (sync_old_nand<mode>): Likewise.
539 (sync_new_nand<mode>): Likewise.
540 (sync_lock_test_and_set<mode>): Likewise.
541 (test_and_set_12): Likewise.
542 (atomic_compare_and_swap<mode>): Likewise.
543 (atomic_exchange<mode>_llsc): Likewise.
544 (atomic_fetch_add<mode>_llsc): Likewise.
545 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
546 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
547 (umips_save_restore_pattern_p): Likewise.
548 (umips_load_store_pair_p): Likewise.
549 (umips_output_load_store_pair): Likewise.
550 (umips_movep_target_p): Likewise.
551 (umips_12bit_offset_address_p): Likewise.
552 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
553 (mips_base_mips16): Rename this...
554 (mips_base_compression_flags): ...to this. Update all uses.
555 (mips_attribute_table): Add micromips, nomicromips and nocompression.
556 (mips_mips16_decl_p): Delete.
557 (mips_nomips16_decl_p): Delete.
558 (mips_get_compress_on_flags): New function.
559 (mips_get_compress_off_flags): New function.
560 (mips_get_compress_mode): New function.
561 (mips_get_compress_on_name): New function.
562 (mips_get_compress_off_name): New function.
563 (mips_insert_attributes): Support multiple compression types.
564 (mips_merge_decl_attributes): Likewise.
565 (umips_12bit_offset_address_p): New function.
566 (mips_start_function_definition): Emit .set micromips directive.
567 (mips_call_may_need_jalx_p): New function.
568 (mips_function_ok_for_sibcall): Add microMIPS support.
569 (mips_print_operand_punctuation): Support short delay slots and
571 (umips_swm_mask, umips_swm_encoding): New.
572 (umips_build_save_restore): New function.
573 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
574 (was_mips16_p): Remove.
575 (old_compression_mode): New.
576 (mips_set_compression_mode): New function.
577 (mips_set_current_function): Add microMIPS support.
578 (mips_option_override): Likewise.
579 (umips_save_restore_pattern_p): New function.
580 (umips_output_save_restore): New function.
581 (umips_load_store_pair_p_1): New function.
582 (umips_load_store_pair_p): New function.
583 (umips_output_load_store_pair_1): New function.
584 (umips_output_load_store_pair): New function.
585 (umips_movep_target_p) New function.
586 (mips_prepare_pch_save): Add microMIPS support.
587 * config/mips/mips.h (TARGET_COMPRESSION): New.
588 (TARGET_CPU_CPP_BUILTINS): Update macro
589 to use new compression flags and to support microMIPS.
590 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
591 (MIPS_ARCH_FLOAT_SPEC): Likewise.
592 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
593 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
594 (ASM_SPEC): Support mmicromips and mno-micromips.
595 (M16STORE_REG_P): New macro.
596 (MIPS_CALL): Support TARGET_MICROMIPS.
597 (MICROMIPS_J): New macro.
598 (mips_base_mips16): Rename this...
599 (mips_base_compression_flags): ...to this.
600 (UMIPS_12BIT_OFFSET_P): New macro.
601 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
602 (MULTILIB_DIRNAMES): Likewise.
603 2013-03-20 Richard Biener <rguenther@suse.de>
605 PR tree-optimization/56661
606 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
607 the result does not have to be distinct.
609 2013-03-20 Richard Biener <rguenther@suse.de>
611 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
614 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
615 Steven Bosscher <steven@gcc.gnu.org>
617 PR rtl-optimization/56605
618 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
620 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
623 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
624 that require movd instead of movq.
626 2013-03-20 Richard Biener <rguenther@suse.de>
628 * tree-ssa-structalias.c (struct variable_info): Add pointer
629 to the first field of an aggregate with sub-vars. Make
630 this and the pointer to the next subfield its ID.
631 (vi_next): New function.
632 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
633 storedanything_id, integer_id): Increment by one.
634 (new_var_info, get_call_vi, lookup_call_clobber_vi,
635 get_call_clobber_vi): Adjust.
636 (solution_set_expand): Simplify and speedup.
637 (solution_set_add): Inline into ...
638 (set_union_with_increment): ... this. Adjust accordingly.
639 (do_sd_constraint): Likewise.
640 (do_ds_constraint): Likewise.
641 (do_complex_constraint): Simplify.
642 (build_pred_graph): Adjust.
643 (solve_graph): Likewise. Simplify and speedup.
644 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
645 get_constraint_for_component_ref, get_constraint_for_1,
646 first_vi_for_offset, first_or_preceding_vi_for_offset,
647 create_function_info_for, create_variable_info_for_1,
648 create_variable_info_for, intra_create_variable_infos): Adjust.
649 (init_base_vars): Push NULL for ID zero.
650 (compute_points_to_sets): Adjust.
652 2013-03-20 Richard Biener <rguenther@suse.de>
654 * cfgloop.c (verify_loop_structure): Streamline and avoid
655 ICEing on corrupt loop tree.
656 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
659 2013-03-20 Richard Biener <rguenther@suse.de>
661 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
662 check whether an SSA update is needed.
664 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
666 * config/mips/constraints.md (T): Rename to...
670 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
671 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
673 2013-03-19 Ian Bolton <ian.bolton@arm.com>
675 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
676 (*subsi3_carryin_uxtw): Likewise.
678 2013-03-19 Ian Bolton <ian.bolton@arm.com>
680 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
681 (*rorsi3_insn_uxtw): Likewise.
683 2013-03-19 Ian Bolton <ian.bolton@arm.com>
685 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
686 (*extrsi5_insn_uxtw): Likewise.
688 2013-03-19 Richard Biener <rguenther@suse.de>
690 PR tree-optimization/56273
691 * passes.c (init_optimization_passes): Move second VRP after DOM.
693 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
695 * config/i386/i386.md (*movti_internal): Merge from
696 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
697 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
698 nox64 isa attributes.
700 2013-03-18 Richard Biener <rguenther@suse.de>
702 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
704 (merge_node_constraints): Likewise.
705 (build_succ_graph): Likewise.
706 (valid_graph_edge): Inline into single caller.
707 (unify_nodes): Likewise. Use bitmap_set_bit return value
709 (scc_visit): Fix formatting and variable use.
710 (do_sd_constraint): Use gcc_checking_assert.
711 (do_ds_constraint): Likewise.
712 (do_complex_constraint): Likewise.
713 (condense_visit): Likewise. Cleanup.
714 (dump_pred_graph): New function.
715 (perform_var_substitution): Dump the pred-graph before
716 variable substitution.
717 (find_equivalent_node): Use gcc_checking_assert.
718 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
720 2013-03-18 Richard Biener <rguenther@suse.de>
722 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
723 Remove cond_expr_stmt_list argument and do not gimplify the
725 (vect_loop_versioning): Adjust.
726 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
727 Cleanup to use less temporaries.
728 (vect_create_data_ref_ptr): Cleanup.
730 2013-03-18 Jakub Jelinek <jakub@redhat.com>
732 PR tree-optimization/56635
733 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
734 require types_compatible_p types.
736 2013-03-18 Nick Clifton <nickc@redhat.com>
738 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
741 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
742 Add missing line to comment describing function.
744 2013-03-18 Richard Biener <rguenther@suse.de>
746 PR tree-optimization/56210
747 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
748 Handle string / character search functions.
749 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
751 2013-03-18 Richard Biener <rguenther@suse.de>
754 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
755 and implement properly.
756 * gimple.h (gimple_cond_single_var_p): Remove.
758 2013-03-18 Richard Biener <rguenther@suse.de>
760 * tree-data-ref.h (find_data_references_in_loop): Declare.
761 * tree-data-ref.c (get_references_in_stmt): Use a stack
762 vector pre-allocated in the callers.
763 (find_data_references_in_stmt): Adjust.
764 (graphite_find_data_references_in_stmt): Likewise.
765 (create_rdg_vertices): Likewise.
766 (find_data_references_in_loop): Export.
767 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
768 Compute dependences here...
769 (vect_analyze_data_refs): ...not here. When we encounter
770 a non-vectorizable data reference in basic-block vectorization
771 truncate the data reference vector. Do not bother to
772 fixup data-dependence information for gather loads.
773 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
774 of data references, as reported.
776 2013-03-18 Richard Biener <rguenther@suse.de>
778 PR tree-optimization/3713
779 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
780 has_constants and expr.
781 (stmt_has_constants): Properly valueize SSA names when deciding
782 whether the stmt has constants.
784 2013-03-18 Richard Biener <rguenther@suse.de>
786 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
787 whole function when there is nothing to do.
788 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
789 * tree-vectorizer.c (vectorize_loops): Update virtual and
790 loop-closed SSA once.
791 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
793 2013-03-18 Richard Biener <rguenther@suse.de>
796 * domwalk.c (bb_postorder): New global static.
797 (cmp_bb_postorder): New function.
798 (walk_dominator_tree): Replace scheme imposing an order for
799 visiting dominator sons by one sorting them at the time they
800 are pushed on the stack.
802 2013-03-18 Richard Biener <rguenther@suse.de>
804 PR tree-optimization/39326
805 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
806 (struct mem_ref): Replace mem member with ao_ref typed member.
807 (MEM_ANALYZABLE): Adjust.
808 (memref_eq): Likewise.
809 (mem_ref_alloc): Likewise.
810 (gather_mem_refs_stmt): Likewise.
811 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
812 (execute_sm_if_changed_flag_set): Adjust.
813 (execute_sm): Likewise.
814 (ref_always_accessed_p): Likewise.
815 (refs_independent_p): Likewise.
816 (can_sm_ref_p): Likewise.
818 2013-03-18 Jakub Jelinek <jakub@redhat.com>
821 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
822 return 1 even for !unsignedp.
824 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
826 * config/i386/i386.md (isa): Add x64 and nox64.
827 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
828 (*pushtf): Enable *roF alternative for x64 isa only.
829 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
830 mode attribute of integer alternatives to DImode for TARGET_64BIT.
831 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
832 (*movtf_internal): Merge from *movtf_internal_rex64 and
833 *movtf_internal_sse. Use x64 and nox64 isa attributes.
834 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
835 nox64 isa attributes.
836 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
837 nox64 isa attributes.
838 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
840 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
842 * config/alpha/alpha.c (TARGET_LRA_P): New define.
844 2013-03-17 Jakub Jelinek <jakub@redhat.com>
847 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
848 class names. Remove trailing comma after "ALL_REGS".
850 2013-03-16 Jan Hubicka <jh@suse.cz>
852 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
853 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
854 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
855 of cgraph_get_create_node.
856 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
858 2013-03-16 Jason Merrill <jason@redhat.com>
861 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
862 with DW_AT_default_value.
864 2013-03-16 Jakub Jelinek <jakub@redhat.com>
866 * BASE-VER: Set to 4.9.0.
868 2013-03-14 Andi Kleen <ak@linux.intel.com>
871 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
872 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
873 Document _x* TSX intrinsics.
875 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
876 David Holsgrove <david.holsgrove@xilinx.com>
878 * configure.ac: Add MicroBlaze TLS support detection.
879 * configure: Regenerate.
880 * config/microblaze/microblaze-protos.h
881 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
882 symbol_mentioned_p, label_mentioned_p): Add prototypes.
883 * config/microblaze/microblaze.c (microblaze_address_type): Add
884 ADDRESS_TLS and tls_reloc address types.
885 (microblaze_address_info): Add tls_reloc.
886 (TARGET_HAVE_TLS): Define.
887 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
888 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
889 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
890 load_tls_operand, microblaze_call_tls_get_addr,
891 microblaze_legitimize_tls_address): New functions.
892 (microblaze_classify_unspec): Handle UNSPEC_TLS.
893 (get_base_reg): Use microblaze_tls_symbol_p.
894 (microblaze_classify_address): Handle TLS.
895 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
896 label_mentioned_p and microblaze_tls_referenced_p.
897 (microblaze_legitimize_address): Handle TLS.
898 (microblaze_address_insns): Handle ADDRESS_TLS.
899 (pic_address_needs_scratch): Handle TLS.
900 (print_operand_address): Handle TLS.
901 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
902 (microblaze_expand_move): Handle TLS.
903 (microblaze_legitimate_constant_p): Check
904 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
905 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
906 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
907 (PIC_OFFSET_TABLE_REGNUM): Set.
908 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
909 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
910 (addsi3, movsi_internal2, movdf_internal): Update constraints
911 * config/microblaze/predicates.md (arith_plus_operand): Define
912 (move_operand): Redefine as move_src_operand,
913 check microblaze_tls_referenced_p.
915 2013-03-14 Ian Bolton <ian.bolton@arm.com>
917 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
918 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
920 2013-03-14 Ian Bolton <ian.bolton@arm.com>
922 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
925 2013-03-14 Jakub Jelinek <jakub@redhat.com>
927 PR tree-optimization/53265
928 * common.opt (Waggressive-loop-optimizations): New option.
929 * tree-ssa-loop-niter.c: Include tree-pass.h.
930 (do_warn_aggressive_loop_optimizations): New function.
931 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
932 if number_of_latch_executions returned constant.
933 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
934 early. If number_of_latch_executions returned constant, set
935 nb_iterations_upper_bound back to it.
936 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
938 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
939 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
941 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
942 (MULTILIB_OSDIRNAMES): Set.
943 * genmultilib: If defaultosdirname doesn't start with :: , set
944 defaultosdirname2 instead, clear it and emit two . multilib_raw
945 entries instead of just one.
947 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
949 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
950 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
951 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
952 (SUBTARGET_OVERRIDE_OPTIONS): New.
954 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
957 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
958 (musermode): Convert to Var(TARGET_USERMODE).
959 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
960 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
961 * config/sh/sh.c (sh_option_override): Use
962 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
963 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
965 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
967 (udivsi3_i4_single, divsi3_i4_single): Use
968 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
970 2013-03-13 Dave Korn <dave.korn.cygwin@....>
972 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
975 2013-03-13 Richard Biener <rguenther@suse.de>
977 PR tree-optimization/56608
978 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
979 calls when vectorizing basic-blocks.
981 2013-03-13 Jakub Jelinek <jakub@redhat.com>
984 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
987 2013-03-12 Jakub Jelinek <jakub@redhat.com>
989 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
991 2013-03-11 Jan Hubicka <jh@suse.cz>
994 * lto-streamer-out.c (output_symbol_p): Skip references from
995 constructors of external variables.
997 2013-03-11 Jan Hubicka <jh@suse.cz>
1000 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
1002 * emit-rtl.c (verify_rtx_sharing): Likewise.
1003 (copy_insn_1): Likewise.
1004 * rtl.c (copy_rtx): Likewise.
1006 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
1009 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
1010 output_operand_lossage message.
1012 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
1015 * arm.c (shift_op): Validate RTL pattern on the fly.
1016 (arm_print_operand, case 'S'): Don't use shift_operator to validate
1019 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1022 * config/pa/pa.md (call_value): Check for calls to powf and direct to
1023 new call patterns that clobber %fr12.
1024 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
1025 split and postreload patterns.
1026 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
1027 registers %fr12 and %fr12R as call used.
1029 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
1031 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
1032 (canon_address, record_store, replace_read, check_mem_read_rtx,
1033 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
1034 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
1035 rest_of_handle_dse): Likewise.
1037 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
1040 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
1042 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
1043 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
1044 (save_optabs_if_changed): Replace with...
1045 (init_tree_optimization_optabs): ...this.
1046 * optabs.c (save_optabs_if_changed): Rename to...
1047 (init_tree_optimization_optabs): ...this. Take the optimization node
1048 as argument. Do nothing if the base optabs are already correct.
1049 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
1050 to recompute optabs.
1051 * function.h (function): Remove optabs field.
1052 * function.c (invoke_set_current_function_hook): Call
1053 init_tree_optimization_optabs. Use the result to initialize
1056 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
1058 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
1059 if GTMA_HAS_NO_INSTRUMENTATION.
1060 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
1061 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
1062 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
1063 * gimple-pretty-print.c (dump_gimple_transaction): Handle
1064 GTMA_HAS_NO_INSTRUMENTATION.
1066 2013-03-08 Jakub Jelinek <jakub@redhat.com>
1068 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
1071 2013-03-08 Marek Polacek <polacek@redhat.com>
1072 Jakub Jelinek <jakub@redhat.com>
1074 PR tree-optimization/56478
1075 * predict.c (is_comparison_with_loop_invariant_p): Change the
1076 type of loop_step to tree.
1077 (predict_loops): Adjust.
1078 (predict_iv_comparison): Perform the computations on double_ints.
1080 2013-03-08 Richard Biener <rguenther@suse.de>
1082 PR tree-optimization/56570
1083 * tree-cfg.c (verify_expr_location_1): Verify locations for
1085 * tree-sra.c (create_access_replacement): Strip locations
1086 from DECL_DEBUG_EXPRs.
1088 2013-03-08 Richard Biener <rguenther@suse.de>
1090 * tree-inline.c (expand_call_inline): Do not associate
1091 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
1092 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
1094 2013-03-08 Richard Biener <rguenther@suse.de>
1096 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
1097 or block changes with -Og. Fix for location / block encoding
1098 changes and PHI arguments with locations.
1100 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
1102 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
1104 (struct output_info): Likewise.
1105 (register_overhead): Remove bad gcc_assert.
1106 (bitmap_find_bit): If there is only a single bitmap element, do not
1107 count a miss as a search.
1108 (print_statistics): Update for counter type changes.
1109 (dump_bitmap_statistics): Likewise. Print headers such that they
1110 are properly lined up with the printed counters.
1112 2013-03-07 Jakub Jelinek <jakub@redhat.com>
1114 PR tree-optimization/56559
1115 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
1116 check that it has only a single use.
1118 2013-03-07 Richard Biener <rguenther@suse.de>
1120 * doc/invoke.texi (fwhole-program): Discourage use in combination
1123 2013-03-06 Jakub Jelinek <jakub@redhat.com>
1125 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
1127 PR tree-optimization/56539
1128 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
1129 instead of GSI_CONTINUE_LINKING as last argument to
1130 force_gimple_operand_gsi. Adjust function comment.
1132 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
1136 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
1137 promoted mode, convert the result back to the original mode.
1139 2013-03-06 Richard Biener <rguenther@suse.de>
1142 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
1143 (insert_updated_phi_nodes_compare_uids): New function.
1144 (update_ssa): Sort symbols_to_rename after UID before
1145 traversing it to insert PHI nodes.
1147 2013-03-06 Richard Biener <rguenther@suse.de>
1150 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
1151 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
1154 2013-02-13 Richard Biener <rguenther@suse.de>
1157 * varasm.c (output_constant_def_1): Get the decl representing
1158 the constant as argument.
1159 (output_constant_def): Wrap output_constant_def_1.
1160 (make_decl_rtl): Use output_constant_def_1 with the decl
1161 representing the constant.
1162 (build_constant_desc): Optionally re-use a decl already
1163 representing the constant.
1164 (tree_output_constant_def): Adjust.
1166 2013-03-06 Joey Ye <joey.ye@arm.com>
1169 * gcc.c (convert_white_space): New function.
1170 (main): Handles white space in function name.
1172 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
1175 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
1176 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
1177 to SH_DIV_CALL_TABLE for TARGET_SH2.
1178 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
1180 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
1183 2013-03-05 Sterling Augustine <saugustine@google.com>
1184 Cary Coutant <ccoutant@google.com>
1187 * dwarf2out.c (resolve_addr): Don't call
1188 remove_loc_list_addr_table_entries a second time for the same
1191 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1194 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
1195 (avoid_complex_debug_insns): New function.
1196 (expand_debug_locations): Call it.
1198 PR rtl-optimization/56484
1199 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
1200 lifetimes of hard registers on small register class machines.
1202 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
1204 * config/microblaze/microblaze-protos.h: Rename
1205 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
1206 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
1208 (microblaze_fast_interrupt_function_p): New function.
1209 (microblaze_is_interrupt_handler): Rename to
1210 microblaze_is_interrupt_variant and add fast_interrupt check.
1211 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
1212 (save_restore_insns): Likewise.
1213 (compute_frame_size): Likewise.
1214 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
1215 (microblaze_globalize_label): Likewise.
1216 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
1217 * config/microblaze/microblaze.md: Use wrapper
1218 microblaze_is_interrupt_variant.
1220 2013-03-05 Kai Tietz <ktietz@redhat.com>
1222 * sdbout.c (sdbout_one_type): Switch to current function's section
1223 supporting cold/hot.
1225 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
1227 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
1230 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1233 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
1234 if VALGRIND_GET_VBITS is defined, temporarily make object
1235 memory all defined, and restore previous valgrind addressability
1236 and definability afterwards. Free this_object at the end.
1239 * lra.c (lra): Call lra_clear_live_ranges if live_p,
1240 right before calling lra_create_live_ranges, also call it
1241 when clearing live_p. Only call lra_clear_live_ranges
1242 at the end if live_p.
1245 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
1247 2013-03-05 Richard Biener <rguenther@suse.de>
1249 PR tree-optimization/56521
1250 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
1253 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
1256 * except.h (remove_unreachable_eh_regions): New prototype.
1257 * except.c (remove_eh_handler_splicer): New function, split out
1258 of remove_eh_handler.
1259 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
1260 warning about running it on many EH regions one at a time.
1261 (remove_unreachable_eh_regions_worker): New function, walk the
1262 EH tree in depth-first order and remove non-marked regions.
1263 (remove_unreachable_eh_regions): New function.
1264 * tree-eh.c (mark_reachable_handlers): New function, split out
1265 from remove_unreachable_handlers.
1266 (remove_unreachable_handlers): Use mark_reachable_handlers and
1267 remove_unreachable_eh_regions.
1268 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
1269 and remove_unreachable_eh_regions.
1271 2013-03-05 Richard Biener <rguenther@suse.de>
1274 * loop-init.c (fix_loop_structure): Remove loops in two stages,
1275 not freeing them until the end.
1277 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1279 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
1281 2013-03-05 Richard Biener <rguenther@suse.de>
1283 PR tree-optimization/56270
1284 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
1285 of loads after scheduling an SLP instance.
1287 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1289 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
1291 (check_gcc_parallelize): Run guality.exp as a separate job from
1292 vect.exp with unsorted.exp and $(dg_target_exps) separately from
1293 struct-layout-1.exp with stackalign.exp.
1295 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
1298 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
1299 load_index sbitmap even if some bit in it isn't set.
1302 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
1303 (discover_iteration_bound_by_body_walk): Change queues to
1304 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
1305 spelling in comment. Call safe_push on queues[bound_index] directly.
1306 Release queues[queue_index] in every iteration unconditionally.
1307 Release bounds vector.
1310 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
1311 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
1312 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
1315 2013-03-05 Richard Biener <rguenther@suse.de>
1318 * tree-inline.c (remap_blocks_to_null): New function.
1319 (expand_call_inline): When expanding a call stmt without
1320 an associated block inline remap all callee blocks to NULL.
1322 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1324 PR rtl-optimization/56494
1325 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
1326 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
1327 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
1330 * sel-sched-ir.c (free_sched_pools): Release
1331 succs_info_pool.stack[succs_info_pool.max_top] vectors too
1332 if succs_info_pool.max_top isn't -1.
1335 * opts.c (opts_obstack, opts_concat): Moved to...
1336 * opts-common.c (opts_obstack, opts_concat): ... here.
1338 2013-03-04 Jakub Jelinek <jakub@redhat.com>
1341 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
1343 2013-03-04 Martin Jambor <mjambor@suse.cz>
1345 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
1346 all appropriate places.
1348 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
1350 PR tree-optimization/56424
1351 * ipa-split.c (split_function): Do not set the RSO flag if result is
1352 not by reference and its type is a register type.
1354 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
1356 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
1357 (microblaze_legitimate_pic_operand): Likewise
1358 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
1359 new function microblaze_legitimate_pic_operand
1360 * config/microblaze/microblaze-protos.h
1361 (microblaze_legitimate_pic_operand): Declare.
1363 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1365 * config/microblaze/predicates.md (call_insn_simple_operand):
1366 New predicate for supported rtx code types.
1367 * config/microblaze/microblaze.md (call_internal1): Use
1368 call_insn_simple_operand predicate.
1370 2013-03-04 Jakub Jelinek <jakub@redhat.com>
1373 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
1374 partitions.ordered_remove.
1377 * tree-vect-stmts.c (vectorizable_conversion): Don't call
1378 vec_oprnds0.create (1) for modifier == NONE.
1381 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
1382 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
1383 vec_oprnds1 right before pushing anything to it for
1387 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
1388 set nbbs to 0 instead of having separate code path.
1389 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
1390 instead of false as last argument if returning NULL.
1392 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
1394 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
1395 the attribute is now called "target" instead of "option".
1396 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
1397 * doc/tm.texi.in (Target Attributes): Likewise document the correct
1398 attribute/pragma name for TARGET_OPTION_VALID_P and
1399 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
1400 * doc/tm.texi: Regenerated.
1402 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
1404 * config/microblaze/microblaze.c:
1405 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
1406 * config/microblaze/microblaze.h: Add -mxl-reorder to
1408 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
1409 instructions emitted if TARGET_REORDER.
1410 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
1411 or 0 for -m/-mno case, but initialises as 2 to detect default use case
1414 2013-03-01 Xinliang David Li <davidxl@google.com>
1416 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
1419 2013-03-01 Jakub Jelinek <jakub@redhat.com>
1422 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
1423 vector even when returning true. Fix up function comment formatting.
1426 * ira-build.c (ira_loop_nodes_count): New variable.
1427 (create_loop_tree_nodes): Initialize it.
1428 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
1431 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
1432 method on dr_chain and result_chain.
1433 * tree-vect-stmts.c (vectorizable_store): Only call
1434 result_chain.create if j == 0.
1437 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
1438 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
1439 before overwriting it.
1441 2013-03-01 Tobias Burnus <burnus@net-b.de>
1443 * doc/extended.texi (C Extensions): Change order in @menu
1445 (Other MIPS Built-in Functions): Move last MIPS entry before
1446 "picoChip Built-in Functions".
1447 (SH Built-in Functions): Move after RX Built-in Functions.
1448 * doc/gcc.texi (Introduction): Change order in @menu
1450 * doc/md.texi (Constraints): Ditto.
1451 * gty.texi (Type Information): Ditto.
1452 (User-provided marking routines for template types): Make
1454 * doc/invoke.texi (AArch64 Options): Move before
1455 "Adapteva Epiphany Options".
1457 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
1458 Jakub Jelinek <jakub@redhat.com>
1461 * asan.c (gate_asan): Lookup no_sanitize_address instead of
1462 no_address_safety_analysis attribute.
1463 * doc/extend.texi (no_address_safety_attribute): Rename to
1464 no_sanitize_address attribute, mention no_address_safety_analysis
1465 attribute as deprecated alias.
1467 2013-02-28 Jakub Jelinek <jakub@redhat.com>
1470 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
1471 type to vec<vec<tree> > *.
1472 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
1473 to be vec<tree> instead of vec<tree> *, set vec_defs
1474 to vNULL and call vec_defs.create (number_of_vects), adjust other
1476 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
1477 vectorizable_condition): Adjust vect_get_slp_defs callers.
1479 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1481 * config/aarch64/aarch64.c
1482 (aarch64_float_const_representable): Remove unused variable.
1484 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1486 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
1488 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1490 * config/aarch64/aarch64-builtins.c
1491 (aarch64_init_simd_builtins): Make static.
1493 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1495 * config/aarch64/aarch64.c
1496 (aarch64_simd_make_constant): Make static.
1498 2013-02-28 Martin Jambor <mjambor@suse.cz>
1500 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
1501 with no initialization to the RHS of debug statements.
1503 2013-02-28 Martin Jambor <mjambor@suse.cz>
1505 PR tree-optimization/56294
1506 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
1508 (get_access_replacement): Do not call create_access_replacement.
1509 Assert a replacement exists.
1510 (get_repl_default_def_ssa_name): Create the replacement declaration
1513 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1515 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
1518 2013-02-28 Marek Polacek <polacek@redhat.com>
1520 PR rtl-optimization/56466
1521 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
1522 if we're changing a loop.
1523 (peel_loops_completely): Likewise.
1525 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1528 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
1530 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1533 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
1534 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
1535 INTX_FTYPE_FX, FX_FTYPE_INTX.
1536 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
1538 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1540 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
1541 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
1542 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
1543 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
1544 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
1545 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
1546 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
1547 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
1548 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
1549 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
1550 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
1551 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
1552 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
1553 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
1554 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
1555 (avrxmega6): Increase max flash segments from 5 to 6.
1556 * config/avr/t-multilib: Regenerate.
1557 * config/avr/avr-tables.opt: Regenerate.
1558 * doc/avr-mmcu.texi: Regenerate.
1560 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1562 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
1563 (avr_device_to_arch): Rename to avr_device_to_ld.
1564 (avr_device_to_as): New prototype.
1565 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
1566 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
1567 * config/avr/driver-avr.c (avr_device_to_as): New.
1568 (avr_device_to_arch): Rename to avr_device_to_ld.
1570 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1573 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
1574 method on dr_chain and result_chain.
1577 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
1578 pointer_set_destroy on not_executed_last_iteration.
1581 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
1584 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
1585 FOR_EACH_DEFINED_FUNCTION when freeing state.
1588 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
1590 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
1594 * ipa-cp.c (decide_whether_version_node): Call vec_free on
1595 known_aggs[i].items and release known_aggs vector.
1598 * ipa-reference.c (propagate): Free node_info even for alias nodes.
1600 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1602 * config/microblaze/microblaze.c (microblaze_emit_compare):
1603 Use xor for EQ/NE comparisions.
1604 * config/microblaze/microblaze.md (cstoresf4): Add constraints
1605 (cbranchsf4): Adjust operator to comparison_operator.
1607 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1610 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
1612 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
1613 vec_safe_push, always update *slot.
1614 (redirect_edge_var_map_clear): Use vec_free.
1615 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
1616 (free_var_map_entry): Use vec_free.
1617 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
1618 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
1620 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
1623 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
1624 when the may_trap_p bit of the exprs being merged differs.
1625 Reorder tests for speculativeness in the logical and operator.
1627 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1629 * incpath.c (add_standard_paths): Use reconcat instead of concat
1630 where appropriate and avoid leaking memory.
1632 * opts.h: Include obstack.h.
1633 (opts_concat): New prototype.
1634 (opts_obstack): New declaration.
1635 * opts.c (opts_concat): New function.
1636 (opts_obstack): New variable.
1637 (init_options_struct): Call gcc_init_obstack on opts_obstack.
1638 (finish_options): Use opts_concat instead of concat
1639 and XOBNEWVEC instead of XNEWVEC.
1640 * opts-common.c (generate_canonical_option, decode_cmdline_option,
1641 generate_option): Likewise.
1642 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
1643 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
1646 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
1647 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
1649 2013-02-26 Jakub Jelinek <jakub@redhat.com>
1652 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
1654 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
1656 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
1657 (arm_block_move_unaligned_straight): Likewise.
1658 (arm_adjust_block_mem): Likewise.
1660 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
1663 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
1664 temp, cond and label.
1665 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
1668 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
1669 * config/c6x/c6x.h (dbx_register_map): Update declaration.
1672 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
1673 of prologue/epilogue functions.
1676 * config/tilegx/tilegx.c (tilegx_expand_prologue):
1677 Remove unused variable cfa_offset.
1678 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
1681 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
1682 type promotion to unsigned.
1685 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
1686 for HOST_WIDE_INT of 32 bit / same size as int.
1687 (arm_block_move_unaligned_straight): Likewise.
1688 (arm_adjust_block_mem): Likewise.
1691 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
1694 2013-02-26 Marek Polacek <polacek@redhat.com>
1696 PR tree-optimization/56426
1697 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
1699 2013-02-26 Richard Biener <rguenther@suse.de>
1702 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
1703 unused variable loops.
1705 2013-02-26 Jakub Jelinek <jakub@redhat.com>
1707 PR tree-optimization/56448
1708 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
1709 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
1710 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
1711 later operands of the references, or even first operand for
1712 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
1714 PR tree-optimization/56443
1715 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
1716 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
1717 to type_for_mode langhook.
1719 2013-02-25 Matt Turner <mattst88@gmail.com>
1721 * doc/invoke.texi: Document r4700.
1723 2013-02-25 Richard Biener <rguenther@suse.de>
1725 PR tree-optimization/56175
1726 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
1728 (simplify_bitwise_binary): ... here. Also guard the conversion
1729 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
1731 2013-02-25 Catherine Moore <clm@codesourcery.com>
1734 2013-02-24 Catherine Moore <clm@codesourcery.com>
1735 Maciej W. Rozycki <macro@codesourcery.com>
1736 Tom de Vries <tom@codesourcery.com>
1737 Nathan Sidwell <nathan@codesourcery.com>
1738 Iain Sandoe <iain@codesourcery.com>
1739 Nathan Froyd <froydnj@codesourcery.com>
1740 Chao-ying Fu <fu@mips.com>
1742 * doc/extend.texi: (micromips, nomicromips, nocompression):
1743 Document new function attributes.
1744 * doc/invoke.texi (minterlink-compressed, mmicromips,
1745 m14k, m14ke, m14kec): Document new options.
1746 (minterlink-mips16): Update documentation.
1747 * doc/md.texi (ZC, ZD): Document new constraints.
1748 * configure.ac (gcc_cv_as_micromips): Check if linker
1749 supports the .set micromips directive.
1750 * configure: Regenerate.
1751 * config.in: Regenerate.
1752 * config/mips/mips-tables.opt: Regenerate.
1753 * config/mips/micromips.md: New file.
1754 * constraints.md (ZC, AD): New constraints.
1755 * config/mips/predicates.md (movep_src_register): New predicate.
1756 (movep_src_operand): New predicate.
1757 (non_volatile_mem_operand): New predicate.
1758 * config/mips/mips.md (multimem): New type.
1759 (length): Differentiate between 17-bit and 18-bit branch offsets.
1760 (MOVEP1, MOVEP2): New mode iterator.
1761 (mov_<load>l): Use ZC constraint.
1762 (mov_<load>r): Likewise.
1763 (mov_<store>l): Likewise.
1764 (mov_<store>r): Likewise.
1765 (*branch_equality<mode>_inverted): Add microMIPS support.
1766 (*branch_equality<mode>): Likewise.
1767 (*jump_absolute): Likewise.
1768 (indirect_jump_<mode>): Likewise.
1769 (tablejump_<mode>): Likewise.
1770 (<optab>_internal): Likewise.
1771 (sibcall_internal): Likewise.
1772 (sibcall_value_internal): Likewise.
1773 (prefetch): Use constraint ZD.
1774 * config/mips/mips.opt (minterlink-compressed): New option.
1775 (minterlink-mips16): Now an alias for minterlink-compressed.
1776 (mmicromips): New option.
1777 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
1778 (compare_and_swap_12): Likewise.
1779 (sync_add<mode>): Likewise.
1780 (sync_<optab>_12): Likewise.
1781 (sync_old_<optab>_12): Likewise.
1782 (sync_new_<optab>_12): Likewise.
1783 (sync_nand_12): Likewise.
1784 (sync_old_nand_12): Likewise.
1785 (sync_new_nand_12): Likewise.
1786 (sync_sub<mode>): Likewise.
1787 (sync_old_add<mode>): Likewise.
1788 (sync_old_sub<mode>): Likewise.
1789 (sync_new_add<mode>): Likewise.
1790 (sync_new_sub<mode>): Likewise.
1791 (sync_<optab><mode>): Likewise.
1792 (sync_old_<optab><mode>): Likewise.
1793 (sync_new_<optab><mode>): Likewise.
1794 (sync_nand<mode>): Likewise.
1795 (sync_old_nand<mode>): Likewise.
1796 (sync_new_nand<mode>): Likewise.
1797 (sync_lock_test_and_set<mode>): Likewise.
1798 (test_and_set_12): Likewise.
1799 (atomic_compare_and_swap<mode>): Likewise.
1800 (atomic_exchange<mode>_llsc): Likewise.
1801 (atomic_fetch_add<mode>_llsc): Likewise.
1802 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
1803 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
1804 (umips_save_restore_pattern_p): Likewise.
1805 (umips_load_store_pair_p): Likewise.
1806 (umips_output_load_store_pair): Likewise.
1807 (umips_movep_target_p): Likewise.
1808 (umips_12bit_offset_address_p): Likewise.
1809 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
1810 (mips_base_mips16): Rename this...
1811 (mips_base_compression_flags): ...to this. Update all uses.
1812 (mips_attribute_table): Add micromips, nomicromips and nocompression.
1813 (mips_mips16_decl_p): Delete.
1814 (mips_nomips16_decl_p): Delete.
1815 (mips_get_compress_on_flags): New function.
1816 (mips_get_compress_off_flags): New function.
1817 (mips_get_compress_mode): New function.
1818 (mips_get_compress_on_name): New function.
1819 (mips_get_compress_off_name): New function.
1820 (mips_insert_attributes): Support multiple compression types.
1821 (mips_merge_decl_attributes): Likewise.
1822 (umips_12bit_offset_address_p): New function.
1823 (mips_start_function_definition): Emit .set micromips directive.
1824 (mips_call_may_need_jalx_p): New function.
1825 (mips_function_ok_for_sibcall): Add microMIPS support.
1826 (mips_print_operand_punctuation): Support short delay slots and
1828 (umips_swm_mask, umips_swm_encoding): New.
1829 (umips_build_save_restore): New function.
1830 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
1831 (was_mips16_p): Remove.
1832 (old_compression_mode): New.
1833 (mips_set_compression_mode): New function.
1834 (mips_set_current_function): Add microMIPS support.
1835 (mips_option_override): Likewise.
1836 (umips_save_restore_pattern_p): New function.
1837 (umips_output_save_restore): New function.
1838 (umips_load_store_pair_p_1): New function.
1839 (umips_load_store_pair_p): New function.
1840 (umips_output_load_store_pair_1): New function.
1841 (umips_output_load_store_pair): New function.
1842 (umips_movep_target_p) New function.
1843 (mips_prepare_pch_save): Add microMIPS support.
1844 * config/mips/mips.h (TARGET_COMPRESSION): New.
1845 (TARGET_CPU_CPP_BUILTINS): Update macro
1846 to use new compression flags and to support microMIPS.
1847 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
1848 (MIPS_ARCH_FLOAT_SPEC): Likewise.
1849 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
1850 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
1851 (ASM_SPEC): Support mmicromips and mno-micromips.
1852 (M16STORE_REG_P): New macro.
1853 (MIPS_CALL): Support TARGET_MICROMIPS.
1854 (MICROMIPS_J): New macro.
1855 (mips_base_mips16): Rename this...
1856 (mips_base_compression_flags): ...to this.
1857 (UMIPS_12BIT_OFFSET_P): New macro.
1858 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
1859 (MULTILIB_DIRNAMES): Likewise.
1861 2013-02-25 Tom de Vries <tom@codesourcery.com>
1863 PR rtl-optimization/56131
1864 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
1865 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
1866 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
1868 2013-02-25 Tobias Burnus <burnus@net-b.de>
1870 * doc/invoke.texi (-fsanitize=): Move from optimization
1871 to debugging options.
1873 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
1875 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
1877 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
1878 Alexander Monakov <amonakov@ispras.ru>
1881 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
1882 flush pending lists also on non-jumps. Adjust comment.
1884 2013-02-24 Catherine Moore <clm@codesourcery.com>
1885 Maciej W. Rozycki <macro@codesourcery.com>
1886 Tom de Vries <tom@codesourcery.com>
1887 Nathan Sidwell <nathan@codesourcery.com>
1888 Iain Sandoe <iain@codesourcery.com>
1889 Nathan Froyd <froydnj@codesourcery.com>
1890 Chao-ying Fu <fu@mips.com>
1892 * doc/extend.texi: (micromips, nomicromips, nocompression):
1893 Document new function attributes.
1894 * doc/invoke.texi (minterlink-compressed, mmicromips,
1895 m14k, m14ke, m14kec): Document new options.
1896 (minterlink-mips16): Update documentation.
1897 * doc/md.texi (ZC, ZD): Document new constraints.
1898 * configure.ac (gcc_cv_as_micromips): Check if linker
1899 supports the .set micromips directive.
1900 * configure: Regenerate.
1901 * config.in: Regenerate.
1902 * config/mips/mips-tables.opt: Regenerate.
1903 * config/mips/micromips.md: New file.
1904 * constraints.md (ZC, AD): New constraints.
1905 * config/mips/predicates.md (movep_src_register): New predicate.
1906 (movep_src_operand): New predicate.
1907 (non_volatile_mem_operand): New predicate.
1908 * config/mips/mips.md (multimem): New type.
1909 (length): Differentiate between 17-bit and 18-bit branch offsets.
1910 (MOVEP1, MOVEP2): New mode iterator.
1911 (mov_<load>l): Use ZC constraint.
1912 (mov_<load>r): Likewise.
1913 (mov_<store>l): Likewise.
1914 (mov_<store>r): Likewise.
1915 (*branch_equality<mode>_inverted): Add microMIPS support.
1916 (*branch_equality<mode>): Likewise.
1917 (*jump_absolute): Likewise.
1918 (indirect_jump_<mode>): Likewise.
1919 (tablejump_<mode>): Likewise.
1920 (<optab>_internal): Likewise.
1921 (sibcall_internal): Likewise.
1922 (sibcall_value_internal): Likewise.
1923 (prefetch): Use constraint ZD.
1924 * config/mips/mips.opt (minterlink-compressed): New option.
1925 (minterlink-mips16): Now an alias for minterlink-compressed.
1926 (mmicromips): New option.
1927 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
1928 (compare_and_swap_12): Likewise.
1929 (sync_add<mode>): Likewise.
1930 (sync_<optab>_12): Likewise.
1931 (sync_old_<optab>_12): Likewise.
1932 (sync_new_<optab>_12): Likewise.
1933 (sync_nand_12): Likewise.
1934 (sync_old_nand_12): Likewise.
1935 (sync_new_nand_12): Likewise.
1936 (sync_sub<mode>): Likewise.
1937 (sync_old_add<mode>): Likewise.
1938 (sync_old_sub<mode>): Likewise.
1939 (sync_new_add<mode>): Likewise.
1940 (sync_new_sub<mode>): Likewise.
1941 (sync_<optab><mode>): Likewise.
1942 (sync_old_<optab><mode>): Likewise.
1943 (sync_new_<optab><mode>): Likewise.
1944 (sync_nand<mode>): Likewise.
1945 (sync_old_nand<mode>): Likewise.
1946 (sync_new_nand<mode>): Likewise.
1947 (sync_lock_test_and_set<mode>): Likewise.
1948 (test_and_set_12): Likewise.
1949 (atomic_compare_and_swap<mode>): Likewise.
1950 (atomic_exchange<mode>_llsc): Likewise.
1951 (atomic_fetch_add<mode>_llsc): Likewise.
1952 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
1953 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
1954 (umips_save_restore_pattern_p): Likewise.
1955 (umips_load_store_pair_p): Likewise.
1956 (umips_output_load_store_pair): Likewise.
1957 (umips_movep_target_p): Likewise.
1958 (umips_12bit_offset_address_p): Likewise.
1959 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
1960 (mips_base_mips16): Rename this...
1961 (mips_base_compression_flags): ...to this. Update all uses.
1962 (mips_attribute_table): Add micromips, nomicromips and nocompression.
1963 (mips_mips16_decl_p): Delete.
1964 (mips_nomips16_decl_p): Delete.
1965 (mips_get_compress_on_flags): New function.
1966 (mips_get_compress_off_flags): New function.
1967 (mips_get_compress_mode): New function.
1968 (mips_get_compress_on_name): New function.
1969 (mips_get_compress_off_name): New function.
1970 (mips_insert_attributes): Support multiple compression types.
1971 (mips_merge_decl_attributes): Likewise.
1972 (umips_12bit_offset_address_p): New function.
1973 (mips_start_function_definition): Emit .set micromips directive.
1974 (mips_call_may_need_jalx_p): New function.
1975 (mips_function_ok_for_sibcall): Add microMIPS support.
1976 (mips_print_operand_punctuation): Support short delay slots and
1978 (umips_swm_mask, umips_swm_encoding): New.
1979 (umips_build_save_restore): New function.
1980 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
1981 (was_mips16_p): Remove.
1982 (old_compression_mode): New.
1983 (mips_set_compression_mode): New function.
1984 (mips_set_current_function): Add microMIPS support.
1985 (mips_option_override): Likewise.
1986 (umips_save_restore_pattern_p): New function.
1987 (umips_output_save_restore): New function.
1988 (umips_load_store_pair_p_1): New function.
1989 (umips_load_store_pair_p): New function.
1990 (umips_output_load_store_pair_1): New function.
1991 (umips_output_load_store_pair): New function.
1992 (umips_movep_target_p) New function.
1993 (mips_prepare_pch_save): Add microMIPS support.
1994 * config/mips/mips.h (TARGET_COMPRESSION): New.
1995 (TARGET_CPU_CPP_BUILTINS): Update macro
1996 to use new compression flags and to support microMIPS.
1997 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
1998 (MIPS_ARCH_FLOAT_SPEC): Likewise.
1999 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
2000 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
2001 (ASM_SPEC): Support mmicromips and mno-micromips.
2002 (M16STORE_REG_P): New macro.
2003 (MIPS_CALL): Support TARGET_MICROMIPS.
2004 (MICROMIPS_J): New macro.
2005 (mips_base_mips16): Rename this...
2006 (mips_base_compression_flags): ...to this.
2007 (UMIPS_12BIT_OFFSET_P): New macro.
2008 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
2009 (MULTILIB_DIRNAMES): Likewise.
2011 2013-02-24 Jakub Jelinek <jakub@redhat.com>
2014 * target-globals.c (save_target_globals): For init_reg_sets and
2015 target_reinit remporarily set this_fn_optabs to this_target_optabs.
2017 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
2019 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
2020 * config/aarch64/t-aarch64
2021 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
2023 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
2026 * lra-constraints.c (process_alt_operands): Reload operand
2027 conflicting with earlier clobber only if no more other conflicting
2030 2013-02-22 Jakub Jelinek <jakub@redhat.com>
2033 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
2034 if not linking a shared library.
2036 2013-02-22 Seth LaForge <sethml@google.com>
2038 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
2040 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
2042 * config/arm/arm.md (split for extendsidi): Update condition.
2043 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
2044 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
2045 (qhs_zextenddi_cstr): Likewise.
2047 2013-02-21 Jakub Jelinek <jakub@redhat.com>
2050 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
2051 avoid signed wrapping.
2052 (expand_mult): Handle properly multiplication by
2053 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
2054 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
2055 in the compiler if coeff is HOST_WIDE_INT_MIN.
2056 (expand_divmod): Don't make ext_op1 static, change it's type to
2057 uhwi. Avoid undefined behavior in -INTVAL (op1).
2059 PR rtl-optimization/50339
2060 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
2062 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
2063 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
2064 into splitting_ashiftrt field.
2065 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
2067 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
2070 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
2073 * trans-mem.c (execute_tm_mark): Do not expand transactions that
2074 are sure to go irrevocable.
2076 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
2078 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
2079 scalars are valid operands.
2081 2013-02-21 Martin Jambor <mjambor@suse.cz>
2083 PR tree-optimization/56310
2084 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
2085 only matching indices and non-negative final offsets.
2086 (intersect_aggregates_with_edge): Pass src_idx to
2087 agg_replacements_to_vector. Pass src_idx insstead of index to
2088 intersect_with_agg_replacements.
2090 2013-02-21 Martin Jambor <mjambor@suse.cz>
2092 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
2093 instead of hard-wired defaults.
2095 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
2097 * doc/invoke.texi (MIPS Options): Update documentation of the
2098 floating-point multiply-accumulate instruction restrictions.
2100 2013-02-21 Kostya Serebryany <kcc@google.com>
2102 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
2103 asan_shadow_offset on x86_64 linux.
2105 2013-02-21 Richard Biener <rguenther@suse.de>
2107 PR tree-optimization/56415
2109 2013-02-11 Richard Biener <rguenther@suse.de>
2111 PR tree-optimization/56273
2112 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
2115 2013-02-21 Jakub Jelinek <jakub@redhat.com>
2118 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
2122 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
2123 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
2125 2013-02-20 Jan Hubicka <jh@suse.cz>
2127 PR tree-optimization/56265
2128 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
2129 when target is referenced for first time.
2131 2013-02-20 Richard Biener <rguenther@suse.de>
2133 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
2134 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
2135 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
2136 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
2137 not return anything.
2138 (rename_ssa_copies): Do not remove unused locals.
2139 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
2140 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
2141 * passes.c (execute_function_todo): Do not schedule unused locals
2142 removal if cleanup_tree_cfg did something.
2143 * tree-ssa-live.c (remove_unused_locals): Dump statistics
2144 about the number of removed locals.
2146 2013-02-20 Richard Biener <rguenther@suse.de>
2148 PR tree-optimization/56398
2149 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
2151 2013-02-20 Martin Jambor <mjambor@suse.cz>
2153 PR tree-optimization/55334
2154 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
2155 restricted pointers to arrays.
2157 2013-02-20 Richard Biener <rguenther@suse.de>
2158 Jakub Jelinek <jakub@redhat.com>
2160 PR tree-optimization/56396
2161 * tree-ssa-ccp.c (n_const_val): New static variable.
2162 (get_value): Return NULL for SSA names we don't have a lattice
2164 (ccp_initialize): Initialize n_const_val.
2165 * tree-ssa-copy.c (n_copy_of): New static variable.
2166 (init_copy_prop): Initialize n_copy_of.
2167 (get_value): Return NULL_TREE for SSA names we don't have a
2170 2013-02-20 Martin Jambor <mjambor@suse.cz>
2172 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
2174 2013-02-20 Richard Biener <rguenther@suse.de>
2176 * genpreds.c (write_lookup_constraint): Do not compare first
2177 letter of the constraint again.
2179 2013-02-20 Richard Biener <rguenther@suse.de>
2181 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
2183 (get_use_iv_cost): Terminate hashtable walk when coming across
2186 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
2188 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
2189 reassociation for avx2 targets.
2191 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2193 * config/microblaze/microblaze.c: microblaze_has_clz = 0
2194 Add version check for v8.10.a to enable microblaze_has_clz
2195 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
2196 version and TARGET_PATTERN_COMPARE check
2197 * config/microblaze/microblaze.md: New clzsi2 instruction
2199 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2201 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
2202 function before branching.
2204 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
2206 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
2208 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
2210 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
2213 * sel-sched.c: Include ira.h.
2214 (implicit_clobber_conflict_p): New function.
2215 (moveup_expr): Use it.
2216 * Makefile.in (sel-sched.o): Depend on ira.h.
2218 2013-02-19 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/56384
2221 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
2222 (vn_hash_type): Split out from ...
2223 (vn_hash_constant_with_type): ... here.
2224 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
2225 (vn_phi_eq): Compare types from vn_phi_s structure.
2226 (vn_phi_lookup): Populate vn_phi_s type.
2227 (vn_phi_insert): Likewise.
2229 2013-02-19 Jakub Jelinek <jakub@redhat.com>
2231 PR tree-optimization/56350
2232 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
2233 if haven't found reduction or nested cycle operand, rather than
2234 asserting we must find it.
2236 PR tree-optimization/56381
2237 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
2240 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
2241 Jakub Jelinek <jakub@redhat.com>
2244 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
2245 (swap_optab_enable): Same.
2246 (init_all_optabs): Use argument instead of global.
2247 * tree.h (struct tree_optimization_option): New field target_optabs.
2248 * expr.h (init_all_optabs): Add argument to prototype.
2249 (TREE_OPTIMIZATION_OPTABS): New.
2250 (save_optabs_if_changed): Protoize.
2251 * optabs.h: Declare this_fn_optabs.
2252 * optabs.c (save_optabs_if_changed): New.
2253 Declare this_fn_optabs.
2254 (init_optabs): Add argument to init_all_optabs() call.
2255 * function.c (invoke_set_current_function_hook): Handle per
2257 * function.h (struct function): New field optabs.
2258 * config/mips/mips.c (mips_set_mips16_mode): Handle when
2259 optimization_current_node has changed.
2260 * target-globals.h (save_target_globals_default_opts): Protoize.
2261 * target-globals.c (save_target_globals_default_opts): New.
2263 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2266 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
2267 registers %fr12 and %fr12R as call used.
2270 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
2271 and HImode, require all displacements to be an integer multiple of
2273 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
2274 only allow QImode and HImode when reload is in progress and strict is
2275 true. Likewise for symbolic addresses. Use base14_operand to check
2276 displacements in REG+BASE addresses.
2278 2013-02-18 Richard Biener <rguenther@suse.de>
2280 PR tree-optimization/56366
2281 * tree-vect-loop.c (get_initial_def_for_induction): Properly
2282 handle sign-conversion of outer-loop initial induction value.
2284 2013-02-18 Richard Biener <rguenther@suse.de>
2287 * cfghooks.c (merge_blocks): If we merge a latch into another
2288 block adjust references to it.
2289 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
2290 (verify_loop_structure): Verify that a recorded latch is in fact
2293 2013-02-18 Richard Biener <rguenther@suse.de>
2295 PR tree-optimization/56321
2296 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
2297 order SSA name release and virtual operand unlinking.
2299 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2301 * config/microblaze/microblaze.md (save_stack_block): Define.
2302 (restore_stack_block): Likewise.
2304 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2306 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
2307 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
2308 * config/microblaze/microblaze.c (microblaze_option_override):
2309 Bail out early for PIC modes when target does not support PIC.
2311 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2313 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
2314 Replace with a microblaze version.
2315 (microblaze_trampoline_init): Adapt for microblaze.
2316 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
2319 2013-02-16 Jakub Jelinek <jakub@redhat.com>
2320 Dodji Seketeli <dodji@redhat.com>
2323 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
2324 (instrument_mem_region_access): Do not forget to always put
2325 instrumentation of the of 'base' and 'base + len' in a "if (len !=
2326 0) statement, even for cases where either 'base' or 'base + len'
2327 are not instrumented -- because they have been previously
2328 instrumented. Simplify the logic by putting all the statements
2329 instrument 'base + len' inside a sequence, and then insert that
2330 sequence right before the current insertion point. Then, to
2331 instrument 'base + len', just get an iterator on that statement.
2332 And do not forget to update the pointer to iterator the function
2333 received as argument.
2335 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
2337 PR rtl-optimization/56348
2338 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
2340 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
2342 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
2343 (clean_graph_dump_file): Pass base to start_graph_dump.
2345 2013-02-14 Richard Henderson <rth@redhat.com>
2348 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
2350 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
2352 * collect2-aix.h: Define F_LOADONLY.
2354 2013-02-14 Richard Biener <rguenther@suse.de>
2357 * varasm.c (output_constant_def_1): Get the decl representing
2358 the constant as argument.
2359 (output_constant_def): Wrap output_constant_def_1.
2360 (make_decl_rtl): Use output_constant_def_1 with the decl
2361 representing the constant.
2362 (build_constant_desc): Optionally re-use a decl already
2363 representing the constant.
2364 (tree_output_constant_def): Adjust.
2366 2013-02-14 Dodji Seketeli <dodji@redhat.com>
2369 * asan.c (instrument_builtin_call): Really put the length of the
2370 second source argument into src1_len.
2372 2013-02-13 Jakub Jelinek <jakub@redhat.com>
2374 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
2375 argument. If it is false, don't create edge from then_bb to
2377 (insert_if_then_before_iter): Pass true to it.
2378 (build_check_stmt): Pass false to it.
2379 (transform_statements): Flush hash table only on extended basic
2380 block boundaries, rather than at the beginning of every bb.
2381 Don't flush hash table on nonfreeing_call_p calls.
2382 * tree-flow.h (nonfreeing_call_p): New prototype.
2383 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
2385 2013-02-13 David S. Miller <davem@davemloft.net>
2387 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
2389 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
2392 * ira.c (max_regno_before_ira): Move from ...
2394 (fix_reg_equiv_init): Use max_regno_before_ira instead of
2397 2013-02-13 Jakub Jelinek <jakub@redhat.com>
2399 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
2401 2013-02-13 Richard Biener <rguenther@suse.de>
2404 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
2405 globals in MEM_REFs.
2407 2013-02-13 Richard Biener <rguenther@suse.de>
2409 * loop-init.c (loop_optimizer_init): Clear loop state when
2410 re-initializing preserved loops.
2411 * loop-unswitch.c (unswitch_single_loop): Return whether
2412 we unswitched the loop. Do not verify loop state here.
2413 (unswitch_loops): When we unswitched a loop discover new loops.
2415 2013-02-13 Kostya Serebryany <kcc@google.com>
2417 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
2419 * sanitizer.def: Rename __asan_init to __asan_init_v1.
2421 2013-02-12 Dodji Seketeli <dodji@redhat.com>
2423 Avoid instrumenting duplicated memory access in the same basic block
2424 * Makefile.in (asan.o): Add new dependency on hash-table.h
2425 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
2426 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
2427 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
2428 (free_mem_ref_resources, has_mem_ref_been_instrumented)
2429 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
2430 (get_mem_ref_of_assignment): New functions.
2431 (get_mem_refs_of_builtin_call): Extract from
2432 instrument_builtin_call and tweak a little bit to make it fit with
2434 (instrument_builtin_call): Use the new
2435 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
2436 of is_gimple_builtin_call.
2437 (instrument_derefs, instrument_mem_region_access): Insert the
2438 instrumented memory reference into the hash table.
2439 (maybe_instrument_assignment): Renamed instrument_assignment into
2440 this, and change it to advance the iterator when instrumentation
2441 actually happened and return true in that case. This makes it
2442 homogeneous with maybe_instrument_assignment, and thus give a
2443 chance to callers to be more 'regular'.
2444 (transform_statements): Clear the memory reference hash table
2445 whenever we enter a new BB, when we cross a function call, or when
2446 we are done transforming statements. Use
2447 maybe_instrument_assignment instead of instrumentation. No more
2448 need to special case maybe_instrument_assignment and advance the
2449 iterator after calling it; it's now handled just like
2450 maybe_instrument_call. Update comment.
2452 2013-02-13 Richard Biener <rguenther@suse.de>
2454 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
2455 Fix loop discovery code.
2457 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
2460 * lra-constraints.c (process_alt_operands): Match early clobber
2461 operand with itself. Check conflicts with earlyclobber only if
2462 the operand is not reloaded. Prefer to reload conflicting operand
2463 if earlyclobber and matching operands are the same.
2465 2013-02-12 Richard Biener <rguenther@suse.de>
2468 * lto-streamer-out.c (write_symbol): Do not output symbols
2469 for hard register variables.
2471 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
2474 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
2475 (umulsidi3_insn, mulsidi3_insn): New insns.
2477 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
2479 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
2480 (struct tune_params): Add vec_costs field.
2481 * config/arm/arm.c (arm_builtin_vectorization_cost)
2482 (arm_add_stmt_cost): New functions.
2483 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
2484 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
2485 (arm_default_vec_cost): New struct of type cpu_vec_costs.
2486 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
2487 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
2488 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
2489 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
2491 2013-02-12 Richard Biener <rguenther@suse.de>
2494 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
2495 decls again if possible.
2497 2013-02-12 Richard Biener <rguenther@suse.de>
2500 * tree-ssa.c (verify_ssa_name): Fix check, move
2501 SSA_NAME_IN_FREE_LIST check up.
2503 2013-02-12 Jakub Jelinek <jakub@redhat.com>
2504 Steven Bosscher <steven@gcc.gnu.org>
2506 PR rtl-optimization/56151
2507 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
2508 equal to op0 or op1, and last_insn pattern is CODE operation
2509 with MEM dest and one of the operands matches that MEM.
2511 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
2513 * doc/extend.texi: Document Function Multiversioning and "default"
2514 parameter string to target attribute.
2515 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
2516 target attribute parameter is "default".
2517 (ix86_compare_version_priority): Remove checks for target attribute.
2518 (ix86_mangle_function_version_assembler_name): Change error to sorry.
2519 Remove check for target attribute equal to NULL. Add assert.
2520 (ix86_generate_version_dispatcher_body): Change error to sorry.
2522 2013-02-11 Iain Sandoe <iain@codesourcery.com>
2523 Jack Howarth <howarth@bromo.med.uc.edu>
2524 Patrick Marlier <patrick.marlier@gmail.com>
2527 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
2528 define ENDFILE_SPEC as TM_DESTRUCTOR.
2529 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
2531 2013-02-11 Alexander Potapenko <glider@google.com>
2532 Jack Howarth <howarth@bromo.med.uc.edu>
2533 Jakub Jelinek <jakub@redhat.com>
2536 * config/darwin.c (cdtor_record): Rename ctor_record.
2537 (sort_cdtor_records): Rename sort_ctor_records.
2538 (finalize_dtors): New routine to sort destructors by
2539 priority before use in assemble_integer.
2540 (machopic_asm_out_destructor): Use finalize_dtors if needed.
2542 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
2544 PR rtl-optimization/56275
2545 * simplify-rtx.c (avoid_constant_pool_reference): Check that
2546 offset is non-negative and less than cmode size before
2547 calling simplify_subreg.
2549 2013-02-11 Richard Biener <rguenther@suse.de>
2551 PR tree-optimization/56264
2552 * cfgloop.h (fix_loop_structure): Adjust prototype.
2553 * loop-init.c (fix_loop_structure): Return the number of
2554 newly discovered loops.
2555 * tree-cfgcleanup.c (repair_loop_structures): When new loops
2556 are discovered, do a full loop-closed SSA rewrite.
2558 2013-02-11 Richard Biener <rguenther@suse.de>
2560 PR tree-optimization/56273
2561 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
2563 (check_array_ref): Fix missing newline in dumps.
2564 (search_for_addr_array): Likewise.
2566 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
2568 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
2570 2013-02-09 Jakub Jelinek <jakub@redhat.com>
2573 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
2575 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
2577 PR rtl-optimization/56246
2578 * lra-constraints.c (simplify_operand_subreg): Try to reuse
2580 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
2581 constraints are satisfied.
2583 2013-02-08 Jeff Law <law@redhat.com>
2586 * emit-rtl.c (reg_is_parm_p): New function.
2587 * regs.h (reg_is_parm_p): New prototype.
2588 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
2589 callee-clobbered registers.
2591 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
2594 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
2595 If there is no implicit builtin declaration, just return NULL.
2597 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2599 * config/i386/sse.md (FMAMODEM): New mode iterator.
2600 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
2601 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
2603 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2605 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
2606 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
2607 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
2609 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2611 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
2612 (microblaze*-*-elf): Likewise.
2613 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
2615 * config/microblaze/microblaze-c.c: Add builtin defines for
2616 _LITTLE_ENDIAN and _BIG_ENDIAN.
2617 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
2618 add to TARGET_DEFAULT flags.
2619 Expand ASM_SPEC and LINK_SPEC.
2620 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
2621 * config/microblaze/microblaze.md: Update extendsidi2 and
2622 movdi_internal instructions to use low-order / high-order reg
2624 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
2625 options and inversemask / mask of LITTLE_ENDIAN.
2626 * config/microblaze/t-microblaze: Expand multilib options to
2627 include mlittle-endian (le) and update exceptions patterns.
2629 2013-02-08 Jakub Jelinek <jakub@redhat.com>
2631 PR rtl-optimization/56195
2632 * lra-constraints.c (get_reload_reg): Don't reuse regs
2633 if they have smaller mode than requested, if they have
2634 wider mode than requested, try to return a SUBREG.
2636 PR tree-optimization/56250
2637 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
2638 if type is unsigned and code isn't MULT_EXPR.
2640 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
2642 PR tree-optimization/56064
2643 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
2644 bits according to mode.
2645 * fixed-value.h (fixed_from_double_int)
2646 (const_fixed_from_double_int): Adjust comments.
2648 2013-02-08 Richard Biener <rguenther@suse.de>
2651 * lto-streamer.h (struct data_in): Remove current_file, current_line
2652 and current_col members.
2653 * lto-streamer-out.c (lto_output_location): Stream changed bits
2654 en-block for efficiency.
2655 * lto-streamer-in.c (clear_line_info): Remove.
2656 (lto_input_location): Cache current file, line and column
2657 globally via local statics. Read changed bits en-block.
2658 (input_function): Do not call clear_line_info.
2659 (lto_read_body): Likewise.
2660 (lto_input_toplevel_asms): Likewise.
2662 2013-02-08 Michael Matz <matz@suse.de>
2664 PR tree-optimization/52448
2665 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
2666 (nt_call_phase): New static.
2667 (add_or_mark_expr): Only mark accesses with newer phase than any
2669 (nonfreeing_call_p): New.
2670 (nt_init_block): Update nt_call_phase, mark blocks as visited.
2671 (nt_fini_block): Keep blocks marked as visited.
2672 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
2674 2013-02-08 Richard Biener <rguenther@suse.de>
2676 * ira.c (ira): Free broken dominator information.
2678 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2680 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
2682 2013-02-08 Marek Polacek <polacek@redhat.com>
2684 * cfgloop.c (verify_loop_structure): Add more checking of headers.
2686 2013-02-08 Richard Biener <rguenther@suse.de>
2689 * cfgloop.h (flow_loops_find): Adjust.
2690 (bb_loop_header_p): Declare.
2691 * cfgloop.c (bb_loop_header_p): New function split out from ...
2692 (flow_loops_find): ... here. Adjust function signature,
2693 support incremental loop structure update.
2694 (verify_loop_structure): Cleanup. Verify a loop is a loop.
2695 * cfgloopmanip.c (fix_loop_structure): Move ...
2696 * loop-init.c (fix_loop_structure): ... here.
2697 (apply_loop_flags): Split out from ...
2698 (loop_optimizer_init): ... here.
2699 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
2700 in incremental mode, only remove dead loops here.
2702 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
2705 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
2706 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
2707 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
2708 (*round<mode>3.libgcc): New insns for fixed-modes.
2709 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
2710 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
2711 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
2712 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
2713 implementations. Define to __builtin_avr_absFX,
2714 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
2715 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
2716 __builtin_avr_countlsFX, respectively.
2717 * config/avr/avr-c.c (target.h): Include it.
2718 (enum avr_builtin_id): New enum.
2719 (avr_resolve_overloaded_builtin): New static function.
2720 (avr_register_target_pragmas): Use it to set
2721 targetm.resolve_overloaded_builtin.
2722 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
2723 tree nodes used by DEF_BUILTIN.
2724 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
2725 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
2726 <AVR_BUILTIN_xxBITS>: Same.
2728 2013-02-08 Richard Biener <rguenther@suse.de>
2730 * cfgloop.c (verify_loop_structure): Properly handle
2731 a loop exiting to another loop header.
2732 * ira-int.h (ira_loops): Remove.
2733 * ira.c (ira_loops): Remove.
2734 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
2735 (do_reload): Use loop_optimizer_finalize.
2736 * ira-build.c (create_loop_tree_nodes): Use get_loops and
2737 number_of_loops to access the loop tree.
2738 (more_one_region_p): Likewise.
2739 (finish_loop_tree_nodes): Likewise.
2740 (rebuild_regno_allocno_maps): Likewise.
2741 (mark_loops_for_removal): Likewise.
2742 (mark_all_loops_for_removal): Likewise.
2743 (remove_unnecessary_regions): Likewise.
2744 (ira_build): Likewise.
2745 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
2747 2013-02-08 Richard Biener <rguenther@suse.de>
2749 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
2750 * ipa-pure-const.c (analyze_function): Avoid calling
2751 mark_irreducible_loops twice.
2752 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
2754 2013-02-07 David S. Miller <davem@davemloft.net>
2756 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
2758 * var-tracking.c (vt_add_function_parameter): Test the presence of
2759 HAVE_window_save properly and do not remap argument registers when
2760 we have a leaf function.
2762 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
2765 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
2767 * config/i386/i386.c (ix86_print_operand): Ditto.
2769 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
2771 * lra-constraints.c (process_alt_operands): Fix recently added comment.
2773 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
2775 PR rtl-optimization/56225
2776 * lra-constraints.c (process_alt_operands): Check that reload hard
2777 reg can hold value for strict_low_part.
2779 2013-02-07 Jakub Jelinek <jakub@redhat.com>
2782 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
2783 dwarf2out_end_function.
2784 (in_first_function_p, maybe_at_text_label_p,
2785 first_loclabel_num_not_at_text_label): New variables.
2786 (dwarf2out_var_location): In the first function find out
2787 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
2788 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
2791 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
2793 PR rtl-optimization/56178
2794 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
2795 SUBREG of a register. Tidy up related block of code.
2796 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
2797 note if the source is a register or a SUBREG of a register.
2799 2013-02-07 Jakub Jelinek <jakub@redhat.com>
2802 * config/rs6000/rs6000.md (ptrm): New mode attr.
2803 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
2804 call_value_indirect_aix<pttrsize>,
2805 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
2808 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
2810 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
2811 if -bnortl. Convert to strcmp and strncmp.
2813 2013-02-07 Alan Modra <amodra@gmail.com>
2816 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
2817 addresses won't wrap when offsetting.
2818 (rs6000_secondary_reload): Provide secondary reloads needed for
2819 wrapping LO_SUM addresses.
2821 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
2823 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
2824 MACH, just __MACH__.
2826 2013-02-06 Richard Biener <rguenther@suse.de>
2828 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
2829 instead of calling fix_loop_structure.
2831 2013-02-06 Jakub Jelinek <jakub@redhat.com>
2834 * omp-low.c (use_pointer_for_field): Return false if
2835 lower_send_shared_vars doesn't generate any copy-out code.
2837 2013-02-06 Tom de Vries <tom@codesourcery.com>
2839 PR rtl-optimization/56131
2840 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
2841 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
2842 of the label is NULL. Add comment.
2844 2013-02-05 Jakub Jelinek <jakub@redhat.com>
2846 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
2849 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
2850 (STATIC_LIBTSAN_LIBS): Likewise.
2851 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
2852 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
2853 is defined, don't add anything else beyond that.
2854 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
2855 (LINK_COMMAND_SPEC): Use them.
2857 PR tree-optimization/56205
2858 * tree-stdarg.c (check_all_va_list_escapes): Return true if
2859 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
2860 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
2862 2013-02-05 Richard Biener <rguenther@suse.de>
2864 PR tree-optimization/53342
2865 PR tree-optimization/53185
2866 * tree-vectorizer.h (vect_check_strided_load): Remove.
2867 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
2868 not disallow peeling for vectorized strided loads.
2869 (vect_check_strided_load): Make static and simplify.
2870 (vect_analyze_data_refs): Adjust.
2871 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
2872 correctly when vectorizing strided loads.
2874 2013-02-05 Richard Biener <rguenther@suse.de>
2876 * doc/install.texi: Refer to ISL, not PPL.
2878 2013-02-05 Jan Hubicka <jh@suse.cz>
2880 PR tree-optimization/55789
2881 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
2883 2013-02-05 Jan Hubicka <jh@suse.cz>
2885 PR tree-optimization/55789
2886 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
2887 the dead call anyway.
2889 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
2892 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
2894 2013-02-04 Alexander Potapenko <glider@google.com>
2895 Jack Howarth <howarth@bromo.med.uc.edu>
2896 Jakub Jelinek <jakub@redhat.com>
2899 * config/darwin.c (sort_ctor_records): Stabilized qsort
2900 on constructor priority by using original position.
2901 (finalize_ctors): New routine to sort constructors by
2902 priority before use in assemble_integer.
2903 (machopic_asm_out_constructor): Use finalize_ctors if needed.
2905 2013-02-04 Jakub Jelinek <jakub@redhat.com>
2908 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
2909 about visibility on artificial decls.
2910 * config/sol2.c (solaris_assemble_visibility): Likewise.
2912 2013-02-04 Kai Tietz <ktietz@redhat.com>
2915 * config/i386/i386.c (function_value_ms_64): Add additional valtype
2916 argument and improve checking of return-argument types for 16-byte
2918 (ix86_function_value_1): Add additional valtype argument on call
2919 of function_value_64.
2920 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
2921 handling infunction_value_64 function.
2923 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2925 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
2927 2013-02-04 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/56188
2930 * tree-ssa-structalias.c (label_visit): Consider case with
2931 initially non-empty points-to set.
2932 (perform_var_substitution): Dump node mapping and clean up.
2934 2013-02-04 Richard Guenther <rguenther@suse.de>
2937 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
2938 node prevail as last resort.
2939 (lto_symtab_merge_decls): Remove guard on LTRANS here.
2940 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
2942 2013-02-04 Richard Biener <rguenther@suse.de>
2944 PR tree-optimization/56113
2945 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
2947 (equiv_class_lookup_or_add): ... this.
2948 (label_visit): Adjust and fix error in previous patch.
2949 (perform_var_substitution): Adjust.
2951 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
2953 * config/sh/divtab.c: Fix formatting and comments throughout the file.
2954 * config/sh/sh4-300.md: Likewise.
2955 * config/sh/sh4a.md: Likewise.
2956 * config/sh/constraints.md: Likewise.
2957 * config/sh/sh.md: Likewise.
2958 * config/sh/netbsd-elf.h: Likewise.
2959 * config/sh/predicates.md: Likewise.
2960 * config/sh/sh-protos.h: Likewise.
2961 * config/sh/ushmedia.h: Likewise.
2962 * config/sh/linux.h: Likewise.
2963 * config/sh/sh.c: Likewise.
2964 * config/sh/superh.h: Likewise.
2965 * config/sh/elf.h: Likewise.
2966 * config/sh/sh4.md: Likewise.
2967 * config/sh/sh.h: Likewise.
2969 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2971 * config/pa/constraints.md: Adjust unused letters. Change "T"
2972 constraint to match_test floating_point_store_memory_operand().
2973 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
2974 (base14_operand): New.
2975 (floating_point_store_memory_operand): New.
2976 (integer_store_memory_operand): Revise to use base14_operand and
2977 reg_plus_base_memory_operand.
2978 (move_dest_operand): Allow symbolic_memory_operands.
2979 (symbolic_memory_operand): Check for LO_SOM.
2980 (symbolic_operand): Change default case to break.
2981 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
2982 CONST_DOUBLE values to be reloaded by putting them into memory when
2983 the destination is a floating point register.
2984 (movdf): Remove code to handle CONST_DOUBLE.
2986 (reload_indf_r1): New.
2987 (reload_insf_r1): New.
2988 Consistently use "Q" and "T" constraints with integer and floating
2989 point move instructions, respectively.
2990 (movdi): Remove FAIL.
2991 Change predicate for source operand unamed DImode move from
2992 general_operand to move_src_operand.
2993 (umulsidi3): Change predicate for destination operand to
2995 Likewise for similar unamed patterns.
2996 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
2997 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
2998 (hppa_legitimize_address): Simplify mask calculation.
2999 (pa_emit_move_sequence): Revised handling of secondary reloads from
3000 REG+D addresses for floating point loads and stores. Directly handle
3001 loading CONST0_RTX (mode) to a floating point register.
3002 (pa_secondary_reload): Handle reloading DF and SFmode constant values
3003 to floating point registers. Don't restrict secondary reloads to
3004 floating point registers to integer modes. Revise some comments and
3006 (TARGET_LEGITIMATE_ADDRESS_P): Define.
3007 (pa_legitimate_address_p): New.
3008 (pa_legitimize_reload_address): New.
3009 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
3010 (STRICT_REG_OK_FOR_BASE_P): New.
3011 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
3012 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
3014 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
3015 Andrew Dixie <andrewd@gentrack.com>
3017 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
3020 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
3022 * expmed.c (extract_bit_field_1): Pass the full width of the
3023 structure to get_best_reg_extraction_insn.
3025 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
3028 * configure.ac (use_cxa_atexit): Add AIX.
3029 * configure: Regenerate.
3031 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
3033 2013-02-01 Jakub Jelinek <jakub@redhat.com>
3036 * final.c (need_profile_function): New variable.
3037 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
3038 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
3039 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
3040 notes, targetm.asm_out.function_prologue doesn't emit anything,
3041 HAVE_prologue and profiler should be emitted before prologue,
3042 set need_profile_function instead of emitting it.
3043 (final_scan_insn): If need_profile_function, emit
3044 profile_function on the first NOTE_INSN_BASIC_BLOCK or
3045 NOTE_INSN_FUNCTION_BEG note.
3047 2013-02-01 Richard Henderson <rth@redhat.com>
3049 * config/rs6000/rs6000.md (smulditi3): New.
3052 * config/alpha/alpha.md (umulditi3): New.
3054 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
3056 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
3057 (ASM_OUTPUT_ALIGNED_LOCAL): New.
3059 2013-02-01 Richard Biener <rguenther@suse.de>
3061 PR tree-optimization/56113
3062 * tree-ssa-structalias.c (label_visit): Reduce work for
3063 single-predecessor nodes.
3065 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
3067 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
3068 range isn't testing for zero.
3070 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
3073 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
3075 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
3076 Nick Clifton <nickc@redhat.com>
3078 * config/v850/constraints.md (Q): Define as a memory constraint.
3079 * config/v850/predicates.md (label_ref_operand): New predicate.
3080 (e3v5_shift_operand): New predicate.
3081 (ior_operator): New predicate.
3082 * config/v850/t-v850: Add e3v5 multilib.
3083 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
3084 (v850_gen_movdi): Prototype.
3085 * config/v850/v850.c: Add support for e3v5 architecture.
3086 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
3088 (construct_save_jarl): Add e3v5 long JARL support.
3089 (v850_adjust_insn_length): New function. Adjust length of call
3090 insns when using e3v5 instructions.
3091 (v850_gen_movdi): New function: Generate instructions to move a
3093 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
3094 (CPP_SPEC): Define __v850e3v5__ as appropriate.
3095 (TARGET_USE_FPU): Enable for e3v5.
3096 (CONST_OK_FOR_W): New macro.
3097 (ADJUST_INSN_LENGTH): Define.
3098 * config/v850/v850.md (UNSPEC_LOOP): Define.
3099 (attr cpu): Add v850e3v5.
3100 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
3101 (movdi): New pattern.
3102 (movdi_internal): New pattern.
3103 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
3104 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
3105 (cstoresf4): Likewise.
3106 (cstoredf4): Likewise.
3107 (insv): New pattern.
3108 (rotlso3_a): New pattern.
3109 (rotlsi3_b): New pattern
3110 (rotlsi3_v850e3v5): New pattern.
3111 (doloop_begin): New pattern.
3112 (fix_loop_counter): New pattern.
3113 (doloop_end): New pattern.
3114 (branch_normal): Add e3v5 long branch support.
3115 (branch_invert): Likewise.
3116 (branch_z_normal): Likewise.
3117 (branch_z_invert): Likewise.
3118 (branch_nz_normal): Likewise.
3119 (branch_nz_invert): Likewise.
3120 (call_internal_short): Add e3v5 register-indirect JARL support.
3121 (call_internal_long): Likewise.
3122 (call_value_internal_short): Likewise.
3123 (call_value_internal_long): Likewise.
3124 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
3125 (mloop): New option.
3126 * config.gcc: Add support for configuring v840e3v5 target.
3127 * doc/invoke.texi: Document new v850 specific command line options.
3129 2013-01-31 Paul Koning <ni1d@arrl.net>
3133 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
3134 children if parent is a class.
3135 (prune_unused_types_prune): Don't add DW_AT_declaration.
3137 2013-01-31 Richard Biener <rguenther@suse.de>
3139 PR tree-optimization/56157
3140 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
3141 match up operand with SLP child.
3143 2013-01-31 Jason Merrill <jason@redhat.com>
3146 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
3147 parameters the first time.
3148 (gen_scheduled_generic_parms_dies): Check completeness here.
3150 2013-01-31 Richard Biener <rguenther@suse.de>
3153 * common.opt (faggressive-loop-optimizations): New flag,
3155 * doc/invoke.texi (faggressive-loop-optimizations): Document.
3156 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
3157 infer_loop_bounds_from_undefined by it.
3159 2013-01-31 Richard Biener <rguenther@suse.de>
3161 PR tree-optimization/56150
3162 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
3163 visit virtual operands.
3164 (find_uses_to_rename_bb): Likewise.
3166 2013-01-31 Richard Biener <rguenther@suse.de>
3168 PR tree-optimization/56150
3169 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
3170 mixed store non-store stmts.
3172 2013-01-30 Jakub Jelinek <jakub@redhat.com>
3175 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
3176 LIBASAN_EARLY_SPEC is defined.
3177 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
3178 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
3180 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
3183 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
3184 invalid args instead of ICEing on it.
3185 (ix86_valid_target_attribute_tree): Return error_mark_node if
3186 ix86_valid_target_attribute_inner_p failed.
3187 (ix86_valid_target_attribute_p): Return false only if
3188 ix86_valid_target_attribute_tree returned error_mark_node. Allow
3189 target("default") attribute.
3190 (sorted_attr_string): Change argument from const char * to tree,
3191 merge in all target attribute arguments rather than just one.
3192 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
3193 instead of free. Avoid using strcat.
3194 (ix86_mangle_function_version_assembler_name): Mangle
3195 target("default") as if no target attribute is present. Adjust
3196 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
3197 instead of xmalloc and XDELETEVEC instead of free.
3198 (ix86_function_versions): Don't return true if one of the decls
3199 doesn't have target attribute. If they don't and one of the decls
3200 is DECL_FUNCTION_VERSIONED, report an error. Adjust
3201 sorted_attr_string caller. Use XDELETEVEC instead of free.
3202 (ix86_supports_function_versions): Remove.
3203 (make_name): Fix up formatting.
3204 (make_dispatcher_decl): Remove resolver_name and its initialization.
3205 Avoid leaking memory.
3206 (is_function_default_version): Return true if there is
3207 target("default") attribute rather than no target attribute at all.
3208 (make_resolver_func): Avoid leaking memory.
3209 (ix86_generate_version_dispatcher_body): Likewise.
3210 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
3211 * target.def (supports_function_versions): Remove.
3212 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
3213 * doc/tm.texi: Regenerated.
3215 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
3217 PR rtl-optimization/56144
3218 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
3219 for values with side effects.
3221 2013-01-30 Richard Biener <rguenther@suse.de>
3223 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
3224 (sparseset_pop): Likewise.
3225 * cfganal.c (compute_idf): Likewise. Increase work-stack size
3226 to be able to use quick_push in the worker loop.
3228 2013-01-30 Marek Polacek <polacek@redhat.com>
3230 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
3232 2013-01-30 Richard Biener <rguenther@suse.de>
3235 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
3237 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
3239 PR tree-optimization/56064
3240 * fixed-value.c (fixed_from_double_int): New function.
3241 * fixed-value.h (fixed_from_double_int): New prototype.
3242 (const_fixed_from_double_int): New static inline function.
3243 * fold-const.c (native_interpret_fixed): New static function.
3244 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
3245 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
3246 (native_encode_fixed): New static function.
3247 (native_encode_expr) <FIXED_CST>: Use it.
3248 (native_interpret_int): Move double_int worker code to...
3249 * double-int.c (double_int::from_buffer): ...this new static method.
3250 * double-int.h (double_int::from_buffer): Prototype it.
3252 2013-01-30 Richard Biener <rguenther@suse.de>
3254 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
3255 New pointer-map and obstack.
3256 (init_alias_vars): Allocate pointer-map and obstack.
3257 (delete_points_to_sets): Free them.
3258 (find_what_var_points_to): Cache result.
3259 (find_what_p_points_to): Adjust for changed interface of
3260 find_what_var_points_to.
3261 (compute_points_to_sets): Likewise.
3262 (ipa_pta_execute): Likewise.
3264 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3266 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
3267 * configure: Regenerate.
3268 * config.in: Regenerate.
3269 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
3270 #nobits/#progbits if supported.
3272 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
3275 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
3276 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
3277 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
3279 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3281 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
3282 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
3284 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3286 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
3287 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
3289 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3291 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
3293 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
3294 * config/arm/cortex-a7.md: New bypasses using
3295 arm_mac_accumulator_is_result.
3297 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3299 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
3300 (cortex_a7_neon_mla): Likewise.
3301 (cortex_a7_fpfmad): New reservation.
3302 (cortex_a7_fpmacs): Use ffmas and update required units.
3303 (cortex_a7_fpmuld): Update required units and latency.
3304 (cortex_a7_fpmacd): Likewise.
3305 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
3306 (cortex_a7_neon). Likewise.
3307 (bypass) Update participating units.
3309 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3311 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
3312 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
3314 * config/arm/vfp11.md (vfp_farith): Use ffmas.
3315 (vfp_fmul): Use ffmad.
3316 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
3317 (cortex_r4_fmacd): Use ffmad.
3318 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
3319 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
3320 (cortex_a9_fmacd): Use ffmad.
3321 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
3322 (cortex_a8_vfp_macd): Use ffmad.
3323 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
3324 (cortex_a5_fpmacd): Use ffmad.
3325 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
3326 (cortex_a15_vfp_macd): Use ffmad.
3327 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
3329 2013-01-29 Jason Merrill <jason@redhat.com>
3332 * varasm.c (default_assemble_visibility): Don't warn about
3333 visibility on artificial decls.
3335 2013-01-29 Richard Biener <rguenther@suse.de>
3337 PR tree-optimization/56113
3338 * tree-ssa-structalias.c (equiv_class_lookup): Also return
3340 (label_visit): Free duplicate bitmaps and record the leader instead.
3341 (perform_var_substitution): Adjust.
3343 2013-01-29 Richard Biener <rguenther@suse.de>
3345 PR tree-optimization/55270
3346 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
3347 the CFG, schedule loops for fixup.
3349 2013-01-29 Nick Clifton <nickc@redhat.com>
3351 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
3354 2013-01-28 Leif Ekblad <leif@rdos.net>
3356 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
3357 * config/i386/i386.h (TARGET_RDOS): New macro.
3358 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
3359 * config/i386/i386.c (ix86_option_override_internal): For 64bit
3360 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
3361 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
3362 DEFAULT_LARGE_SECTION_THRESHOLD.
3363 * config/i386/i386.md (R14_REG, R15_REG): New constants.
3364 * config/i386/rdos.h: New file.
3365 * config/i386/rdos64.h: New file.
3367 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
3370 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
3373 2013-01-28 Jakub Jelinek <jakub@redhat.com>
3375 PR rtl-optimization/56117
3376 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
3377 call cselib_lookup_from_insn on the MEM before calling
3378 add_insn_mem_dependence.
3380 2013-01-28 Richard Biener <rguenther@suse.de>
3382 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
3383 to a stmt that didn't have one.
3384 (copy_phis_for_bb): Likewise for PHI arguments.
3385 (copy_debug_stmt): Likewise for debug stmts.
3387 2013-01-28 Richard Biener <rguenther@suse.de>
3389 PR tree-optimization/56034
3390 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
3391 (partition_builtin_p): Adjust.
3392 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
3393 it is the last partition.
3394 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
3395 up the vertex for the definition.
3396 (classify_partition): Classify whether a partition is a
3397 PKIND_REDUCTION, thus has uses outside of the loop.
3398 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
3399 Merge all PKIND_REDUCTION partitions into the last partition.
3400 (tree_loop_distribution): Seed partitions from reductions as well.
3402 2013-01-28 Jakub Jelinek <jakub@redhat.com>
3404 PR tree-optimization/56125
3405 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
3406 pow(x,c) into sqrt(x) * powi(x, n/2) or
3407 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
3408 optimizing for size.
3409 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
3410 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
3413 PR tree-optimization/56094
3414 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
3415 to UNKNOWN_LOCATION while gimplifying expr.
3417 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
3420 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
3421 operand 0 in movabs insn template for -masm=intel asm alternative.
3422 (*movabs<mode>_2): Ditto for operand 1.
3424 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
3427 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
3430 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
3432 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
3435 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
3437 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
3438 Undef to avoid warning.
3440 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
3442 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
3443 * configure: Regenerate.
3445 2013-01-25 Jakub Jelinek <jakub@redhat.com>
3447 PR tree-optimization/56098
3448 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
3449 for stmts with volatile ops.
3450 (cond_store_replacement): Don't optimize if assign has volatile ops.
3451 (cond_if_else_store_replacement_1): Don't optimize if either
3452 then_assign or else_assign have volatile ops.
3453 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
3456 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3458 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
3460 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3462 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
3463 missing ':' in asm example.
3465 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
3467 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
3468 entries into lane and laneq entries.
3469 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
3470 Remove AdvSIMD scalar modes.
3471 (aarch64_sq<r>dmulh_laneq<mode>): New.
3472 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
3474 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
3475 builtin implementations to relfect changes in RTL in aarch64-simd.md.
3476 * config/aarch64/iterators.md (VCOND): New.
3479 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3482 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
3483 Add NULL LIBNAME argument to existing definitions.
3484 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
3485 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
3486 * config/avr/avr.c (DEF_BUILTIN): Same.
3487 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
3488 (avr_expand_builtin): Expand to a vanilla call if a libgcc
3489 implementation is available (DECL_ASSEMBLER_NAME is set).
3490 (avr_fold_absfx): New static function.
3491 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
3492 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
3493 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
3495 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
3496 (abshk, absk, abslk, absllk): Provide as static inline functions.
3498 2013-01-25 Marek Polacek <polacek@redhat.com>
3500 PR tree-optimization/56035
3501 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
3503 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
3505 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
3506 (*movtf_internal_rex64): Add (!o,C) alternative
3507 (*movxf_internal_rex64): Ditto.
3508 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
3510 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
3512 * doc/invoke.texi: fix typo.
3513 * doc/objc.texi: fix typo.
3515 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
3517 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
3518 for the first two alternatives.
3520 2013-01-24 Diego Novillo <dnovillo@google.com>
3522 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
3523 (ggc-zone.o): Remove.
3524 * configure.ac: Remove option --with-gc.
3525 * configure: Re-generate.
3526 * doc/install.texi: Remove documentation for --with-gc.
3527 * gengtype.c (write_enum_defn): Remove. Update all users.
3528 (write_Types_process_field): Remove generation of gt_e_* argument.
3529 (output_type_enum): Remove. Update all users.
3530 (write_enum_defn): Remove. Update all users.
3531 (enum alloc_zone): Remove. Update all users.
3532 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
3533 * ggc-common.c (ggc_splay_alloc): Remove first argument.
3535 (struct ptr_data): Remove field TYPE. Update all users.
3536 (gt_pch_note_object): Remove argument TYPE. Update all users.
3537 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
3539 * ggc-none.c (ggc_alloc_typed_stat): Remove.
3540 (struct alloc_zone): Remove.
3541 (ggc_internal_alloc_zone_stat): Remove.
3542 (ggc_internal_cleared_alloc_zone_stat): Remove.
3543 * ggc-page.c (ggc_alloc_typed_stat): Remove.
3544 (ggc_pch_count_object): Remove last argument. Update all users.
3545 (ggc_pch_alloc_object): Remove last argument. Update all users.
3546 (struct alloc_zone): Remove.
3547 * ggc-zone.c: Remove.
3548 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
3549 (struct alloc_zone): Remove.
3550 (ggc_alloc_typed_stat): Remove.
3551 (ggc_alloc_typed): Remove.
3552 (ggc_splay_alloc): Remove first argument.
3553 (rtl_zone): Remove. Update all users.
3554 (tree_zone): Remove. Update all users.
3555 (tree_id_zone): Remove. Update all users.
3556 (ggc_internal_zone_alloc_stat): Remove. Update all users.
3557 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
3558 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
3559 * tree-ssanames.c: Remove references to zone allocator in comments.
3561 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
3563 * config/avr/avr.c (avr_out_fract): Make register numbers that
3564 might be outside of source operand signed.
3566 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
3568 * config/i386/constraints.md (Yf): New constraint.
3569 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
3570 of f constraint to conditionaly disable x87 register preferences.
3571 (*movdf_internal): Ditto.
3572 (*movsf_internal): Ditto.
3574 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
3577 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
3578 that have operands with impossible constraints.
3579 Add a FIXME for a speed-up opportunity.
3580 * lra-constraints.c (process_alt_operands): Verify that a class
3581 selected from constraints on asms is valid for the operand mode.
3582 (curr_insn_transform): Remove incorrect comment.
3584 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
3586 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
3587 TOC operand is a valid symbol ref in the constant pool.
3589 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
3591 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
3593 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
3596 * config/avr/stdfix.h: New file.
3597 * t-avr (stdfix-gcc.h): New rule to build it.
3598 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
3600 2013-01-23 Kostya Serebryany <kcc@google.com>
3602 * config/darwin.h: remove dependency on
3603 CoreFoundation (asan on Mac OS).
3605 2013-01-23 Jakub Jelinek <jakub@redhat.com>
3608 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
3609 instead of cmpdi_operand for first comparison operand.
3610 Don't assert that comparison operands aren't both constants.
3612 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
3614 * doc/install.texi (Downloading the Source): Update references to
3615 downloading separate components.
3617 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
3619 * doc/extend.texi (__int128): Improve grammar.
3621 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
3624 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
3625 alternative to (o,r).
3626 (*movdi_internal_rex64): Remove (!o,n) alternative.
3627 (DImode immediate->memory splitter): Remove.
3628 (DImode immediate->memory peephole2): Remove.
3629 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
3630 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
3631 alternative to (!o,*r).
3632 (*movtf_internal_sse): New pattern.
3633 (*movxf_internal_rex64): New pattern.
3634 (*movxf_internal): Disable for TARGET_64BIT.
3635 (*movdf_internal_rex64): Remove (!o,F) alternative.
3637 2013-01-22 Jakub Jelinek <jakub@redhat.com>
3640 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
3641 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
3642 * tree-vect-loop-manip.c (find_loop_location): Also ignore
3643 stmt locations where LOCATION_LOCUS of the stmt location is
3644 UNKNOWN_LOCATION or BUILTINS_LOCATION.
3647 * config/i386/i386.md (UNSPEC_STOS): New.
3648 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
3649 *strsetqi_1): Add UNSPEC_STOS.
3651 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
3654 * doc/invoke.texi: Remove left over -Wsynth example.
3656 2013-01-21 Jakub Jelinek <jakub@redhat.com>
3658 PR tree-optimization/56051
3659 * fold-const.c (fold_binary_loc): Don't fold
3660 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
3661 a narrowing conversion, or widening conversion from signed
3664 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
3666 PR rtl-optimization/56023
3667 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
3668 dependent on debug instruction.
3670 2013-01-21 Martin Jambor <mjambor@suse.cz>
3673 * function.c (allocate_struct_function): Call
3674 invoke_set_current_function_hook earlier.
3676 2013-01-21 Jakub Jelinek <jakub@redhat.com>
3678 * reload1.c (init_reload): Only initialize reload_obstack
3679 during the first call.
3681 2013-01-21 Marek Polacek <polacek@redhat.com>
3683 * cfgloop.c (verify_loop_structure): Fix up grammar.
3685 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
3687 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
3688 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
3690 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3693 * config/arm/marvell-pj4.md: Update copyright year.
3694 Fix up use of alu to alu_reg and simple_alu_imm.
3696 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
3698 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
3700 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
3703 * lra-constraints.c (curr_insn_transform): Don't reuse original
3704 insn for secondary memory move when memory mode should be different.
3706 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3708 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
3709 atomic_storedi_1): New patterns.
3711 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
3713 btver2 pipeline descriptions.
3714 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
3716 * config/i386/i386.md (btver2_decode): New type attributes.
3717 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
3719 * config/i386/btver2.md: New file describing btver2 pipelines.
3721 2013-01-19 Andrew Pinski <apinski@cavium.com>
3723 PR tree-optimization/52631
3724 * tree-ssa-sccvn (visit_use): Before looking up the original
3725 statement, try looking up the simplified expression.
3727 2013-01-19 Anthony Green <green@moxielogic.com>
3729 * config/moxie/moxie.c (moxie_expand_prologue): Set
3730 current_function_static_stack_size.
3732 2013-01-18 Jakub Jelinek <jakub@redhat.com>
3734 PR tree-optimization/56029
3735 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
3736 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
3738 2013-01-18 Sharad Singhai <singhai@google.com>
3740 PR tree-optimization/55995
3741 * dumpfile.c (dump_loc): Print location only if available.
3742 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
3744 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
3747 * lra-constraints.c (curr_insn_transform): Reuse original insn for
3748 secondary memory move.
3749 (inherit_reload_reg): Use rclass instead of cl for
3750 check_secondary_memory_needed_p.
3752 2013-01-18 Jakub Jelinek <jakub@redhat.com>
3755 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
3756 the case where writing real complex part of target modifies op1.
3758 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
3760 * config/aarch64/aarch64-simd.md
3761 (aarch64_vcond_internal<mode>): Handle unordered cases.
3762 * config/aarch64/iterators.md (v_cmp_result): New.
3764 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
3765 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3767 * config/arm/marvell-pj4.md: New file.
3768 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
3769 * config/arm/arm.md (generic_sched): Add marvell_pj4.
3770 (generic_vfp): Likewise.
3771 * config/arm/arm-cores.def: Add marvell-pj4.
3772 * config/arm/arm-tune.md: Regenerate.
3773 * config/arm/arm-tables.opt: Regenerate.
3774 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
3775 * doc/invoke.texi: Document marvell-pj4.
3777 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
3779 * config/aarch64/arm_neon.h: Map scalar types to standard types.
3781 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3786 * var-tracking.c (negative_power_of_two_p): New.
3787 (global_get_addr_cache, local_get_addr_cache): New.
3788 (get_addr_from_global_cache, get_addr_from_local_cache): New.
3789 (vt_canonicalize_addr): Rewrite using the above. Adjust the
3791 (vt_stack_offset_p): Remove.
3792 (vt_canon_true_dep): Always canonicalize loc's address.
3793 (clobber_overlapping_mems): Make sure we have a MEM.
3794 (local_get_addr_clear_given_value): New.
3795 (val_reset): Clear local cached entries.
3796 (compute_bb_dataflow): Create and release the local cache.
3797 Disable duplicate MEMs clobbering.
3798 (emit_notes_in_bb): Clobber MEMs likewise.
3799 (vt_emit_notes): Create and release the local cache.
3800 (vt_initialize, vt_finalize): Create and release the global
3801 cache, respectively.
3802 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
3804 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3807 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
3808 not found in the symtab.
3810 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3813 PR rtl-optimization/55547
3814 PR rtl-optimization/53827
3817 * alias.c (offset_overlap_p): New, factored out of...
3818 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
3819 the conservative special case for symbolic constants. Don't
3820 adjust zero sizes on alignment.
3822 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
3824 PR rtl-optimization/52573
3825 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
3826 REG_UNUSED for the same register.
3828 2013-01-17 Richard Biener <rguenther@suse.de>
3829 Marek Polacek <polacek@redhat.com>
3831 PR rtl-optimization/55833
3832 * loop-unswitch.c (unswitch_loops): Move loop verification...
3833 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
3834 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
3835 Set it to true when we're removing a loop from hierarchy tree in
3836 an irreducible region.
3837 (fix_bb_placements): Adjust caller.
3838 (fix_loop_placements): Likewise.
3840 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
3842 * config/avr/builtins.def (DEF_BUILTIN): Factor out
3843 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
3844 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
3845 Remove ID. Adjust comments.
3846 * config/avr/avr-c.c (avr_builtin_name): Remove.
3847 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
3848 * config/avr/avr.c (avr_tolower): New static function.
3849 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
3850 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
3851 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
3854 2013-01-17 Jan Hubicka <jh@suse.cz>
3856 PR tree-optimization/55273
3857 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
3859 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
3862 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
3863 store through atomic_store<mode>_1.
3864 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
3866 2013-01-17 Martin Jambor <mjambor@suse.cz>
3868 PR tree-optimizations/55264
3869 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
3870 for virtual methods.
3871 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
3872 virtual methods before inlining is over.
3873 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
3875 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
3878 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
3880 PR rtl-optimization/56005
3881 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
3882 pending reads for prefetch.
3884 2013-01-16 Ian Bolton <ian.bolton@arm.com>
3886 * config/aarch64/aarch64.md
3887 (*cstoresi_neg_uxtw): New pattern.
3888 (*cmovsi_insn_uxtw): New pattern.
3889 (*<optab>si3_uxtw): New pattern.
3890 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
3891 (*<optab>si3_insn_uxtw): New pattern.
3892 (*bswapsi2_uxtw): New pattern.
3894 2013-01-16 Richard Biener <rguenther@suse.de>
3896 * tree-inline.c (tree_function_versioning): Remove set but
3897 never used variable.
3899 2013-01-16 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/55964
3902 * tree-flow.h (rename_variables_in_loop): Remove.
3903 (rename_variables_in_bb): Likewise.
3904 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
3905 (copy_loop_before): Adjust and delete update-ssa status.
3906 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
3907 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
3908 (rename_variables_in_loop): Remove.
3909 (slpeel_update_phis_for_duplicate_loop): Likewise.
3910 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
3911 use available cfg machinery instead of duplicating it.
3912 Update PHI nodes and perform poor-mans SSA update here.
3913 (slpeel_tree_peel_loop_to_edge): Adjust.
3915 2013-01-16 Richard Biener <rguenther@suse.de>
3917 PR tree-optimization/54767
3918 PR tree-optimization/53465
3919 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
3920 (vrp_visit_phi_node): For PHI arguments coming via backedges
3921 drop all symbolical range information.
3922 (execute_vrp): Compute backedges.
3924 2013-01-16 Richard Biener <rguenther@suse.de>
3926 * doc/install.texi: Update CLooG and ISL requirements to
3929 2013-01-16 Christian Bruel <christian.bruel@st.com>
3932 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
3933 (broken_move): Handle UNSPECV_SP_SWITCH_B.
3934 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
3936 2013-01-16 DJ Delorie <dj@redhat.com>
3938 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
3939 (UNSPECV_SP_SWITCH_E): New.
3940 (sp_switch_1): Change to an unspec.
3941 (sp_switch_2): Change to an unspec. Don't use post-inc when we
3944 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
3946 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
3947 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
3948 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
3949 (expand_mem_thread_fence): Ditto.
3950 (expand_mem_signal_fence): Ditto.
3951 (expand_atomic_load): Ditto.
3952 (expand_atomic_store): Ditto.
3954 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
3956 PR rtl-optimization/55547
3957 PR rtl-optimization/53827
3960 * alias.c (memrefs_conflict_p): Set sizes to negative after
3963 2013-01-15 Jakub Jelinek <jakub@redhat.com>
3966 * function.c (thread_prologue_and_epilogue_insns): Always
3967 add crtl->drap_reg to set_up_by_prologue.set, even if
3968 stack_realign_drap is false.
3970 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3972 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
3973 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
3974 *call): Fix indention.
3976 2013-01-15 Tom de Vries <tom@codesourcery.com>
3979 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
3982 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
3984 PR rtl-optimization/55153
3985 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
3987 2013-01-15 Martin Jambor <mjambor@suse.cz>
3989 PR tree-optimization/55920
3990 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
3991 accesses as grp_to_be_debug_replaced.
3993 2013-01-15 Jakub Jelinek <jakub@redhat.com>
3995 PR tree-optimization/55920
3996 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
3997 there is non-useless type conversion needed from debug rhs to lhs,
3998 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
4000 2013-01-15 Joseph Myers <joseph@codesourcery.com>
4001 Mikael Pettersson <mikpe@it.uu.se>
4004 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
4006 (ASM_OUTPUT_CASE_LABEL): Remove.
4007 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
4008 * final.c (shorten_branches): Update alignment of labels before
4009 jump tables if CASE_VECTOR_SHORTEN_MODE.
4011 2013-01-15 Richard Biener <rguenther@suse.de>
4014 * system.h: Do not include gmp.h for building host tools.
4016 2013-01-15 Richard Biener <rguenther@suse.de>
4019 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
4020 account for bitpos when computing alignment.
4022 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
4024 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
4025 (ix86_target_macros_internal): Likewise.
4027 * config/i386/i386.c (m_CORE2I7): Removed.
4028 (m_CORE_HASWELL): New macro.
4029 (m_CORE_ALL): Likewise.
4030 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
4031 (initial_ix86_arch_features): Likewise.
4032 (processor_target_table): Initializations for Core avx2.
4033 (cpu_names): New names "core-avx2".
4034 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
4035 PROCESSOR_CORE_HASWELL.
4036 (ix86_issue_rate): New case.
4037 (ia32_multipass_dfa_lookahead): Likewise.
4038 (ix86_sched_init_global): Likewise.
4040 * config/i386/i386.h (TARGET_HASWELL): New macro.
4041 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
4042 (processor_type): New PROCESSOR_HASWELL.
4044 2013-01-15 Jakub Jelinek <jakub@redhat.com>
4046 PR tree-optimization/55955
4047 * tree-vect-loop.c (vectorizable_reduction): Give up early on
4048 *SHIFT_EXPR and *ROTATE_EXPR codes.
4050 PR tree-optimization/48766
4051 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
4052 -ftrapv disable -fwrapv.
4054 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4057 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
4058 etc. to 1 and not to __flash.
4059 Use LL suffix for __INT24_MAX__ with -mint8.
4060 Use ULL suffix for __UINT24_MAX__ with -mint8.
4062 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4064 * config/avr/avr-arch.h
4065 (struct base_arch_s): Use typedef avr_arch_t instead.
4066 (struct arch_info_s): Use typedef avr_arch_info_t instead.
4067 (struct mcu_type_s): Use typedef avr_mcu_t instead.
4068 * config/avr/avr.c: Same.
4069 * config/avr/avr-devices.c: Same.
4070 * config/avr/driver-avr.c: Same.
4071 * config/avr/gen-avr-mmcu-texi.c: Same.
4072 * config/avr/avr-mcus.def: Adjust comment.
4074 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
4076 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
4077 * config/aarch64/iterators.md (VALLDI): New.
4079 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
4080 Andi Kleen <ak@linux.intel.com>
4083 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
4084 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
4087 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4089 * config/avr/avr-stdint.h: Remove trailing blanks.
4090 * config/avr/avr-log.h: Same.
4091 * config/avr/avr-arch.h: Same.
4092 * config/avr/avr-devices.c: Same.
4093 * config/avr/avr-dimode.md: Same.
4094 * config/avr/predicates.md: Same.
4095 * config/avr/avr-c.c: Same. And fix typo.
4097 * config/avr/avr-protos.h: Same. And:
4098 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
4099 (init_cumulative_args): Rename to avr_init_cumulative_args.
4100 (expand_prologue): Rename to avr_expand_prologue.
4101 (expand_epilogue): Rename to avr_expand_epilogue.
4102 (adjust_insn_length): Rename to avr_adjust_insn_length.
4103 (notice_update_cc): Rename to avr_notice_update_cc.
4104 (final_prescan_insn): Rename to avr_final_prescan_insn.
4105 * config/avr/avr.c: Same.
4106 * config/avr/avr.h: Same.
4107 * config/avr/avr.md: Remove trailing blanks.
4108 (prologue): Use avr_expand_prologue.
4109 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
4111 2013-01-14 Richard Biener <rguenther@suse.de>
4113 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
4114 verify_location, collect_subblocks): New functions.
4115 (verify_gimple_in_cfg): Verify that locations only reference
4116 BLOCKs in the functions BLOCK tree.
4118 2013-01-14 Richard Biener <rguenther@suse.de>
4120 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
4122 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
4124 (insert_out_of_ssa_copy_on_edge): Likewise.
4125 (rewrite_close_phi_out_of_ssa): Likewise.
4126 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
4128 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
4129 propagated constants.
4130 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
4133 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4135 * config/avr/avr-modes.def: Add GPL copyright notice.
4137 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
4139 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
4140 MEMMODEL_MASK to determine memory model.
4141 (atomic_store<mode>): Ditto from operands[2].
4142 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
4144 2013-01-13 Jakub Jelinek <jakub@redhat.com>
4147 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
4148 (fold_gimple_assign): Don't call unshare_expr here.
4149 (fold_ctor_reference): Call unshare_expr.
4151 2013-01-13 Terry Guo <terry.guo@arm.com>
4153 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
4154 * doc/fragments.texi: Document MULTILIB_REUSE.
4155 * gcc.c (multilib_reuse): New internal spec.
4156 (set_multilib_dir): Also search multilib from multilib_reuse.
4157 * genmultilib (tmpmultilib3): Refactor code.
4158 (tmpmultilib4): Ditto.
4159 (multilib_reuse): New multilib argument.
4161 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
4163 * Makefile.in: Update copyright.
4165 2013-01-12 Tom de Vries <tom@codesourcery.com>
4168 * calls.c (expand_call): Check if arg_nr is valid.
4170 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4172 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
4173 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
4174 documentation. Add missing '__' in front of
4175 __builtin_ia32_packssdw256.
4177 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4180 * config/s390/s390.c (s390_preferred_reload_class): Do not return
4181 NO_REGS for larl operands.
4182 (s390_reload_larl_operand): Use s390_load_address instead of
4185 2013-01-11 Richard Biener <rguenther@suse.de>
4187 * tree-cfg.c (verify_node_sharing_1): Split out from ...
4188 (verify_node_sharing): ... here.
4189 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
4191 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
4193 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
4194 Substitute TREECHECKING.
4195 * configure: Regenerate.
4196 * Makefile.in (TREECHECKING): New.
4198 2013-01-11 Richard Guenther <rguenther@suse.de>
4200 PR tree-optimization/44061
4201 * tree-vrp.c (extract_range_basic): Compute zero as
4202 value-range for __builtin_constant_p of function parameters.
4204 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
4206 Update copyright years.
4208 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
4210 PR rtl-optimization/55672
4211 * lra-eliminations.c (mark_not_eliminable): Permit addition with
4212 const to be eliminable.
4214 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
4216 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
4217 * configure: Regenerate.
4219 2013-01-10 Richard Biener <rguenther@suse.de>
4221 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
4223 2013-01-10 Richard Biener <rguenther@suse.de>
4226 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
4227 locations for virtual PHI arguments.
4228 (rewrite_update_phi_arguments): Likewise.
4230 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
4232 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
4235 2013-01-10 Jakub Jelinek <jakub@redhat.com>
4237 PR tree-optimization/55921
4238 * tree-complex.c (expand_complex_asm): New function.
4239 (expand_complex_operations_1): Call it for GIMPLE_ASM.
4241 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4244 * config/s390/s390.c (s390_symref_operand_p)
4245 (s390_loadrelative_operand_p): Merge the two functions.
4246 (s390_check_qrst_address, print_operand_address): Add parameters
4247 to s390_loadrelative_operand_p invokation.
4248 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
4249 (s390_reload_larl_operand, s390_secondary_reload): Use
4250 s390_loadrelative_operand_p instead of s390_symref_operand_p.
4251 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
4253 2013-01-09 Mike Stump <mikestump@comcast.net>
4255 * dse.c (record_store): Remove unnecessary assert.
4257 2013-01-09 Jan Hubicka <jh@suse.cz>
4259 PR tree-optimization/55569
4260 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
4261 * cfgloop.h (scale_loop_profile): Likewise.
4263 2013-01-09 Jan Hubicka <jh@suse.cz>
4266 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
4268 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
4270 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
4273 * expr.h (maybe_emit_group_store): Declare.
4274 * expr.c (maybe_emit_group_store): New function.
4275 * builtins.c (expand_builtin_int_roundingfn): Call it.
4276 (expand_builtin_int_roundingfn_2): Likewise.
4278 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
4280 PR rtl-optimization/55829
4281 * lra-constraints.c (match_reload): Add code for absent output.
4282 (curr_insn_transform): Add code for reloads of matched inputs
4285 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
4287 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
4288 attribute of movddup insn to DF.
4289 (*vec_interleave_lowv2df): Ditto.
4290 (vec_dupv2df): Ditto.
4292 2013-01-09 Jan Hubicka <jh@suse.cz>
4294 PR tree-optimiation/55875
4295 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
4296 EVERY_ITERATION parameter.
4297 (number_of_iterations_exit): Check if exit is executed every iteration.
4298 (idx_infer_loop_bounds): Similarly here.
4299 (n_of_executions_at_most): Simplify
4300 to only test for cases where statement is dominated by the
4301 particular bound; handle correctly the "postdominance" test.
4302 (scev_probably_wraps_p): Use max loop iterations info
4303 as a global bound first.
4305 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
4306 Nick Clifton <nickc@redhat.com>
4308 * config/v850/v850.md (cbranchsf4): New pattern.
4309 (cstoresf4): New pattern.
4310 (cbranchdf4): New pattern.
4311 (cstoredf4): New pattern.
4312 (movsicc): Disallow floating point comparisons.
4313 (cmpsf_le_insn): Fix order of operators.
4314 (cmpsf_lt_insn): Likewise.
4315 (cmpsf_eq_insn): Likewise.
4316 (cmpdf_le_insn): Likewise.
4317 (cmpdf_lt_insn): Likewise.
4318 (cmpdf_eq_insn): Likewise.
4319 (cmpsf_ge_insn): Use LE comparison.
4320 (cmpdf_ge_insn): Likewise.
4321 (cmpsf_gt_insn): Use LT comparison.
4322 (cmpdf_gt_insn): Likewise.
4323 (cmpsf_ne_insn): Delete pattern.
4324 (cmpdf_ne_insn): Delete pattern.
4325 * config/v850/v850.c (v850_gen_float_compare): Use
4326 gen_cmpdf_eq_insn for NE comparison.
4327 (v850_float_z_comparison_operator)
4328 (v850_float_nz_comparison_operator): Move from here ...
4329 * config/v850/predicates.md: ... to here. Move GT and GE
4330 comparisons into v850_float_z_comparison_operator.
4331 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
4333 (v850_float_nz_comparison_operator): Likewise.
4335 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4337 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
4338 with calls to gen_insvsi/gen_insvdi.
4340 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4342 * config/i386/i386.c (initial_ix86_tune_features): Set up
4343 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
4345 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
4346 Jakub Jelinek <jakub@redhat.com>
4348 PR tree-optimization/48189
4349 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
4350 If nitercst is 0, don't predict the exit edge.
4352 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
4354 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
4355 in asm_fprintf with reg_names.
4356 (aarch64_print_operand_address): Likewise.
4357 (aarch64_return_addr): Likewise.
4358 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
4360 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4362 * config/pa/pa.h (VAL_U6_BITS_P): Define.
4363 (INT_U6_BITS): Likewise.
4364 * config/pa/predicates.md (uint6_operand): New predicate.
4365 (shift5_operand, shift6_operand): Likewise.
4366 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
4368 (lshrdi3): Use shift6_operand.
4369 (shrpsi4, shrpdi4): New insn patterns.
4370 (extzv): Delete expander.
4371 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
4372 predicates in unamed zero extract patterns. Tighten common constraint.
4373 (extv): Delete expander.
4374 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
4375 predicates in unamed sign extract patterns. Tighten common constraint.
4376 (insv): Delete expander.
4377 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
4378 predicates in unamed insert patterns. Tighten common constraint.
4379 Change uint32_operand predicate to uint6_operand predicate in unamed
4380 DImode pattern to insert constant values of type 1...1xxxx.
4382 2013-01-04 Jan Hubicka <jh@suse.cz>
4384 PR tree-optimization/55823
4385 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
4388 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4389 Uros Bizjak <ubizjak@gmail.com>
4391 PR rtl-optimization/55845
4392 * df-problems.c (can_move_insns_across): Stop scanning at
4393 volatile_insn_p source instruction or give up if
4394 across_from .. across_to range contains any volatile_insn_p
4397 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
4399 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
4400 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
4402 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
4403 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
4405 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4408 * asan.c (asan_clear_shadow): New function.
4409 (asan_emit_stack_protection): Use it.
4411 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
4413 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
4414 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
4415 with tab instead of space.
4417 2013-01-08 Nick Clifton <nickc@redhat.com>
4419 * config/rl78/rl78.c (rl78_expand_prologue): Always select
4420 register bank 0 at the start of an interrupt handler.
4421 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
4424 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
4426 * config/aarch64/aarch64-simd.md
4427 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
4428 (aarch64_simd_bsl): Likewise.
4429 (aarch64_vcond_internal<mode>): Likewise.
4430 (vcond<mode><mode>): Likewise.
4431 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
4432 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
4434 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
4436 * config/aarch64/aarch64-builtins.c
4437 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
4439 2013-01-08 Martin Jambor <mjambor@suse.cz>
4442 * tree-sra.c (analyze_access_subtree): Return true also after
4443 potentially creating a debug-only replacement.
4445 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4448 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
4450 PR tree-optimization/54120
4451 * tree-vrp.c (range_fits_type_p): Don't allow
4452 src_precision < precision from signed vr to unsigned_p
4453 if vr->min or vr->max is negative.
4454 (simplify_float_conversion_using_ranges): Test can_float_p
4455 against CODE_FOR_nothing.
4457 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4458 Richard Biener <rguenther@suse.de>
4461 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
4462 types instead of just INTEGER_TYPE types.
4464 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
4466 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
4469 2013-01-07 Steve Ellcey <sellcey@mips.com>
4472 * config/mips/mips.opt: Change mad to mmad to match documentation.
4474 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4477 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
4480 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4483 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
4484 (avr_addrspace_t): Add .section_name field.
4485 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
4487 (avr_addrspace): Same. Initialize .section_name. Remove last
4488 NULL entry. Put __memx into .progmemx.data.
4489 (progmem_section_prefix): Remove.
4490 (avr_asm_init_sections): No need to initialize progmem_section.
4491 (avr_asm_named_section): Use avr_addrspace[].section_name to get
4492 section name prefix.
4493 (avr_asm_select_section): Ditto. And use get_unnamed_section to
4494 retrieve the progmem section.
4495 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
4496 boundary to run over avr_addrspace[].
4497 (avr_register_target_pragmas): Ditto.
4499 2013-01-06 Jakub Jelinek <jakub@redhat.com>
4501 * varasm.c (output_constant_def_contents): For asan_protect_global
4502 protected strings, adjust DECL_ALIGN if needed, before testing for
4504 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
4505 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
4507 (output_object_block): For asan protected decls, emit asan padding
4508 after their contents.
4509 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
4510 (asan_finish_file): Test it here instead.
4512 2013-01-07 Nick Clifton <nickc@redhat.com>
4513 Matthias Klose <doko@debian.org>
4514 Doug Kwan <dougkwan@google.com>
4515 H.J. Lu <hongjiu.lu@intel.com>
4518 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
4520 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
4522 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
4524 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
4526 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
4528 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4531 * doc/install.texi (Cross-Compiler-Specific Options): Document
4534 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
4536 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
4537 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
4538 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
4539 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
4540 vqmovun_high_s64): Fix source operand number and update copyright.
4542 2013-01-07 Richard Biener <rguenther@suse.de>
4545 * gimple.h (gimple_call_builtin_p): New overload.
4546 * gimple.c (validate_call): New function.
4547 (gimple_call_builtin_p): Likewise.
4548 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4549 Use gimple_call_builtin_p.
4550 (find_func_clobbers): Likewise.
4551 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
4552 (strlen_optimize_stmt): Likewise.
4554 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4556 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
4557 (vld1q_dup_*): Likewise.
4559 (vld1q_*): Likewise.
4560 (vld1_lane_*): Likewise.
4561 (vld1q_lane_*): Likewise.
4563 2013-01-07 Richard Biener <rguenther@suse.de>
4565 * lto-streamer.h (LTO_minor_version): Bump to 2.
4567 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4569 * config/aarch64/aarch64-protos.h
4570 (aarch64_const_double_zero_rtx_p): Rename to...
4571 (aarch64_float_const_zero_rtx_p): ...this.
4572 (aarch64_float_const_representable_p): New.
4573 (aarch64_output_simd_mov_immediate): Likewise.
4574 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
4575 move immediate case.
4576 * config/aarch64/aarch64.c
4577 (aarch64_const_double_zero_rtx_p): Rename to...
4578 (aarch64_float_const_zero_rtx_p): ...this.
4579 (aarch64_print_operand): Allow printing of new constants.
4580 (aarch64_valid_floating_const): New.
4581 (aarch64_legitimate_constant_p): Check for valid floating-point
4583 (aarch64_simd_valid_immediate): Likewise.
4584 (aarch64_vect_float_const_representable_p): New.
4585 (aarch64_float_const_representable_p): Likewise.
4586 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
4587 (aarch64_output_simd_mov_immediate): New.
4588 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
4589 (*movdf_aarch64): Likewise.
4590 * config/aarch64/constraints.md (Ufc): New.
4591 (Y): call aarch64_float_const_zero_rtx.
4592 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
4594 2013-01-07 Richard Biener <rguenther@suse.de>
4596 PR tree-optimization/55888
4597 PR tree-optimization/55862
4598 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
4599 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
4600 not if it is contained therein.
4602 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4604 * config/avr/t-avr: Typo.
4606 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4609 * config/avr/t-avr: Don't automatically rebuild
4610 $(srcdir)/config/avr/t-multilib
4611 $(srcdir)/config/avr/avr-tables.opt
4612 $(srcdir)/doc/avr-mmcu.texi
4613 (avr-mcus): New phony target to build them on request.
4614 (s-avr-mlib, s-avr-mmcu-texi): Remove.
4615 * avr/avr-mcus.def: Adjust comments.
4617 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
4619 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
4621 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
4623 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
4625 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
4627 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
4629 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
4631 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
4632 to generate profiling.
4633 * config/rs6000/aix64.h (LIB_SPEC): Same.
4635 2013-01-04 Andrew Pinski <apinski@cavium.com>
4637 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
4639 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
4641 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
4643 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
4645 (ix86_expand_move): Ditto.
4646 (ix86_zero_extend_to_Pmode): Ditto.
4647 (ix86_expand_call): Ditto.
4648 (ix86_expand_special_args_builtin): Ditto.
4649 (ix86_expand_builtin): Ditto.
4651 2013-01-04 Richard Biener <rguenther@suse.de>
4653 PR tree-optimization/55862
4654 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
4655 translating them through PHI nodes.
4657 2013-01-04 Martin Jambor <mjambor@suse.cz>
4659 PR tree-optimization/55755
4660 * tree-sra.c (sra_modify_assign): Do not check that an access has no
4661 children when trying to avoid producing a VIEW_CONVERT_EXPR.
4663 2013-01-04 Marek Polacek <polacek@redhat.com>
4666 * opts.c (default_options_optimization): Clarify error message.
4668 2013-01-04 Richard Biener <rguenther@suse.de>
4671 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
4674 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4677 * config/pa/pa.md (movsi): Revert previous change.
4678 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
4681 2013-01-03 Richard Henderson <rth@redhat.com>
4683 * config/i386/i386.c (ix86_expand_move): Always assign to op1
4684 after eliminating TLS symbols.
4686 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
4689 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
4690 * graphite-poly.c (debug_gmp_value): Likewise.
4692 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
4695 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
4696 selected code model, define __code_mode_small__, __code_model_medium__,
4697 __code_model_large__, __code_model_32__ or __code_model_kernel__.
4698 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
4699 xchg temporary register with %k. Declare temporary register as
4701 [__x86_64__]: For medium and large code models, preserve %rbx register.
4703 2013-01-03 Richard Biener <rguenther@suse.de>
4705 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
4706 (dump_subscript): Adjust.
4707 (finalize_ddr_dependent): Do not dump redundant info.
4708 (analyze_siv_subscript): Adjust.
4709 (subscript_dependence_tester): Likewise.
4710 (compute_affine_dependence): Likewise.
4712 2013-01-03 Richard Biener <rguenther@suse.de>
4715 2013-01-03 Richard Biener <rguenther@suse.de>
4717 PR tree-optimization/55857
4718 * tree-vect-stmts.c (vectorizable_load): Do not setup
4719 re-alignment for invariant loads.
4721 2013-01-02 Richard Biener <rguenther@suse.de>
4723 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
4724 invariant load do not generate a vector load from the scalar location.
4726 2013-01-03 Richard Biener <rguenther@suse.de>
4728 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
4729 for not vectorizing.
4730 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
4731 not build INDIRECT_REFs, call get_name once only.
4732 (vect_create_data_ref_ptr): Likewise. Dump base object kind
4733 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
4735 2013-01-03 Richard Biener <rguenther@suse.de>
4737 PR tree-optimization/55857
4738 * tree-vect-stmts.c (vectorizable_load): Do not setup
4739 re-alignment for invariant loads.
4741 2013-01-03 Richard Biener <rguenther@suse.de>
4744 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
4745 prefer a built-in decl.
4747 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4749 * gcc.c (process_command): Update copyright notice dates.
4750 * gcov.c (print_version): Likewise.
4751 * gcov-dump.c (print_version): Likewise.
4753 PR rtl-optimization/55838
4754 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
4755 iv0.step, iv1.step and step.
4757 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4758 Marc Glisse <marc.glisse@inria.fr>
4760 PR tree-optimization/55832
4761 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
4762 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
4763 integer_{one,zero}_node.
4765 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4768 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
4769 * var-tracking.c (reverse_op): Don't add reverse ops to
4770 VALUEs that have already
4771 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
4773 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
4775 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
4777 2013-01-02 Teresa Johnson <tejohnson@google.com>
4779 * dumpfile.c (dump_loc): Print filename with location.
4780 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
4781 new location_t parameter to emit complete unroll message with
4783 (canonicalize_loop_induction_variables): Compute loops location
4784 and pass to try_unroll_loop_completely.
4785 * loop-unroll.c (report_unroll_peel): New function.
4786 (peel_loops_completely): Use new dump format with location
4787 for main dumpfile message, and invoke report_unroll_peel on success.
4788 (decide_unrolling_and_peeling): Ditto.
4789 (decide_peel_once_rolling): Remove old dumpfile message subsumed
4790 by report_unroll_peel.
4791 (decide_peel_completely): Ditto.
4792 (decide_unroll_constant_iterations): Ditto.
4793 (decide_unroll_runtime_iterations): Ditto.
4794 (decide_peel_simple): Ditto.
4795 (decide_unroll_stupid): Ditto.
4796 * cfgloop.c (get_loop_location): New function.
4797 * cfgloop.h (get_loop_location): Declare.
4799 2013-01-02 Sriraman Tallam <tmsriram@google.com>
4801 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
4804 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4807 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
4808 BLKmode objects when EXPAND_MEMORY is specified.
4810 2013-01-02 Sriraman Tallam <tmsriram@google.com>
4812 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
4814 (fold_builtin_cpu): Do not share cpu model decls across statements.
4816 2013-01-02 Jason Merrill <jason@redhat.com>
4819 * tree.c (build_array_type_1): Revert earlier change.
4821 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
4823 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
4825 * config/aarch64/aarch64-tune.md: Re-generate.
4827 2013-01-02 Richard Biener <rguenther@suse.de>
4829 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
4830 invariant load do not generate a vector load from the scalar location.
4832 2013-01-02 Richard Biener <rguenther@suse.de>
4835 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
4836 * configure: Regenerate.
4838 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4840 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
4841 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
4842 (expand_builtin_int_roundingfn_2): Keep the original target around
4843 for the fallback case.
4845 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4847 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
4848 to be clear for sign changes.
4850 2013-01-01 Jan Hubicka <jh@suse.cz>
4852 * ipa-inline-analysis.c: Fix formatting.
4854 2013-01-01 Jakub Jelinek <jakub@redhat.com>
4856 PR tree-optimization/55831
4857 * tree-vect-loop.c (get_initial_def_for_induction): Use
4858 gsi_after_labels instead of gsi_start_bb.
4860 Copyright (C) 2013 Free Software Foundation, Inc.
4862 Copying and distribution of this file, with or without modification,
4863 are permitted in any medium without royalty provided the copyright
4864 notice and this notice are preserved.