1 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
3 * tree.h (VECTOR_TYPE_P): New macro.
4 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
6 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
7 VEC_COND_EXPR cannot be lvalues.
8 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
10 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
12 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13 Restrict the transformation to equal modes.
15 2013-03-21 Richard Biener <rguenther@suse.de>
17 PR tree-optimization/39326
18 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
19 (MEM_ANALYZABLE): Adjust.
20 (record_mem_ref_loc): Move bitmap ops ...
21 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
22 unanalyzable refs, do not record locations for it.
23 (analyze_memory_references): Allocate ref zero as shared
25 (refs_independent_p): Do not test for unanalyzed mems here.
26 (ref_indep_loop_p_1): Special-case disambiguation against
28 (ref_indep_loop_p): Assert we are not queried for the
31 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
33 * config/arm/arm-protos.h (tune_params): Add
34 prefer_neon_for_64bits field.
35 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
36 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
37 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
38 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
39 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
40 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
41 (arm_option_override): Handle -mneon-for-64bits new option.
42 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
43 (prefer_neon_for_64bits): Declare new variable.
44 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
45 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and
47 (arch_enabled): Handle new arch types. Remove support for onlya8
49 (one_cmpldi2): Use new arch names.
50 (zero_extend<mode>di2, extend<mode>di2): Ditto.
51 * config/arm/arm.opt (mneon-for-64bits): Add option.
52 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
53 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
54 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
56 * doc/invoke.texi (-mneon-for-64bits): Document.
58 2013-03-21 Richard Biener <rguenther@suse.de>
60 PR tree-optimization/39326
61 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
62 (sort_bbs_in_loop_postorder_cmp): New function.
63 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop
66 2013-03-21 Richard Biener <rguenther@suse.de>
68 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
69 (vect_insert_into_interleaving_chain): Likewise.
70 (vect_drs_dependent_in_basic_block): Inline ...
71 (vect_slp_analyze_data_ref_dependence): ... here. New function,
73 (vect_analyze_data_ref_dependence): ... here. Simplify.
74 (vect_check_interleaving): Simplify.
75 (vect_analyze_data_ref_dependences): Likewise. Split out ...
76 (vect_slp_analyze_data_ref_dependences): ... this new function.
77 (dr_group_sort_cmp): New function.
78 (vect_analyze_data_ref_accesses): Compute data-reference groups
79 here instead of in vect_analyze_data_ref_dependence. Use
80 a more efficient algorithm.
81 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
82 vect_slp_analyze_data_ref_dependences. Call
83 vect_analyze_data_ref_accesses earlier.
84 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
85 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
86 (vect_slp_analyze_data_ref_dependences): New prototype.
88 2013-03-21 Richard Biener <rguenther@suse.de>
90 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
91 ref is stored in the loop.
92 (find_refs_for_sm): Walk only over all stores.
93 (store_motion_loop): Allocate from lim_bitmap_obstack.
94 (store_motion): Likewise.
96 2013-03-21 Richard Biener <rguenther@suse.de>
98 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
99 Update virtual SSA form.
101 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
103 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
104 * configure: Regenerate.
105 * config.in: Regenerate.
106 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
107 if !HAVE_LD_EH_FRAME_CIEV3.
109 2013-03-21 Richard Biener <rguenther@suse.de>
111 * tree-cfg.c (verify_expr_no_block): New function.
112 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
113 nor DECL_VALUE_EXPR have locations with associated blocks.
114 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
115 (clear_unused_block_pointer): Remove code dealing with
116 blocks in DECL_DEBUG_EXPR locations.
118 2013-03-21 Richard Biener <rguenther@suse.de>
120 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
121 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
122 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
123 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
124 instead of DECL_DEBUG_EXPR_IS_FROM.
125 * gimplify.c (gimplify_modify_expr): Likewise.
126 * tree-cfg.c (verify_expr_location_1): Likewise.
127 * tree-complex.c (create_one_component_var): Likewise.
128 * tree-sra.c (create_access_replacement): Likewise.
129 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
130 (clear_unused_block_pointer): Likewise.
131 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
132 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
133 * var-tracking.c (var_debug_decl): Likewise.
134 (track_expr_p): Likewise.
135 * tree-inline.c (add_local_variables): Likewise. Set
136 DECL_HAS_DEBUG_EXPR_P after copying it.
137 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
138 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
140 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
143 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
144 * configure: Regenerate.
145 * config.in: Regenerate.
146 * config/i386/i386.md (*movdf_internal): Use
147 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
148 movd instead of movq mnemonic for interunit moves.
149 (*movdi_internal): Ditto.
151 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
153 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
154 (abd<mode>_3): New pattern.
155 (aba<mode>_3): New pattern.
156 (fabd<mode>_3): New pattern.
158 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
160 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
161 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
162 occurrence of REGISTER_PREFIX as its empty string.
164 2013-03-20 Jeff Law <law@redhat.com>
166 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
167 addititional equivalences for equality comparisons between an SSA_NAME
168 and a constant where the SSA_NAME was set from a widening conversion.
170 2013-03-20 Walter Lee <walt@tilera.com>
172 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
174 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
176 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
177 depending on type attribute.
178 (*movti_internal): Ditto.
179 (*movtf_internal): Ditto.
180 (*movxf_internal): Ditto.
181 (*movdf_internal): Ditto.
182 (*movsf_internal): Ditto.
184 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
186 * config/i386/i386.md (*movti_internal): Set prefix attribute to
187 maybe_vex for sselog1 and ssemov types.
188 (*movdi_internal): Reorder operand constraints.
189 (*movsi_internal): Ditto. Set prefix attribute to
190 maybe_vex for sselog1 and ssemov types.
191 (*movtf_internal): Set prefix attribute to maybe_vex
192 for sselog1 and ssemov types.
193 (*movdf_internal): Ditto. Set prefix_data16 attribute for
194 DImode ssemov types. Reorder operand constraints.
195 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
196 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
197 attribute for SImode ssemov types. Reorder operand constraints.
199 2013-03-20 Martin Jambor <mjambor@suse.cz>
201 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
202 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
204 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
206 * config/rs6000/predicates.md (indexed_address, update_address_mem
207 update_indexed_address_mem): New predicates.
208 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
209 attribute for load/store instructions.
210 * config/rs6000/dfp.md (movsd_store): Likewise.
211 (movsd_load): Likewise.
212 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
213 (unnamed HI->DI extend define_insn): Likewise.
214 (unnamed SI->DI extend define_insn): Likewise.
215 (unnamed QI->SI extend define_insn): Likewise.
216 (unnamed QI->HI extend define_insn): Likewise.
217 (unnamed HI->SI extend define_insn): Likewise.
218 (unnamed HI->SI extend define_insn): Likewise.
219 (extendsfdf2_fpr): Likewise.
220 (movsi_internal1): Likewise.
221 (movsi_internal1_single): Likewise.
222 (movhi_internal): Likewise.
223 (movqi_internal): Likewise.
224 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
225 attribute for load/store instructions.
226 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
228 (mov<mode>_softfloat): Likewise.
229 (mov<mode>_hardfloat32): Likewise.
230 (mov<mode>_hardfloat64): Likewise.
231 (mov<mode>_softfloat64): Likewise.
232 (movdi_internal32): Likewise.
233 (movdi_internal64): Likewise.
234 (probe_stack_<mode>): Likewise.
236 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
238 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
239 floating point, and decimal floating point to reload iterator.
241 * config/rs6000/constraints.md (wl constraint): New constraints to
242 return FLOAT_REGS if certain options are used to reduce the number
243 of separate patterns that exist in the file.
244 (wx constraint): Likewise.
245 (wz constraint): Likewise.
247 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
248 -mdebug=reg, print wg, wl, wx, and wz constraints.
249 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
250 Initialize the reload functions for 64-bit binary/decimal floating
252 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
253 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
254 create the buffer on the stack to overcome not having a 32-bit
256 (rs6000_emit_move): Likewise.
257 (rs6000_secondary_memory_needed_rtx): Likewise.
258 (rs6000_alloc_sdmode_stack_slot): Likewise.
259 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
260 via xxlxor, just like DFmode 0.0.
262 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
263 define as 1 if we are running on a power7 or newer.
264 (enum r6000_reg_class_enum): Add new constraints.
266 * config/rs6000/dfp.md (movsd): Delete, combine with binary
267 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
268 with other moves by using conditional constraits (wg). Use LFIWZX
269 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
270 (movsd splitter): Likewise.
271 (movsd_hardfloat): Likewise.
272 (movsd_softfloat): Likewise.
274 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
275 binary and decimal floating point moves.
276 (fmove_ok): New attributes to combine binary and decimal floating
277 point moves, and to combine power6x (mfpgpr) moves along normal
279 (real_value_to_target): Likewise.
286 (movsf): Combine binary and decimal floating point moves. Combine
287 power6x (mfpgpr) moves with other moves by using conditional
288 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
289 (mov<mode> for SFmode/SDmode); Likewise.
290 (SFmode/SDmode splitters): Likewise.
291 (movsf_hardfloat): Likewise.
292 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
293 (movsf_softfloat): Likewise.
294 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
296 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
297 wx and wz constraints.
299 * config/rs6000/constraints.md (wg constraint): New constraint to
300 return FLOAT_REGS if -mmfpgpr (power6x) was used.
302 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
305 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
306 -mdebug=reg, print wg, wl, wx, and wz constraints.
307 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
308 Initialize the reload functions for 64-bit binary/decimal floating
310 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
311 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
312 create the buffer on the stack to overcome not having a 32-bit
314 (rs6000_emit_move): Likewise.
315 (rs6000_secondary_memory_needed_rtx): Likewise.
316 (rs6000_alloc_sdmode_stack_slot): Likewise.
317 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
318 via xxlxor, just like DFmode 0.0.
320 * config/rs6000/dfp.md (movdd): Delete, combine with binary
321 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
322 with other moves by using conditional constraits (wg). Use LFIWZX
323 and STFIWX for loading SDmode on power7.
324 (movdd splitters): Likewise.
325 (movdd_hardfloat32): Likewise.
326 (movdd_softfloat32): Likewise.
327 (movdd_hardfloat64_mfpgpr): Likewise.
328 (movdd_hardfloat64): Likewise.
329 (movdd_softfloat64): Likewise.
331 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
332 64-bit binary and decimal floating point moves.
333 (FMOVE64X): Likewise.
334 (movdf): Combine 64-bit binary and decimal floating point moves.
335 Combine power6x (mfpgpr) moves with other moves by using
336 conditional constraits (wg).
337 (mov<mode> for DFmode/DDmode): Likewise.
338 (DFmode/DDmode splitters): Likewise.
339 (movdf_hardfloat32): Likewise.
340 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
341 (movdf_softfloat32): Likewise.
342 (movdf_hardfloat64_mfpgpr): Likewise.
343 (movdf_hardfloat64): Likewise.
344 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
345 (movdf_softfloat64): Likewise.
346 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
347 (reload_<mode>_load): Move to later in the file so they aren't in
348 the middle of the floating point move insns.
349 (reload_<mode>_store): Likewise.
351 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
354 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
355 constraint if -mdebug=reg.
356 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
357 Enable using dd reload support if needed.
359 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
360 binary and decimal floating point moves in rs6000.md.
361 (movtd_internal): Likewise.
363 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
364 decimal floating point moves.
366 (movtf_internal): Likewise.
367 (mov<mode>_internal, TDmode/TFmode): Likewise.
368 (movtf_softfloat): Likewise.
369 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
371 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
372 movdi_internal64, using wg constraint for move direct operations.
373 (movdi_internal64): Likewise.
375 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
376 MODES_TIEABLE_P for selected modes. Print the numerical value of
377 the various virtual registers. Use GPR/FPR first/last values,
378 instead of hard coding the register numbers. Print which modes
379 have reload functions registered.
380 (rs6000_option_override_internal): If -mdebug=reg, trace the options
381 settings before/after setting cpu, target and subtarget settings.
382 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
383 and for secondary reload failures in rs6000_secondary_reload_inner.
384 (rs6000_secondary_reload_fail): Likewise.
385 (rs6000_secondary_reload_inner): Likewise.
387 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
388 macros for first/last GPR and FPR registers.
389 (LAST_GPR_REGNO): Likewise.
390 (FIRST_FPR_REGNO): Likewise.
391 (LAST_FPR_REGNO): Likewise.
393 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
394 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
395 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
396 (vcond<mode><mode>): Likewise.
397 (vcondu<mode><mode>): Likewise.
398 (vector_gtu<mode>): Likewise.
399 (vector_gte<mode>): Likewise.
400 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
401 to prevent the compiler from converting DImode operations to TImode.
402 (ior<mode>3): Likewise.
403 (and<mode>3): Likewise.
404 (one_cmpl<mode>2): Likewise.
405 (nor<mode>3): Likewise.
406 (andc<mode>3): Likewise.
408 * config/rs6000/constraints.md (wt constraint): New constraint
409 that returns VSX_REGS if TImode is allowed in VSX registers.
411 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
414 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
415 similar to TImode, but it is restricted to being in the GPRs.
417 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
418 TImode to occupy a single VSX register.
420 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
421 -mvsx-timode for power7/power8.
422 (power7 cpu): Likewise.
423 (power8 cpu): Likewise.
425 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
426 sure that TFmode/TDmode take up two registers if they are ever
427 allowed in the upper VSX registers.
428 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
430 (rs6000_init_hard_regno_mode_ok): Likewise.
431 (rs6000_debug_reg_global): Add debugging for PTImode and wt
432 constraint. Print if LRA is turned on.
433 (rs6000_option_override_internal): Give an error if -mvsx-timode
434 and VSX is not enabled.
435 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
436 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
437 to reg+offset addressing. Use PTImode when checking offset
438 addresses for validity.
439 (reg_offset_addressing_ok_p): Likewise.
440 (rs6000_legitimate_offset_address_p): Likewise.
441 (rs6000_legitimize_address): Likewise.
442 (rs6000_legitimize_reload_address): Likewise.
443 (rs6000_legitimate_address_p): Likewise.
444 (rs6000_eliminate_indexed_memrefs): Likewise.
445 (rs6000_emit_move): Likewise.
446 (rs6000_secondary_reload): Likewise.
447 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
448 reloads to fpr registers to continue to use reg+offset addressing,
449 but 64-bit reloads to altivec registers need reg+reg addressing.
450 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
451 it. Treat LO_SUM like a PLUS operation.
452 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
453 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
454 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
455 registers to share a register with a smaller sized type, since VSX
456 puts scalars in the upper 64-bits.
457 (print_operand): Add support for PTImode.
458 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
459 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
460 registers, but don't have arithmetic support.
461 (rs6000_memory_move_cost): Add test for VSX.
462 (rs6000_opt_masks): Add -mvsx-timode.
464 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
467 (VSr): Use wt constraint for TImode.
468 (VSv): Drop TImode support.
469 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
470 (vsx_movti_64bit): Likewise.
471 (vsx_movti_32bit): Likewise.
472 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
473 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
474 one '?' on the appropriate output constraint. Do not allow TImode
475 logical operations on 32-bit systems.
476 (vsx_ior<mode>3): Likewise.
477 (vsx_xor<mode>3): Likewise.
478 (vsx_one_cmpl<mode>2): Likewise.
479 (vsx_nor<mode>3): Likewise.
480 (vsx_andc<mode>3): Likewise.
481 (vsx_concat_<mode>): Likewise.
482 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
484 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
485 OPTION_MASK_VSX_TIMODE.
486 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
487 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
489 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
490 (TI2 iterator): New iterator for TImode, PTImode.
491 (wd mode attribute): Add values for vector types.
492 (movti_string): Replace TI move operations with operations for TImode
493 and PTImode. Add support for TImode being allowed in VSX registers.
494 (mov<mode>_string, TImode/PTImode): Likewise.
495 (movti_ppc64): Likewise.
496 (mov<mode>_ppc64, TImode/PTImode): Likewise.
497 (TI mode splitters): Likewise.
499 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
502 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
504 PR tree-optimization/56355
505 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
506 Also handle integers with undefined overflow.
508 2013-03-20 Catherine Moore <clm@codesourcery.com>
509 Maciej W. Rozycki <macro@codesourcery.com>
510 Tom de Vries <tom@codesourcery.com>
511 Nathan Sidwell <nathan@codesourcery.com>
512 Iain Sandoe <iain@codesourcery.com>
513 Nathan Froyd <froydnj@codesourcery.com>
514 Chao-ying Fu <fu@mips.com>
516 * doc/extend.texi: (micromips, nomicromips, nocompression):
517 Document new function attributes.
518 * doc/invoke.texi (minterlink-compressed, mmicromips,
519 m14k, m14ke, m14kec): Document new options.
520 (minterlink-mips16): Update documentation.
521 * doc/md.texi (ZC, ZD): Document new constraints.
522 * configure.ac (gcc_cv_as_micromips): Check if linker
523 supports the .set micromips directive.
524 * configure: Regenerate.
525 * config.in: Regenerate.
526 * config/mips/mips-tables.opt: Regenerate.
527 * config/mips/micromips.md: New file.
528 * constraints.md (ZC, ZD): New constraints.
529 * config/mips/predicates.md (movep_src_register): New predicate.
530 (movep_src_operand): New predicate.
531 (non_volatile_mem_operand): New predicate.
532 * config/mips/mips.md (multimem): New type.
533 (length): Differentiate between 17-bit and 18-bit branch offsets.
534 (MOVEP1, MOVEP2): New mode iterator.
535 (mov_<load>l): Use ZC constraint.
536 (mov_<load>r): Likewise.
537 (mov_<store>l): Likewise.
538 (mov_<store>r): Likewise.
539 (*branch_equality<mode>_inverted): Add microMIPS support.
540 (*branch_equality<mode>): Likewise.
541 (*jump_absolute): Likewise.
542 (indirect_jump_<mode>): Likewise.
543 (tablejump_<mode>): Likewise.
544 (<optab>_internal): Likewise.
545 (sibcall_internal): Likewise.
546 (sibcall_value_internal): Likewise.
547 (prefetch): Use constraint ZD.
548 * config/mips/mips.opt (minterlink-compressed): New option.
549 (minterlink-mips16): Now an alias for minterlink-compressed.
550 (mmicromips): New option.
551 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
552 (compare_and_swap_12): Likewise.
553 (sync_add<mode>): Likewise.
554 (sync_<optab>_12): Likewise.
555 (sync_old_<optab>_12): Likewise.
556 (sync_new_<optab>_12): Likewise.
557 (sync_nand_12): Likewise.
558 (sync_old_nand_12): Likewise.
559 (sync_new_nand_12): Likewise.
560 (sync_sub<mode>): Likewise.
561 (sync_old_add<mode>): Likewise.
562 (sync_old_sub<mode>): Likewise.
563 (sync_new_add<mode>): Likewise.
564 (sync_new_sub<mode>): Likewise.
565 (sync_<optab><mode>): Likewise.
566 (sync_old_<optab><mode>): Likewise.
567 (sync_new_<optab><mode>): Likewise.
568 (sync_nand<mode>): Likewise.
569 (sync_old_nand<mode>): Likewise.
570 (sync_new_nand<mode>): Likewise.
571 (sync_lock_test_and_set<mode>): Likewise.
572 (test_and_set_12): Likewise.
573 (atomic_compare_and_swap<mode>): Likewise.
574 (atomic_exchange<mode>_llsc): Likewise.
575 (atomic_fetch_add<mode>_llsc): Likewise.
576 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
577 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
578 (umips_save_restore_pattern_p): Likewise.
579 (umips_load_store_pair_p): Likewise.
580 (umips_output_load_store_pair): Likewise.
581 (umips_movep_target_p): Likewise.
582 (umips_12bit_offset_address_p): Likewise.
583 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
584 (mips_base_mips16): Rename this...
585 (mips_base_compression_flags): ...to this. Update all uses.
586 (mips_attribute_table): Add micromips, nomicromips and nocompression.
587 (mips_mips16_decl_p): Delete.
588 (mips_nomips16_decl_p): Delete.
589 (mips_get_compress_on_flags): New function.
590 (mips_get_compress_off_flags): New function.
591 (mips_get_compress_mode): New function.
592 (mips_get_compress_on_name): New function.
593 (mips_get_compress_off_name): New function.
594 (mips_insert_attributes): Support multiple compression types.
595 (mips_merge_decl_attributes): Likewise.
596 (umips_12bit_offset_address_p): New function.
597 (mips_start_function_definition): Emit .set micromips directive.
598 (mips_call_may_need_jalx_p): New function.
599 (mips_function_ok_for_sibcall): Add microMIPS support.
600 (mips_print_operand_punctuation): Support short delay slots and
602 (umips_swm_mask, umips_swm_encoding): New.
603 (umips_build_save_restore): New function.
604 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
605 (was_mips16_p): Remove.
606 (old_compression_mode): New.
607 (mips_set_compression_mode): New function.
608 (mips_set_current_function): Add microMIPS support.
609 (mips_option_override): Likewise.
610 (umips_save_restore_pattern_p): New function.
611 (umips_output_save_restore): New function.
612 (umips_load_store_pair_p_1): New function.
613 (umips_load_store_pair_p): New function.
614 (umips_output_load_store_pair_1): New function.
615 (umips_output_load_store_pair): New function.
616 (umips_movep_target_p) New function.
617 (mips_prepare_pch_save): Add microMIPS support.
618 * config/mips/mips.h (TARGET_COMPRESSION): New.
619 (TARGET_CPU_CPP_BUILTINS): Update macro
620 to use new compression flags and to support microMIPS.
621 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
622 (MIPS_ARCH_FLOAT_SPEC): Likewise.
623 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
624 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
625 (ASM_SPEC): Support mmicromips and mno-micromips.
626 (M16STORE_REG_P): New macro.
627 (MIPS_CALL): Support TARGET_MICROMIPS.
628 (MICROMIPS_J): New macro.
629 (mips_base_mips16): Rename this...
630 (mips_base_compression_flags): ...to this.
631 (UMIPS_12BIT_OFFSET_P): New macro.
632 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
633 (MULTILIB_DIRNAMES): Likewise.
634 2013-03-20 Richard Biener <rguenther@suse.de>
636 PR tree-optimization/56661
637 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
638 the result does not have to be distinct.
640 2013-03-20 Richard Biener <rguenther@suse.de>
642 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
645 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
646 Steven Bosscher <steven@gcc.gnu.org>
648 PR rtl-optimization/56605
649 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
651 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
654 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
655 that require movd instead of movq.
657 2013-03-20 Richard Biener <rguenther@suse.de>
659 * tree-ssa-structalias.c (struct variable_info): Add pointer
660 to the first field of an aggregate with sub-vars. Make
661 this and the pointer to the next subfield its ID.
662 (vi_next): New function.
663 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
664 storedanything_id, integer_id): Increment by one.
665 (new_var_info, get_call_vi, lookup_call_clobber_vi,
666 get_call_clobber_vi): Adjust.
667 (solution_set_expand): Simplify and speedup.
668 (solution_set_add): Inline into ...
669 (set_union_with_increment): ... this. Adjust accordingly.
670 (do_sd_constraint): Likewise.
671 (do_ds_constraint): Likewise.
672 (do_complex_constraint): Simplify.
673 (build_pred_graph): Adjust.
674 (solve_graph): Likewise. Simplify and speedup.
675 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
676 get_constraint_for_component_ref, get_constraint_for_1,
677 first_vi_for_offset, first_or_preceding_vi_for_offset,
678 create_function_info_for, create_variable_info_for_1,
679 create_variable_info_for, intra_create_variable_infos): Adjust.
680 (init_base_vars): Push NULL for ID zero.
681 (compute_points_to_sets): Adjust.
683 2013-03-20 Richard Biener <rguenther@suse.de>
685 * cfgloop.c (verify_loop_structure): Streamline and avoid
686 ICEing on corrupt loop tree.
687 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
690 2013-03-20 Richard Biener <rguenther@suse.de>
692 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
693 check whether an SSA update is needed.
695 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
697 * config/mips/constraints.md (T): Rename to...
701 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
702 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
704 2013-03-19 Ian Bolton <ian.bolton@arm.com>
706 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
707 (*subsi3_carryin_uxtw): Likewise.
709 2013-03-19 Ian Bolton <ian.bolton@arm.com>
711 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
712 (*rorsi3_insn_uxtw): Likewise.
714 2013-03-19 Ian Bolton <ian.bolton@arm.com>
716 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
717 (*extrsi5_insn_uxtw): Likewise.
719 2013-03-19 Richard Biener <rguenther@suse.de>
721 PR tree-optimization/56273
722 * passes.c (init_optimization_passes): Move second VRP after DOM.
724 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
726 * config/i386/i386.md (*movti_internal): Merge from
727 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
728 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
729 nox64 isa attributes.
731 2013-03-18 Richard Biener <rguenther@suse.de>
733 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
735 (merge_node_constraints): Likewise.
736 (build_succ_graph): Likewise.
737 (valid_graph_edge): Inline into single caller.
738 (unify_nodes): Likewise. Use bitmap_set_bit return value
740 (scc_visit): Fix formatting and variable use.
741 (do_sd_constraint): Use gcc_checking_assert.
742 (do_ds_constraint): Likewise.
743 (do_complex_constraint): Likewise.
744 (condense_visit): Likewise. Cleanup.
745 (dump_pred_graph): New function.
746 (perform_var_substitution): Dump the pred-graph before
747 variable substitution.
748 (find_equivalent_node): Use gcc_checking_assert.
749 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
751 2013-03-18 Richard Biener <rguenther@suse.de>
753 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
754 Remove cond_expr_stmt_list argument and do not gimplify the
756 (vect_loop_versioning): Adjust.
757 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
758 Cleanup to use less temporaries.
759 (vect_create_data_ref_ptr): Cleanup.
761 2013-03-18 Jakub Jelinek <jakub@redhat.com>
763 PR tree-optimization/56635
764 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
765 require types_compatible_p types.
767 2013-03-18 Nick Clifton <nickc@redhat.com>
769 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
772 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
773 Add missing line to comment describing function.
775 2013-03-18 Richard Biener <rguenther@suse.de>
777 PR tree-optimization/56210
778 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
779 Handle string / character search functions.
780 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
782 2013-03-18 Richard Biener <rguenther@suse.de>
785 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
786 and implement properly.
787 * gimple.h (gimple_cond_single_var_p): Remove.
789 2013-03-18 Richard Biener <rguenther@suse.de>
791 * tree-data-ref.h (find_data_references_in_loop): Declare.
792 * tree-data-ref.c (get_references_in_stmt): Use a stack
793 vector pre-allocated in the callers.
794 (find_data_references_in_stmt): Adjust.
795 (graphite_find_data_references_in_stmt): Likewise.
796 (create_rdg_vertices): Likewise.
797 (find_data_references_in_loop): Export.
798 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
799 Compute dependences here...
800 (vect_analyze_data_refs): ...not here. When we encounter
801 a non-vectorizable data reference in basic-block vectorization
802 truncate the data reference vector. Do not bother to
803 fixup data-dependence information for gather loads.
804 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
805 of data references, as reported.
807 2013-03-18 Richard Biener <rguenther@suse.de>
809 PR tree-optimization/3713
810 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
811 has_constants and expr.
812 (stmt_has_constants): Properly valueize SSA names when deciding
813 whether the stmt has constants.
815 2013-03-18 Richard Biener <rguenther@suse.de>
817 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
818 whole function when there is nothing to do.
819 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
820 * tree-vectorizer.c (vectorize_loops): Update virtual and
821 loop-closed SSA once.
822 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
824 2013-03-18 Richard Biener <rguenther@suse.de>
827 * domwalk.c (bb_postorder): New global static.
828 (cmp_bb_postorder): New function.
829 (walk_dominator_tree): Replace scheme imposing an order for
830 visiting dominator sons by one sorting them at the time they
831 are pushed on the stack.
833 2013-03-18 Richard Biener <rguenther@suse.de>
835 PR tree-optimization/39326
836 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
837 (struct mem_ref): Replace mem member with ao_ref typed member.
838 (MEM_ANALYZABLE): Adjust.
839 (memref_eq): Likewise.
840 (mem_ref_alloc): Likewise.
841 (gather_mem_refs_stmt): Likewise.
842 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
843 (execute_sm_if_changed_flag_set): Adjust.
844 (execute_sm): Likewise.
845 (ref_always_accessed_p): Likewise.
846 (refs_independent_p): Likewise.
847 (can_sm_ref_p): Likewise.
849 2013-03-18 Jakub Jelinek <jakub@redhat.com>
852 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
853 return 1 even for !unsignedp.
855 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
857 * config/i386/i386.md (isa): Add x64 and nox64.
858 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
859 (*pushtf): Enable *roF alternative for x64 isa only.
860 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
861 mode attribute of integer alternatives to DImode for TARGET_64BIT.
862 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
863 (*movtf_internal): Merge from *movtf_internal_rex64 and
864 *movtf_internal_sse. Use x64 and nox64 isa attributes.
865 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
866 nox64 isa attributes.
867 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
868 nox64 isa attributes.
869 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
871 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
873 * config/alpha/alpha.c (TARGET_LRA_P): New define.
875 2013-03-17 Jakub Jelinek <jakub@redhat.com>
878 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
879 class names. Remove trailing comma after "ALL_REGS".
881 2013-03-16 Jan Hubicka <jh@suse.cz>
883 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
884 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
885 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
886 of cgraph_get_create_node.
887 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
889 2013-03-16 Jason Merrill <jason@redhat.com>
892 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
893 with DW_AT_default_value.
895 2013-03-16 Jakub Jelinek <jakub@redhat.com>
897 * BASE-VER: Set to 4.9.0.
899 2013-03-14 Andi Kleen <ak@linux.intel.com>
902 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
903 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
904 Document _x* TSX intrinsics.
906 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
907 David Holsgrove <david.holsgrove@xilinx.com>
909 * configure.ac: Add MicroBlaze TLS support detection.
910 * configure: Regenerate.
911 * config/microblaze/microblaze-protos.h
912 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
913 symbol_mentioned_p, label_mentioned_p): Add prototypes.
914 * config/microblaze/microblaze.c (microblaze_address_type): Add
915 ADDRESS_TLS and tls_reloc address types.
916 (microblaze_address_info): Add tls_reloc.
917 (TARGET_HAVE_TLS): Define.
918 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
919 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
920 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
921 load_tls_operand, microblaze_call_tls_get_addr,
922 microblaze_legitimize_tls_address): New functions.
923 (microblaze_classify_unspec): Handle UNSPEC_TLS.
924 (get_base_reg): Use microblaze_tls_symbol_p.
925 (microblaze_classify_address): Handle TLS.
926 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
927 label_mentioned_p and microblaze_tls_referenced_p.
928 (microblaze_legitimize_address): Handle TLS.
929 (microblaze_address_insns): Handle ADDRESS_TLS.
930 (pic_address_needs_scratch): Handle TLS.
931 (print_operand_address): Handle TLS.
932 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
933 (microblaze_expand_move): Handle TLS.
934 (microblaze_legitimate_constant_p): Check
935 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
936 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
937 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
938 (PIC_OFFSET_TABLE_REGNUM): Set.
939 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
940 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
941 (addsi3, movsi_internal2, movdf_internal): Update constraints
942 * config/microblaze/predicates.md (arith_plus_operand): Define
943 (move_operand): Redefine as move_src_operand,
944 check microblaze_tls_referenced_p.
946 2013-03-14 Ian Bolton <ian.bolton@arm.com>
948 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
949 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
951 2013-03-14 Ian Bolton <ian.bolton@arm.com>
953 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
956 2013-03-14 Jakub Jelinek <jakub@redhat.com>
958 PR tree-optimization/53265
959 * common.opt (Waggressive-loop-optimizations): New option.
960 * tree-ssa-loop-niter.c: Include tree-pass.h.
961 (do_warn_aggressive_loop_optimizations): New function.
962 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
963 if number_of_latch_executions returned constant.
964 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
965 early. If number_of_latch_executions returned constant, set
966 nb_iterations_upper_bound back to it.
967 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
969 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
970 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
972 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
973 (MULTILIB_OSDIRNAMES): Set.
974 * genmultilib: If defaultosdirname doesn't start with :: , set
975 defaultosdirname2 instead, clear it and emit two . multilib_raw
976 entries instead of just one.
978 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
980 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
981 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
982 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
983 (SUBTARGET_OVERRIDE_OPTIONS): New.
985 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
988 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
989 (musermode): Convert to Var(TARGET_USERMODE).
990 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
991 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
992 * config/sh/sh.c (sh_option_override): Use
993 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
994 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
996 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
998 (udivsi3_i4_single, divsi3_i4_single): Use
999 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
1001 2013-03-13 Dave Korn <dave.korn.cygwin@....>
1003 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
1006 2013-03-13 Richard Biener <rguenther@suse.de>
1008 PR tree-optimization/56608
1009 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
1010 calls when vectorizing basic-blocks.
1012 2013-03-13 Jakub Jelinek <jakub@redhat.com>
1015 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
1018 2013-03-12 Jakub Jelinek <jakub@redhat.com>
1020 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
1022 2013-03-11 Jan Hubicka <jh@suse.cz>
1025 * lto-streamer-out.c (output_symbol_p): Skip references from
1026 constructors of external variables.
1028 2013-03-11 Jan Hubicka <jh@suse.cz>
1031 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
1033 * emit-rtl.c (verify_rtx_sharing): Likewise.
1034 (copy_insn_1): Likewise.
1035 * rtl.c (copy_rtx): Likewise.
1037 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
1040 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
1041 output_operand_lossage message.
1043 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
1046 * arm.c (shift_op): Validate RTL pattern on the fly.
1047 (arm_print_operand, case 'S'): Don't use shift_operator to validate
1050 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1053 * config/pa/pa.md (call_value): Check for calls to powf and direct to
1054 new call patterns that clobber %fr12.
1055 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
1056 split and postreload patterns.
1057 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
1058 registers %fr12 and %fr12R as call used.
1060 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
1062 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
1063 (canon_address, record_store, replace_read, check_mem_read_rtx,
1064 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
1065 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
1066 rest_of_handle_dse): Likewise.
1068 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
1071 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
1073 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
1074 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
1075 (save_optabs_if_changed): Replace with...
1076 (init_tree_optimization_optabs): ...this.
1077 * optabs.c (save_optabs_if_changed): Rename to...
1078 (init_tree_optimization_optabs): ...this. Take the optimization node
1079 as argument. Do nothing if the base optabs are already correct.
1080 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
1081 to recompute optabs.
1082 * function.h (function): Remove optabs field.
1083 * function.c (invoke_set_current_function_hook): Call
1084 init_tree_optimization_optabs. Use the result to initialize
1087 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
1089 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
1090 if GTMA_HAS_NO_INSTRUMENTATION.
1091 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
1092 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
1093 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
1094 * gimple-pretty-print.c (dump_gimple_transaction): Handle
1095 GTMA_HAS_NO_INSTRUMENTATION.
1097 2013-03-08 Jakub Jelinek <jakub@redhat.com>
1099 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
1102 2013-03-08 Marek Polacek <polacek@redhat.com>
1103 Jakub Jelinek <jakub@redhat.com>
1105 PR tree-optimization/56478
1106 * predict.c (is_comparison_with_loop_invariant_p): Change the
1107 type of loop_step to tree.
1108 (predict_loops): Adjust.
1109 (predict_iv_comparison): Perform the computations on double_ints.
1111 2013-03-08 Richard Biener <rguenther@suse.de>
1113 PR tree-optimization/56570
1114 * tree-cfg.c (verify_expr_location_1): Verify locations for
1116 * tree-sra.c (create_access_replacement): Strip locations
1117 from DECL_DEBUG_EXPRs.
1119 2013-03-08 Richard Biener <rguenther@suse.de>
1121 * tree-inline.c (expand_call_inline): Do not associate
1122 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
1123 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
1125 2013-03-08 Richard Biener <rguenther@suse.de>
1127 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
1128 or block changes with -Og. Fix for location / block encoding
1129 changes and PHI arguments with locations.
1131 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
1133 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
1135 (struct output_info): Likewise.
1136 (register_overhead): Remove bad gcc_assert.
1137 (bitmap_find_bit): If there is only a single bitmap element, do not
1138 count a miss as a search.
1139 (print_statistics): Update for counter type changes.
1140 (dump_bitmap_statistics): Likewise. Print headers such that they
1141 are properly lined up with the printed counters.
1143 2013-03-07 Jakub Jelinek <jakub@redhat.com>
1145 PR tree-optimization/56559
1146 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
1147 check that it has only a single use.
1149 2013-03-07 Richard Biener <rguenther@suse.de>
1151 * doc/invoke.texi (fwhole-program): Discourage use in combination
1154 2013-03-06 Jakub Jelinek <jakub@redhat.com>
1156 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
1158 PR tree-optimization/56539
1159 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
1160 instead of GSI_CONTINUE_LINKING as last argument to
1161 force_gimple_operand_gsi. Adjust function comment.
1163 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
1167 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
1168 promoted mode, convert the result back to the original mode.
1170 2013-03-06 Richard Biener <rguenther@suse.de>
1173 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
1174 (insert_updated_phi_nodes_compare_uids): New function.
1175 (update_ssa): Sort symbols_to_rename after UID before
1176 traversing it to insert PHI nodes.
1178 2013-03-06 Richard Biener <rguenther@suse.de>
1181 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
1182 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
1185 2013-02-13 Richard Biener <rguenther@suse.de>
1188 * varasm.c (output_constant_def_1): Get the decl representing
1189 the constant as argument.
1190 (output_constant_def): Wrap output_constant_def_1.
1191 (make_decl_rtl): Use output_constant_def_1 with the decl
1192 representing the constant.
1193 (build_constant_desc): Optionally re-use a decl already
1194 representing the constant.
1195 (tree_output_constant_def): Adjust.
1197 2013-03-06 Joey Ye <joey.ye@arm.com>
1200 * gcc.c (convert_white_space): New function.
1201 (main): Handles white space in function name.
1203 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
1206 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
1207 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
1208 to SH_DIV_CALL_TABLE for TARGET_SH2.
1209 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
1211 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
1214 2013-03-05 Sterling Augustine <saugustine@google.com>
1215 Cary Coutant <ccoutant@google.com>
1218 * dwarf2out.c (resolve_addr): Don't call
1219 remove_loc_list_addr_table_entries a second time for the same
1222 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1225 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
1226 (avoid_complex_debug_insns): New function.
1227 (expand_debug_locations): Call it.
1229 PR rtl-optimization/56484
1230 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
1231 lifetimes of hard registers on small register class machines.
1233 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
1235 * config/microblaze/microblaze-protos.h: Rename
1236 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
1237 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
1239 (microblaze_fast_interrupt_function_p): New function.
1240 (microblaze_is_interrupt_handler): Rename to
1241 microblaze_is_interrupt_variant and add fast_interrupt check.
1242 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
1243 (save_restore_insns): Likewise.
1244 (compute_frame_size): Likewise.
1245 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
1246 (microblaze_globalize_label): Likewise.
1247 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
1248 * config/microblaze/microblaze.md: Use wrapper
1249 microblaze_is_interrupt_variant.
1251 2013-03-05 Kai Tietz <ktietz@redhat.com>
1253 * sdbout.c (sdbout_one_type): Switch to current function's section
1254 supporting cold/hot.
1256 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
1258 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
1261 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1264 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
1265 if VALGRIND_GET_VBITS is defined, temporarily make object
1266 memory all defined, and restore previous valgrind addressability
1267 and definability afterwards. Free this_object at the end.
1270 * lra.c (lra): Call lra_clear_live_ranges if live_p,
1271 right before calling lra_create_live_ranges, also call it
1272 when clearing live_p. Only call lra_clear_live_ranges
1273 at the end if live_p.
1276 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
1278 2013-03-05 Richard Biener <rguenther@suse.de>
1280 PR tree-optimization/56521
1281 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
1284 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
1287 * except.h (remove_unreachable_eh_regions): New prototype.
1288 * except.c (remove_eh_handler_splicer): New function, split out
1289 of remove_eh_handler.
1290 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
1291 warning about running it on many EH regions one at a time.
1292 (remove_unreachable_eh_regions_worker): New function, walk the
1293 EH tree in depth-first order and remove non-marked regions.
1294 (remove_unreachable_eh_regions): New function.
1295 * tree-eh.c (mark_reachable_handlers): New function, split out
1296 from remove_unreachable_handlers.
1297 (remove_unreachable_handlers): Use mark_reachable_handlers and
1298 remove_unreachable_eh_regions.
1299 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
1300 and remove_unreachable_eh_regions.
1302 2013-03-05 Richard Biener <rguenther@suse.de>
1305 * loop-init.c (fix_loop_structure): Remove loops in two stages,
1306 not freeing them until the end.
1308 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1310 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
1312 2013-03-05 Richard Biener <rguenther@suse.de>
1314 PR tree-optimization/56270
1315 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
1316 of loads after scheduling an SLP instance.
1318 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1320 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
1322 (check_gcc_parallelize): Run guality.exp as a separate job from
1323 vect.exp with unsorted.exp and $(dg_target_exps) separately from
1324 struct-layout-1.exp with stackalign.exp.
1326 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
1329 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
1330 load_index sbitmap even if some bit in it isn't set.
1333 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
1334 (discover_iteration_bound_by_body_walk): Change queues to
1335 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
1336 spelling in comment. Call safe_push on queues[bound_index] directly.
1337 Release queues[queue_index] in every iteration unconditionally.
1338 Release bounds vector.
1341 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
1342 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
1343 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
1346 2013-03-05 Richard Biener <rguenther@suse.de>
1349 * tree-inline.c (remap_blocks_to_null): New function.
1350 (expand_call_inline): When expanding a call stmt without
1351 an associated block inline remap all callee blocks to NULL.
1353 2013-03-05 Jakub Jelinek <jakub@redhat.com>
1355 PR rtl-optimization/56494
1356 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
1357 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
1358 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
1361 * sel-sched-ir.c (free_sched_pools): Release
1362 succs_info_pool.stack[succs_info_pool.max_top] vectors too
1363 if succs_info_pool.max_top isn't -1.
1366 * opts.c (opts_obstack, opts_concat): Moved to...
1367 * opts-common.c (opts_obstack, opts_concat): ... here.
1369 2013-03-04 Jakub Jelinek <jakub@redhat.com>
1372 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
1374 2013-03-04 Martin Jambor <mjambor@suse.cz>
1376 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
1377 all appropriate places.
1379 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
1381 PR tree-optimization/56424
1382 * ipa-split.c (split_function): Do not set the RSO flag if result is
1383 not by reference and its type is a register type.
1385 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
1387 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
1388 (microblaze_legitimate_pic_operand): Likewise
1389 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
1390 new function microblaze_legitimate_pic_operand
1391 * config/microblaze/microblaze-protos.h
1392 (microblaze_legitimate_pic_operand): Declare.
1394 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1396 * config/microblaze/predicates.md (call_insn_simple_operand):
1397 New predicate for supported rtx code types.
1398 * config/microblaze/microblaze.md (call_internal1): Use
1399 call_insn_simple_operand predicate.
1401 2013-03-04 Jakub Jelinek <jakub@redhat.com>
1404 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
1405 partitions.ordered_remove.
1408 * tree-vect-stmts.c (vectorizable_conversion): Don't call
1409 vec_oprnds0.create (1) for modifier == NONE.
1412 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
1413 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
1414 vec_oprnds1 right before pushing anything to it for
1418 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
1419 set nbbs to 0 instead of having separate code path.
1420 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
1421 instead of false as last argument if returning NULL.
1423 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
1425 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
1426 the attribute is now called "target" instead of "option".
1427 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
1428 * doc/tm.texi.in (Target Attributes): Likewise document the correct
1429 attribute/pragma name for TARGET_OPTION_VALID_P and
1430 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
1431 * doc/tm.texi: Regenerated.
1433 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
1435 * config/microblaze/microblaze.c:
1436 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
1437 * config/microblaze/microblaze.h: Add -mxl-reorder to
1439 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
1440 instructions emitted if TARGET_REORDER.
1441 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
1442 or 0 for -m/-mno case, but initialises as 2 to detect default use case
1445 2013-03-01 Xinliang David Li <davidxl@google.com>
1447 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
1450 2013-03-01 Jakub Jelinek <jakub@redhat.com>
1453 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
1454 vector even when returning true. Fix up function comment formatting.
1457 * ira-build.c (ira_loop_nodes_count): New variable.
1458 (create_loop_tree_nodes): Initialize it.
1459 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
1462 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
1463 method on dr_chain and result_chain.
1464 * tree-vect-stmts.c (vectorizable_store): Only call
1465 result_chain.create if j == 0.
1468 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
1469 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
1470 before overwriting it.
1472 2013-03-01 Tobias Burnus <burnus@net-b.de>
1474 * doc/extended.texi (C Extensions): Change order in @menu
1476 (Other MIPS Built-in Functions): Move last MIPS entry before
1477 "picoChip Built-in Functions".
1478 (SH Built-in Functions): Move after RX Built-in Functions.
1479 * doc/gcc.texi (Introduction): Change order in @menu
1481 * doc/md.texi (Constraints): Ditto.
1482 * gty.texi (Type Information): Ditto.
1483 (User-provided marking routines for template types): Make
1485 * doc/invoke.texi (AArch64 Options): Move before
1486 "Adapteva Epiphany Options".
1488 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
1489 Jakub Jelinek <jakub@redhat.com>
1492 * asan.c (gate_asan): Lookup no_sanitize_address instead of
1493 no_address_safety_analysis attribute.
1494 * doc/extend.texi (no_address_safety_attribute): Rename to
1495 no_sanitize_address attribute, mention no_address_safety_analysis
1496 attribute as deprecated alias.
1498 2013-02-28 Jakub Jelinek <jakub@redhat.com>
1501 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
1502 type to vec<vec<tree> > *.
1503 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
1504 to be vec<tree> instead of vec<tree> *, set vec_defs
1505 to vNULL and call vec_defs.create (number_of_vects), adjust other
1507 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
1508 vectorizable_condition): Adjust vect_get_slp_defs callers.
1510 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1512 * config/aarch64/aarch64.c
1513 (aarch64_float_const_representable): Remove unused variable.
1515 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1517 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
1519 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1521 * config/aarch64/aarch64-builtins.c
1522 (aarch64_init_simd_builtins): Make static.
1524 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
1526 * config/aarch64/aarch64.c
1527 (aarch64_simd_make_constant): Make static.
1529 2013-02-28 Martin Jambor <mjambor@suse.cz>
1531 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
1532 with no initialization to the RHS of debug statements.
1534 2013-02-28 Martin Jambor <mjambor@suse.cz>
1536 PR tree-optimization/56294
1537 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
1539 (get_access_replacement): Do not call create_access_replacement.
1540 Assert a replacement exists.
1541 (get_repl_default_def_ssa_name): Create the replacement declaration
1544 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1546 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
1549 2013-02-28 Marek Polacek <polacek@redhat.com>
1551 PR rtl-optimization/56466
1552 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
1553 if we're changing a loop.
1554 (peel_loops_completely): Likewise.
1556 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1559 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
1561 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1564 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
1565 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
1566 INTX_FTYPE_FX, FX_FTYPE_INTX.
1567 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
1569 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1571 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
1572 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
1573 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
1574 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
1575 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
1576 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
1577 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
1578 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
1579 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
1580 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
1581 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
1582 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
1583 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
1584 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
1585 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
1586 (avrxmega6): Increase max flash segments from 5 to 6.
1587 * config/avr/t-multilib: Regenerate.
1588 * config/avr/avr-tables.opt: Regenerate.
1589 * doc/avr-mmcu.texi: Regenerate.
1591 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
1593 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
1594 (avr_device_to_arch): Rename to avr_device_to_ld.
1595 (avr_device_to_as): New prototype.
1596 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
1597 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
1598 * config/avr/driver-avr.c (avr_device_to_as): New.
1599 (avr_device_to_arch): Rename to avr_device_to_ld.
1601 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1604 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
1605 method on dr_chain and result_chain.
1608 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
1609 pointer_set_destroy on not_executed_last_iteration.
1612 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
1615 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
1616 FOR_EACH_DEFINED_FUNCTION when freeing state.
1619 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
1621 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
1625 * ipa-cp.c (decide_whether_version_node): Call vec_free on
1626 known_aggs[i].items and release known_aggs vector.
1629 * ipa-reference.c (propagate): Free node_info even for alias nodes.
1631 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
1633 * config/microblaze/microblaze.c (microblaze_emit_compare):
1634 Use xor for EQ/NE comparisions.
1635 * config/microblaze/microblaze.md (cstoresf4): Add constraints
1636 (cbranchsf4): Adjust operator to comparison_operator.
1638 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1641 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
1643 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
1644 vec_safe_push, always update *slot.
1645 (redirect_edge_var_map_clear): Use vec_free.
1646 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
1647 (free_var_map_entry): Use vec_free.
1648 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
1649 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
1651 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
1654 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
1655 when the may_trap_p bit of the exprs being merged differs.
1656 Reorder tests for speculativeness in the logical and operator.
1658 2013-02-27 Jakub Jelinek <jakub@redhat.com>
1660 * incpath.c (add_standard_paths): Use reconcat instead of concat
1661 where appropriate and avoid leaking memory.
1663 * opts.h: Include obstack.h.
1664 (opts_concat): New prototype.
1665 (opts_obstack): New declaration.
1666 * opts.c (opts_concat): New function.
1667 (opts_obstack): New variable.
1668 (init_options_struct): Call gcc_init_obstack on opts_obstack.
1669 (finish_options): Use opts_concat instead of concat
1670 and XOBNEWVEC instead of XNEWVEC.
1671 * opts-common.c (generate_canonical_option, decode_cmdline_option,
1672 generate_option): Likewise.
1673 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
1674 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
1677 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
1678 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
1680 2013-02-26 Jakub Jelinek <jakub@redhat.com>
1683 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
1685 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
1687 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
1688 (arm_block_move_unaligned_straight): Likewise.
1689 (arm_adjust_block_mem): Likewise.
1691 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
1694 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
1695 temp, cond and label.
1696 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
1699 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
1700 * config/c6x/c6x.h (dbx_register_map): Update declaration.
1703 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
1704 of prologue/epilogue functions.
1707 * config/tilegx/tilegx.c (tilegx_expand_prologue):
1708 Remove unused variable cfa_offset.
1709 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
1712 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
1713 type promotion to unsigned.
1716 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
1717 for HOST_WIDE_INT of 32 bit / same size as int.
1718 (arm_block_move_unaligned_straight): Likewise.
1719 (arm_adjust_block_mem): Likewise.
1722 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
1725 2013-02-26 Marek Polacek <polacek@redhat.com>
1727 PR tree-optimization/56426
1728 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
1730 2013-02-26 Richard Biener <rguenther@suse.de>
1733 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
1734 unused variable loops.
1736 2013-02-26 Jakub Jelinek <jakub@redhat.com>
1738 PR tree-optimization/56448
1739 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
1740 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
1741 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
1742 later operands of the references, or even first operand for
1743 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
1745 PR tree-optimization/56443
1746 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
1747 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
1748 to type_for_mode langhook.
1750 2013-02-25 Matt Turner <mattst88@gmail.com>
1752 * doc/invoke.texi: Document r4700.
1754 2013-02-25 Richard Biener <rguenther@suse.de>
1756 PR tree-optimization/56175
1757 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
1759 (simplify_bitwise_binary): ... here. Also guard the conversion
1760 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
1762 2013-02-25 Catherine Moore <clm@codesourcery.com>
1765 2013-02-24 Catherine Moore <clm@codesourcery.com>
1766 Maciej W. Rozycki <macro@codesourcery.com>
1767 Tom de Vries <tom@codesourcery.com>
1768 Nathan Sidwell <nathan@codesourcery.com>
1769 Iain Sandoe <iain@codesourcery.com>
1770 Nathan Froyd <froydnj@codesourcery.com>
1771 Chao-ying Fu <fu@mips.com>
1773 * doc/extend.texi: (micromips, nomicromips, nocompression):
1774 Document new function attributes.
1775 * doc/invoke.texi (minterlink-compressed, mmicromips,
1776 m14k, m14ke, m14kec): Document new options.
1777 (minterlink-mips16): Update documentation.
1778 * doc/md.texi (ZC, ZD): Document new constraints.
1779 * configure.ac (gcc_cv_as_micromips): Check if linker
1780 supports the .set micromips directive.
1781 * configure: Regenerate.
1782 * config.in: Regenerate.
1783 * config/mips/mips-tables.opt: Regenerate.
1784 * config/mips/micromips.md: New file.
1785 * constraints.md (ZC, AD): New constraints.
1786 * config/mips/predicates.md (movep_src_register): New predicate.
1787 (movep_src_operand): New predicate.
1788 (non_volatile_mem_operand): New predicate.
1789 * config/mips/mips.md (multimem): New type.
1790 (length): Differentiate between 17-bit and 18-bit branch offsets.
1791 (MOVEP1, MOVEP2): New mode iterator.
1792 (mov_<load>l): Use ZC constraint.
1793 (mov_<load>r): Likewise.
1794 (mov_<store>l): Likewise.
1795 (mov_<store>r): Likewise.
1796 (*branch_equality<mode>_inverted): Add microMIPS support.
1797 (*branch_equality<mode>): Likewise.
1798 (*jump_absolute): Likewise.
1799 (indirect_jump_<mode>): Likewise.
1800 (tablejump_<mode>): Likewise.
1801 (<optab>_internal): Likewise.
1802 (sibcall_internal): Likewise.
1803 (sibcall_value_internal): Likewise.
1804 (prefetch): Use constraint ZD.
1805 * config/mips/mips.opt (minterlink-compressed): New option.
1806 (minterlink-mips16): Now an alias for minterlink-compressed.
1807 (mmicromips): New option.
1808 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
1809 (compare_and_swap_12): Likewise.
1810 (sync_add<mode>): Likewise.
1811 (sync_<optab>_12): Likewise.
1812 (sync_old_<optab>_12): Likewise.
1813 (sync_new_<optab>_12): Likewise.
1814 (sync_nand_12): Likewise.
1815 (sync_old_nand_12): Likewise.
1816 (sync_new_nand_12): Likewise.
1817 (sync_sub<mode>): Likewise.
1818 (sync_old_add<mode>): Likewise.
1819 (sync_old_sub<mode>): Likewise.
1820 (sync_new_add<mode>): Likewise.
1821 (sync_new_sub<mode>): Likewise.
1822 (sync_<optab><mode>): Likewise.
1823 (sync_old_<optab><mode>): Likewise.
1824 (sync_new_<optab><mode>): Likewise.
1825 (sync_nand<mode>): Likewise.
1826 (sync_old_nand<mode>): Likewise.
1827 (sync_new_nand<mode>): Likewise.
1828 (sync_lock_test_and_set<mode>): Likewise.
1829 (test_and_set_12): Likewise.
1830 (atomic_compare_and_swap<mode>): Likewise.
1831 (atomic_exchange<mode>_llsc): Likewise.
1832 (atomic_fetch_add<mode>_llsc): Likewise.
1833 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
1834 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
1835 (umips_save_restore_pattern_p): Likewise.
1836 (umips_load_store_pair_p): Likewise.
1837 (umips_output_load_store_pair): Likewise.
1838 (umips_movep_target_p): Likewise.
1839 (umips_12bit_offset_address_p): Likewise.
1840 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
1841 (mips_base_mips16): Rename this...
1842 (mips_base_compression_flags): ...to this. Update all uses.
1843 (mips_attribute_table): Add micromips, nomicromips and nocompression.
1844 (mips_mips16_decl_p): Delete.
1845 (mips_nomips16_decl_p): Delete.
1846 (mips_get_compress_on_flags): New function.
1847 (mips_get_compress_off_flags): New function.
1848 (mips_get_compress_mode): New function.
1849 (mips_get_compress_on_name): New function.
1850 (mips_get_compress_off_name): New function.
1851 (mips_insert_attributes): Support multiple compression types.
1852 (mips_merge_decl_attributes): Likewise.
1853 (umips_12bit_offset_address_p): New function.
1854 (mips_start_function_definition): Emit .set micromips directive.
1855 (mips_call_may_need_jalx_p): New function.
1856 (mips_function_ok_for_sibcall): Add microMIPS support.
1857 (mips_print_operand_punctuation): Support short delay slots and
1859 (umips_swm_mask, umips_swm_encoding): New.
1860 (umips_build_save_restore): New function.
1861 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
1862 (was_mips16_p): Remove.
1863 (old_compression_mode): New.
1864 (mips_set_compression_mode): New function.
1865 (mips_set_current_function): Add microMIPS support.
1866 (mips_option_override): Likewise.
1867 (umips_save_restore_pattern_p): New function.
1868 (umips_output_save_restore): New function.
1869 (umips_load_store_pair_p_1): New function.
1870 (umips_load_store_pair_p): New function.
1871 (umips_output_load_store_pair_1): New function.
1872 (umips_output_load_store_pair): New function.
1873 (umips_movep_target_p) New function.
1874 (mips_prepare_pch_save): Add microMIPS support.
1875 * config/mips/mips.h (TARGET_COMPRESSION): New.
1876 (TARGET_CPU_CPP_BUILTINS): Update macro
1877 to use new compression flags and to support microMIPS.
1878 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
1879 (MIPS_ARCH_FLOAT_SPEC): Likewise.
1880 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
1881 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
1882 (ASM_SPEC): Support mmicromips and mno-micromips.
1883 (M16STORE_REG_P): New macro.
1884 (MIPS_CALL): Support TARGET_MICROMIPS.
1885 (MICROMIPS_J): New macro.
1886 (mips_base_mips16): Rename this...
1887 (mips_base_compression_flags): ...to this.
1888 (UMIPS_12BIT_OFFSET_P): New macro.
1889 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
1890 (MULTILIB_DIRNAMES): Likewise.
1892 2013-02-25 Tom de Vries <tom@codesourcery.com>
1894 PR rtl-optimization/56131
1895 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
1896 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
1897 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
1899 2013-02-25 Tobias Burnus <burnus@net-b.de>
1901 * doc/invoke.texi (-fsanitize=): Move from optimization
1902 to debugging options.
1904 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
1906 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
1908 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
1909 Alexander Monakov <amonakov@ispras.ru>
1912 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
1913 flush pending lists also on non-jumps. Adjust comment.
1915 2013-02-24 Catherine Moore <clm@codesourcery.com>
1916 Maciej W. Rozycki <macro@codesourcery.com>
1917 Tom de Vries <tom@codesourcery.com>
1918 Nathan Sidwell <nathan@codesourcery.com>
1919 Iain Sandoe <iain@codesourcery.com>
1920 Nathan Froyd <froydnj@codesourcery.com>
1921 Chao-ying Fu <fu@mips.com>
1923 * doc/extend.texi: (micromips, nomicromips, nocompression):
1924 Document new function attributes.
1925 * doc/invoke.texi (minterlink-compressed, mmicromips,
1926 m14k, m14ke, m14kec): Document new options.
1927 (minterlink-mips16): Update documentation.
1928 * doc/md.texi (ZC, ZD): Document new constraints.
1929 * configure.ac (gcc_cv_as_micromips): Check if linker
1930 supports the .set micromips directive.
1931 * configure: Regenerate.
1932 * config.in: Regenerate.
1933 * config/mips/mips-tables.opt: Regenerate.
1934 * config/mips/micromips.md: New file.
1935 * constraints.md (ZC, AD): New constraints.
1936 * config/mips/predicates.md (movep_src_register): New predicate.
1937 (movep_src_operand): New predicate.
1938 (non_volatile_mem_operand): New predicate.
1939 * config/mips/mips.md (multimem): New type.
1940 (length): Differentiate between 17-bit and 18-bit branch offsets.
1941 (MOVEP1, MOVEP2): New mode iterator.
1942 (mov_<load>l): Use ZC constraint.
1943 (mov_<load>r): Likewise.
1944 (mov_<store>l): Likewise.
1945 (mov_<store>r): Likewise.
1946 (*branch_equality<mode>_inverted): Add microMIPS support.
1947 (*branch_equality<mode>): Likewise.
1948 (*jump_absolute): Likewise.
1949 (indirect_jump_<mode>): Likewise.
1950 (tablejump_<mode>): Likewise.
1951 (<optab>_internal): Likewise.
1952 (sibcall_internal): Likewise.
1953 (sibcall_value_internal): Likewise.
1954 (prefetch): Use constraint ZD.
1955 * config/mips/mips.opt (minterlink-compressed): New option.
1956 (minterlink-mips16): Now an alias for minterlink-compressed.
1957 (mmicromips): New option.
1958 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
1959 (compare_and_swap_12): Likewise.
1960 (sync_add<mode>): Likewise.
1961 (sync_<optab>_12): Likewise.
1962 (sync_old_<optab>_12): Likewise.
1963 (sync_new_<optab>_12): Likewise.
1964 (sync_nand_12): Likewise.
1965 (sync_old_nand_12): Likewise.
1966 (sync_new_nand_12): Likewise.
1967 (sync_sub<mode>): Likewise.
1968 (sync_old_add<mode>): Likewise.
1969 (sync_old_sub<mode>): Likewise.
1970 (sync_new_add<mode>): Likewise.
1971 (sync_new_sub<mode>): Likewise.
1972 (sync_<optab><mode>): Likewise.
1973 (sync_old_<optab><mode>): Likewise.
1974 (sync_new_<optab><mode>): Likewise.
1975 (sync_nand<mode>): Likewise.
1976 (sync_old_nand<mode>): Likewise.
1977 (sync_new_nand<mode>): Likewise.
1978 (sync_lock_test_and_set<mode>): Likewise.
1979 (test_and_set_12): Likewise.
1980 (atomic_compare_and_swap<mode>): Likewise.
1981 (atomic_exchange<mode>_llsc): Likewise.
1982 (atomic_fetch_add<mode>_llsc): Likewise.
1983 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
1984 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
1985 (umips_save_restore_pattern_p): Likewise.
1986 (umips_load_store_pair_p): Likewise.
1987 (umips_output_load_store_pair): Likewise.
1988 (umips_movep_target_p): Likewise.
1989 (umips_12bit_offset_address_p): Likewise.
1990 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
1991 (mips_base_mips16): Rename this...
1992 (mips_base_compression_flags): ...to this. Update all uses.
1993 (mips_attribute_table): Add micromips, nomicromips and nocompression.
1994 (mips_mips16_decl_p): Delete.
1995 (mips_nomips16_decl_p): Delete.
1996 (mips_get_compress_on_flags): New function.
1997 (mips_get_compress_off_flags): New function.
1998 (mips_get_compress_mode): New function.
1999 (mips_get_compress_on_name): New function.
2000 (mips_get_compress_off_name): New function.
2001 (mips_insert_attributes): Support multiple compression types.
2002 (mips_merge_decl_attributes): Likewise.
2003 (umips_12bit_offset_address_p): New function.
2004 (mips_start_function_definition): Emit .set micromips directive.
2005 (mips_call_may_need_jalx_p): New function.
2006 (mips_function_ok_for_sibcall): Add microMIPS support.
2007 (mips_print_operand_punctuation): Support short delay slots and
2009 (umips_swm_mask, umips_swm_encoding): New.
2010 (umips_build_save_restore): New function.
2011 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
2012 (was_mips16_p): Remove.
2013 (old_compression_mode): New.
2014 (mips_set_compression_mode): New function.
2015 (mips_set_current_function): Add microMIPS support.
2016 (mips_option_override): Likewise.
2017 (umips_save_restore_pattern_p): New function.
2018 (umips_output_save_restore): New function.
2019 (umips_load_store_pair_p_1): New function.
2020 (umips_load_store_pair_p): New function.
2021 (umips_output_load_store_pair_1): New function.
2022 (umips_output_load_store_pair): New function.
2023 (umips_movep_target_p) New function.
2024 (mips_prepare_pch_save): Add microMIPS support.
2025 * config/mips/mips.h (TARGET_COMPRESSION): New.
2026 (TARGET_CPU_CPP_BUILTINS): Update macro
2027 to use new compression flags and to support microMIPS.
2028 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
2029 (MIPS_ARCH_FLOAT_SPEC): Likewise.
2030 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
2031 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
2032 (ASM_SPEC): Support mmicromips and mno-micromips.
2033 (M16STORE_REG_P): New macro.
2034 (MIPS_CALL): Support TARGET_MICROMIPS.
2035 (MICROMIPS_J): New macro.
2036 (mips_base_mips16): Rename this...
2037 (mips_base_compression_flags): ...to this.
2038 (UMIPS_12BIT_OFFSET_P): New macro.
2039 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
2040 (MULTILIB_DIRNAMES): Likewise.
2042 2013-02-24 Jakub Jelinek <jakub@redhat.com>
2045 * target-globals.c (save_target_globals): For init_reg_sets and
2046 target_reinit remporarily set this_fn_optabs to this_target_optabs.
2048 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
2050 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
2051 * config/aarch64/t-aarch64
2052 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
2054 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
2057 * lra-constraints.c (process_alt_operands): Reload operand
2058 conflicting with earlier clobber only if no more other conflicting
2061 2013-02-22 Jakub Jelinek <jakub@redhat.com>
2064 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
2065 if not linking a shared library.
2067 2013-02-22 Seth LaForge <sethml@google.com>
2069 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
2071 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
2073 * config/arm/arm.md (split for extendsidi): Update condition.
2074 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
2075 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
2076 (qhs_zextenddi_cstr): Likewise.
2078 2013-02-21 Jakub Jelinek <jakub@redhat.com>
2081 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
2082 avoid signed wrapping.
2083 (expand_mult): Handle properly multiplication by
2084 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
2085 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
2086 in the compiler if coeff is HOST_WIDE_INT_MIN.
2087 (expand_divmod): Don't make ext_op1 static, change it's type to
2088 uhwi. Avoid undefined behavior in -INTVAL (op1).
2090 PR rtl-optimization/50339
2091 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
2093 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
2094 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
2095 into splitting_ashiftrt field.
2096 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
2098 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
2101 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
2104 * trans-mem.c (execute_tm_mark): Do not expand transactions that
2105 are sure to go irrevocable.
2107 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
2109 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
2110 scalars are valid operands.
2112 2013-02-21 Martin Jambor <mjambor@suse.cz>
2114 PR tree-optimization/56310
2115 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
2116 only matching indices and non-negative final offsets.
2117 (intersect_aggregates_with_edge): Pass src_idx to
2118 agg_replacements_to_vector. Pass src_idx insstead of index to
2119 intersect_with_agg_replacements.
2121 2013-02-21 Martin Jambor <mjambor@suse.cz>
2123 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
2124 instead of hard-wired defaults.
2126 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
2128 * doc/invoke.texi (MIPS Options): Update documentation of the
2129 floating-point multiply-accumulate instruction restrictions.
2131 2013-02-21 Kostya Serebryany <kcc@google.com>
2133 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
2134 asan_shadow_offset on x86_64 linux.
2136 2013-02-21 Richard Biener <rguenther@suse.de>
2138 PR tree-optimization/56415
2140 2013-02-11 Richard Biener <rguenther@suse.de>
2142 PR tree-optimization/56273
2143 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
2146 2013-02-21 Jakub Jelinek <jakub@redhat.com>
2149 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
2153 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
2154 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
2156 2013-02-20 Jan Hubicka <jh@suse.cz>
2158 PR tree-optimization/56265
2159 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
2160 when target is referenced for first time.
2162 2013-02-20 Richard Biener <rguenther@suse.de>
2164 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
2165 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
2166 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
2167 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
2168 not return anything.
2169 (rename_ssa_copies): Do not remove unused locals.
2170 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
2171 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
2172 * passes.c (execute_function_todo): Do not schedule unused locals
2173 removal if cleanup_tree_cfg did something.
2174 * tree-ssa-live.c (remove_unused_locals): Dump statistics
2175 about the number of removed locals.
2177 2013-02-20 Richard Biener <rguenther@suse.de>
2179 PR tree-optimization/56398
2180 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
2182 2013-02-20 Martin Jambor <mjambor@suse.cz>
2184 PR tree-optimization/55334
2185 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
2186 restricted pointers to arrays.
2188 2013-02-20 Richard Biener <rguenther@suse.de>
2189 Jakub Jelinek <jakub@redhat.com>
2191 PR tree-optimization/56396
2192 * tree-ssa-ccp.c (n_const_val): New static variable.
2193 (get_value): Return NULL for SSA names we don't have a lattice
2195 (ccp_initialize): Initialize n_const_val.
2196 * tree-ssa-copy.c (n_copy_of): New static variable.
2197 (init_copy_prop): Initialize n_copy_of.
2198 (get_value): Return NULL_TREE for SSA names we don't have a
2201 2013-02-20 Martin Jambor <mjambor@suse.cz>
2203 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
2205 2013-02-20 Richard Biener <rguenther@suse.de>
2207 * genpreds.c (write_lookup_constraint): Do not compare first
2208 letter of the constraint again.
2210 2013-02-20 Richard Biener <rguenther@suse.de>
2212 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
2214 (get_use_iv_cost): Terminate hashtable walk when coming across
2217 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
2219 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
2220 reassociation for avx2 targets.
2222 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2224 * config/microblaze/microblaze.c: microblaze_has_clz = 0
2225 Add version check for v8.10.a to enable microblaze_has_clz
2226 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
2227 version and TARGET_PATTERN_COMPARE check
2228 * config/microblaze/microblaze.md: New clzsi2 instruction
2230 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2232 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
2233 function before branching.
2235 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
2237 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
2239 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
2241 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
2244 * sel-sched.c: Include ira.h.
2245 (implicit_clobber_conflict_p): New function.
2246 (moveup_expr): Use it.
2247 * Makefile.in (sel-sched.o): Depend on ira.h.
2249 2013-02-19 Richard Biener <rguenther@suse.de>
2251 PR tree-optimization/56384
2252 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
2253 (vn_hash_type): Split out from ...
2254 (vn_hash_constant_with_type): ... here.
2255 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
2256 (vn_phi_eq): Compare types from vn_phi_s structure.
2257 (vn_phi_lookup): Populate vn_phi_s type.
2258 (vn_phi_insert): Likewise.
2260 2013-02-19 Jakub Jelinek <jakub@redhat.com>
2262 PR tree-optimization/56350
2263 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
2264 if haven't found reduction or nested cycle operand, rather than
2265 asserting we must find it.
2267 PR tree-optimization/56381
2268 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
2271 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
2272 Jakub Jelinek <jakub@redhat.com>
2275 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
2276 (swap_optab_enable): Same.
2277 (init_all_optabs): Use argument instead of global.
2278 * tree.h (struct tree_optimization_option): New field target_optabs.
2279 * expr.h (init_all_optabs): Add argument to prototype.
2280 (TREE_OPTIMIZATION_OPTABS): New.
2281 (save_optabs_if_changed): Protoize.
2282 * optabs.h: Declare this_fn_optabs.
2283 * optabs.c (save_optabs_if_changed): New.
2284 Declare this_fn_optabs.
2285 (init_optabs): Add argument to init_all_optabs() call.
2286 * function.c (invoke_set_current_function_hook): Handle per
2288 * function.h (struct function): New field optabs.
2289 * config/mips/mips.c (mips_set_mips16_mode): Handle when
2290 optimization_current_node has changed.
2291 * target-globals.h (save_target_globals_default_opts): Protoize.
2292 * target-globals.c (save_target_globals_default_opts): New.
2294 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2297 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
2298 registers %fr12 and %fr12R as call used.
2301 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
2302 and HImode, require all displacements to be an integer multiple of
2304 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
2305 only allow QImode and HImode when reload is in progress and strict is
2306 true. Likewise for symbolic addresses. Use base14_operand to check
2307 displacements in REG+BASE addresses.
2309 2013-02-18 Richard Biener <rguenther@suse.de>
2311 PR tree-optimization/56366
2312 * tree-vect-loop.c (get_initial_def_for_induction): Properly
2313 handle sign-conversion of outer-loop initial induction value.
2315 2013-02-18 Richard Biener <rguenther@suse.de>
2318 * cfghooks.c (merge_blocks): If we merge a latch into another
2319 block adjust references to it.
2320 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
2321 (verify_loop_structure): Verify that a recorded latch is in fact
2324 2013-02-18 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/56321
2327 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
2328 order SSA name release and virtual operand unlinking.
2330 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2332 * config/microblaze/microblaze.md (save_stack_block): Define.
2333 (restore_stack_block): Likewise.
2335 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2337 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
2338 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
2339 * config/microblaze/microblaze.c (microblaze_option_override):
2340 Bail out early for PIC modes when target does not support PIC.
2342 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2344 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
2345 Replace with a microblaze version.
2346 (microblaze_trampoline_init): Adapt for microblaze.
2347 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
2350 2013-02-16 Jakub Jelinek <jakub@redhat.com>
2351 Dodji Seketeli <dodji@redhat.com>
2354 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
2355 (instrument_mem_region_access): Do not forget to always put
2356 instrumentation of the of 'base' and 'base + len' in a "if (len !=
2357 0) statement, even for cases where either 'base' or 'base + len'
2358 are not instrumented -- because they have been previously
2359 instrumented. Simplify the logic by putting all the statements
2360 instrument 'base + len' inside a sequence, and then insert that
2361 sequence right before the current insertion point. Then, to
2362 instrument 'base + len', just get an iterator on that statement.
2363 And do not forget to update the pointer to iterator the function
2364 received as argument.
2366 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
2368 PR rtl-optimization/56348
2369 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
2371 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
2373 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
2374 (clean_graph_dump_file): Pass base to start_graph_dump.
2376 2013-02-14 Richard Henderson <rth@redhat.com>
2379 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
2381 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
2383 * collect2-aix.h: Define F_LOADONLY.
2385 2013-02-14 Richard Biener <rguenther@suse.de>
2388 * varasm.c (output_constant_def_1): Get the decl representing
2389 the constant as argument.
2390 (output_constant_def): Wrap output_constant_def_1.
2391 (make_decl_rtl): Use output_constant_def_1 with the decl
2392 representing the constant.
2393 (build_constant_desc): Optionally re-use a decl already
2394 representing the constant.
2395 (tree_output_constant_def): Adjust.
2397 2013-02-14 Dodji Seketeli <dodji@redhat.com>
2400 * asan.c (instrument_builtin_call): Really put the length of the
2401 second source argument into src1_len.
2403 2013-02-13 Jakub Jelinek <jakub@redhat.com>
2405 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
2406 argument. If it is false, don't create edge from then_bb to
2408 (insert_if_then_before_iter): Pass true to it.
2409 (build_check_stmt): Pass false to it.
2410 (transform_statements): Flush hash table only on extended basic
2411 block boundaries, rather than at the beginning of every bb.
2412 Don't flush hash table on nonfreeing_call_p calls.
2413 * tree-flow.h (nonfreeing_call_p): New prototype.
2414 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
2416 2013-02-13 David S. Miller <davem@davemloft.net>
2418 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
2420 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
2423 * ira.c (max_regno_before_ira): Move from ...
2425 (fix_reg_equiv_init): Use max_regno_before_ira instead of
2428 2013-02-13 Jakub Jelinek <jakub@redhat.com>
2430 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
2432 2013-02-13 Richard Biener <rguenther@suse.de>
2435 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
2436 globals in MEM_REFs.
2438 2013-02-13 Richard Biener <rguenther@suse.de>
2440 * loop-init.c (loop_optimizer_init): Clear loop state when
2441 re-initializing preserved loops.
2442 * loop-unswitch.c (unswitch_single_loop): Return whether
2443 we unswitched the loop. Do not verify loop state here.
2444 (unswitch_loops): When we unswitched a loop discover new loops.
2446 2013-02-13 Kostya Serebryany <kcc@google.com>
2448 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
2450 * sanitizer.def: Rename __asan_init to __asan_init_v1.
2452 2013-02-12 Dodji Seketeli <dodji@redhat.com>
2454 Avoid instrumenting duplicated memory access in the same basic block
2455 * Makefile.in (asan.o): Add new dependency on hash-table.h
2456 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
2457 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
2458 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
2459 (free_mem_ref_resources, has_mem_ref_been_instrumented)
2460 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
2461 (get_mem_ref_of_assignment): New functions.
2462 (get_mem_refs_of_builtin_call): Extract from
2463 instrument_builtin_call and tweak a little bit to make it fit with
2465 (instrument_builtin_call): Use the new
2466 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
2467 of is_gimple_builtin_call.
2468 (instrument_derefs, instrument_mem_region_access): Insert the
2469 instrumented memory reference into the hash table.
2470 (maybe_instrument_assignment): Renamed instrument_assignment into
2471 this, and change it to advance the iterator when instrumentation
2472 actually happened and return true in that case. This makes it
2473 homogeneous with maybe_instrument_assignment, and thus give a
2474 chance to callers to be more 'regular'.
2475 (transform_statements): Clear the memory reference hash table
2476 whenever we enter a new BB, when we cross a function call, or when
2477 we are done transforming statements. Use
2478 maybe_instrument_assignment instead of instrumentation. No more
2479 need to special case maybe_instrument_assignment and advance the
2480 iterator after calling it; it's now handled just like
2481 maybe_instrument_call. Update comment.
2483 2013-02-13 Richard Biener <rguenther@suse.de>
2485 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
2486 Fix loop discovery code.
2488 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
2491 * lra-constraints.c (process_alt_operands): Match early clobber
2492 operand with itself. Check conflicts with earlyclobber only if
2493 the operand is not reloaded. Prefer to reload conflicting operand
2494 if earlyclobber and matching operands are the same.
2496 2013-02-12 Richard Biener <rguenther@suse.de>
2499 * lto-streamer-out.c (write_symbol): Do not output symbols
2500 for hard register variables.
2502 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
2505 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
2506 (umulsidi3_insn, mulsidi3_insn): New insns.
2508 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
2510 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
2511 (struct tune_params): Add vec_costs field.
2512 * config/arm/arm.c (arm_builtin_vectorization_cost)
2513 (arm_add_stmt_cost): New functions.
2514 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
2515 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
2516 (arm_default_vec_cost): New struct of type cpu_vec_costs.
2517 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
2518 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
2519 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
2520 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
2522 2013-02-12 Richard Biener <rguenther@suse.de>
2525 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
2526 decls again if possible.
2528 2013-02-12 Richard Biener <rguenther@suse.de>
2531 * tree-ssa.c (verify_ssa_name): Fix check, move
2532 SSA_NAME_IN_FREE_LIST check up.
2534 2013-02-12 Jakub Jelinek <jakub@redhat.com>
2535 Steven Bosscher <steven@gcc.gnu.org>
2537 PR rtl-optimization/56151
2538 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
2539 equal to op0 or op1, and last_insn pattern is CODE operation
2540 with MEM dest and one of the operands matches that MEM.
2542 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
2544 * doc/extend.texi: Document Function Multiversioning and "default"
2545 parameter string to target attribute.
2546 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
2547 target attribute parameter is "default".
2548 (ix86_compare_version_priority): Remove checks for target attribute.
2549 (ix86_mangle_function_version_assembler_name): Change error to sorry.
2550 Remove check for target attribute equal to NULL. Add assert.
2551 (ix86_generate_version_dispatcher_body): Change error to sorry.
2553 2013-02-11 Iain Sandoe <iain@codesourcery.com>
2554 Jack Howarth <howarth@bromo.med.uc.edu>
2555 Patrick Marlier <patrick.marlier@gmail.com>
2558 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
2559 define ENDFILE_SPEC as TM_DESTRUCTOR.
2560 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
2562 2013-02-11 Alexander Potapenko <glider@google.com>
2563 Jack Howarth <howarth@bromo.med.uc.edu>
2564 Jakub Jelinek <jakub@redhat.com>
2567 * config/darwin.c (cdtor_record): Rename ctor_record.
2568 (sort_cdtor_records): Rename sort_ctor_records.
2569 (finalize_dtors): New routine to sort destructors by
2570 priority before use in assemble_integer.
2571 (machopic_asm_out_destructor): Use finalize_dtors if needed.
2573 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
2575 PR rtl-optimization/56275
2576 * simplify-rtx.c (avoid_constant_pool_reference): Check that
2577 offset is non-negative and less than cmode size before
2578 calling simplify_subreg.
2580 2013-02-11 Richard Biener <rguenther@suse.de>
2582 PR tree-optimization/56264
2583 * cfgloop.h (fix_loop_structure): Adjust prototype.
2584 * loop-init.c (fix_loop_structure): Return the number of
2585 newly discovered loops.
2586 * tree-cfgcleanup.c (repair_loop_structures): When new loops
2587 are discovered, do a full loop-closed SSA rewrite.
2589 2013-02-11 Richard Biener <rguenther@suse.de>
2591 PR tree-optimization/56273
2592 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
2594 (check_array_ref): Fix missing newline in dumps.
2595 (search_for_addr_array): Likewise.
2597 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
2599 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
2601 2013-02-09 Jakub Jelinek <jakub@redhat.com>
2604 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
2606 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
2608 PR rtl-optimization/56246
2609 * lra-constraints.c (simplify_operand_subreg): Try to reuse
2611 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
2612 constraints are satisfied.
2614 2013-02-08 Jeff Law <law@redhat.com>
2617 * emit-rtl.c (reg_is_parm_p): New function.
2618 * regs.h (reg_is_parm_p): New prototype.
2619 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
2620 callee-clobbered registers.
2622 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
2625 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
2626 If there is no implicit builtin declaration, just return NULL.
2628 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2630 * config/i386/sse.md (FMAMODEM): New mode iterator.
2631 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
2632 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
2634 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2636 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
2637 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
2638 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
2640 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
2642 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
2643 (microblaze*-*-elf): Likewise.
2644 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
2646 * config/microblaze/microblaze-c.c: Add builtin defines for
2647 _LITTLE_ENDIAN and _BIG_ENDIAN.
2648 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
2649 add to TARGET_DEFAULT flags.
2650 Expand ASM_SPEC and LINK_SPEC.
2651 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
2652 * config/microblaze/microblaze.md: Update extendsidi2 and
2653 movdi_internal instructions to use low-order / high-order reg
2655 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
2656 options and inversemask / mask of LITTLE_ENDIAN.
2657 * config/microblaze/t-microblaze: Expand multilib options to
2658 include mlittle-endian (le) and update exceptions patterns.
2660 2013-02-08 Jakub Jelinek <jakub@redhat.com>
2662 PR rtl-optimization/56195
2663 * lra-constraints.c (get_reload_reg): Don't reuse regs
2664 if they have smaller mode than requested, if they have
2665 wider mode than requested, try to return a SUBREG.
2667 PR tree-optimization/56250
2668 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
2669 if type is unsigned and code isn't MULT_EXPR.
2671 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
2673 PR tree-optimization/56064
2674 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
2675 bits according to mode.
2676 * fixed-value.h (fixed_from_double_int)
2677 (const_fixed_from_double_int): Adjust comments.
2679 2013-02-08 Richard Biener <rguenther@suse.de>
2682 * lto-streamer.h (struct data_in): Remove current_file, current_line
2683 and current_col members.
2684 * lto-streamer-out.c (lto_output_location): Stream changed bits
2685 en-block for efficiency.
2686 * lto-streamer-in.c (clear_line_info): Remove.
2687 (lto_input_location): Cache current file, line and column
2688 globally via local statics. Read changed bits en-block.
2689 (input_function): Do not call clear_line_info.
2690 (lto_read_body): Likewise.
2691 (lto_input_toplevel_asms): Likewise.
2693 2013-02-08 Michael Matz <matz@suse.de>
2695 PR tree-optimization/52448
2696 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
2697 (nt_call_phase): New static.
2698 (add_or_mark_expr): Only mark accesses with newer phase than any
2700 (nonfreeing_call_p): New.
2701 (nt_init_block): Update nt_call_phase, mark blocks as visited.
2702 (nt_fini_block): Keep blocks marked as visited.
2703 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
2705 2013-02-08 Richard Biener <rguenther@suse.de>
2707 * ira.c (ira): Free broken dominator information.
2709 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
2711 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
2713 2013-02-08 Marek Polacek <polacek@redhat.com>
2715 * cfgloop.c (verify_loop_structure): Add more checking of headers.
2717 2013-02-08 Richard Biener <rguenther@suse.de>
2720 * cfgloop.h (flow_loops_find): Adjust.
2721 (bb_loop_header_p): Declare.
2722 * cfgloop.c (bb_loop_header_p): New function split out from ...
2723 (flow_loops_find): ... here. Adjust function signature,
2724 support incremental loop structure update.
2725 (verify_loop_structure): Cleanup. Verify a loop is a loop.
2726 * cfgloopmanip.c (fix_loop_structure): Move ...
2727 * loop-init.c (fix_loop_structure): ... here.
2728 (apply_loop_flags): Split out from ...
2729 (loop_optimizer_init): ... here.
2730 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
2731 in incremental mode, only remove dead loops here.
2733 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
2736 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
2737 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
2738 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
2739 (*round<mode>3.libgcc): New insns for fixed-modes.
2740 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
2741 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
2742 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
2743 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
2744 implementations. Define to __builtin_avr_absFX,
2745 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
2746 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
2747 __builtin_avr_countlsFX, respectively.
2748 * config/avr/avr-c.c (target.h): Include it.
2749 (enum avr_builtin_id): New enum.
2750 (avr_resolve_overloaded_builtin): New static function.
2751 (avr_register_target_pragmas): Use it to set
2752 targetm.resolve_overloaded_builtin.
2753 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
2754 tree nodes used by DEF_BUILTIN.
2755 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
2756 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
2757 <AVR_BUILTIN_xxBITS>: Same.
2759 2013-02-08 Richard Biener <rguenther@suse.de>
2761 * cfgloop.c (verify_loop_structure): Properly handle
2762 a loop exiting to another loop header.
2763 * ira-int.h (ira_loops): Remove.
2764 * ira.c (ira_loops): Remove.
2765 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
2766 (do_reload): Use loop_optimizer_finalize.
2767 * ira-build.c (create_loop_tree_nodes): Use get_loops and
2768 number_of_loops to access the loop tree.
2769 (more_one_region_p): Likewise.
2770 (finish_loop_tree_nodes): Likewise.
2771 (rebuild_regno_allocno_maps): Likewise.
2772 (mark_loops_for_removal): Likewise.
2773 (mark_all_loops_for_removal): Likewise.
2774 (remove_unnecessary_regions): Likewise.
2775 (ira_build): Likewise.
2776 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
2778 2013-02-08 Richard Biener <rguenther@suse.de>
2780 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
2781 * ipa-pure-const.c (analyze_function): Avoid calling
2782 mark_irreducible_loops twice.
2783 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
2785 2013-02-07 David S. Miller <davem@davemloft.net>
2787 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
2789 * var-tracking.c (vt_add_function_parameter): Test the presence of
2790 HAVE_window_save properly and do not remap argument registers when
2791 we have a leaf function.
2793 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
2796 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
2798 * config/i386/i386.c (ix86_print_operand): Ditto.
2800 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
2802 * lra-constraints.c (process_alt_operands): Fix recently added comment.
2804 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
2806 PR rtl-optimization/56225
2807 * lra-constraints.c (process_alt_operands): Check that reload hard
2808 reg can hold value for strict_low_part.
2810 2013-02-07 Jakub Jelinek <jakub@redhat.com>
2813 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
2814 dwarf2out_end_function.
2815 (in_first_function_p, maybe_at_text_label_p,
2816 first_loclabel_num_not_at_text_label): New variables.
2817 (dwarf2out_var_location): In the first function find out
2818 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
2819 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
2822 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
2824 PR rtl-optimization/56178
2825 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
2826 SUBREG of a register. Tidy up related block of code.
2827 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
2828 note if the source is a register or a SUBREG of a register.
2830 2013-02-07 Jakub Jelinek <jakub@redhat.com>
2833 * config/rs6000/rs6000.md (ptrm): New mode attr.
2834 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
2835 call_value_indirect_aix<pttrsize>,
2836 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
2839 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
2841 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
2842 if -bnortl. Convert to strcmp and strncmp.
2844 2013-02-07 Alan Modra <amodra@gmail.com>
2847 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
2848 addresses won't wrap when offsetting.
2849 (rs6000_secondary_reload): Provide secondary reloads needed for
2850 wrapping LO_SUM addresses.
2852 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
2854 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
2855 MACH, just __MACH__.
2857 2013-02-06 Richard Biener <rguenther@suse.de>
2859 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
2860 instead of calling fix_loop_structure.
2862 2013-02-06 Jakub Jelinek <jakub@redhat.com>
2865 * omp-low.c (use_pointer_for_field): Return false if
2866 lower_send_shared_vars doesn't generate any copy-out code.
2868 2013-02-06 Tom de Vries <tom@codesourcery.com>
2870 PR rtl-optimization/56131
2871 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
2872 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
2873 of the label is NULL. Add comment.
2875 2013-02-05 Jakub Jelinek <jakub@redhat.com>
2877 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
2880 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
2881 (STATIC_LIBTSAN_LIBS): Likewise.
2882 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
2883 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
2884 is defined, don't add anything else beyond that.
2885 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
2886 (LINK_COMMAND_SPEC): Use them.
2888 PR tree-optimization/56205
2889 * tree-stdarg.c (check_all_va_list_escapes): Return true if
2890 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
2891 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
2893 2013-02-05 Richard Biener <rguenther@suse.de>
2895 PR tree-optimization/53342
2896 PR tree-optimization/53185
2897 * tree-vectorizer.h (vect_check_strided_load): Remove.
2898 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
2899 not disallow peeling for vectorized strided loads.
2900 (vect_check_strided_load): Make static and simplify.
2901 (vect_analyze_data_refs): Adjust.
2902 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
2903 correctly when vectorizing strided loads.
2905 2013-02-05 Richard Biener <rguenther@suse.de>
2907 * doc/install.texi: Refer to ISL, not PPL.
2909 2013-02-05 Jan Hubicka <jh@suse.cz>
2911 PR tree-optimization/55789
2912 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
2914 2013-02-05 Jan Hubicka <jh@suse.cz>
2916 PR tree-optimization/55789
2917 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
2918 the dead call anyway.
2920 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
2923 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
2925 2013-02-04 Alexander Potapenko <glider@google.com>
2926 Jack Howarth <howarth@bromo.med.uc.edu>
2927 Jakub Jelinek <jakub@redhat.com>
2930 * config/darwin.c (sort_ctor_records): Stabilized qsort
2931 on constructor priority by using original position.
2932 (finalize_ctors): New routine to sort constructors by
2933 priority before use in assemble_integer.
2934 (machopic_asm_out_constructor): Use finalize_ctors if needed.
2936 2013-02-04 Jakub Jelinek <jakub@redhat.com>
2939 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
2940 about visibility on artificial decls.
2941 * config/sol2.c (solaris_assemble_visibility): Likewise.
2943 2013-02-04 Kai Tietz <ktietz@redhat.com>
2946 * config/i386/i386.c (function_value_ms_64): Add additional valtype
2947 argument and improve checking of return-argument types for 16-byte
2949 (ix86_function_value_1): Add additional valtype argument on call
2950 of function_value_64.
2951 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
2952 handling infunction_value_64 function.
2954 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2956 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
2958 2013-02-04 Richard Biener <rguenther@suse.de>
2960 PR tree-optimization/56188
2961 * tree-ssa-structalias.c (label_visit): Consider case with
2962 initially non-empty points-to set.
2963 (perform_var_substitution): Dump node mapping and clean up.
2965 2013-02-04 Richard Guenther <rguenther@suse.de>
2968 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
2969 node prevail as last resort.
2970 (lto_symtab_merge_decls): Remove guard on LTRANS here.
2971 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
2973 2013-02-04 Richard Biener <rguenther@suse.de>
2975 PR tree-optimization/56113
2976 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
2978 (equiv_class_lookup_or_add): ... this.
2979 (label_visit): Adjust and fix error in previous patch.
2980 (perform_var_substitution): Adjust.
2982 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
2984 * config/sh/divtab.c: Fix formatting and comments throughout the file.
2985 * config/sh/sh4-300.md: Likewise.
2986 * config/sh/sh4a.md: Likewise.
2987 * config/sh/constraints.md: Likewise.
2988 * config/sh/sh.md: Likewise.
2989 * config/sh/netbsd-elf.h: Likewise.
2990 * config/sh/predicates.md: Likewise.
2991 * config/sh/sh-protos.h: Likewise.
2992 * config/sh/ushmedia.h: Likewise.
2993 * config/sh/linux.h: Likewise.
2994 * config/sh/sh.c: Likewise.
2995 * config/sh/superh.h: Likewise.
2996 * config/sh/elf.h: Likewise.
2997 * config/sh/sh4.md: Likewise.
2998 * config/sh/sh.h: Likewise.
3000 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3002 * config/pa/constraints.md: Adjust unused letters. Change "T"
3003 constraint to match_test floating_point_store_memory_operand().
3004 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
3005 (base14_operand): New.
3006 (floating_point_store_memory_operand): New.
3007 (integer_store_memory_operand): Revise to use base14_operand and
3008 reg_plus_base_memory_operand.
3009 (move_dest_operand): Allow symbolic_memory_operands.
3010 (symbolic_memory_operand): Check for LO_SOM.
3011 (symbolic_operand): Change default case to break.
3012 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
3013 CONST_DOUBLE values to be reloaded by putting them into memory when
3014 the destination is a floating point register.
3015 (movdf): Remove code to handle CONST_DOUBLE.
3017 (reload_indf_r1): New.
3018 (reload_insf_r1): New.
3019 Consistently use "Q" and "T" constraints with integer and floating
3020 point move instructions, respectively.
3021 (movdi): Remove FAIL.
3022 Change predicate for source operand unamed DImode move from
3023 general_operand to move_src_operand.
3024 (umulsidi3): Change predicate for destination operand to
3026 Likewise for similar unamed patterns.
3027 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
3028 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
3029 (hppa_legitimize_address): Simplify mask calculation.
3030 (pa_emit_move_sequence): Revised handling of secondary reloads from
3031 REG+D addresses for floating point loads and stores. Directly handle
3032 loading CONST0_RTX (mode) to a floating point register.
3033 (pa_secondary_reload): Handle reloading DF and SFmode constant values
3034 to floating point registers. Don't restrict secondary reloads to
3035 floating point registers to integer modes. Revise some comments and
3037 (TARGET_LEGITIMATE_ADDRESS_P): Define.
3038 (pa_legitimate_address_p): New.
3039 (pa_legitimize_reload_address): New.
3040 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
3041 (STRICT_REG_OK_FOR_BASE_P): New.
3042 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
3043 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
3045 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
3046 Andrew Dixie <andrewd@gentrack.com>
3048 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
3051 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
3053 * expmed.c (extract_bit_field_1): Pass the full width of the
3054 structure to get_best_reg_extraction_insn.
3056 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
3059 * configure.ac (use_cxa_atexit): Add AIX.
3060 * configure: Regenerate.
3062 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
3064 2013-02-01 Jakub Jelinek <jakub@redhat.com>
3067 * final.c (need_profile_function): New variable.
3068 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
3069 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
3070 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
3071 notes, targetm.asm_out.function_prologue doesn't emit anything,
3072 HAVE_prologue and profiler should be emitted before prologue,
3073 set need_profile_function instead of emitting it.
3074 (final_scan_insn): If need_profile_function, emit
3075 profile_function on the first NOTE_INSN_BASIC_BLOCK or
3076 NOTE_INSN_FUNCTION_BEG note.
3078 2013-02-01 Richard Henderson <rth@redhat.com>
3080 * config/rs6000/rs6000.md (smulditi3): New.
3083 * config/alpha/alpha.md (umulditi3): New.
3085 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
3087 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
3088 (ASM_OUTPUT_ALIGNED_LOCAL): New.
3090 2013-02-01 Richard Biener <rguenther@suse.de>
3092 PR tree-optimization/56113
3093 * tree-ssa-structalias.c (label_visit): Reduce work for
3094 single-predecessor nodes.
3096 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
3098 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
3099 range isn't testing for zero.
3101 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
3104 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
3106 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
3107 Nick Clifton <nickc@redhat.com>
3109 * config/v850/constraints.md (Q): Define as a memory constraint.
3110 * config/v850/predicates.md (label_ref_operand): New predicate.
3111 (e3v5_shift_operand): New predicate.
3112 (ior_operator): New predicate.
3113 * config/v850/t-v850: Add e3v5 multilib.
3114 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
3115 (v850_gen_movdi): Prototype.
3116 * config/v850/v850.c: Add support for e3v5 architecture.
3117 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
3119 (construct_save_jarl): Add e3v5 long JARL support.
3120 (v850_adjust_insn_length): New function. Adjust length of call
3121 insns when using e3v5 instructions.
3122 (v850_gen_movdi): New function: Generate instructions to move a
3124 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
3125 (CPP_SPEC): Define __v850e3v5__ as appropriate.
3126 (TARGET_USE_FPU): Enable for e3v5.
3127 (CONST_OK_FOR_W): New macro.
3128 (ADJUST_INSN_LENGTH): Define.
3129 * config/v850/v850.md (UNSPEC_LOOP): Define.
3130 (attr cpu): Add v850e3v5.
3131 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
3132 (movdi): New pattern.
3133 (movdi_internal): New pattern.
3134 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
3135 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
3136 (cstoresf4): Likewise.
3137 (cstoredf4): Likewise.
3138 (insv): New pattern.
3139 (rotlso3_a): New pattern.
3140 (rotlsi3_b): New pattern
3141 (rotlsi3_v850e3v5): New pattern.
3142 (doloop_begin): New pattern.
3143 (fix_loop_counter): New pattern.
3144 (doloop_end): New pattern.
3145 (branch_normal): Add e3v5 long branch support.
3146 (branch_invert): Likewise.
3147 (branch_z_normal): Likewise.
3148 (branch_z_invert): Likewise.
3149 (branch_nz_normal): Likewise.
3150 (branch_nz_invert): Likewise.
3151 (call_internal_short): Add e3v5 register-indirect JARL support.
3152 (call_internal_long): Likewise.
3153 (call_value_internal_short): Likewise.
3154 (call_value_internal_long): Likewise.
3155 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
3156 (mloop): New option.
3157 * config.gcc: Add support for configuring v840e3v5 target.
3158 * doc/invoke.texi: Document new v850 specific command line options.
3160 2013-01-31 Paul Koning <ni1d@arrl.net>
3164 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
3165 children if parent is a class.
3166 (prune_unused_types_prune): Don't add DW_AT_declaration.
3168 2013-01-31 Richard Biener <rguenther@suse.de>
3170 PR tree-optimization/56157
3171 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
3172 match up operand with SLP child.
3174 2013-01-31 Jason Merrill <jason@redhat.com>
3177 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
3178 parameters the first time.
3179 (gen_scheduled_generic_parms_dies): Check completeness here.
3181 2013-01-31 Richard Biener <rguenther@suse.de>
3184 * common.opt (faggressive-loop-optimizations): New flag,
3186 * doc/invoke.texi (faggressive-loop-optimizations): Document.
3187 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
3188 infer_loop_bounds_from_undefined by it.
3190 2013-01-31 Richard Biener <rguenther@suse.de>
3192 PR tree-optimization/56150
3193 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
3194 visit virtual operands.
3195 (find_uses_to_rename_bb): Likewise.
3197 2013-01-31 Richard Biener <rguenther@suse.de>
3199 PR tree-optimization/56150
3200 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
3201 mixed store non-store stmts.
3203 2013-01-30 Jakub Jelinek <jakub@redhat.com>
3206 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
3207 LIBASAN_EARLY_SPEC is defined.
3208 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
3209 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
3211 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
3214 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
3215 invalid args instead of ICEing on it.
3216 (ix86_valid_target_attribute_tree): Return error_mark_node if
3217 ix86_valid_target_attribute_inner_p failed.
3218 (ix86_valid_target_attribute_p): Return false only if
3219 ix86_valid_target_attribute_tree returned error_mark_node. Allow
3220 target("default") attribute.
3221 (sorted_attr_string): Change argument from const char * to tree,
3222 merge in all target attribute arguments rather than just one.
3223 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
3224 instead of free. Avoid using strcat.
3225 (ix86_mangle_function_version_assembler_name): Mangle
3226 target("default") as if no target attribute is present. Adjust
3227 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
3228 instead of xmalloc and XDELETEVEC instead of free.
3229 (ix86_function_versions): Don't return true if one of the decls
3230 doesn't have target attribute. If they don't and one of the decls
3231 is DECL_FUNCTION_VERSIONED, report an error. Adjust
3232 sorted_attr_string caller. Use XDELETEVEC instead of free.
3233 (ix86_supports_function_versions): Remove.
3234 (make_name): Fix up formatting.
3235 (make_dispatcher_decl): Remove resolver_name and its initialization.
3236 Avoid leaking memory.
3237 (is_function_default_version): Return true if there is
3238 target("default") attribute rather than no target attribute at all.
3239 (make_resolver_func): Avoid leaking memory.
3240 (ix86_generate_version_dispatcher_body): Likewise.
3241 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
3242 * target.def (supports_function_versions): Remove.
3243 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
3244 * doc/tm.texi: Regenerated.
3246 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
3248 PR rtl-optimization/56144
3249 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
3250 for values with side effects.
3252 2013-01-30 Richard Biener <rguenther@suse.de>
3254 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
3255 (sparseset_pop): Likewise.
3256 * cfganal.c (compute_idf): Likewise. Increase work-stack size
3257 to be able to use quick_push in the worker loop.
3259 2013-01-30 Marek Polacek <polacek@redhat.com>
3261 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
3263 2013-01-30 Richard Biener <rguenther@suse.de>
3266 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
3268 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
3270 PR tree-optimization/56064
3271 * fixed-value.c (fixed_from_double_int): New function.
3272 * fixed-value.h (fixed_from_double_int): New prototype.
3273 (const_fixed_from_double_int): New static inline function.
3274 * fold-const.c (native_interpret_fixed): New static function.
3275 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
3276 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
3277 (native_encode_fixed): New static function.
3278 (native_encode_expr) <FIXED_CST>: Use it.
3279 (native_interpret_int): Move double_int worker code to...
3280 * double-int.c (double_int::from_buffer): ...this new static method.
3281 * double-int.h (double_int::from_buffer): Prototype it.
3283 2013-01-30 Richard Biener <rguenther@suse.de>
3285 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
3286 New pointer-map and obstack.
3287 (init_alias_vars): Allocate pointer-map and obstack.
3288 (delete_points_to_sets): Free them.
3289 (find_what_var_points_to): Cache result.
3290 (find_what_p_points_to): Adjust for changed interface of
3291 find_what_var_points_to.
3292 (compute_points_to_sets): Likewise.
3293 (ipa_pta_execute): Likewise.
3295 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3297 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
3298 * configure: Regenerate.
3299 * config.in: Regenerate.
3300 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
3301 #nobits/#progbits if supported.
3303 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
3306 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
3307 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
3308 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
3310 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3312 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
3313 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
3315 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3317 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
3318 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
3320 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3322 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
3324 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
3325 * config/arm/cortex-a7.md: New bypasses using
3326 arm_mac_accumulator_is_result.
3328 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3330 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
3331 (cortex_a7_neon_mla): Likewise.
3332 (cortex_a7_fpfmad): New reservation.
3333 (cortex_a7_fpmacs): Use ffmas and update required units.
3334 (cortex_a7_fpmuld): Update required units and latency.
3335 (cortex_a7_fpmacd): Likewise.
3336 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
3337 (cortex_a7_neon). Likewise.
3338 (bypass) Update participating units.
3340 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
3342 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
3343 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
3345 * config/arm/vfp11.md (vfp_farith): Use ffmas.
3346 (vfp_fmul): Use ffmad.
3347 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
3348 (cortex_r4_fmacd): Use ffmad.
3349 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
3350 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
3351 (cortex_a9_fmacd): Use ffmad.
3352 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
3353 (cortex_a8_vfp_macd): Use ffmad.
3354 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
3355 (cortex_a5_fpmacd): Use ffmad.
3356 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
3357 (cortex_a15_vfp_macd): Use ffmad.
3358 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
3360 2013-01-29 Jason Merrill <jason@redhat.com>
3363 * varasm.c (default_assemble_visibility): Don't warn about
3364 visibility on artificial decls.
3366 2013-01-29 Richard Biener <rguenther@suse.de>
3368 PR tree-optimization/56113
3369 * tree-ssa-structalias.c (equiv_class_lookup): Also return
3371 (label_visit): Free duplicate bitmaps and record the leader instead.
3372 (perform_var_substitution): Adjust.
3374 2013-01-29 Richard Biener <rguenther@suse.de>
3376 PR tree-optimization/55270
3377 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
3378 the CFG, schedule loops for fixup.
3380 2013-01-29 Nick Clifton <nickc@redhat.com>
3382 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
3385 2013-01-28 Leif Ekblad <leif@rdos.net>
3387 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
3388 * config/i386/i386.h (TARGET_RDOS): New macro.
3389 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
3390 * config/i386/i386.c (ix86_option_override_internal): For 64bit
3391 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
3392 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
3393 DEFAULT_LARGE_SECTION_THRESHOLD.
3394 * config/i386/i386.md (R14_REG, R15_REG): New constants.
3395 * config/i386/rdos.h: New file.
3396 * config/i386/rdos64.h: New file.
3398 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
3401 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
3404 2013-01-28 Jakub Jelinek <jakub@redhat.com>
3406 PR rtl-optimization/56117
3407 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
3408 call cselib_lookup_from_insn on the MEM before calling
3409 add_insn_mem_dependence.
3411 2013-01-28 Richard Biener <rguenther@suse.de>
3413 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
3414 to a stmt that didn't have one.
3415 (copy_phis_for_bb): Likewise for PHI arguments.
3416 (copy_debug_stmt): Likewise for debug stmts.
3418 2013-01-28 Richard Biener <rguenther@suse.de>
3420 PR tree-optimization/56034
3421 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
3422 (partition_builtin_p): Adjust.
3423 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
3424 it is the last partition.
3425 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
3426 up the vertex for the definition.
3427 (classify_partition): Classify whether a partition is a
3428 PKIND_REDUCTION, thus has uses outside of the loop.
3429 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
3430 Merge all PKIND_REDUCTION partitions into the last partition.
3431 (tree_loop_distribution): Seed partitions from reductions as well.
3433 2013-01-28 Jakub Jelinek <jakub@redhat.com>
3435 PR tree-optimization/56125
3436 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
3437 pow(x,c) into sqrt(x) * powi(x, n/2) or
3438 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
3439 optimizing for size.
3440 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
3441 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
3444 PR tree-optimization/56094
3445 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
3446 to UNKNOWN_LOCATION while gimplifying expr.
3448 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
3451 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
3452 operand 0 in movabs insn template for -masm=intel asm alternative.
3453 (*movabs<mode>_2): Ditto for operand 1.
3455 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
3458 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
3461 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
3463 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
3466 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
3468 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
3469 Undef to avoid warning.
3471 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
3473 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
3474 * configure: Regenerate.
3476 2013-01-25 Jakub Jelinek <jakub@redhat.com>
3478 PR tree-optimization/56098
3479 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
3480 for stmts with volatile ops.
3481 (cond_store_replacement): Don't optimize if assign has volatile ops.
3482 (cond_if_else_store_replacement_1): Don't optimize if either
3483 then_assign or else_assign have volatile ops.
3484 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
3487 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3489 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
3491 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3493 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
3494 missing ':' in asm example.
3496 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
3498 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
3499 entries into lane and laneq entries.
3500 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
3501 Remove AdvSIMD scalar modes.
3502 (aarch64_sq<r>dmulh_laneq<mode>): New.
3503 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
3505 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
3506 builtin implementations to relfect changes in RTL in aarch64-simd.md.
3507 * config/aarch64/iterators.md (VCOND): New.
3510 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
3513 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
3514 Add NULL LIBNAME argument to existing definitions.
3515 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
3516 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
3517 * config/avr/avr.c (DEF_BUILTIN): Same.
3518 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
3519 (avr_expand_builtin): Expand to a vanilla call if a libgcc
3520 implementation is available (DECL_ASSEMBLER_NAME is set).
3521 (avr_fold_absfx): New static function.
3522 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
3523 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
3524 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
3526 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
3527 (abshk, absk, abslk, absllk): Provide as static inline functions.
3529 2013-01-25 Marek Polacek <polacek@redhat.com>
3531 PR tree-optimization/56035
3532 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
3534 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
3536 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
3537 (*movtf_internal_rex64): Add (!o,C) alternative
3538 (*movxf_internal_rex64): Ditto.
3539 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
3541 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
3543 * doc/invoke.texi: fix typo.
3544 * doc/objc.texi: fix typo.
3546 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
3548 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
3549 for the first two alternatives.
3551 2013-01-24 Diego Novillo <dnovillo@google.com>
3553 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
3554 (ggc-zone.o): Remove.
3555 * configure.ac: Remove option --with-gc.
3556 * configure: Re-generate.
3557 * doc/install.texi: Remove documentation for --with-gc.
3558 * gengtype.c (write_enum_defn): Remove. Update all users.
3559 (write_Types_process_field): Remove generation of gt_e_* argument.
3560 (output_type_enum): Remove. Update all users.
3561 (write_enum_defn): Remove. Update all users.
3562 (enum alloc_zone): Remove. Update all users.
3563 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
3564 * ggc-common.c (ggc_splay_alloc): Remove first argument.
3566 (struct ptr_data): Remove field TYPE. Update all users.
3567 (gt_pch_note_object): Remove argument TYPE. Update all users.
3568 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
3570 * ggc-none.c (ggc_alloc_typed_stat): Remove.
3571 (struct alloc_zone): Remove.
3572 (ggc_internal_alloc_zone_stat): Remove.
3573 (ggc_internal_cleared_alloc_zone_stat): Remove.
3574 * ggc-page.c (ggc_alloc_typed_stat): Remove.
3575 (ggc_pch_count_object): Remove last argument. Update all users.
3576 (ggc_pch_alloc_object): Remove last argument. Update all users.
3577 (struct alloc_zone): Remove.
3578 * ggc-zone.c: Remove.
3579 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
3580 (struct alloc_zone): Remove.
3581 (ggc_alloc_typed_stat): Remove.
3582 (ggc_alloc_typed): Remove.
3583 (ggc_splay_alloc): Remove first argument.
3584 (rtl_zone): Remove. Update all users.
3585 (tree_zone): Remove. Update all users.
3586 (tree_id_zone): Remove. Update all users.
3587 (ggc_internal_zone_alloc_stat): Remove. Update all users.
3588 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
3589 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
3590 * tree-ssanames.c: Remove references to zone allocator in comments.
3592 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
3594 * config/avr/avr.c (avr_out_fract): Make register numbers that
3595 might be outside of source operand signed.
3597 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
3599 * config/i386/constraints.md (Yf): New constraint.
3600 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
3601 of f constraint to conditionaly disable x87 register preferences.
3602 (*movdf_internal): Ditto.
3603 (*movsf_internal): Ditto.
3605 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
3608 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
3609 that have operands with impossible constraints.
3610 Add a FIXME for a speed-up opportunity.
3611 * lra-constraints.c (process_alt_operands): Verify that a class
3612 selected from constraints on asms is valid for the operand mode.
3613 (curr_insn_transform): Remove incorrect comment.
3615 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
3617 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
3618 TOC operand is a valid symbol ref in the constant pool.
3620 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
3622 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
3624 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
3627 * config/avr/stdfix.h: New file.
3628 * t-avr (stdfix-gcc.h): New rule to build it.
3629 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
3631 2013-01-23 Kostya Serebryany <kcc@google.com>
3633 * config/darwin.h: remove dependency on
3634 CoreFoundation (asan on Mac OS).
3636 2013-01-23 Jakub Jelinek <jakub@redhat.com>
3639 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
3640 instead of cmpdi_operand for first comparison operand.
3641 Don't assert that comparison operands aren't both constants.
3643 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
3645 * doc/install.texi (Downloading the Source): Update references to
3646 downloading separate components.
3648 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
3650 * doc/extend.texi (__int128): Improve grammar.
3652 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
3655 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
3656 alternative to (o,r).
3657 (*movdi_internal_rex64): Remove (!o,n) alternative.
3658 (DImode immediate->memory splitter): Remove.
3659 (DImode immediate->memory peephole2): Remove.
3660 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
3661 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
3662 alternative to (!o,*r).
3663 (*movtf_internal_sse): New pattern.
3664 (*movxf_internal_rex64): New pattern.
3665 (*movxf_internal): Disable for TARGET_64BIT.
3666 (*movdf_internal_rex64): Remove (!o,F) alternative.
3668 2013-01-22 Jakub Jelinek <jakub@redhat.com>
3671 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
3672 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
3673 * tree-vect-loop-manip.c (find_loop_location): Also ignore
3674 stmt locations where LOCATION_LOCUS of the stmt location is
3675 UNKNOWN_LOCATION or BUILTINS_LOCATION.
3678 * config/i386/i386.md (UNSPEC_STOS): New.
3679 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
3680 *strsetqi_1): Add UNSPEC_STOS.
3682 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
3685 * doc/invoke.texi: Remove left over -Wsynth example.
3687 2013-01-21 Jakub Jelinek <jakub@redhat.com>
3689 PR tree-optimization/56051
3690 * fold-const.c (fold_binary_loc): Don't fold
3691 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
3692 a narrowing conversion, or widening conversion from signed
3695 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
3697 PR rtl-optimization/56023
3698 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
3699 dependent on debug instruction.
3701 2013-01-21 Martin Jambor <mjambor@suse.cz>
3704 * function.c (allocate_struct_function): Call
3705 invoke_set_current_function_hook earlier.
3707 2013-01-21 Jakub Jelinek <jakub@redhat.com>
3709 * reload1.c (init_reload): Only initialize reload_obstack
3710 during the first call.
3712 2013-01-21 Marek Polacek <polacek@redhat.com>
3714 * cfgloop.c (verify_loop_structure): Fix up grammar.
3716 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
3718 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
3719 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
3721 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3724 * config/arm/marvell-pj4.md: Update copyright year.
3725 Fix up use of alu to alu_reg and simple_alu_imm.
3727 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
3729 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
3731 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
3734 * lra-constraints.c (curr_insn_transform): Don't reuse original
3735 insn for secondary memory move when memory mode should be different.
3737 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3739 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
3740 atomic_storedi_1): New patterns.
3742 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
3744 btver2 pipeline descriptions.
3745 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
3747 * config/i386/i386.md (btver2_decode): New type attributes.
3748 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
3750 * config/i386/btver2.md: New file describing btver2 pipelines.
3752 2013-01-19 Andrew Pinski <apinski@cavium.com>
3754 PR tree-optimization/52631
3755 * tree-ssa-sccvn (visit_use): Before looking up the original
3756 statement, try looking up the simplified expression.
3758 2013-01-19 Anthony Green <green@moxielogic.com>
3760 * config/moxie/moxie.c (moxie_expand_prologue): Set
3761 current_function_static_stack_size.
3763 2013-01-18 Jakub Jelinek <jakub@redhat.com>
3765 PR tree-optimization/56029
3766 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
3767 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
3769 2013-01-18 Sharad Singhai <singhai@google.com>
3771 PR tree-optimization/55995
3772 * dumpfile.c (dump_loc): Print location only if available.
3773 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
3775 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
3778 * lra-constraints.c (curr_insn_transform): Reuse original insn for
3779 secondary memory move.
3780 (inherit_reload_reg): Use rclass instead of cl for
3781 check_secondary_memory_needed_p.
3783 2013-01-18 Jakub Jelinek <jakub@redhat.com>
3786 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
3787 the case where writing real complex part of target modifies op1.
3789 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
3791 * config/aarch64/aarch64-simd.md
3792 (aarch64_vcond_internal<mode>): Handle unordered cases.
3793 * config/aarch64/iterators.md (v_cmp_result): New.
3795 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
3796 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3798 * config/arm/marvell-pj4.md: New file.
3799 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
3800 * config/arm/arm.md (generic_sched): Add marvell_pj4.
3801 (generic_vfp): Likewise.
3802 * config/arm/arm-cores.def: Add marvell-pj4.
3803 * config/arm/arm-tune.md: Regenerate.
3804 * config/arm/arm-tables.opt: Regenerate.
3805 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
3806 * doc/invoke.texi: Document marvell-pj4.
3808 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
3810 * config/aarch64/arm_neon.h: Map scalar types to standard types.
3812 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3817 * var-tracking.c (negative_power_of_two_p): New.
3818 (global_get_addr_cache, local_get_addr_cache): New.
3819 (get_addr_from_global_cache, get_addr_from_local_cache): New.
3820 (vt_canonicalize_addr): Rewrite using the above. Adjust the
3822 (vt_stack_offset_p): Remove.
3823 (vt_canon_true_dep): Always canonicalize loc's address.
3824 (clobber_overlapping_mems): Make sure we have a MEM.
3825 (local_get_addr_clear_given_value): New.
3826 (val_reset): Clear local cached entries.
3827 (compute_bb_dataflow): Create and release the local cache.
3828 Disable duplicate MEMs clobbering.
3829 (emit_notes_in_bb): Clobber MEMs likewise.
3830 (vt_emit_notes): Create and release the local cache.
3831 (vt_initialize, vt_finalize): Create and release the global
3832 cache, respectively.
3833 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
3835 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3838 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
3839 not found in the symtab.
3841 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
3844 PR rtl-optimization/55547
3845 PR rtl-optimization/53827
3848 * alias.c (offset_overlap_p): New, factored out of...
3849 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
3850 the conservative special case for symbolic constants. Don't
3851 adjust zero sizes on alignment.
3853 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
3855 PR rtl-optimization/52573
3856 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
3857 REG_UNUSED for the same register.
3859 2013-01-17 Richard Biener <rguenther@suse.de>
3860 Marek Polacek <polacek@redhat.com>
3862 PR rtl-optimization/55833
3863 * loop-unswitch.c (unswitch_loops): Move loop verification...
3864 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
3865 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
3866 Set it to true when we're removing a loop from hierarchy tree in
3867 an irreducible region.
3868 (fix_bb_placements): Adjust caller.
3869 (fix_loop_placements): Likewise.
3871 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
3873 * config/avr/builtins.def (DEF_BUILTIN): Factor out
3874 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
3875 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
3876 Remove ID. Adjust comments.
3877 * config/avr/avr-c.c (avr_builtin_name): Remove.
3878 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
3879 * config/avr/avr.c (avr_tolower): New static function.
3880 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
3881 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
3882 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
3885 2013-01-17 Jan Hubicka <jh@suse.cz>
3887 PR tree-optimization/55273
3888 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
3890 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
3893 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
3894 store through atomic_store<mode>_1.
3895 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
3897 2013-01-17 Martin Jambor <mjambor@suse.cz>
3899 PR tree-optimizations/55264
3900 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
3901 for virtual methods.
3902 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
3903 virtual methods before inlining is over.
3904 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
3906 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
3909 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
3911 PR rtl-optimization/56005
3912 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
3913 pending reads for prefetch.
3915 2013-01-16 Ian Bolton <ian.bolton@arm.com>
3917 * config/aarch64/aarch64.md
3918 (*cstoresi_neg_uxtw): New pattern.
3919 (*cmovsi_insn_uxtw): New pattern.
3920 (*<optab>si3_uxtw): New pattern.
3921 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
3922 (*<optab>si3_insn_uxtw): New pattern.
3923 (*bswapsi2_uxtw): New pattern.
3925 2013-01-16 Richard Biener <rguenther@suse.de>
3927 * tree-inline.c (tree_function_versioning): Remove set but
3928 never used variable.
3930 2013-01-16 Richard Biener <rguenther@suse.de>
3932 PR tree-optimization/55964
3933 * tree-flow.h (rename_variables_in_loop): Remove.
3934 (rename_variables_in_bb): Likewise.
3935 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
3936 (copy_loop_before): Adjust and delete update-ssa status.
3937 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
3938 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
3939 (rename_variables_in_loop): Remove.
3940 (slpeel_update_phis_for_duplicate_loop): Likewise.
3941 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
3942 use available cfg machinery instead of duplicating it.
3943 Update PHI nodes and perform poor-mans SSA update here.
3944 (slpeel_tree_peel_loop_to_edge): Adjust.
3946 2013-01-16 Richard Biener <rguenther@suse.de>
3948 PR tree-optimization/54767
3949 PR tree-optimization/53465
3950 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
3951 (vrp_visit_phi_node): For PHI arguments coming via backedges
3952 drop all symbolical range information.
3953 (execute_vrp): Compute backedges.
3955 2013-01-16 Richard Biener <rguenther@suse.de>
3957 * doc/install.texi: Update CLooG and ISL requirements to
3960 2013-01-16 Christian Bruel <christian.bruel@st.com>
3963 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
3964 (broken_move): Handle UNSPECV_SP_SWITCH_B.
3965 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
3967 2013-01-16 DJ Delorie <dj@redhat.com>
3969 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
3970 (UNSPECV_SP_SWITCH_E): New.
3971 (sp_switch_1): Change to an unspec.
3972 (sp_switch_2): Change to an unspec. Don't use post-inc when we
3975 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
3977 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
3978 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
3979 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
3980 (expand_mem_thread_fence): Ditto.
3981 (expand_mem_signal_fence): Ditto.
3982 (expand_atomic_load): Ditto.
3983 (expand_atomic_store): Ditto.
3985 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
3987 PR rtl-optimization/55547
3988 PR rtl-optimization/53827
3991 * alias.c (memrefs_conflict_p): Set sizes to negative after
3994 2013-01-15 Jakub Jelinek <jakub@redhat.com>
3997 * function.c (thread_prologue_and_epilogue_insns): Always
3998 add crtl->drap_reg to set_up_by_prologue.set, even if
3999 stack_realign_drap is false.
4001 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4003 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
4004 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
4005 *call): Fix indention.
4007 2013-01-15 Tom de Vries <tom@codesourcery.com>
4010 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
4013 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
4015 PR rtl-optimization/55153
4016 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
4018 2013-01-15 Martin Jambor <mjambor@suse.cz>
4020 PR tree-optimization/55920
4021 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
4022 accesses as grp_to_be_debug_replaced.
4024 2013-01-15 Jakub Jelinek <jakub@redhat.com>
4026 PR tree-optimization/55920
4027 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
4028 there is non-useless type conversion needed from debug rhs to lhs,
4029 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
4031 2013-01-15 Joseph Myers <joseph@codesourcery.com>
4032 Mikael Pettersson <mikpe@it.uu.se>
4035 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
4037 (ASM_OUTPUT_CASE_LABEL): Remove.
4038 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
4039 * final.c (shorten_branches): Update alignment of labels before
4040 jump tables if CASE_VECTOR_SHORTEN_MODE.
4042 2013-01-15 Richard Biener <rguenther@suse.de>
4045 * system.h: Do not include gmp.h for building host tools.
4047 2013-01-15 Richard Biener <rguenther@suse.de>
4050 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
4051 account for bitpos when computing alignment.
4053 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
4055 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
4056 (ix86_target_macros_internal): Likewise.
4058 * config/i386/i386.c (m_CORE2I7): Removed.
4059 (m_CORE_HASWELL): New macro.
4060 (m_CORE_ALL): Likewise.
4061 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
4062 (initial_ix86_arch_features): Likewise.
4063 (processor_target_table): Initializations for Core avx2.
4064 (cpu_names): New names "core-avx2".
4065 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
4066 PROCESSOR_CORE_HASWELL.
4067 (ix86_issue_rate): New case.
4068 (ia32_multipass_dfa_lookahead): Likewise.
4069 (ix86_sched_init_global): Likewise.
4071 * config/i386/i386.h (TARGET_HASWELL): New macro.
4072 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
4073 (processor_type): New PROCESSOR_HASWELL.
4075 2013-01-15 Jakub Jelinek <jakub@redhat.com>
4077 PR tree-optimization/55955
4078 * tree-vect-loop.c (vectorizable_reduction): Give up early on
4079 *SHIFT_EXPR and *ROTATE_EXPR codes.
4081 PR tree-optimization/48766
4082 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
4083 -ftrapv disable -fwrapv.
4085 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4088 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
4089 etc. to 1 and not to __flash.
4090 Use LL suffix for __INT24_MAX__ with -mint8.
4091 Use ULL suffix for __UINT24_MAX__ with -mint8.
4093 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4095 * config/avr/avr-arch.h
4096 (struct base_arch_s): Use typedef avr_arch_t instead.
4097 (struct arch_info_s): Use typedef avr_arch_info_t instead.
4098 (struct mcu_type_s): Use typedef avr_mcu_t instead.
4099 * config/avr/avr.c: Same.
4100 * config/avr/avr-devices.c: Same.
4101 * config/avr/driver-avr.c: Same.
4102 * config/avr/gen-avr-mmcu-texi.c: Same.
4103 * config/avr/avr-mcus.def: Adjust comment.
4105 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
4107 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
4108 * config/aarch64/iterators.md (VALLDI): New.
4110 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
4111 Andi Kleen <ak@linux.intel.com>
4114 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
4115 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
4118 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4120 * config/avr/avr-stdint.h: Remove trailing blanks.
4121 * config/avr/avr-log.h: Same.
4122 * config/avr/avr-arch.h: Same.
4123 * config/avr/avr-devices.c: Same.
4124 * config/avr/avr-dimode.md: Same.
4125 * config/avr/predicates.md: Same.
4126 * config/avr/avr-c.c: Same. And fix typo.
4128 * config/avr/avr-protos.h: Same. And:
4129 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
4130 (init_cumulative_args): Rename to avr_init_cumulative_args.
4131 (expand_prologue): Rename to avr_expand_prologue.
4132 (expand_epilogue): Rename to avr_expand_epilogue.
4133 (adjust_insn_length): Rename to avr_adjust_insn_length.
4134 (notice_update_cc): Rename to avr_notice_update_cc.
4135 (final_prescan_insn): Rename to avr_final_prescan_insn.
4136 * config/avr/avr.c: Same.
4137 * config/avr/avr.h: Same.
4138 * config/avr/avr.md: Remove trailing blanks.
4139 (prologue): Use avr_expand_prologue.
4140 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
4142 2013-01-14 Richard Biener <rguenther@suse.de>
4144 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
4145 verify_location, collect_subblocks): New functions.
4146 (verify_gimple_in_cfg): Verify that locations only reference
4147 BLOCKs in the functions BLOCK tree.
4149 2013-01-14 Richard Biener <rguenther@suse.de>
4151 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
4153 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
4155 (insert_out_of_ssa_copy_on_edge): Likewise.
4156 (rewrite_close_phi_out_of_ssa): Likewise.
4157 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
4159 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
4160 propagated constants.
4161 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
4164 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
4166 * config/avr/avr-modes.def: Add GPL copyright notice.
4168 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
4170 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
4171 MEMMODEL_MASK to determine memory model.
4172 (atomic_store<mode>): Ditto from operands[2].
4173 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
4175 2013-01-13 Jakub Jelinek <jakub@redhat.com>
4178 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
4179 (fold_gimple_assign): Don't call unshare_expr here.
4180 (fold_ctor_reference): Call unshare_expr.
4182 2013-01-13 Terry Guo <terry.guo@arm.com>
4184 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
4185 * doc/fragments.texi: Document MULTILIB_REUSE.
4186 * gcc.c (multilib_reuse): New internal spec.
4187 (set_multilib_dir): Also search multilib from multilib_reuse.
4188 * genmultilib (tmpmultilib3): Refactor code.
4189 (tmpmultilib4): Ditto.
4190 (multilib_reuse): New multilib argument.
4192 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
4194 * Makefile.in: Update copyright.
4196 2013-01-12 Tom de Vries <tom@codesourcery.com>
4199 * calls.c (expand_call): Check if arg_nr is valid.
4201 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4203 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
4204 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
4205 documentation. Add missing '__' in front of
4206 __builtin_ia32_packssdw256.
4208 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4211 * config/s390/s390.c (s390_preferred_reload_class): Do not return
4212 NO_REGS for larl operands.
4213 (s390_reload_larl_operand): Use s390_load_address instead of
4216 2013-01-11 Richard Biener <rguenther@suse.de>
4218 * tree-cfg.c (verify_node_sharing_1): Split out from ...
4219 (verify_node_sharing): ... here.
4220 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
4222 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
4224 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
4225 Substitute TREECHECKING.
4226 * configure: Regenerate.
4227 * Makefile.in (TREECHECKING): New.
4229 2013-01-11 Richard Guenther <rguenther@suse.de>
4231 PR tree-optimization/44061
4232 * tree-vrp.c (extract_range_basic): Compute zero as
4233 value-range for __builtin_constant_p of function parameters.
4235 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
4237 Update copyright years.
4239 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
4241 PR rtl-optimization/55672
4242 * lra-eliminations.c (mark_not_eliminable): Permit addition with
4243 const to be eliminable.
4245 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
4247 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
4248 * configure: Regenerate.
4250 2013-01-10 Richard Biener <rguenther@suse.de>
4252 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
4254 2013-01-10 Richard Biener <rguenther@suse.de>
4257 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
4258 locations for virtual PHI arguments.
4259 (rewrite_update_phi_arguments): Likewise.
4261 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
4263 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
4266 2013-01-10 Jakub Jelinek <jakub@redhat.com>
4268 PR tree-optimization/55921
4269 * tree-complex.c (expand_complex_asm): New function.
4270 (expand_complex_operations_1): Call it for GIMPLE_ASM.
4272 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4275 * config/s390/s390.c (s390_symref_operand_p)
4276 (s390_loadrelative_operand_p): Merge the two functions.
4277 (s390_check_qrst_address, print_operand_address): Add parameters
4278 to s390_loadrelative_operand_p invokation.
4279 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
4280 (s390_reload_larl_operand, s390_secondary_reload): Use
4281 s390_loadrelative_operand_p instead of s390_symref_operand_p.
4282 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
4284 2013-01-09 Mike Stump <mikestump@comcast.net>
4286 * dse.c (record_store): Remove unnecessary assert.
4288 2013-01-09 Jan Hubicka <jh@suse.cz>
4290 PR tree-optimization/55569
4291 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
4292 * cfgloop.h (scale_loop_profile): Likewise.
4294 2013-01-09 Jan Hubicka <jh@suse.cz>
4297 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
4299 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
4301 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
4304 * expr.h (maybe_emit_group_store): Declare.
4305 * expr.c (maybe_emit_group_store): New function.
4306 * builtins.c (expand_builtin_int_roundingfn): Call it.
4307 (expand_builtin_int_roundingfn_2): Likewise.
4309 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
4311 PR rtl-optimization/55829
4312 * lra-constraints.c (match_reload): Add code for absent output.
4313 (curr_insn_transform): Add code for reloads of matched inputs
4316 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
4318 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
4319 attribute of movddup insn to DF.
4320 (*vec_interleave_lowv2df): Ditto.
4321 (vec_dupv2df): Ditto.
4323 2013-01-09 Jan Hubicka <jh@suse.cz>
4325 PR tree-optimiation/55875
4326 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
4327 EVERY_ITERATION parameter.
4328 (number_of_iterations_exit): Check if exit is executed every iteration.
4329 (idx_infer_loop_bounds): Similarly here.
4330 (n_of_executions_at_most): Simplify
4331 to only test for cases where statement is dominated by the
4332 particular bound; handle correctly the "postdominance" test.
4333 (scev_probably_wraps_p): Use max loop iterations info
4334 as a global bound first.
4336 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
4337 Nick Clifton <nickc@redhat.com>
4339 * config/v850/v850.md (cbranchsf4): New pattern.
4340 (cstoresf4): New pattern.
4341 (cbranchdf4): New pattern.
4342 (cstoredf4): New pattern.
4343 (movsicc): Disallow floating point comparisons.
4344 (cmpsf_le_insn): Fix order of operators.
4345 (cmpsf_lt_insn): Likewise.
4346 (cmpsf_eq_insn): Likewise.
4347 (cmpdf_le_insn): Likewise.
4348 (cmpdf_lt_insn): Likewise.
4349 (cmpdf_eq_insn): Likewise.
4350 (cmpsf_ge_insn): Use LE comparison.
4351 (cmpdf_ge_insn): Likewise.
4352 (cmpsf_gt_insn): Use LT comparison.
4353 (cmpdf_gt_insn): Likewise.
4354 (cmpsf_ne_insn): Delete pattern.
4355 (cmpdf_ne_insn): Delete pattern.
4356 * config/v850/v850.c (v850_gen_float_compare): Use
4357 gen_cmpdf_eq_insn for NE comparison.
4358 (v850_float_z_comparison_operator)
4359 (v850_float_nz_comparison_operator): Move from here ...
4360 * config/v850/predicates.md: ... to here. Move GT and GE
4361 comparisons into v850_float_z_comparison_operator.
4362 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
4364 (v850_float_nz_comparison_operator): Likewise.
4366 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4368 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
4369 with calls to gen_insvsi/gen_insvdi.
4371 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4373 * config/i386/i386.c (initial_ix86_tune_features): Set up
4374 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
4376 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
4377 Jakub Jelinek <jakub@redhat.com>
4379 PR tree-optimization/48189
4380 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
4381 If nitercst is 0, don't predict the exit edge.
4383 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
4385 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
4386 in asm_fprintf with reg_names.
4387 (aarch64_print_operand_address): Likewise.
4388 (aarch64_return_addr): Likewise.
4389 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
4391 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4393 * config/pa/pa.h (VAL_U6_BITS_P): Define.
4394 (INT_U6_BITS): Likewise.
4395 * config/pa/predicates.md (uint6_operand): New predicate.
4396 (shift5_operand, shift6_operand): Likewise.
4397 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
4399 (lshrdi3): Use shift6_operand.
4400 (shrpsi4, shrpdi4): New insn patterns.
4401 (extzv): Delete expander.
4402 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
4403 predicates in unamed zero extract patterns. Tighten common constraint.
4404 (extv): Delete expander.
4405 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
4406 predicates in unamed sign extract patterns. Tighten common constraint.
4407 (insv): Delete expander.
4408 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
4409 predicates in unamed insert patterns. Tighten common constraint.
4410 Change uint32_operand predicate to uint6_operand predicate in unamed
4411 DImode pattern to insert constant values of type 1...1xxxx.
4413 2013-01-04 Jan Hubicka <jh@suse.cz>
4415 PR tree-optimization/55823
4416 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
4419 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4420 Uros Bizjak <ubizjak@gmail.com>
4422 PR rtl-optimization/55845
4423 * df-problems.c (can_move_insns_across): Stop scanning at
4424 volatile_insn_p source instruction or give up if
4425 across_from .. across_to range contains any volatile_insn_p
4428 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
4430 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
4431 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
4433 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
4434 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
4436 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4439 * asan.c (asan_clear_shadow): New function.
4440 (asan_emit_stack_protection): Use it.
4442 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
4444 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
4445 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
4446 with tab instead of space.
4448 2013-01-08 Nick Clifton <nickc@redhat.com>
4450 * config/rl78/rl78.c (rl78_expand_prologue): Always select
4451 register bank 0 at the start of an interrupt handler.
4452 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
4455 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
4457 * config/aarch64/aarch64-simd.md
4458 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
4459 (aarch64_simd_bsl): Likewise.
4460 (aarch64_vcond_internal<mode>): Likewise.
4461 (vcond<mode><mode>): Likewise.
4462 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
4463 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
4465 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
4467 * config/aarch64/aarch64-builtins.c
4468 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
4470 2013-01-08 Martin Jambor <mjambor@suse.cz>
4473 * tree-sra.c (analyze_access_subtree): Return true also after
4474 potentially creating a debug-only replacement.
4476 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4479 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
4481 PR tree-optimization/54120
4482 * tree-vrp.c (range_fits_type_p): Don't allow
4483 src_precision < precision from signed vr to unsigned_p
4484 if vr->min or vr->max is negative.
4485 (simplify_float_conversion_using_ranges): Test can_float_p
4486 against CODE_FOR_nothing.
4488 2013-01-08 Jakub Jelinek <jakub@redhat.com>
4489 Richard Biener <rguenther@suse.de>
4492 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
4493 types instead of just INTEGER_TYPE types.
4495 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
4497 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
4500 2013-01-07 Steve Ellcey <sellcey@mips.com>
4503 * config/mips/mips.opt: Change mad to mmad to match documentation.
4505 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4508 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
4511 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4514 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
4515 (avr_addrspace_t): Add .section_name field.
4516 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
4518 (avr_addrspace): Same. Initialize .section_name. Remove last
4519 NULL entry. Put __memx into .progmemx.data.
4520 (progmem_section_prefix): Remove.
4521 (avr_asm_init_sections): No need to initialize progmem_section.
4522 (avr_asm_named_section): Use avr_addrspace[].section_name to get
4523 section name prefix.
4524 (avr_asm_select_section): Ditto. And use get_unnamed_section to
4525 retrieve the progmem section.
4526 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
4527 boundary to run over avr_addrspace[].
4528 (avr_register_target_pragmas): Ditto.
4530 2013-01-06 Jakub Jelinek <jakub@redhat.com>
4532 * varasm.c (output_constant_def_contents): For asan_protect_global
4533 protected strings, adjust DECL_ALIGN if needed, before testing for
4535 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
4536 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
4538 (output_object_block): For asan protected decls, emit asan padding
4539 after their contents.
4540 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
4541 (asan_finish_file): Test it here instead.
4543 2013-01-07 Nick Clifton <nickc@redhat.com>
4544 Matthias Klose <doko@debian.org>
4545 Doug Kwan <dougkwan@google.com>
4546 H.J. Lu <hongjiu.lu@intel.com>
4549 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
4551 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
4553 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
4555 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
4557 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
4559 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4562 * doc/install.texi (Cross-Compiler-Specific Options): Document
4565 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
4567 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
4568 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
4569 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
4570 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
4571 vqmovun_high_s64): Fix source operand number and update copyright.
4573 2013-01-07 Richard Biener <rguenther@suse.de>
4576 * gimple.h (gimple_call_builtin_p): New overload.
4577 * gimple.c (validate_call): New function.
4578 (gimple_call_builtin_p): Likewise.
4579 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4580 Use gimple_call_builtin_p.
4581 (find_func_clobbers): Likewise.
4582 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
4583 (strlen_optimize_stmt): Likewise.
4585 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4587 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
4588 (vld1q_dup_*): Likewise.
4590 (vld1q_*): Likewise.
4591 (vld1_lane_*): Likewise.
4592 (vld1q_lane_*): Likewise.
4594 2013-01-07 Richard Biener <rguenther@suse.de>
4596 * lto-streamer.h (LTO_minor_version): Bump to 2.
4598 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4600 * config/aarch64/aarch64-protos.h
4601 (aarch64_const_double_zero_rtx_p): Rename to...
4602 (aarch64_float_const_zero_rtx_p): ...this.
4603 (aarch64_float_const_representable_p): New.
4604 (aarch64_output_simd_mov_immediate): Likewise.
4605 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
4606 move immediate case.
4607 * config/aarch64/aarch64.c
4608 (aarch64_const_double_zero_rtx_p): Rename to...
4609 (aarch64_float_const_zero_rtx_p): ...this.
4610 (aarch64_print_operand): Allow printing of new constants.
4611 (aarch64_valid_floating_const): New.
4612 (aarch64_legitimate_constant_p): Check for valid floating-point
4614 (aarch64_simd_valid_immediate): Likewise.
4615 (aarch64_vect_float_const_representable_p): New.
4616 (aarch64_float_const_representable_p): Likewise.
4617 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
4618 (aarch64_output_simd_mov_immediate): New.
4619 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
4620 (*movdf_aarch64): Likewise.
4621 * config/aarch64/constraints.md (Ufc): New.
4622 (Y): call aarch64_float_const_zero_rtx.
4623 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
4625 2013-01-07 Richard Biener <rguenther@suse.de>
4627 PR tree-optimization/55888
4628 PR tree-optimization/55862
4629 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
4630 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
4631 not if it is contained therein.
4633 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4635 * config/avr/t-avr: Typo.
4637 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
4640 * config/avr/t-avr: Don't automatically rebuild
4641 $(srcdir)/config/avr/t-multilib
4642 $(srcdir)/config/avr/avr-tables.opt
4643 $(srcdir)/doc/avr-mmcu.texi
4644 (avr-mcus): New phony target to build them on request.
4645 (s-avr-mlib, s-avr-mmcu-texi): Remove.
4646 * avr/avr-mcus.def: Adjust comments.
4648 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
4650 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
4652 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
4654 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
4656 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
4658 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
4660 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
4662 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
4663 to generate profiling.
4664 * config/rs6000/aix64.h (LIB_SPEC): Same.
4666 2013-01-04 Andrew Pinski <apinski@cavium.com>
4668 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
4670 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
4672 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
4674 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
4676 (ix86_expand_move): Ditto.
4677 (ix86_zero_extend_to_Pmode): Ditto.
4678 (ix86_expand_call): Ditto.
4679 (ix86_expand_special_args_builtin): Ditto.
4680 (ix86_expand_builtin): Ditto.
4682 2013-01-04 Richard Biener <rguenther@suse.de>
4684 PR tree-optimization/55862
4685 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
4686 translating them through PHI nodes.
4688 2013-01-04 Martin Jambor <mjambor@suse.cz>
4690 PR tree-optimization/55755
4691 * tree-sra.c (sra_modify_assign): Do not check that an access has no
4692 children when trying to avoid producing a VIEW_CONVERT_EXPR.
4694 2013-01-04 Marek Polacek <polacek@redhat.com>
4697 * opts.c (default_options_optimization): Clarify error message.
4699 2013-01-04 Richard Biener <rguenther@suse.de>
4702 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
4705 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4708 * config/pa/pa.md (movsi): Revert previous change.
4709 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
4712 2013-01-03 Richard Henderson <rth@redhat.com>
4714 * config/i386/i386.c (ix86_expand_move): Always assign to op1
4715 after eliminating TLS symbols.
4717 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
4720 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
4721 * graphite-poly.c (debug_gmp_value): Likewise.
4723 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
4726 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
4727 selected code model, define __code_mode_small__, __code_model_medium__,
4728 __code_model_large__, __code_model_32__ or __code_model_kernel__.
4729 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
4730 xchg temporary register with %k. Declare temporary register as
4732 [__x86_64__]: For medium and large code models, preserve %rbx register.
4734 2013-01-03 Richard Biener <rguenther@suse.de>
4736 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
4737 (dump_subscript): Adjust.
4738 (finalize_ddr_dependent): Do not dump redundant info.
4739 (analyze_siv_subscript): Adjust.
4740 (subscript_dependence_tester): Likewise.
4741 (compute_affine_dependence): Likewise.
4743 2013-01-03 Richard Biener <rguenther@suse.de>
4746 2013-01-03 Richard Biener <rguenther@suse.de>
4748 PR tree-optimization/55857
4749 * tree-vect-stmts.c (vectorizable_load): Do not setup
4750 re-alignment for invariant loads.
4752 2013-01-02 Richard Biener <rguenther@suse.de>
4754 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
4755 invariant load do not generate a vector load from the scalar location.
4757 2013-01-03 Richard Biener <rguenther@suse.de>
4759 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
4760 for not vectorizing.
4761 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
4762 not build INDIRECT_REFs, call get_name once only.
4763 (vect_create_data_ref_ptr): Likewise. Dump base object kind
4764 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
4766 2013-01-03 Richard Biener <rguenther@suse.de>
4768 PR tree-optimization/55857
4769 * tree-vect-stmts.c (vectorizable_load): Do not setup
4770 re-alignment for invariant loads.
4772 2013-01-03 Richard Biener <rguenther@suse.de>
4775 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
4776 prefer a built-in decl.
4778 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4780 * gcc.c (process_command): Update copyright notice dates.
4781 * gcov.c (print_version): Likewise.
4782 * gcov-dump.c (print_version): Likewise.
4784 PR rtl-optimization/55838
4785 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
4786 iv0.step, iv1.step and step.
4788 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4789 Marc Glisse <marc.glisse@inria.fr>
4791 PR tree-optimization/55832
4792 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
4793 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
4794 integer_{one,zero}_node.
4796 2013-01-03 Jakub Jelinek <jakub@redhat.com>
4799 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
4800 * var-tracking.c (reverse_op): Don't add reverse ops to
4801 VALUEs that have already
4802 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
4804 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
4806 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
4808 2013-01-02 Teresa Johnson <tejohnson@google.com>
4810 * dumpfile.c (dump_loc): Print filename with location.
4811 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
4812 new location_t parameter to emit complete unroll message with
4814 (canonicalize_loop_induction_variables): Compute loops location
4815 and pass to try_unroll_loop_completely.
4816 * loop-unroll.c (report_unroll_peel): New function.
4817 (peel_loops_completely): Use new dump format with location
4818 for main dumpfile message, and invoke report_unroll_peel on success.
4819 (decide_unrolling_and_peeling): Ditto.
4820 (decide_peel_once_rolling): Remove old dumpfile message subsumed
4821 by report_unroll_peel.
4822 (decide_peel_completely): Ditto.
4823 (decide_unroll_constant_iterations): Ditto.
4824 (decide_unroll_runtime_iterations): Ditto.
4825 (decide_peel_simple): Ditto.
4826 (decide_unroll_stupid): Ditto.
4827 * cfgloop.c (get_loop_location): New function.
4828 * cfgloop.h (get_loop_location): Declare.
4830 2013-01-02 Sriraman Tallam <tmsriram@google.com>
4832 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
4835 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
4838 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
4839 BLKmode objects when EXPAND_MEMORY is specified.
4841 2013-01-02 Sriraman Tallam <tmsriram@google.com>
4843 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
4845 (fold_builtin_cpu): Do not share cpu model decls across statements.
4847 2013-01-02 Jason Merrill <jason@redhat.com>
4850 * tree.c (build_array_type_1): Revert earlier change.
4852 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
4854 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
4856 * config/aarch64/aarch64-tune.md: Re-generate.
4858 2013-01-02 Richard Biener <rguenther@suse.de>
4860 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
4861 invariant load do not generate a vector load from the scalar location.
4863 2013-01-02 Richard Biener <rguenther@suse.de>
4866 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
4867 * configure: Regenerate.
4869 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4871 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
4872 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
4873 (expand_builtin_int_roundingfn_2): Keep the original target around
4874 for the fallback case.
4876 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4878 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
4879 to be clear for sign changes.
4881 2013-01-01 Jan Hubicka <jh@suse.cz>
4883 * ipa-inline-analysis.c: Fix formatting.
4885 2013-01-01 Jakub Jelinek <jakub@redhat.com>
4887 PR tree-optimization/55831
4888 * tree-vect-loop.c (get_initial_def_for_induction): Use
4889 gsi_after_labels instead of gsi_start_bb.
4891 Copyright (C) 2013 Free Software Foundation, Inc.
4893 Copying and distribution of this file, with or without modification,
4894 are permitted in any medium without royalty provided the copyright
4895 notice and this notice are preserved.