1 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/60023
4 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
6 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
7 has been in some EH region and vec_stmt could throw, add
8 vec_stmt into the same EH region.
9 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
10 has no lhs, ignore it.
11 * internal-fn.c (expand_MASK_LOAD): Likewise.
14 * tree-inline.c (copy_forbidden): Fail for
15 __attribute__((optimize (0))) functions.
18 * omp-low.c (simd_clone_struct_copy): If from->inbranch
19 is set, copy one less argument.
20 (expand_simd_clones): Don't subtract clone_info->inbranch
21 from simd_clone_struct_alloc argument.
23 PR rtl-optimization/57915
24 * recog.c (simplify_while_replacing): If all unary/binary/relational
25 operation arguments are constant, attempt to simplify those.
28 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
29 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
31 2014-02-04 Richard Biener <rguenther@suse.de>
33 PR tree-optimization/60012
34 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
35 TBAA disambiguation to all DDRs.
37 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
41 (LINK_SPEC): Use it for -shared, -shared-libgcc.
43 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
46 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
48 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
50 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
51 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
53 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
56 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
57 to figure out targets of polymorphic calls with known decl.
58 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
59 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
60 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
61 (get_polymorphic_call_info): ... here.
62 (get_polymorphic_call_info_from_invariant): New function.
64 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
66 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
67 lookup via vtable pointer; check for type consistency
68 and turn inconsitent facts into UNREACHABLE.
69 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
70 * gimple-fold.c (gimple_get_virt_method_for_vtable):
71 Do not ICE on type inconsistent querries; return UNREACHABLE
74 2014-02-03 Richard Henderson <rth@twiddle.net>
77 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
78 already processed this node.
79 (normalize_one_pred_1): Pass along mark_set.
80 (normalize_one_pred): Create and destroy a pointer_set_t.
81 (normalize_one_pred_chain): Likewise.
83 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
86 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
88 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
91 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
92 Try to devirtualize by the knowledge of virtual table pointer given by
93 aggregate propagation.
94 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
95 ipa_print_node_jump_functions): Dump also offset that
96 is relevant for polymorphic calls.
97 (determine_known_aggregate_parts): Add arg_type parameter; use it
98 instead of determining the type from pointer type.
99 (ipa_compute_jump_functions_for_edge): Update call of
100 determine_known_aggregate_parts.
101 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
102 (gimple_get_virt_method_for_binfo): ... here; simplify using
103 vtable_pointer_value_to_vtable.
104 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
105 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
106 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
107 (vtable_pointer_value_to_vtable): Break out from ...; handle also
109 (vtable_pointer_value_to_binfo): ... here.
110 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
112 2014-02-03 Teresa Johnson <tejohnson@google.com>
114 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
115 redef of outer loop index variable.
117 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
121 * doc/extend.texi (Function Attributes): Typo.
123 2014-02-03 Cong Hou <congh@google.com>
125 PR tree-optimization/60000
126 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
127 if the vectorized statement is a store. A store statement can only
128 appear at the end of pattern statements.
130 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
132 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
133 (ix86_option_override_internal): Default long double to 64-bit for
134 32-bit Bionic and to 128-bit for 64-bit Bionic.
136 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
137 TARGET_LONG_DOUBLE_128 is true.
138 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
140 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
141 (mlong-double-64): Negate -mlong-double-128.
142 (mlong-double-128): New option.
144 * config/i386/i386-c.c (ix86_target_macros): Define
145 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
147 * doc/invoke.texi: Document -mlong-double-128.
149 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
151 PR rtl-optimization/60024
152 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
154 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
156 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
158 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
160 PR rtl-optimization/57662
161 * sel-sched.c (code_motion_path_driver): Do not mark already not
162 existing blocks in the visiting bitmap.
164 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
166 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
167 on the insn being emitted.
169 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
170 Will Deacon <will.deacon@arm.com>
172 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
174 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
176 * config/arm/arm-tables.opt: Regenerate.
178 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
180 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
181 for vector types other than V16QImode.
182 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
183 define_expand, and call altivec_expand_vec_perm_le when producing
184 code with little endian element order.
185 (*altivec_vperm_<mode>_internal): New insn having previous
186 behavior of altivec_vperm_<mode>.
187 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
188 altivec_expand_vec_perm_le when producing code with little endian
190 (*altivec_vperm_<mode>_uns_internal): New insn having previous
191 behavior of altivec_vperm_<mode>_uns.
193 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
195 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
196 (altivec_vsumsws): Add handling for -maltivec=be with a little
198 (altivec_vsumsws_direct): New.
199 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
202 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
204 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
205 vtable_pointer_value_to_binfo): New functions.
206 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
207 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
209 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
211 * config/nios2/nios2.md (load_got_register): Initialize GOT
212 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
213 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
215 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
217 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
218 preserverd by passthrough, do not propagate the type.
220 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
222 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
223 (mips_atomic_assign_expand_fenv): New function.
224 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
226 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
228 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
229 (__builtin_mips_set_fcsr): Likewise.
230 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
232 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
233 (mips16_expand_set_fcsr): Likewise.
234 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
235 (mips16_set_fcsr_stub): Likewise.
236 (mips16_get_fcsr_one_only_stub): New class.
237 (mips16_set_fcsr_one_only_stub): Likewise.
238 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
239 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
240 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
241 (hard_float): New availability predicate.
242 (mips_builtins): Add get_fcsr and set_fcsr.
243 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
244 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
245 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
246 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
247 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
250 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
252 * config/mips/mips.c (mips_one_only_stub): New class.
253 (mips_need_mips16_rdhwr_p): Replace with...
254 (mips16_rdhwr_stub): ...this new variable.
255 (mips16_stub_call_address): New function.
256 (mips16_rdhwr_one_only_stub): New class.
257 (mips_expand_thread_pointer): Use mips16_stub_call_address.
258 (mips_output_mips16_rdhwr): Delete.
259 (mips_finish_stub): New function.
260 (mips_code_end): Use it to handle rdhwr stubs.
262 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
265 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
266 when calculating size of integer atomic types.
268 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
270 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
272 2014-02-01 Jakub Jelinek <jakub@redhat.com>
274 PR tree-optimization/60003
275 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
276 * profile.c (branch_prob): Use gimple_call_builtin_p
277 to check for BUILT_IN_SETJMP_RECEIVER.
278 * tree-inline.c (copy_bb): Call notice_special_calls.
280 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
283 * lra-constraints.c (process_alt_operands): Update reload_sum only
286 2014-01-31 Richard Henderson <rth@redhat.com>
289 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
290 until after else_eh is processed.
292 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
294 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
295 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
296 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
297 in smmintrin.h, remove them.
298 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
299 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
300 * config/i386/i386.md (ROUND_SAE): Fix value.
301 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
302 (const48_operand): New.
303 * config/i386/subst.md (round), (round_expand): Use
304 const_4_or_8_to_11_operand.
305 (round_saeonly), (round_saeonly_expand): Use const48_operand.
307 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
309 * config/i386/constraints.md (Yk): Swap meaning with k.
310 * config/i386/i386.md (movhi_internal): Change Yk to k.
311 (movqi_internal): Ditto.
312 (*k<logic><mode>): Ditto.
315 (kandn<mode>): Ditto.
316 (*<code>hi_1): Ditto.
317 (*<code>qi_1): Ditto.
318 (kxnor<mode>): Ditto.
322 (*one_cmplhi2_1): Ditto.
323 (*one_cmplqi2_1): Ditto.
324 * config/i386/sse.md (): Change k to Yk.
325 (avx512f_load<mode>_mask): Ditto.
326 (avx512f_blendm<mode>): Ditto.
327 (avx512f_store<mode>_mask): Ditto.
328 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
329 (avx512f_storedqu<mode>_mask): Ditto.
330 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
332 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
333 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
334 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
335 (avx512f_maskcmp<mode>3): Ditto.
336 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
337 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
338 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
339 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
340 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
341 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
342 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
343 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
344 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
345 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
346 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
347 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
348 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
349 (vec_extract_lo_<mode>_maskm): Ditto.
350 (vec_extract_hi_<mode>_maskm): Ditto.
351 (avx512f_vternlog<mode>_mask): Ditto.
352 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
353 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
354 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
355 (avx512f_<code>v8div16qi2_mask): Ditto.
356 (avx512f_<code>v8div16qi2_mask_store): Ditto.
357 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
358 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
359 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
360 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
361 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
362 (*avx512pf_gatherpf<mode>df_mask): Ditto.
363 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
364 (*avx512pf_scatterpf<mode>df_mask): Ditto.
365 (avx512cd_maskb_vec_dupv8di): Ditto.
366 (avx512cd_maskw_vec_dupv16si): Ditto.
367 (avx512f_vpermi2var<mode>3_maskz): Ditto.
368 (avx512f_vpermi2var<mode>3_mask): Ditto.
369 (avx512f_vpermi2var<mode>3_mask): Ditto.
370 (avx512f_vpermt2var<mode>3_maskz): Ditto.
371 (*avx512f_gathersi<mode>): Ditto.
372 (*avx512f_gathersi<mode>_2): Ditto.
373 (*avx512f_gatherdi<mode>): Ditto.
374 (*avx512f_gatherdi<mode>_2): Ditto.
375 (*avx512f_scattersi<mode>): Ditto.
376 (*avx512f_scatterdi<mode>): Ditto.
377 (avx512f_compress<mode>_mask): Ditto.
378 (avx512f_compressstore<mode>_mask): Ditto.
379 (avx512f_expand<mode>_mask): Ditto.
380 * config/i386/subst.md (mask): Change k to Yk.
381 (mask_scalar_merge): Ditto.
384 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
386 * doc/extend.texi (Vector Extensions): Document ?: in C++.
388 2014-01-31 Richard Biener <rguenther@suse.de>
391 * builtins.c (fold_builtin_memory_op): Make sure to not
392 use a floating-point mode or a boolean or enumeral type for
395 2014-01-30 DJ Delorie <dj@redhat.com>
397 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
398 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
399 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
400 whenever main() has an epilogue.
402 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
404 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
405 unused variable "field".
406 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
407 (vsx_mergeh_<mode>): Likewise.
408 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
409 (altivec_vmrghh): Likewise.
410 (altivec_vmrghw): Likewise.
411 (altivec_vmrglb): Likewise.
412 (altivec_vmrglh): Likewise.
413 (altivec_vmrglw): Likewise.
414 (altivec_vspltb): Add missing uses.
415 (altivec_vsplth): Likewise.
416 (altivec_vspltw): Likewise.
417 (altivec_vspltsf): Likewise.
419 2014-01-30 Jakub Jelinek <jakub@redhat.com>
422 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
423 frame related instructions.
425 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
427 PR rtl-optimization/59959
428 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
429 any reload of register whose subreg is invalid.
431 2014-01-30 Jakub Jelinek <jakub@redhat.com>
433 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
434 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
435 Add missing return type - void.
437 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
439 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
440 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
441 remove element index adjustment for endian (now handled in vsx.md
443 (altivec_expand_vec_perm_const): Use
444 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
445 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
446 (vsx_xxspltw_<mode>): Adjust element index for little endian.
447 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
448 define_expand and a new define_insn *altivec_vspltb_internal;
449 adjust for -maltivec=be on a little endian target.
450 (altivec_vspltb_direct): New.
451 (altivec_vsplth): Divide into a define_expand and a new
452 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
453 little endian target.
454 (altivec_vsplth_direct): New.
455 (altivec_vspltw): Divide into a define_expand and a new
456 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
457 little endian target.
458 (altivec_vspltw_direct): New.
459 (altivec_vspltsf): Divide into a define_expand and a new
460 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
461 a little endian target.
463 2014-01-30 Richard Biener <rguenther@suse.de>
465 PR tree-optimization/59993
466 * tree-ssa-forwprop.c (associate_pointerplus): Check we
467 can propagate form the earlier stmt and avoid the transform
468 when the intermediate result is needed.
470 2014-01-30 Alangi Derick <alangiderick@gmail.com>
472 * README.Portability: Fix typo.
474 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
476 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
477 comparison_operator with ordered_comparison_operator.
479 2014-01-30 Nick Clifton <nickc@redhat.com>
481 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
482 Rename to mn10300_store_multiple_regs.
483 * config/mn10300/mn10300.c: Likewise.
484 * config/mn10300/mn10300.md (store_movm): Fix typo: call
486 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
487 Call mn10300_store_multiple_regs.
489 2014-01-30 Nick Clifton <nickc@redhat.com>
490 DJ Delorie <dj@redhat.com>
492 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
493 %fp 2 to keep registers after it properly word-aligned.
494 (rl78_alloc_physical_registers_umul): Handle the case where both
495 input operands are the same.
497 2014-01-30 Richard Biener <rguenther@suse.de>
499 PR tree-optimization/59903
500 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
503 2014-01-30 Jason Merrill <jason@redhat.com>
506 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
509 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
511 2014-01-30 Richard Biener <rguenther@suse.de>
513 PR tree-optimization/59951
514 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
516 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
519 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
520 SFmode to DFmode case.
522 2014-01-29 DJ Delorie <dj@redhat.com>
524 * config/msp430/msp430.opt (-minrt): New.
525 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
527 (ENDFILE_SPEC): Likewise.
529 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
531 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
532 (estimate_function_body_sizes): Use it.
534 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
537 * dwarf2out.c (is_cxx_auto): New.
538 (is_base_type): Use it.
539 (gen_type_die_with_usage): Likewise.
541 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
543 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
544 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
545 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
546 -maltivec=be with LE targets.
547 (vsx_mergeh_<mode>): Likewise.
548 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
549 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
550 (altivec_vmrghb): Replace with define_expand and new
551 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
552 (altivec_vmrghb_direct): New define_insn.
553 (altivec_vmrghh): Replace with define_expand and new
554 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
555 (altivec_vmrghh_direct): New define_insn.
556 (altivec_vmrghw): Replace with define_expand and new
557 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
558 (altivec_vmrghw_direct): New define_insn.
559 (*altivec_vmrghsf): Adjust for endianness.
560 (altivec_vmrglb): Replace with define_expand and new
561 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
562 (altivec_vmrglb_direct): New define_insn.
563 (altivec_vmrglh): Replace with define_expand and new
564 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
565 (altivec_vmrglh_direct): New define_insn.
566 (altivec_vmrglw): Replace with define_expand and new
567 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
568 (altivec_vmrglw_direct): New define_insn.
569 (*altivec_vmrglsf): Adjust for endianness.
570 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
571 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
572 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
573 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
574 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
575 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
576 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
577 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
579 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
581 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
582 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
585 2014-01-29 Richard Biener <rguenther@suse.de>
587 PR tree-optimization/58742
588 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
589 associate_pointerplus_align.
590 (associate_pointerplus_diff): New function.
591 (associate_pointerplus): Likewise. Call associate_pointerplus_align
592 and associate_pointerplus_diff.
594 2014-01-29 Richard Biener <rguenther@suse.de>
596 * lto-streamer.h (LTO_major_version): Bump to 3.
597 (LTO_minor_version): Reset to 0.
599 2014-01-29 Renlin Li <Renlin.Li@arm.com>
601 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
602 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
603 (arm_file_start): Generate correct asm header for armv7ve.
604 * config/arm/bpabi.h: Add multilib support for armv7ve.
605 * config/arm/driver-arm.c: Change the architectures of cortex-a7
606 and cortex-a15 to armv7ve.
607 * config/arm/t-aprofile: Add multilib support for armv7ve.
608 * doc/invoke.texi: Document -march=armv7ve.
610 2014-01-29 Richard Biener <rguenther@suse.de>
612 PR tree-optimization/58742
613 * tree-ssa-forwprop.c (associate_plusminus): Return true
614 if we changed sth, defer EH cleanup to ...
615 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
616 (simplify_mult): New function.
618 2014-01-29 Jakub Jelinek <jakub@redhat.com>
621 PR tree-optimization/59920
622 * tree.c (build_common_builtin_nodes): Remove
623 __builtin_setjmp_dispatcher initialization.
624 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
625 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
626 instead of gsi_after_labels + manually skipping debug stmts.
627 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
628 ignore bbs with IFN_ABNORMAL_DISPATCHER.
629 * tree-inline.c (copy_edges_for_bb): Remove
630 can_make_abnormal_goto argument, instead add abnormal_goto_dest
631 argument. Ignore computed_goto_p stmts. Don't call
632 make_abnormal_goto_edges. If a call might need abnormal edges
633 for non-local gotos, see if it already has an edge to
634 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
635 with true argument, don't do anything then, otherwise add
636 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
637 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
639 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
640 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
641 (lower_stmt): Don't set data->calls_builtin_setjmp.
642 (lower_builtin_setjmp): Adjust comment.
643 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
644 * tree-cfg.c (found_computed_goto): Remove.
645 (factor_computed_gotos): Remove.
646 (make_goto_expr_edges): Return bool, true for computed gotos.
647 Don't call make_abnormal_goto_edges.
648 (build_gimple_cfg): Don't set found_computed_goto, don't call
649 factor_computed_gotos.
650 (computed_goto_p): No longer static.
651 (make_blocks): Don't set found_computed_goto.
652 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
653 (make_edges): If make_goto_expr_edges returns true, push bb
654 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
655 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
656 vector. Record mapping between bbs and OpenMP regions if there
657 are any, adjust make_gimple_omp_edges caller. Call
658 handle_abnormal_edges.
659 (make_abnormal_goto_edges): Remove.
660 * tree-cfg.h (make_abnormal_goto_edges): Remove.
661 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
662 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
663 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
664 * internal-fn.def (ABNORMAL_DISPATCHER): New.
665 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
666 filling *region also set *region_idx to (*region)->entry->index.
669 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
670 For REGs set ORIGINAL_REGNO.
672 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
674 * doc/md.texi: Mention that a target shouldn't implement
675 vec_widen_(s|u)mul_even/odd pair if it is less efficient
678 2014-01-29 Jakub Jelinek <jakub@redhat.com>
680 PR tree-optimization/59594
681 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
682 a copy of the datarefs vector rather than the vector itself.
684 2014-01-28 Jason Merrill <jason@redhat.com>
687 * dwarf2out.c (auto_die): New static.
688 (gen_type_die_with_usage): Handle C++1y 'auto'.
689 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
692 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
695 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
696 (SPEC_X32): Likewise.
698 * config/i386/i386.c (ix86_option_override_internal): Turn off
699 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
701 (x86_file_start): Output .code16gcc for TARGET_16BIT.
702 * config/i386/i386.h (TARGET_16BIT): New macro.
703 (TARGET_16BIT_P): Likewise.
704 * config/i386/i386.opt: Add m16.
705 * doc/invoke.texi: Document -m16.
707 2014-01-28 Jakub Jelinek <jakub@redhat.com>
709 PR preprocessor/59935
710 * input.c (location_get_source_line): Bail out on when line number
711 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
713 2014-01-28 Richard Biener <rguenther@suse.de>
715 PR tree-optimization/58742
716 * tree-ssa-forwprop.c (associate_plusminus): Handle
717 pointer subtraction of the form (T)(P + A) - (T)P.
719 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
721 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
724 2014-01-28 Richard Biener <rguenther@suse.de>
727 2014-01-28 Richard Biener <rguenther@suse.de>
729 PR rtl-optimization/45364
730 PR rtl-optimization/59890
731 * var-tracking.c (local_get_addr_clear_given_value): Handle
732 already cleared slot.
733 (val_reset): Handle not allocated local_get_addr_cache.
734 (vt_find_locations): Use post-order on the inverted CFG.
736 2014-01-28 Richard Biener <rguenther@suse.de>
738 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
740 2014-01-28 Richard Biener <rguenther@suse.de>
742 PR rtl-optimization/45364
743 PR rtl-optimization/59890
744 * var-tracking.c (local_get_addr_clear_given_value): Handle
745 already cleared slot.
746 (val_reset): Handle not allocated local_get_addr_cache.
747 (vt_find_locations): Use post-order on the inverted CFG.
749 2014-01-28 Alan Modra <amodra@gmail.com>
751 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
752 * configure.ac <recursive call for build != host>: Define
753 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
754 and LD_FOR_BUILD too.
755 * configure: Regenerate.
757 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
759 * config/i386/i386.c (get_builtin_code_for_version): Separate
760 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
761 Broadwell from Haswell.
763 2014-01-27 Steve Ellcey <sellcey@mips.com>
765 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
766 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
767 * config/mips/mips.c (mips_option_override): Change setting
769 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
770 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
771 Change from Mask to Var.
773 2014-01-27 Jeff Law <law@redhat.com>
775 * ipa-inline.c (inline_small_functions): Fix typo.
777 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
779 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
780 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
781 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
782 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
783 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
784 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
785 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
786 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
787 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
788 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
789 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
790 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
791 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
792 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
793 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
794 (_mm512_storeu_epi64): Ditto.
795 (_mm512_cmpge_epi32_mask): Ditto.
796 (_mm512_cmpge_epu32_mask): Ditto.
797 (_mm512_cmpge_epi64_mask): Ditto.
798 (_mm512_cmpge_epu64_mask): Ditto.
799 (_mm512_cmple_epi32_mask): Ditto.
800 (_mm512_cmple_epu32_mask): Ditto.
801 (_mm512_cmple_epi64_mask): Ditto.
802 (_mm512_cmple_epu64_mask): Ditto.
803 (_mm512_cmplt_epi32_mask): Ditto.
804 (_mm512_cmplt_epu32_mask): Ditto.
805 (_mm512_cmplt_epi64_mask): Ditto.
806 (_mm512_cmplt_epu64_mask): Ditto.
807 (_mm512_cmpneq_epi32_mask): Ditto.
808 (_mm512_cmpneq_epu32_mask): Ditto.
809 (_mm512_cmpneq_epi64_mask): Ditto.
810 (_mm512_cmpneq_epu64_mask): Ditto.
811 (_mm512_expand_pd): Ditto.
812 (_mm512_expand_ps): Ditto.
813 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
814 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
815 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
816 * config/i386/i386.c (ix86_builtins): Add
817 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
818 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
819 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
820 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
821 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
822 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
823 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
824 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
825 IX86_BUILTIN_PMOVUSQW512_MEM.
826 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
827 __builtin_ia32_pmovsqd512mem_mask,
828 __builtin_ia32_pmovqd512mem_mask,
829 __builtin_ia32_pmovusqw512mem_mask,
830 __builtin_ia32_pmovsqw512mem_mask,
831 __builtin_ia32_pmovqw512mem_mask,
832 __builtin_ia32_pmovusdw512mem_mask,
833 __builtin_ia32_pmovsdw512mem_mask,
834 __builtin_ia32_pmovdw512mem_mask,
835 __builtin_ia32_pmovqb512mem_mask,
836 __builtin_ia32_pmovusqb512mem_mask,
837 __builtin_ia32_pmovsqb512mem_mask,
838 __builtin_ia32_pmovusdb512mem_mask,
839 __builtin_ia32_pmovsdb512mem_mask,
840 __builtin_ia32_pmovdb512mem_mask.
841 (bdesc_args): Add __builtin_ia32_expanddf512,
842 __builtin_ia32_expandsf512.
843 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
844 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
845 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
846 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
847 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
848 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
849 (avx512f_<code>v8div16qi2_mask_store): This.
850 (avx512f_expand<mode>): New.
852 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
854 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
856 (_mm512_mask_prefetch_i64gather_pd): Ditto.
857 (_mm512_prefetch_i32scatter_pd): Ditto.
858 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
859 (_mm512_prefetch_i64scatter_pd): Ditto.
860 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
861 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
862 (_mm512_mask_prefetch_i64gather_ps): Ditto.
863 (_mm512_prefetch_i32scatter_ps): Ditto.
864 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
865 (_mm512_prefetch_i64scatter_ps): Ditto.
866 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
867 * config/i386/i386-builtin-types.def: Define
868 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
869 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
870 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
871 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
872 IX86_BUILTIN_SCATTERPFQPD.
873 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
874 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
875 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
876 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
877 __builtin_ia32_scatterpfqps.
878 (ix86_expand_builtin): Expand new built-ins.
879 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
880 fix memory access data type.
881 (*avx512pf_gatherpf<mode>_mask): Ditto.
882 (*avx512pf_gatherpf<mode>): Ditto.
883 (avx512pf_scatterpf<mode>): Ditto.
884 (*avx512pf_scatterpf<mode>_mask): Ditto.
885 (*avx512pf_scatterpf<mode>): Ditto.
886 (GATHER_SCATTER_SF_MEM_MODE): New.
887 (avx512pf_gatherpf<mode>df): Ditto.
888 (*avx512pf_gatherpf<mode>df_mask): Ditto.
889 (*avx512pf_scatterpf<mode>df): Ditto.
891 2014-01-27 Jakub Jelinek <jakub@redhat.com>
894 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
895 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
898 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
900 * common/config/arm/arm-common.c
901 (arm_rewrite_mcpu): Handle multiple names.
903 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
905 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
907 * gimple-builder.h (create_gimple_tmp): Delete.
909 2014-01-27 Christian Bruel <christian.bruel@st.com>
911 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
914 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
916 * config/pa/pa.md (call): Generate indirect long calls to non-local
917 functions when outputing 32-bit code.
918 (call_value): Likewise except for special call to buggy powf function.
920 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
921 portable runtime and PIC indirect calls.
922 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
923 and PIC call sequences. Use ldo instead of blr to set return register
924 in PIC call sequence.
926 2014-01-25 Walter Lee <walt@tilera.com>
928 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
929 avoid clobbering a live register.
931 2014-01-25 Walter Lee <walt@tilera.com>
933 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
934 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
935 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
936 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
938 2014-01-25 Walter Lee <walt@tilera.com>
940 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
941 arguments on even registers.
942 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
944 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
945 (BIGGEST_ALIGNMENT): Ditto.
946 (BIGGEST_FIELD_ALIGNMENT): Ditto.
948 2014-01-25 Walter Lee <walt@tilera.com>
950 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
951 insns before bundling.
952 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
954 2014-01-25 Walter Lee <walt@tilera.com>
956 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
957 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
958 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
960 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
962 * config/mips/constraints.md (kl): Delete.
963 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
964 define expands, using...
965 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
966 instructions for MIPS16.
967 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
968 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
970 2014-01-25 Walter Lee <walt@tilera.com>
972 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
976 2014-01-25 Walter Lee <walt@tilera.com>
978 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
979 (TARGET_EXPAND_TO_RTL_HOOK): Define.
981 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
983 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
986 2014-01-25 Jakub Jelinek <jakub@redhat.com>
988 * print-rtl.c (in_call_function_usage): New var.
989 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
990 EXPR_LIST mode as mode and not as reg note name.
993 * cfgloopmanip.c (copy_loop_info): If
994 loop->warned_aggressive_loop_optimizations, make sure
995 the flag is set in target loop too.
997 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
999 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1001 * builtins.def: Likewise.
1002 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1003 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1004 * ira.c (ira_setup_eliminable_regset): Likewise.
1005 * omp-low.c (gate_expand_omp): Likewise.
1006 (execute_lower_omp): Likewise.
1007 (diagnose_sb_0): Likewise.
1008 (gate_diagnose_omp_blocks): Likewise.
1009 (simd_clone_clauses_extract): Likewise.
1012 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1014 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1015 correction for little endian...
1016 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1019 2014-01-24 Jeff Law <law@redhat.com>
1021 PR tree-optimization/59919
1022 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1023 for non-returning calls.
1025 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1027 * common/config/aarch64/aarch64-common.c
1028 (aarch64_rewrite_mcpu): Handle multiple names.
1029 * config/aarch64/aarch64.h
1030 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1032 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1034 * input.c (add_file_to_cache_tab): Handle the case where fopen
1037 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1040 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1041 from push operand if code of push isn't PRE_DEC.
1043 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1046 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1047 -mquad-memory-atomic. Update -mquad-memory documentation to say
1048 it is only used for non-atomic loads/stores.
1050 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1051 -mquad-memory or -mquad-memory-atomic switches.
1053 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1054 -mquad-memory-atomic to ISA 2.07 support.
1056 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1057 to separate support of normal quad word memory operations (ldq, stq)
1058 from the atomic quad word memory operations.
1060 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1061 support to separate non-atomic quad word operations from atomic
1062 quad word operations. Disable non-atomic quad word operations in
1063 little endian mode so that we don't have to swap words after the
1064 load and before the store.
1065 (quad_load_store_p): Add comment about atomic quad word support.
1066 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1067 options printed with -mdebug=reg.
1069 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1070 -mquad-memory-atomic as the test for whether we have quad word
1071 atomic instructions.
1072 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1073 or -mp8-vector are used, allow byte/half-word atomic operations.
1075 * config/rs6000/sync.md (load_lockedti): Insure that the address
1076 is a proper indexed or indirect address for the lqarx instruction.
1077 On little endian systems, swap the hi/lo registers after the lqarx
1079 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1080 insure the address is valid for the lqarx instruction.
1081 (store_conditionalti): Insure that the address is a proper indexed
1082 or indirect address for the stqcrx. instruction. On little endian
1083 systems, swap the hi/lo registers before doing the stqcrx.
1085 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1086 insure the address is valid for the stqcrx. instruction.
1088 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1089 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1090 type of quad memory support is available.
1092 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1095 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1096 there is a danger of looping.
1098 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1100 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1101 force flag_ira_loop_pressure if set via command line.
1103 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1105 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1106 (ashr_simd): New builtin handling DI mode.
1107 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1108 (aarch64_sshr_simddi): New match pattern.
1109 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1110 (vshrd_n_s64): Likewise.
1111 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1113 2014-01-23 Nick Clifton <nickc@redhat.com>
1115 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1116 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1117 favour of mcu specific scripts.
1118 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1121 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1122 Alex Velenko <Alex.Velenko@arm.com>
1124 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1125 (vaddv_s16): Likewise.
1126 (vaddv_s32): Likewise.
1127 (vaddv_u8): Likewise.
1128 (vaddv_u16): Likewise.
1129 (vaddv_u32): Likewise.
1130 (vaddvq_s8): Likewise.
1131 (vaddvq_s16): Likewise.
1132 (vaddvq_s32): Likewise.
1133 (vaddvq_s64): Likewise.
1134 (vaddvq_u8): Likewise.
1135 (vaddvq_u16): Likewise.
1136 (vaddvq_u32): Likewise.
1137 (vaddvq_u64): Likewise.
1138 (vaddv_f32): Likewise.
1139 (vaddvq_f32): Likewise.
1140 (vaddvq_f64): Likewise.
1141 (vmaxv_f32): Likewise.
1142 (vmaxv_s8): Likewise.
1143 (vmaxv_s16): Likewise.
1144 (vmaxv_s32): Likewise.
1145 (vmaxv_u8): Likewise.
1146 (vmaxv_u16): Likewise.
1147 (vmaxv_u32): Likewise.
1148 (vmaxvq_f32): Likewise.
1149 (vmaxvq_f64): Likewise.
1150 (vmaxvq_s8): Likewise.
1151 (vmaxvq_s16): Likewise.
1152 (vmaxvq_s32): Likewise.
1153 (vmaxvq_u8): Likewise.
1154 (vmaxvq_u16): Likewise.
1155 (vmaxvq_u32): Likewise.
1156 (vmaxnmv_f32): Likewise.
1157 (vmaxnmvq_f32): Likewise.
1158 (vmaxnmvq_f64): Likewise.
1159 (vminv_f32): Likewise.
1160 (vminv_s8): Likewise.
1161 (vminv_s16): Likewise.
1162 (vminv_s32): Likewise.
1163 (vminv_u8): Likewise.
1164 (vminv_u16): Likewise.
1165 (vminv_u32): Likewise.
1166 (vminvq_f32): Likewise.
1167 (vminvq_f64): Likewise.
1168 (vminvq_s8): Likewise.
1169 (vminvq_s16): Likewise.
1170 (vminvq_s32): Likewise.
1171 (vminvq_u8): Likewise.
1172 (vminvq_u16): Likewise.
1173 (vminvq_u32): Likewise.
1174 (vminnmv_f32): Likewise.
1175 (vminnmvq_f32): Likewise.
1176 (vminnmvq_f64): Likewise.
1178 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1180 * config/aarch64/aarch64-simd.md
1181 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1182 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1183 (*aarch64_mul3_elt<mode>): Likewise.
1184 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1185 (*aarch64_mul3_elt_to_64v2df): Likewise.
1186 (*aarch64_mla_elt<mode>): Likewise.
1187 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1188 (*aarch64_mls_elt<mode>): Likewise.
1189 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1190 (*aarch64_fma4_elt<mode>): Likewise.
1191 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1192 (*aarch64_fma4_elt_to_64v2df): Likewise.
1193 (*aarch64_fnma4_elt<mode>): Likewise.
1194 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1195 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1196 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1197 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1198 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1199 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1200 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1201 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1202 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1204 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1206 * config/aarch64/aarch64-simd.md
1207 (aarch64_be_checked_get_lane<mode>): New define_expand.
1208 * config/aarch64/aarch64-simd-builtins.def
1209 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1210 New builtin definition.
1211 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1212 Use new safe be builtin.
1214 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1216 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1218 (aarch64_be_st1<mode>): Likewise.
1219 (aarch_ld1<VALL:mode>): Define_expand modified.
1220 (aarch_st1<VALL:mode>): Likewise.
1221 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1222 (UNSPEC_ST1): Likewise.
1224 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1226 * config/microblaze/microblaze.md: Add trap insn and attribute
1228 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1230 PR preprocessor/58580
1231 * input.h (location_get_source_line): Take an additional line_size
1233 (void diagnostics_file_cache_fini): Declare new function.
1234 * input.c (struct fcache): New type.
1235 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1236 New static constants.
1237 (diagnostic_file_cache_init, total_lines_num)
1238 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1239 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1240 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1241 (get_next_line, read_next_line, goto_next_line, read_line_num):
1242 New static function definitions.
1243 (diagnostic_file_cache_fini): New function.
1244 (location_get_source_line): Take an additional output line_len
1245 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1247 * diagnostic.c (diagnostic_finish): Call
1248 diagnostic_file_cache_fini.
1249 (adjust_line): Take an additional input parameter for the length
1250 of the line, rather than calculating it with strlen.
1251 (diagnostic_show_locus): Adjust the use of
1252 location_get_source_line and adjust_line with respect to their new
1253 signature. While displaying a line now, do not stop at the first
1254 null byte. Rather, display the zero byte as a space and keep
1255 going until we reach the size of the line.
1256 * Makefile.in: Add vec.o to OBJS-libcommon
1258 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1259 Ilya Tocar <ilya.tocar@intel.com>
1261 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1262 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1263 (__builtin_ia32_kmov16): Ditto.
1264 * config/i386/i386.md (UNSPEC_KMOV): New.
1267 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1269 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1270 (_mm512_storeu_si512): Ditto.
1272 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1275 * rtl.h (get_referenced_operands): Declare.
1276 * recog.c (get_referenced_operands): New function.
1277 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1278 operands have been referenced when recording LO_SUM references.
1280 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1282 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1284 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1286 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1287 Enable for generic and recent AMD targets.
1289 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1291 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1292 ARG_SIZE note when adjustment was eliminated.
1294 2014-01-22 Jeff Law <law@redhat.com>
1296 PR tree-optimization/59597
1297 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1298 in file. Accept new argument REGISTERING and use it to modify
1299 dump output appropriately.
1300 (register_jump_thread): Corresponding changes.
1301 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1302 thread paths involving joiner blocks. Add code to dump cancelled
1303 jump threading paths.
1305 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1307 PR rtl-optimization/59477
1308 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1309 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1311 2014-01-22 Tom Tromey <tromey@redhat.com>
1313 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1315 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1317 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1319 PR rtl-optimization/59896
1320 * lra-constraints.c (process_alt_operands): Check unused note for
1321 matched operands of insn with no output reloads.
1323 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1325 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1326 (mips_move_from_gpr_cost): Likewise.
1328 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1330 PR rtl-optimization/59858
1331 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1332 ira_class_hard_regs_num.
1333 (process_alt_operands): Increase reject for dying matched operand.
1335 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1338 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1339 smaller than size, perform several stores or loads and stores
1340 at dst + count - size to store or copy all of size bytes, rather
1341 than just last modesize bytes.
1343 2014-01-20 DJ Delorie <dj@redhat.com>
1345 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1346 that CLOBBERs are REGs before propogating their values.
1348 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1351 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1352 (cgraph_inline_failed_type): New function.
1353 * cgraph.h (DEFCIFCODE): Add type.
1354 (cgraph_inline_failed_type_t): New enum.
1355 (cgraph_inline_failed_type): New prototype.
1356 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1357 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1358 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1359 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1360 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1361 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1362 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1363 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1364 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1365 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1366 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1367 OPTIMIZATION_MISMATCH.
1368 * tree-inline.c (expand_call_inline): Emit errors during
1369 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1371 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1374 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1375 mode attribute in insn output.
1377 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1379 * output.h (output_constant): Delete.
1380 * varasm.c (output_constant): Make private.
1382 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1384 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1386 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1389 * tree.h (fold_builtin_strcat): New prototype.
1390 * builtins.c (fold_builtin_strcat): No longer static. Add len
1391 argument, if non-NULL, don't call c_strlen. Optimize
1392 directly into __builtin_memcpy instead of __builtin_strcpy.
1393 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1394 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1396 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1398 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1399 for SImode_address_operand operands, having only a REG argument.
1401 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1403 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1404 loader name using mbig-endian.
1405 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1407 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1409 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1413 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1415 * config/aarch64/aarch64-protos.h
1416 (aarch64_cannot_change_mode_class_ptr): Declare.
1417 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1418 aarch64_cannot_change_mode_class_ptr): New.
1419 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1420 backend hook aarch64_cannot_change_mode_class.
1422 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1424 * common/config/aarch64/aarch64-common.c
1425 (aarch64_handle_option): Don't handle any option order logic here.
1426 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1427 selected_cpu, warn on architecture version mismatch.
1428 (aarch64_override_options): Fix parsing order for option strings.
1430 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1431 Iain Sandoe <iain@codesourcery.com>
1434 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1435 warning. Amend comment to reflect current functionality.
1437 2014-01-20 Richard Biener <rguenther@suse.de>
1440 * builtins.c (fold_builtin_strcat): Remove case better handled
1441 by tree-ssa-strlen.c.
1443 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1445 * config/aarch64/aarch64.opt
1446 (mcpu, march, mtune): Make case-insensitive.
1448 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1451 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1452 if operands[1] is a REG or ZERO_EXTEND of a REG.
1454 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1456 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1458 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1460 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1461 long non-pic millicode calls.
1463 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1465 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1467 2014-01-19 Kito Cheng <kito@0xlab.org>
1469 * builtins.c (expand_movstr): Check movstr expand done or fail.
1471 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1472 H.J. Lu <hongjiu.lu@intel.com>
1475 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1476 to DImode for zero-extended addresses.
1478 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1480 PR rtl-optimization/57763
1481 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1482 on the new indirect jump_insn and increment LABEL_NUSES (label).
1484 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1488 * config.gcc (x86_archs): New variable.
1489 (x86_64_archs): Likewise.
1490 (x86_cpus): Likewise.
1491 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1492 --with-arch/--with-cpu= options.
1493 Support --with-arch=/--with-cpu={nehalem,westmere,
1494 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1496 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1498 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1499 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1501 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1503 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1505 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1508 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1509 clear cpp_get_options (parse_in)->warn_unused_macros for
1510 ix86_target_macros_internal with cpp_define.
1512 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1514 * jump.c (delete_related_insns): Keep (use (insn))s.
1515 * reorg.c (redundant_insn): Check for barriers too.
1517 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1519 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1521 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1523 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1524 call to $$dyncall when TARGET_LONG_CALLS is true.
1526 2014-01-17 Jeff Law <law@redhat.com>
1528 * ree.c (combine_set_extension): Temporarily disable test for
1529 changing number of hard registers.
1531 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1534 * ipa-inline-analysis.c (inline_free_summary):
1535 Do not free summary of aliases.
1537 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1540 * gimplify.c (gimplify_expr): Use create_tmp_var
1541 instead of create_tmp_var_raw. If cond doesn't have
1542 integral type, don't add the IFN_ANNOTATE builtin at all.
1544 2014-01-17 Martin Jambor <mjambor@suse.cz>
1547 * ipa-cp.c (prev_edge_clone): New variable.
1548 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1549 Also resize prev_edge_clone vector.
1550 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1551 (ipcp_edge_removal_hook): New function.
1552 (ipcp_driver): Register ipcp_edge_removal_hook.
1554 2014-01-17 Andrew Pinski <apinski@cavium.com>
1555 Steve Ellcey <sellcey@mips.com>
1558 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1561 2014-01-17 Jeff Law <law@redhat.com>
1564 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1565 so that pass_ccp runs first.
1567 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1569 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1570 (ix86_adjust_cost): Use !TARGET_XXX.
1571 (do_reorder_for_imul): Likewise.
1572 (swap_top_of_ready_list): Likewise.
1573 (ix86_sched_reorder): Likewise.
1575 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1577 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1578 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1579 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1580 (intel_memset): New. Duplicate slm_memset.
1581 (intel_cost): New. Duplicate slm_cost.
1582 (m_INTEL): New macro.
1583 (processor_target_table): Add "intel".
1584 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1585 with PROCESSOR_INTEL for "intel".
1586 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1587 PROCESSOR_SILVERMONT.
1588 (ix86_issue_rate): Likewise.
1589 (ix86_adjust_cost): Likewise.
1590 (ia32_multipass_dfa_lookahead): Likewise.
1591 (swap_top_of_ready_list): Likewise.
1592 (ix86_sched_reorder): Likewise.
1593 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1594 instead of TARGET_OPT_AGU.
1595 * config/i386/i386.h (TARGET_INTEL): New.
1596 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1597 (processor_type): Add PROCESSOR_INTEL.
1598 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1599 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1601 2014-01-17 Marek Polacek <polacek@redhat.com>
1604 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1607 2014-01-17 Richard Biener <rguenther@suse.de>
1609 PR tree-optimization/46590
1610 * opts.c (default_options_table): Add entries for
1611 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1612 all enabled at -O1 but not for -Og.
1613 * common.opt (fbranch-count-reg): Remove Init(1).
1614 (fmove-loop-invariants): Likewise.
1615 (ftree-pta): Likewise.
1617 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1619 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1620 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1621 decls to at least the GCC 4.8 used alignments.
1624 * tree-nested.c (convert_nonlocal_reference_stmt,
1625 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1626 of GIMPLE_BIND stmts, adjust associated decls.
1628 2014-01-17 Richard Biener <rguenther@suse.de>
1630 PR tree-optimization/46590
1631 * vec.h (vec<>::bseach): New member function implementing
1632 binary search according to C89 bsearch.
1633 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1634 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1635 bitmap pointer again. Make accesses_in_loop a flat array.
1636 (mem_ref_obstack): New global.
1637 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1638 (mark_ref_stored): Likewise.
1639 (ref_indep_loop_p_2): Likewise.
1640 (set_ref_stored_in_loop): New helper function.
1641 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1642 (memref_free): Adjust.
1643 (record_mem_ref_loc): Simplify.
1644 (gather_mem_refs_stmt): Adjust.
1645 (sort_locs_in_loop_postorder_cmp): New function.
1646 (analyze_memory_references): Sort accesses_in_loop after
1647 loop postorder number.
1648 (find_ref_loc_in_loop_cmp): New function.
1649 (for_all_locs_in_loop): Find relevant cluster of locs in
1650 accesses_in_loop and iterate without recursion.
1651 (execute_sm): Avoid uninit warning.
1652 (struct ref_always_accessed): Simplify.
1653 (ref_always_accessed::operator ()): Likewise.
1654 (ref_always_accessed_p): Likewise.
1655 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1656 loop postorder numbers here.
1657 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1660 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1663 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1664 on decls for which assemble_alias has been called.
1666 2014-01-17 Nick Clifton <nickc@redhat.com>
1668 * config/msp430/msp430.opt: (mcpu): New option.
1669 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1670 (msp430_option_override): Parse target_cpu. If the MCU name
1671 matches a generic string, clear target_mcu.
1672 (msp430_attr): Allow numeric interrupt values up to 63.
1673 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1674 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1676 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1678 * config/msp430/msp430.md (popm): Use %J rather than %I.
1679 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1680 (addhi_cy_i): Use immediate_operand for operand 2.
1681 * doc/invoke.texi: Document -mcpu option.
1683 2014-01-17 Richard Biener <rguenther@suse.de>
1685 PR rtl-optimization/38518
1686 * df.h (df_analyze_loop): Declare.
1687 * df-core.c: Include cfgloop.h.
1688 (df_analyze_1): Split out main part of df_analyze.
1689 (df_analyze): Adjust.
1690 (loop_inverted_post_order_compute): New function.
1691 (loop_post_order_compute): Likewise.
1692 (df_analyze_loop): New function avoiding whole-function
1694 * loop-invariant.c (find_defs): Use df_analyze_loop.
1695 (find_invariants): Adjust.
1696 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1698 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1700 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1701 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1703 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1705 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1706 traversal when removing references.
1708 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
1711 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1713 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1716 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1717 pushing a reload for an autoinc when we had previously reloaded an
1718 inner part of the address.
1720 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1722 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1724 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1725 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1726 when not giving up or versioning for alias only because of
1728 (vect_analyze_data_ref_dependences): Set to true.
1729 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1731 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1732 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1736 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1739 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1740 operand 0 predicate for gathers, use a new pseudo as subtarget.
1742 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1745 * lra-constraints.c (process_alt_operands): Add printing debug info.
1746 Check absence of input/output reloads for matched operands too.
1748 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1750 PR rtl-optimization/59835
1751 * ira.c (ira_init_register_move_cost): Increase cost for
1754 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1756 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1758 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1761 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1762 non-register objects. Use gen_(high/low)part more consistently.
1765 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1768 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1769 endian support, remove tests for WORDS_BIG_ENDIAN.
1770 (p8_mfvsrd_3_<mode>): Likewise.
1771 (reload_gpr_from_vsx<mode>): Likewise.
1772 (reload_gpr_from_vsxsf): Likewise.
1773 (p8_mfvsrd_4_disf): Likewise.
1775 2014-01-16 Richard Biener <rguenther@suse.de>
1777 PR rtl-optimization/46590
1778 * lcm.c (compute_antinout_edge): Use postorder iteration.
1779 (compute_laterin): Use inverted postorder iteration.
1781 2014-01-16 Nick Clifton <nickc@redhat.com>
1784 * varasm.c (output_constant): Return the number of bytes actually
1786 (output_constructor_array_range): Update the field size with the
1787 number of bytes emitted by output_constant.
1788 (output_constructor_regular_field): Likewise. Also do not
1789 complain if the total number of bytes emitted is now greater
1790 than the expected fieldpos.
1791 * output.h (output_constant): Update prototype and descriptive comment.
1793 2014-01-16 Marek Polacek <polacek@redhat.com>
1796 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1797 it is error_mark_node.
1799 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1801 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1802 VALID_AVX256_REG_OR_OI_MODE.
1804 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1806 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1807 current procedure should be profiled.
1809 2014-01-15 Andrew Pinski <apinski@cavium.com>
1811 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1812 of moving from/to the STACK_REG register class.
1814 2014-01-15 Richard Henderson <rth@redhat.com>
1817 * reginfo.c (global_regs_decl): Globalize.
1818 * rtl.h (global_regs_decl): Declare.
1819 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1820 reserved via global_regs.
1822 2014-01-15 Teresa Johnson <tejohnson@google.com>
1824 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1826 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1828 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1829 and vmulosh rather than call gen_vec_widen_smult_*.
1830 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1831 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1832 (vec_widen_smult_even_v16qi): Likewise.
1833 (vec_widen_umult_even_v8hi): Likewise.
1834 (vec_widen_smult_even_v8hi): Likewise.
1835 (vec_widen_umult_odd_v16qi): Likewise.
1836 (vec_widen_smult_odd_v16qi): Likewise.
1837 (vec_widen_umult_odd_v8hi): Likewise.
1838 (vec_widen_smult_odd_v8hi): Likewise.
1839 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1840 vmuloub rather than call gen_vec_widen_umult_*.
1841 (vec_widen_umult_lo_v16qi): Likewise.
1842 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1843 vmulosb rather than call gen_vec_widen_smult_*.
1844 (vec_widen_smult_lo_v16qi): Likewise.
1845 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1846 rather than call gen_vec_widen_umult_*.
1847 (vec_widen_umult_lo_v8hi): Likewise.
1848 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1849 rather than call gen_vec_widen_smult_*.
1850 (vec_widen_smult_lo_v8hi): Likewise.
1852 2014-01-15 Jeff Law <law@redhat.com>
1854 PR tree-optimization/59747
1855 * ree.c (find_and_remove_re): Properly handle case where a second
1856 eliminated extension requires widening a copy created for elimination
1857 of a prior extension.
1858 (combine_set_extension): Ensure that the number of hard regs needed
1859 for a destination register does not change when we widen it.
1861 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1863 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1864 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1865 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1866 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1867 (avr-*-rtems*): Likewise.
1868 (bfin*-rtems*): Likewise.
1869 (moxie-*-rtems*): Likewise.
1870 (h8300-*-rtems*): Likewise.
1871 (i[34567]86-*-rtems*): Likewise.
1872 (lm32-*-rtems*): Likewise.
1873 (m32r-*-rtems*): Likewise.
1874 (m68k-*-rtems*): Likewise.
1875 (microblaze*-*-rtems*): Likewise.
1876 (mips*-*-rtems*): Likewise.
1877 (powerpc-*-rtems*): Likewise.
1878 (sh-*-rtems*): Likewise.
1879 (sparc-*-rtems*): Likewise.
1880 (sparc64-*-rtems*): Likewise.
1881 (v850-*-rtems*): Likewise.
1882 (m32c-*-rtems*): Likewise.
1884 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1886 PR rtl-optimization/59511
1887 * ira.c (ira_init_register_move_cost): Use memory costs for some
1888 cases of register move cost calculations.
1889 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1890 instead of BB frequency.
1891 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1892 * lra-assigns.c (find_hard_regno_for): Ditto.
1894 2014-01-15 Richard Biener <rguenther@suse.de>
1896 PR tree-optimization/59822
1897 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1898 (vectorizable_load): Use it to hoist defs of uses of invariant
1899 loads out of the loop.
1901 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1902 Kugan Vivekanandarajah <kuganv@linaro.org>
1905 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1908 2014-01-15 Richard Biener <rguenther@suse.de>
1910 PR rtl-optimization/59802
1911 * lcm.c (compute_available): Use inverted postorder to seed
1912 the initial worklist.
1914 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1917 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1918 ADDR_REGS for invalid symrefs in non-PIC code.
1920 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1923 * builtins.c (determine_block_size): Initialize *probable_max_size
1924 even if len_rtx is CONST_INT.
1926 2014-01-14 Andrew Pinski <apinski@cavium.com>
1928 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1929 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1930 (cortexa53_tunings): Likewise.
1931 (aarch64_sched_issue_rate): New function.
1932 (TARGET_SCHED_ISSUE_RATE): Define.
1934 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1936 * ira-costs.c (find_costs_and_classes): Add missed
1937 ira_init_register_move_cost_if_necessary.
1939 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1942 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1944 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1947 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1948 to indicate if type is used for function return value. Warn ABI
1949 change if the vector mode isn't available for function return value.
1950 (ix86_function_arg_advance): Pass false to type_natural_mode.
1951 (ix86_function_arg): Likewise.
1952 (ix86_gimplify_va_arg): Likewise.
1953 (function_arg_32): Don't warn ABI change.
1954 (ix86_function_value): Pass true to type_natural_mode.
1955 (ix86_return_in_memory): Likewise.
1956 (ix86_struct_value_rtx): Removed.
1957 (TARGET_STRUCT_VALUE_RTX): Likewise.
1959 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1961 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1962 converting a conditional jump into a conditional return.
1964 2014-01-14 Richard Biener <rguenther@suse.de>
1966 PR tree-optimization/58921
1967 PR tree-optimization/59006
1968 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1969 hoisting invariant stmts.
1970 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1971 invariant loads on the preheader edge if possible.
1973 2014-01-14 Joey Ye <joey.ye@arm.com>
1975 * doc/plugin.texi (Building GCC plugins): Update to C++.
1977 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1979 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1980 (_mm_rcp28_round_ss): Ditto.
1981 (_mm_rsqrt28_round_sd): Ditto.
1982 (_mm_rsqrt28_round_ss): Ditto.
1983 (_mm_rcp28_sd): Ditto.
1984 (_mm_rcp28_ss): Ditto.
1985 (_mm_rsqrt28_sd): Ditto.
1986 (_mm_rsqrt28_ss): Ditto.
1987 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1988 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1989 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1990 (IX86_BUILTIN_RCP28SD): Ditto.
1991 (IX86_BUILTIN_RCP28SS): Ditto.
1992 (IX86_BUILTIN_RSQRT28SD): Ditto.
1993 (IX86_BUILTIN_RSQRT28SS): Ditto.
1994 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1995 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1996 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1997 (ix86_expand_special_args_builtin): Expand new FTYPE.
1998 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1999 (srcp14<mode>): Make insn unary.
2000 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2001 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2002 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2003 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2004 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2005 Fix rounding: make it SAE only.
2006 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2008 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2010 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2011 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2012 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2013 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2014 (round_saeonly_mask_scalar_operand4): Ditto.
2015 (round_saeonly_mask_scalar_op3): Ditto.
2016 (round_saeonly_mask_scalar_op4): Ditto.
2018 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2020 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2021 Implement -maltivec=be for vec_insert and vec_extract.
2023 2014-01-10 DJ Delorie <dj@redhat.com>
2025 * config/msp430/msp430.md (call_internal): Don't allow memory
2026 references with SP as the base register.
2027 (call_value_internal): Likewise.
2028 * config/msp430/constraints.md (Yc): New. For memory references
2029 that don't use SP as a base register.
2031 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2032 "an integer without a # prefix"
2033 * config/msp430/msp430.md (epilogue_helper): Use it.
2035 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2038 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2039 AVX512F gather builtins.
2040 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2041 on gather decls with INTEGER_TYPE masktype.
2042 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2043 directly into the builtin rather than hoisting it before loop.
2045 PR tree-optimization/59387
2046 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2047 (scev_const_prop): If folded_casts and type has undefined overflow,
2048 use force_gimple_operand instead of force_gimple_operand_gsi and
2049 for each added stmt if it is assign with
2050 arith_code_with_undefined_signed_overflow, call
2051 rewrite_to_defined_overflow.
2052 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2053 gimple-fold.h instead.
2054 (arith_code_with_undefined_signed_overflow,
2055 rewrite_to_defined_overflow): Moved to ...
2056 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2057 rewrite_to_defined_overflow): ... here. No longer static.
2058 Include gimplify-me.h.
2059 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2060 rewrite_to_defined_overflow): New prototypes.
2062 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2064 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2066 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2068 * builtins.c (get_object_alignment_2): Minor tweak.
2069 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2071 2014-01-13 Christian Bruel <christian.bruel@st.com>
2073 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2074 optimized non constant lengths.
2076 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2079 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2080 load as __atomic_load_N if possible.
2082 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2084 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2086 (rs6000_expand_builtin): Adjust call.
2088 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2091 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2092 * config/rs6000/rs6000.c: Include target-globals.h.
2093 (rs6000_set_current_function): Instead of doing target_reinit
2094 unconditionally, use save_target_globals_default_opts and
2095 restore_target_globals.
2097 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2099 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2100 (rs6000_expand_builtin): Handle mffs and mtfsf.
2101 (rs6000_init_builtins): Define mffs and mtfsf.
2102 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2103 (rs6000_mffs): New pattern.
2104 (rs6000_mtfsf): New pattern.
2106 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2108 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2109 Start narrowing with START. Apply candidate-use pair
2110 and check overall cost in narrowing.
2111 (iv_ca_prune): Pass new argument.
2113 2014-01-10 Jeff Law <law@redhat.com>
2116 * ree.c (combine_reaching_defs): Ensure the defining statement
2117 occurs before the extension when optimizing extensions with
2118 different source and destination hard registers.
2120 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2123 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2124 vtables into the type inheritance graph.
2126 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2128 PR rtl-optimization/59754
2129 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2130 modes in the REGNO != REGNO case.
2132 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2134 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2136 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2138 PR tree-optimization/59745
2139 * tree-predcom.c (tree_predictive_commoning_loop): Call
2140 free_affine_expand_cache if giving up because components is NULL.
2142 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2143 GC in payload of target_globals struct instead of allocating them on
2144 the heap and the larger structs separately using GC.
2145 * target-globals.h (struct target_globals): Make regs, hard_regs,
2146 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2147 of GTY((skip)) and change type to void *.
2148 (reset_target_globals): Cast loads from those fields to corresponding
2151 2014-01-10 Steve Ellcey <sellcey@mips.com>
2154 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2155 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2156 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2158 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2161 * aarch64-modes.def (CC_Zmode): New flags mode.
2162 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2163 represents an equality.
2164 (aarch64_get_condition_code): Handle CC_Zmode.
2165 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2167 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2169 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2170 extraction in good case.
2172 2014-01-10 Richard Biener <rguenther@suse.de>
2174 PR tree-optimization/59374
2175 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2176 checking after SLP discovery. Mark stmts not participating
2177 in any SLP instance properly.
2179 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2181 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2182 when handling a SET rtx.
2184 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2186 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2187 (cortex-a57): Likewise.
2188 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2190 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2192 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2193 non-iwmmxt builtins.
2195 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2199 * ipa-devirt.c record_target_from_binfo): Take as argument
2200 stack of binfos and lookup matching one for virtual inheritance.
2201 (possible_polymorphic_call_targets_1): Update.
2203 2014-01-10 Huacai Chen <chenhc@lemote.com>
2205 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2206 kernel strings for Loongson-2E/2F/3A.
2208 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2211 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2212 is_gimple_call before calling gimple_call_internal_p.
2214 2014-01-09 Steve Ellcey <sellcey@mips.com>
2216 * Makefile.in (TREE_FLOW_H): Remove.
2217 (TREE_SSA_H): Add file names from tree-flow.h.
2218 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2219 * tree.h: Remove tree-flow.h reference.
2220 * hash-table.h: Remove tree-flow.h reference.
2221 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2222 reference with tree-ssa-loop.h.
2224 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2226 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2227 default element-order behavior for -maltivec.
2228 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2230 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2231 when targeting big endian, at least for now.
2232 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2234 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2237 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2238 var satisfies use_register_for_decl, just take into account type
2239 alignment, rather than decl alignment.
2241 PR tree-optimization/59622
2242 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2243 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2244 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2245 Don't devirtualize for inplace at all. For targets.length () == 1,
2246 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2248 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2250 * config/i386/i386.md (cpu): Remove the unused btver1.
2252 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2254 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2256 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2259 * tree-core.h (struct target_globals): New forward declaration.
2260 (struct tree_target_option): Add globals field.
2261 * tree.h (TREE_TARGET_GLOBALS): Define.
2262 (prepare_target_option_nodes_for_pch): New prototype.
2263 * target-globals.h (struct target_globals): Define even if
2265 * tree.c (prepare_target_option_node_for_pch,
2266 prepare_target_option_nodes_for_pch): New functions.
2267 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2268 * config/i386/i386.c: Include target-globals.h.
2269 (ix86_set_current_function): Instead of doing target_reinit
2270 unconditionally, use save_target_globals_default_opts and
2271 restore_target_globals.
2273 2014-01-09 Richard Biener <rguenther@suse.de>
2275 PR tree-optimization/59715
2276 * tree-cfg.h (split_critical_edges): Declare.
2277 * tree-cfg.c (split_critical_edges): Export.
2278 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2280 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2282 * cfgexpand.c (expand_stack_vars): Optionally disable
2283 asan stack protection.
2284 (expand_used_vars): Likewise.
2285 (partition_stack_vars): Likewise.
2286 * asan.c (asan_emit_stack_protection): Optionally disable
2287 after return stack usage.
2288 (instrument_derefs): Optionally disable memory access instrumentation.
2289 (instrument_builtin_call): Likewise.
2290 (instrument_strlen_call): Likewise.
2291 (asan_protect_global): Optionally disable global variables protection.
2292 * doc/invoke.texi: Added doc for new options.
2293 * params.def: Added new options.
2294 * params.h: Likewise.
2296 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2298 PR rtl-optimization/59724
2299 * ifcvt.c (cond_exec_process_if_block): Don't call
2300 flow_find_head_matching_sequence with 0 longest_match.
2301 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2302 non-active insns if !stop_after.
2303 (try_head_merge_bb): Revert 2014-01-07 changes.
2305 2014-01-08 Jeff Law <law@redhat.com>
2307 * ree.c (get_sub_rtx): New function, extracted from...
2308 (merge_def_and_ext): Here.
2309 (combine_reaching_defs): Use get_sub_rtx.
2311 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2313 * cgraph.h (varpool_variable_node): Do not choke on null node.
2315 2014-01-08 Catherine Moore <clm@codesourcery.com>
2317 * config/mips/mips.md (simple_return): Attempt to use JRC
2319 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2321 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2323 PR rtl-optimization/59137
2324 * reorg.c (steal_delay_list_from_target): Call update_block for
2326 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2328 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2330 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2331 two duplicate entries.
2333 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2336 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2338 * config/mips/mips.c (mips_truncated_op_cost): New function.
2339 (mips_rtx_costs): Adjust test for BADDU.
2340 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2342 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2344 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2345 (*baddu_si): ...this new pattern.
2347 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2350 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2352 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2355 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2357 (expand_expr, expand_normal): Adjust.
2358 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2359 inner_reference_p. Use inner_reference_p to expand inner references.
2360 (store_expr): Adjust.
2361 * cfgexpand.c (expand_call_stmt): Adjust.
2363 2014-01-08 Rong Xu <xur@google.com>
2365 * gcov-io.c (gcov_var): Move from gcov-io.h.
2366 (gcov_position): Ditto.
2367 (gcov_is_error): Ditto.
2368 (gcov_rewrite): Ditto.
2369 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2370 only part to libgcc/libgcov.h.
2372 2014-01-08 Marek Polacek <polacek@redhat.com>
2375 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2377 2014-01-08 Marek Polacek <polacek@redhat.com>
2380 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2382 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2384 PR rtl-optimization/59649
2385 * stor-layout.c (get_mode_bounds): For BImode return
2386 0 and STORE_FLAG_VALUE.
2388 2014-01-08 Richard Biener <rguenther@suse.de>
2391 * gimple.h (is_gimple_builtin_call): Remove.
2392 (gimple_builtin_call_types_compatible_p): New.
2393 (gimple_call_builtin_p): New overload.
2394 * gimple.c (is_gimple_builtin_call): Remove.
2395 (validate_call): Rename to ...
2396 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2398 (validate_type): New static function.
2399 (gimple_call_builtin_p): New overload and adjust.
2400 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2401 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2402 (gimple_fold_stmt_to_constant_1): Likewise.
2403 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2405 2014-01-08 Richard Biener <rguenther@suse.de>
2408 * gimplify.c (gimplify_expr): Gimplify register-register type
2409 VIEW_CONVERT_EXPRs to separate stmts.
2411 2014-01-07 Jeff Law <law@redhat.com>
2414 * ree.c (combine_set_extension): Handle case where source
2415 and destination registers in an extension insn are different.
2416 (combine_reaching_defs): Allow source and destination registers
2417 in extension to be different under limited circumstances.
2418 (add_removable_extension): Remove restriction that the
2419 source and destination registers in the extension are the same.
2420 (find_and_remove_re): Emit a copy from the extension's
2421 destination to its source after the defining insn if
2422 the source and destination registers are different.
2425 * ifcvt.c (merge_if_block): If we are merging a block with more than
2426 one successor with a block with no successors, remove any BARRIER
2427 after the second block.
2429 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2431 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2433 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2436 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2437 for 14-bit register offsets when INT14_OK_STRICT is false.
2439 2014-01-07 Roland Stigge <stigge@antcom.de>
2440 Michael Meissner <meissner@linux.vnet.ibm.com>
2443 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2444 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2446 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2448 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2451 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2453 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2454 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2455 rtx is const0_rtx or not.
2457 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2460 * target-globals.c (save_target_globals): Remove this_fn_optab
2462 * toplev.c: Include optabs.h.
2463 (target_reinit): Temporarily restore the global options if another
2464 set of options are in force.
2466 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2468 PR rtl-optimization/58668
2469 * cfgcleanup.c (flow_find_cross_jump): Don't count
2470 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2471 to determine what is counted.
2472 (flow_find_head_matching_sequence): Use active_insn_p to determine
2474 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2476 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2477 determine what is counted.
2479 PR tree-optimization/59643
2480 * tree-predcom.c (split_data_refs_to_components): If one dr is
2481 read and one write, determine_offset fails and the write isn't
2482 in the bad component, just put the read into the bad component.
2484 2014-01-07 Mike Stump <mikestump@comcast.net>
2485 Jakub Jelinek <jakub@redhat.com>
2488 * tree-core.h (struct tree_optimization_option): Change optabs
2489 type from unsigned char * to void *.
2490 * optabs.c (init_tree_optimization_optabs): Adjust
2491 TREE_OPTIMIZATION_OPTABS initialization.
2493 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2496 * config/i386/i386.h (struct machine_function): Add
2497 no_drap_save_restore field.
2498 * config/i386/i386.c (ix86_save_reg): Use
2499 !cfun->machine->no_drap_save_restore instead of
2500 crtl->stack_realign_needed.
2501 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2502 this function clears frame_pointer_needed. Set
2503 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2504 and DRAP reg is needed.
2506 2014-01-06 Marek Polacek <polacek@redhat.com>
2509 * doc/implement-c.texi: Mention that other integer types are
2510 permitted as bit-field types in strictly conforming mode.
2512 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2514 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2517 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2519 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2521 2014-01-06 Martin Jambor <mjambor@suse.cz>
2524 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2526 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2528 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2532 * var-tracking.c (add_stores): Preserve the value of the source even if
2533 we don't record the store.
2535 2014-01-06 Terry Guo <terry.guo@arm.com>
2537 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2539 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2542 * config/darwin.c (darwin_function_section): Adjust return values to
2543 correspond to optimisation changes made in r206070.
2545 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2547 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2548 from prefetch_block tune setting.
2549 (nocona_cost): Correct size of prefetch block to 64.
2551 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2553 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2554 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2555 used to save the static chain register in the computation of the offset
2556 from which the FP registers need to be restored.
2558 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2560 PR tree-optimization/59519
2561 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2562 ICE if get_current_def (current_new_name) is already non-NULL, as long
2563 as it is a phi result of some other phi in *new_exit_bb that has
2566 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2567 or vmovdqu* for misaligned_operand.
2568 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2569 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2570 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2571 aligned_mem for AVX512F masked aligned load and store builtins and for
2574 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2576 PR tree-optimization/59651
2577 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2578 Address range for negative step should be added by TYPE_SIZE_UNIT.
2580 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2582 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2583 overlapping registers also for registers other than the stack pointer.
2585 2014-01-03 Marek Polacek <polacek@redhat.com>
2588 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2591 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2594 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2597 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2598 (push splitter): Use <P:MODE_SIZE> instead of
2599 GET_MODE_SIZE (<P:MODE>mode).
2600 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2601 (mov -1, reg peephole2): Likewise.
2602 * config/i386/sse.md (*mov<mode>_internal,
2603 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2604 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2605 *<code><mode>3, *andnot<mode>3<mask_name>,
2606 <mask_codefor><code><mode>3<mask_name>): Likewise.
2607 * config/i386/subst.md (mask_mode512bit_condition,
2608 sd_mask_mode512bit_condition): Likewise.
2610 2014-01-02 Xinliang David Li <davidxl@google.com>
2612 PR tree-optimization/59303
2613 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2614 (dump_predicates): Better output format.
2615 (pred_equal_p): New function.
2616 (is_neq_relop_p): Ditto.
2617 (is_neq_zero_form_p): Ditto.
2618 (pred_expr_equal_p): Ditto.
2619 (pred_neg_p): Ditto.
2620 (simplify_pred): Ditto.
2621 (simplify_preds_2): Ditto.
2622 (simplify_preds_3): Ditto.
2623 (simplify_preds_4): Ditto.
2624 (simplify_preds): Ditto.
2626 (push_to_worklist): Ditto.
2627 (get_pred_info_from_cmp): Ditto.
2628 (is_degenerated_phi): Ditto.
2629 (normalize_one_pred_1): Ditto.
2630 (normalize_one_pred): Ditto.
2631 (normalize_one_pred_chain): Ditto.
2632 (normalize_preds): Ditto.
2633 (normalize_cond_1): Remove function.
2634 (normalize_cond): Ditto.
2635 (is_gcond_subset_of): Ditto.
2636 (is_subset_of_any): Ditto.
2637 (is_or_set_subset_of): Ditto.
2638 (is_and_set_subset_of): Ditto.
2639 (is_norm_cond_subset_of): Ditto.
2640 (pred_chain_length_cmp): Ditto.
2641 (convert_control_dep_chain_into_preds): Type change.
2642 (find_predicates): Ditto.
2643 (find_def_preds): Ditto.
2644 (destroy_predicates_vecs): Ditto.
2645 (find_matching_predicates_in_rest_chains): Ditto.
2646 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2647 (is_pred_expr_subset): Ditto.
2648 (is_pred_chain_subset_of): Ditto.
2649 (is_included_in): Ditto.
2650 (is_superset_of): Ditto.
2652 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2654 Update copyright years.
2656 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2658 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2659 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2660 config/arc/arc.md, config/arc/arc.opt,
2661 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2662 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2663 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2664 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2665 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2666 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2667 vtable-verify.c, vtable-verify.h: Use the standard form for the
2670 2014-01-02 Tobias Burnus <burnus@net-b.de>
2672 * gcc.c (process_command): Update copyright notice dates.
2673 * gcov-dump.c: Ditto.
2675 * doc/cpp.texi: Bump @copying's copyright year.
2676 * doc/cppinternals.texi: Ditto.
2677 * doc/gcc.texi: Ditto.
2678 * doc/gccint.texi: Ditto.
2679 * doc/gcov.texi: Ditto.
2680 * doc/install.texi: Ditto.
2681 * doc/invoke.texi: Ditto.
2683 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2685 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2687 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2689 * config/i386/sse.md (*mov<mode>_internal): Guard
2690 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2692 PR rtl-optimization/59647
2693 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2694 new_rtx into UNSIGNED_FLOAT rtxes.
2696 Copyright (C) 2014 Free Software Foundation, Inc.
2698 Copying and distribution of this file, with or without modification,
2699 are permitted in any medium without royalty provided the copyright
2700 notice and this notice are preserved.