1 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4 * config/pa/pa.c (legitimize_tls_address): Return original address
5 if not passed a SYMBOL_REF rtx.
6 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
8 (pa_emit_move_sequence): Simplify TLS source operands.
9 (pa_legitimate_constant_p): Reject all TLS constants.
10 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
11 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
13 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
15 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
16 groups when we know they are controlled by LTO.
17 * varasm.c (default_binds_local_p_1): If object is in other partition,
18 it will be resolved locally.
20 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
22 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
23 use SSIZE_MAX because it is not always defined.
25 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
28 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
29 threshold for pseudo splitting.
30 (update_ebb_live_info): Process call argument hard registers and
31 hard registers from insn definition too.
32 (max_small_class_regs_num): New constant.
33 (inherit_in_ebb): Update live hard regs through EBBs. Update
34 reloads_num only for small register classes. Don't split for
37 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
40 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
43 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
45 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
47 2014-02-04 Marek Polacek <polacek@redhat.com>
49 * gdbinit.in (pel): Define.
51 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
53 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
56 2014-02-04 Richard Biener <rguenther@suse.de>
59 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
60 in function context local.
61 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
62 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
63 similar to LTO_imported_decl_ref.
65 2014-02-04 Jakub Jelinek <jakub@redhat.com>
67 PR tree-optimization/60002
68 * cgraphclones.c (build_function_decl_skip_args): Clear
71 PR tree-optimization/60023
72 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
74 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
75 has been in some EH region and vec_stmt could throw, add
76 vec_stmt into the same EH region.
77 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
78 has no lhs, ignore it.
79 * internal-fn.c (expand_MASK_LOAD): Likewise.
82 * tree-inline.c (copy_forbidden): Fail for
83 __attribute__((optimize (0))) functions.
86 * omp-low.c (simd_clone_struct_copy): If from->inbranch
87 is set, copy one less argument.
88 (expand_simd_clones): Don't subtract clone_info->inbranch
89 from simd_clone_struct_alloc argument.
91 PR rtl-optimization/57915
92 * recog.c (simplify_while_replacing): If all unary/binary/relational
93 operation arguments are constant, attempt to simplify those.
96 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
97 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
99 2014-02-04 Richard Biener <rguenther@suse.de>
101 PR tree-optimization/60012
102 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
103 TBAA disambiguation to all DDRs.
105 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
108 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
109 (LINK_SPEC): Use it for -shared, -shared-libgcc.
111 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
114 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
116 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
118 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
119 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
121 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
124 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
125 to figure out targets of polymorphic calls with known decl.
126 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
127 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
128 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
129 (get_polymorphic_call_info): ... here.
130 (get_polymorphic_call_info_from_invariant): New function.
132 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
134 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
135 lookup via vtable pointer; check for type consistency
136 and turn inconsitent facts into UNREACHABLE.
137 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
138 * gimple-fold.c (gimple_get_virt_method_for_vtable):
139 Do not ICE on type inconsistent querries; return UNREACHABLE
142 2014-02-03 Richard Henderson <rth@twiddle.net>
145 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
146 already processed this node.
147 (normalize_one_pred_1): Pass along mark_set.
148 (normalize_one_pred): Create and destroy a pointer_set_t.
149 (normalize_one_pred_chain): Likewise.
151 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
153 PR gcov-profile/58602
154 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
156 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
159 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
160 Try to devirtualize by the knowledge of virtual table pointer given by
161 aggregate propagation.
162 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
163 ipa_print_node_jump_functions): Dump also offset that
164 is relevant for polymorphic calls.
165 (determine_known_aggregate_parts): Add arg_type parameter; use it
166 instead of determining the type from pointer type.
167 (ipa_compute_jump_functions_for_edge): Update call of
168 determine_known_aggregate_parts.
169 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
170 (gimple_get_virt_method_for_binfo): ... here; simplify using
171 vtable_pointer_value_to_vtable.
172 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
173 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
174 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
175 (vtable_pointer_value_to_vtable): Break out from ...; handle also
177 (vtable_pointer_value_to_binfo): ... here.
178 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
180 2014-02-03 Teresa Johnson <tejohnson@google.com>
182 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
183 redef of outer loop index variable.
185 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
189 * doc/extend.texi (Function Attributes): Typo.
191 2014-02-03 Cong Hou <congh@google.com>
193 PR tree-optimization/60000
194 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
195 if the vectorized statement is a store. A store statement can only
196 appear at the end of pattern statements.
198 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
200 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
201 (ix86_option_override_internal): Default long double to 64-bit for
202 32-bit Bionic and to 128-bit for 64-bit Bionic.
204 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
205 TARGET_LONG_DOUBLE_128 is true.
206 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
208 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
209 (mlong-double-64): Negate -mlong-double-128.
210 (mlong-double-128): New option.
212 * config/i386/i386-c.c (ix86_target_macros): Define
213 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
215 * doc/invoke.texi: Document -mlong-double-128.
217 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
219 PR rtl-optimization/60024
220 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
222 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
224 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
226 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
228 PR rtl-optimization/57662
229 * sel-sched.c (code_motion_path_driver): Do not mark already not
230 existing blocks in the visiting bitmap.
232 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
234 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
235 on the insn being emitted.
237 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
238 Will Deacon <will.deacon@arm.com>
240 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
242 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
244 * config/arm/arm-tables.opt: Regenerate.
246 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
248 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
249 for vector types other than V16QImode.
250 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
251 define_expand, and call altivec_expand_vec_perm_le when producing
252 code with little endian element order.
253 (*altivec_vperm_<mode>_internal): New insn having previous
254 behavior of altivec_vperm_<mode>.
255 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
256 altivec_expand_vec_perm_le when producing code with little endian
258 (*altivec_vperm_<mode>_uns_internal): New insn having previous
259 behavior of altivec_vperm_<mode>_uns.
261 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
263 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
264 (altivec_vsumsws): Add handling for -maltivec=be with a little
266 (altivec_vsumsws_direct): New.
267 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
270 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
272 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
273 vtable_pointer_value_to_binfo): New functions.
274 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
275 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
277 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
279 * config/nios2/nios2.md (load_got_register): Initialize GOT
280 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
281 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
283 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
285 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
286 preserverd by passthrough, do not propagate the type.
288 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
290 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
291 (mips_atomic_assign_expand_fenv): New function.
292 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
294 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
296 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
297 (__builtin_mips_set_fcsr): Likewise.
298 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
300 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
301 (mips16_expand_set_fcsr): Likewise.
302 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
303 (mips16_set_fcsr_stub): Likewise.
304 (mips16_get_fcsr_one_only_stub): New class.
305 (mips16_set_fcsr_one_only_stub): Likewise.
306 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
307 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
308 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
309 (hard_float): New availability predicate.
310 (mips_builtins): Add get_fcsr and set_fcsr.
311 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
312 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
313 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
314 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
315 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
318 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
320 * config/mips/mips.c (mips_one_only_stub): New class.
321 (mips_need_mips16_rdhwr_p): Replace with...
322 (mips16_rdhwr_stub): ...this new variable.
323 (mips16_stub_call_address): New function.
324 (mips16_rdhwr_one_only_stub): New class.
325 (mips_expand_thread_pointer): Use mips16_stub_call_address.
326 (mips_output_mips16_rdhwr): Delete.
327 (mips_finish_stub): New function.
328 (mips_code_end): Use it to handle rdhwr stubs.
330 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
333 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
334 when calculating size of integer atomic types.
336 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
338 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
340 2014-02-01 Jakub Jelinek <jakub@redhat.com>
342 PR tree-optimization/60003
343 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
344 * profile.c (branch_prob): Use gimple_call_builtin_p
345 to check for BUILT_IN_SETJMP_RECEIVER.
346 * tree-inline.c (copy_bb): Call notice_special_calls.
348 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
351 * lra-constraints.c (process_alt_operands): Update reload_sum only
354 2014-01-31 Richard Henderson <rth@redhat.com>
357 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
358 until after else_eh is processed.
360 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
362 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
363 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
364 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
365 in smmintrin.h, remove them.
366 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
367 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
368 * config/i386/i386.md (ROUND_SAE): Fix value.
369 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
370 (const48_operand): New.
371 * config/i386/subst.md (round), (round_expand): Use
372 const_4_or_8_to_11_operand.
373 (round_saeonly), (round_saeonly_expand): Use const48_operand.
375 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
377 * config/i386/constraints.md (Yk): Swap meaning with k.
378 * config/i386/i386.md (movhi_internal): Change Yk to k.
379 (movqi_internal): Ditto.
380 (*k<logic><mode>): Ditto.
383 (kandn<mode>): Ditto.
384 (*<code>hi_1): Ditto.
385 (*<code>qi_1): Ditto.
386 (kxnor<mode>): Ditto.
390 (*one_cmplhi2_1): Ditto.
391 (*one_cmplqi2_1): Ditto.
392 * config/i386/sse.md (): Change k to Yk.
393 (avx512f_load<mode>_mask): Ditto.
394 (avx512f_blendm<mode>): Ditto.
395 (avx512f_store<mode>_mask): Ditto.
396 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
397 (avx512f_storedqu<mode>_mask): Ditto.
398 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
400 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
401 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
402 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
403 (avx512f_maskcmp<mode>3): Ditto.
404 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
405 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
406 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
407 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
408 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
409 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
410 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
411 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
412 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
413 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
414 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
415 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
416 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
417 (vec_extract_lo_<mode>_maskm): Ditto.
418 (vec_extract_hi_<mode>_maskm): Ditto.
419 (avx512f_vternlog<mode>_mask): Ditto.
420 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
421 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
422 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
423 (avx512f_<code>v8div16qi2_mask): Ditto.
424 (avx512f_<code>v8div16qi2_mask_store): Ditto.
425 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
426 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
427 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
428 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
429 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
430 (*avx512pf_gatherpf<mode>df_mask): Ditto.
431 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
432 (*avx512pf_scatterpf<mode>df_mask): Ditto.
433 (avx512cd_maskb_vec_dupv8di): Ditto.
434 (avx512cd_maskw_vec_dupv16si): Ditto.
435 (avx512f_vpermi2var<mode>3_maskz): Ditto.
436 (avx512f_vpermi2var<mode>3_mask): Ditto.
437 (avx512f_vpermi2var<mode>3_mask): Ditto.
438 (avx512f_vpermt2var<mode>3_maskz): Ditto.
439 (*avx512f_gathersi<mode>): Ditto.
440 (*avx512f_gathersi<mode>_2): Ditto.
441 (*avx512f_gatherdi<mode>): Ditto.
442 (*avx512f_gatherdi<mode>_2): Ditto.
443 (*avx512f_scattersi<mode>): Ditto.
444 (*avx512f_scatterdi<mode>): Ditto.
445 (avx512f_compress<mode>_mask): Ditto.
446 (avx512f_compressstore<mode>_mask): Ditto.
447 (avx512f_expand<mode>_mask): Ditto.
448 * config/i386/subst.md (mask): Change k to Yk.
449 (mask_scalar_merge): Ditto.
452 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
454 * doc/extend.texi (Vector Extensions): Document ?: in C++.
456 2014-01-31 Richard Biener <rguenther@suse.de>
459 * builtins.c (fold_builtin_memory_op): Make sure to not
460 use a floating-point mode or a boolean or enumeral type for
463 2014-01-30 DJ Delorie <dj@redhat.com>
465 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
466 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
467 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
468 whenever main() has an epilogue.
470 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
472 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
473 unused variable "field".
474 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
475 (vsx_mergeh_<mode>): Likewise.
476 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
477 (altivec_vmrghh): Likewise.
478 (altivec_vmrghw): Likewise.
479 (altivec_vmrglb): Likewise.
480 (altivec_vmrglh): Likewise.
481 (altivec_vmrglw): Likewise.
482 (altivec_vspltb): Add missing uses.
483 (altivec_vsplth): Likewise.
484 (altivec_vspltw): Likewise.
485 (altivec_vspltsf): Likewise.
487 2014-01-30 Jakub Jelinek <jakub@redhat.com>
490 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
491 frame related instructions.
493 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
495 PR rtl-optimization/59959
496 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
497 any reload of register whose subreg is invalid.
499 2014-01-30 Jakub Jelinek <jakub@redhat.com>
501 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
502 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
503 Add missing return type - void.
505 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
507 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
508 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
509 remove element index adjustment for endian (now handled in vsx.md
511 (altivec_expand_vec_perm_const): Use
512 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
513 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
514 (vsx_xxspltw_<mode>): Adjust element index for little endian.
515 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
516 define_expand and a new define_insn *altivec_vspltb_internal;
517 adjust for -maltivec=be on a little endian target.
518 (altivec_vspltb_direct): New.
519 (altivec_vsplth): Divide into a define_expand and a new
520 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
521 little endian target.
522 (altivec_vsplth_direct): New.
523 (altivec_vspltw): Divide into a define_expand and a new
524 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
525 little endian target.
526 (altivec_vspltw_direct): New.
527 (altivec_vspltsf): Divide into a define_expand and a new
528 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
529 a little endian target.
531 2014-01-30 Richard Biener <rguenther@suse.de>
533 PR tree-optimization/59993
534 * tree-ssa-forwprop.c (associate_pointerplus): Check we
535 can propagate form the earlier stmt and avoid the transform
536 when the intermediate result is needed.
538 2014-01-30 Alangi Derick <alangiderick@gmail.com>
540 * README.Portability: Fix typo.
542 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
544 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
545 comparison_operator with ordered_comparison_operator.
547 2014-01-30 Nick Clifton <nickc@redhat.com>
549 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
550 Rename to mn10300_store_multiple_regs.
551 * config/mn10300/mn10300.c: Likewise.
552 * config/mn10300/mn10300.md (store_movm): Fix typo: call
554 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
555 Call mn10300_store_multiple_regs.
557 2014-01-30 Nick Clifton <nickc@redhat.com>
558 DJ Delorie <dj@redhat.com>
560 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
561 %fp 2 to keep registers after it properly word-aligned.
562 (rl78_alloc_physical_registers_umul): Handle the case where both
563 input operands are the same.
565 2014-01-30 Richard Biener <rguenther@suse.de>
567 PR tree-optimization/59903
568 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
571 2014-01-30 Jason Merrill <jason@redhat.com>
574 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
577 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
579 2014-01-30 Richard Biener <rguenther@suse.de>
581 PR tree-optimization/59951
582 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
584 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
587 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
588 SFmode to DFmode case.
590 2014-01-29 DJ Delorie <dj@redhat.com>
592 * config/msp430/msp430.opt (-minrt): New.
593 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
595 (ENDFILE_SPEC): Likewise.
597 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
599 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
600 (estimate_function_body_sizes): Use it.
602 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
605 * dwarf2out.c (is_cxx_auto): New.
606 (is_base_type): Use it.
607 (gen_type_die_with_usage): Likewise.
609 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
611 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
612 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
613 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
614 -maltivec=be with LE targets.
615 (vsx_mergeh_<mode>): Likewise.
616 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
617 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
618 (altivec_vmrghb): Replace with define_expand and new
619 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
620 (altivec_vmrghb_direct): New define_insn.
621 (altivec_vmrghh): Replace with define_expand and new
622 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
623 (altivec_vmrghh_direct): New define_insn.
624 (altivec_vmrghw): Replace with define_expand and new
625 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
626 (altivec_vmrghw_direct): New define_insn.
627 (*altivec_vmrghsf): Adjust for endianness.
628 (altivec_vmrglb): Replace with define_expand and new
629 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
630 (altivec_vmrglb_direct): New define_insn.
631 (altivec_vmrglh): Replace with define_expand and new
632 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
633 (altivec_vmrglh_direct): New define_insn.
634 (altivec_vmrglw): Replace with define_expand and new
635 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
636 (altivec_vmrglw_direct): New define_insn.
637 (*altivec_vmrglsf): Adjust for endianness.
638 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
639 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
640 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
641 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
642 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
643 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
644 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
645 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
647 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
649 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
650 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
653 2014-01-29 Richard Biener <rguenther@suse.de>
655 PR tree-optimization/58742
656 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
657 associate_pointerplus_align.
658 (associate_pointerplus_diff): New function.
659 (associate_pointerplus): Likewise. Call associate_pointerplus_align
660 and associate_pointerplus_diff.
662 2014-01-29 Richard Biener <rguenther@suse.de>
664 * lto-streamer.h (LTO_major_version): Bump to 3.
665 (LTO_minor_version): Reset to 0.
667 2014-01-29 Renlin Li <Renlin.Li@arm.com>
669 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
670 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
671 (arm_file_start): Generate correct asm header for armv7ve.
672 * config/arm/bpabi.h: Add multilib support for armv7ve.
673 * config/arm/driver-arm.c: Change the architectures of cortex-a7
674 and cortex-a15 to armv7ve.
675 * config/arm/t-aprofile: Add multilib support for armv7ve.
676 * doc/invoke.texi: Document -march=armv7ve.
678 2014-01-29 Richard Biener <rguenther@suse.de>
680 PR tree-optimization/58742
681 * tree-ssa-forwprop.c (associate_plusminus): Return true
682 if we changed sth, defer EH cleanup to ...
683 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
684 (simplify_mult): New function.
686 2014-01-29 Jakub Jelinek <jakub@redhat.com>
689 PR tree-optimization/59920
690 * tree.c (build_common_builtin_nodes): Remove
691 __builtin_setjmp_dispatcher initialization.
692 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
693 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
694 instead of gsi_after_labels + manually skipping debug stmts.
695 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
696 ignore bbs with IFN_ABNORMAL_DISPATCHER.
697 * tree-inline.c (copy_edges_for_bb): Remove
698 can_make_abnormal_goto argument, instead add abnormal_goto_dest
699 argument. Ignore computed_goto_p stmts. Don't call
700 make_abnormal_goto_edges. If a call might need abnormal edges
701 for non-local gotos, see if it already has an edge to
702 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
703 with true argument, don't do anything then, otherwise add
704 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
705 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
707 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
708 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
709 (lower_stmt): Don't set data->calls_builtin_setjmp.
710 (lower_builtin_setjmp): Adjust comment.
711 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
712 * tree-cfg.c (found_computed_goto): Remove.
713 (factor_computed_gotos): Remove.
714 (make_goto_expr_edges): Return bool, true for computed gotos.
715 Don't call make_abnormal_goto_edges.
716 (build_gimple_cfg): Don't set found_computed_goto, don't call
717 factor_computed_gotos.
718 (computed_goto_p): No longer static.
719 (make_blocks): Don't set found_computed_goto.
720 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
721 (make_edges): If make_goto_expr_edges returns true, push bb
722 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
723 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
724 vector. Record mapping between bbs and OpenMP regions if there
725 are any, adjust make_gimple_omp_edges caller. Call
726 handle_abnormal_edges.
727 (make_abnormal_goto_edges): Remove.
728 * tree-cfg.h (make_abnormal_goto_edges): Remove.
729 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
730 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
731 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
732 * internal-fn.def (ABNORMAL_DISPATCHER): New.
733 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
734 filling *region also set *region_idx to (*region)->entry->index.
737 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
738 For REGs set ORIGINAL_REGNO.
740 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
742 * doc/md.texi: Mention that a target shouldn't implement
743 vec_widen_(s|u)mul_even/odd pair if it is less efficient
746 2014-01-29 Jakub Jelinek <jakub@redhat.com>
748 PR tree-optimization/59594
749 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
750 a copy of the datarefs vector rather than the vector itself.
752 2014-01-28 Jason Merrill <jason@redhat.com>
755 * dwarf2out.c (auto_die): New static.
756 (gen_type_die_with_usage): Handle C++1y 'auto'.
757 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
760 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
763 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
764 (SPEC_X32): Likewise.
766 * config/i386/i386.c (ix86_option_override_internal): Turn off
767 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
769 (x86_file_start): Output .code16gcc for TARGET_16BIT.
770 * config/i386/i386.h (TARGET_16BIT): New macro.
771 (TARGET_16BIT_P): Likewise.
772 * config/i386/i386.opt: Add m16.
773 * doc/invoke.texi: Document -m16.
775 2014-01-28 Jakub Jelinek <jakub@redhat.com>
777 PR preprocessor/59935
778 * input.c (location_get_source_line): Bail out on when line number
779 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
781 2014-01-28 Richard Biener <rguenther@suse.de>
783 PR tree-optimization/58742
784 * tree-ssa-forwprop.c (associate_plusminus): Handle
785 pointer subtraction of the form (T)(P + A) - (T)P.
787 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
789 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
792 2014-01-28 Richard Biener <rguenther@suse.de>
795 2014-01-28 Richard Biener <rguenther@suse.de>
797 PR rtl-optimization/45364
798 PR rtl-optimization/59890
799 * var-tracking.c (local_get_addr_clear_given_value): Handle
800 already cleared slot.
801 (val_reset): Handle not allocated local_get_addr_cache.
802 (vt_find_locations): Use post-order on the inverted CFG.
804 2014-01-28 Richard Biener <rguenther@suse.de>
806 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
808 2014-01-28 Richard Biener <rguenther@suse.de>
810 PR rtl-optimization/45364
811 PR rtl-optimization/59890
812 * var-tracking.c (local_get_addr_clear_given_value): Handle
813 already cleared slot.
814 (val_reset): Handle not allocated local_get_addr_cache.
815 (vt_find_locations): Use post-order on the inverted CFG.
817 2014-01-28 Alan Modra <amodra@gmail.com>
819 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
820 * configure.ac <recursive call for build != host>: Define
821 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
822 and LD_FOR_BUILD too.
823 * configure: Regenerate.
825 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
827 * config/i386/i386.c (get_builtin_code_for_version): Separate
828 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
829 Broadwell from Haswell.
831 2014-01-27 Steve Ellcey <sellcey@mips.com>
833 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
834 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
835 * config/mips/mips.c (mips_option_override): Change setting
837 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
838 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
839 Change from Mask to Var.
841 2014-01-27 Jeff Law <law@redhat.com>
843 * ipa-inline.c (inline_small_functions): Fix typo.
845 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
847 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
848 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
849 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
850 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
851 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
852 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
853 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
854 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
855 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
856 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
857 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
858 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
859 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
860 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
861 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
862 (_mm512_storeu_epi64): Ditto.
863 (_mm512_cmpge_epi32_mask): Ditto.
864 (_mm512_cmpge_epu32_mask): Ditto.
865 (_mm512_cmpge_epi64_mask): Ditto.
866 (_mm512_cmpge_epu64_mask): Ditto.
867 (_mm512_cmple_epi32_mask): Ditto.
868 (_mm512_cmple_epu32_mask): Ditto.
869 (_mm512_cmple_epi64_mask): Ditto.
870 (_mm512_cmple_epu64_mask): Ditto.
871 (_mm512_cmplt_epi32_mask): Ditto.
872 (_mm512_cmplt_epu32_mask): Ditto.
873 (_mm512_cmplt_epi64_mask): Ditto.
874 (_mm512_cmplt_epu64_mask): Ditto.
875 (_mm512_cmpneq_epi32_mask): Ditto.
876 (_mm512_cmpneq_epu32_mask): Ditto.
877 (_mm512_cmpneq_epi64_mask): Ditto.
878 (_mm512_cmpneq_epu64_mask): Ditto.
879 (_mm512_expand_pd): Ditto.
880 (_mm512_expand_ps): Ditto.
881 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
882 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
883 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
884 * config/i386/i386.c (ix86_builtins): Add
885 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
886 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
887 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
888 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
889 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
890 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
891 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
892 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
893 IX86_BUILTIN_PMOVUSQW512_MEM.
894 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
895 __builtin_ia32_pmovsqd512mem_mask,
896 __builtin_ia32_pmovqd512mem_mask,
897 __builtin_ia32_pmovusqw512mem_mask,
898 __builtin_ia32_pmovsqw512mem_mask,
899 __builtin_ia32_pmovqw512mem_mask,
900 __builtin_ia32_pmovusdw512mem_mask,
901 __builtin_ia32_pmovsdw512mem_mask,
902 __builtin_ia32_pmovdw512mem_mask,
903 __builtin_ia32_pmovqb512mem_mask,
904 __builtin_ia32_pmovusqb512mem_mask,
905 __builtin_ia32_pmovsqb512mem_mask,
906 __builtin_ia32_pmovusdb512mem_mask,
907 __builtin_ia32_pmovsdb512mem_mask,
908 __builtin_ia32_pmovdb512mem_mask.
909 (bdesc_args): Add __builtin_ia32_expanddf512,
910 __builtin_ia32_expandsf512.
911 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
912 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
913 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
914 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
915 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
916 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
917 (avx512f_<code>v8div16qi2_mask_store): This.
918 (avx512f_expand<mode>): New.
920 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
922 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
924 (_mm512_mask_prefetch_i64gather_pd): Ditto.
925 (_mm512_prefetch_i32scatter_pd): Ditto.
926 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
927 (_mm512_prefetch_i64scatter_pd): Ditto.
928 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
929 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
930 (_mm512_mask_prefetch_i64gather_ps): Ditto.
931 (_mm512_prefetch_i32scatter_ps): Ditto.
932 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
933 (_mm512_prefetch_i64scatter_ps): Ditto.
934 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
935 * config/i386/i386-builtin-types.def: Define
936 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
937 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
938 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
939 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
940 IX86_BUILTIN_SCATTERPFQPD.
941 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
942 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
943 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
944 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
945 __builtin_ia32_scatterpfqps.
946 (ix86_expand_builtin): Expand new built-ins.
947 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
948 fix memory access data type.
949 (*avx512pf_gatherpf<mode>_mask): Ditto.
950 (*avx512pf_gatherpf<mode>): Ditto.
951 (avx512pf_scatterpf<mode>): Ditto.
952 (*avx512pf_scatterpf<mode>_mask): Ditto.
953 (*avx512pf_scatterpf<mode>): Ditto.
954 (GATHER_SCATTER_SF_MEM_MODE): New.
955 (avx512pf_gatherpf<mode>df): Ditto.
956 (*avx512pf_gatherpf<mode>df_mask): Ditto.
957 (*avx512pf_scatterpf<mode>df): Ditto.
959 2014-01-27 Jakub Jelinek <jakub@redhat.com>
962 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
963 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
966 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
968 * common/config/arm/arm-common.c
969 (arm_rewrite_mcpu): Handle multiple names.
971 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
973 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
975 * gimple-builder.h (create_gimple_tmp): Delete.
977 2014-01-27 Christian Bruel <christian.bruel@st.com>
979 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
982 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
984 * config/pa/pa.md (call): Generate indirect long calls to non-local
985 functions when outputing 32-bit code.
986 (call_value): Likewise except for special call to buggy powf function.
988 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
989 portable runtime and PIC indirect calls.
990 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
991 and PIC call sequences. Use ldo instead of blr to set return register
992 in PIC call sequence.
994 2014-01-25 Walter Lee <walt@tilera.com>
996 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
997 avoid clobbering a live register.
999 2014-01-25 Walter Lee <walt@tilera.com>
1001 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1002 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1003 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1004 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1006 2014-01-25 Walter Lee <walt@tilera.com>
1008 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1009 arguments on even registers.
1010 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1012 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1013 (BIGGEST_ALIGNMENT): Ditto.
1014 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1016 2014-01-25 Walter Lee <walt@tilera.com>
1018 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1019 insns before bundling.
1020 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1022 2014-01-25 Walter Lee <walt@tilera.com>
1024 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1025 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1026 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1028 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1030 * config/mips/constraints.md (kl): Delete.
1031 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1032 define expands, using...
1033 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1034 instructions for MIPS16.
1035 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1036 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1038 2014-01-25 Walter Lee <walt@tilera.com>
1040 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1044 2014-01-25 Walter Lee <walt@tilera.com>
1046 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1047 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1049 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1051 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1054 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1056 * print-rtl.c (in_call_function_usage): New var.
1057 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1058 EXPR_LIST mode as mode and not as reg note name.
1061 * cfgloopmanip.c (copy_loop_info): If
1062 loop->warned_aggressive_loop_optimizations, make sure
1063 the flag is set in target loop too.
1065 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1067 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1069 * builtins.def: Likewise.
1070 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1071 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1072 * ira.c (ira_setup_eliminable_regset): Likewise.
1073 * omp-low.c (gate_expand_omp): Likewise.
1074 (execute_lower_omp): Likewise.
1075 (diagnose_sb_0): Likewise.
1076 (gate_diagnose_omp_blocks): Likewise.
1077 (simd_clone_clauses_extract): Likewise.
1080 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1082 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1083 correction for little endian...
1084 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1087 2014-01-24 Jeff Law <law@redhat.com>
1089 PR tree-optimization/59919
1090 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1091 for non-returning calls.
1093 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1095 * common/config/aarch64/aarch64-common.c
1096 (aarch64_rewrite_mcpu): Handle multiple names.
1097 * config/aarch64/aarch64.h
1098 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1100 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1102 * input.c (add_file_to_cache_tab): Handle the case where fopen
1105 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1108 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1109 from push operand if code of push isn't PRE_DEC.
1111 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1114 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1115 -mquad-memory-atomic. Update -mquad-memory documentation to say
1116 it is only used for non-atomic loads/stores.
1118 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1119 -mquad-memory or -mquad-memory-atomic switches.
1121 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1122 -mquad-memory-atomic to ISA 2.07 support.
1124 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1125 to separate support of normal quad word memory operations (ldq, stq)
1126 from the atomic quad word memory operations.
1128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1129 support to separate non-atomic quad word operations from atomic
1130 quad word operations. Disable non-atomic quad word operations in
1131 little endian mode so that we don't have to swap words after the
1132 load and before the store.
1133 (quad_load_store_p): Add comment about atomic quad word support.
1134 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1135 options printed with -mdebug=reg.
1137 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1138 -mquad-memory-atomic as the test for whether we have quad word
1139 atomic instructions.
1140 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1141 or -mp8-vector are used, allow byte/half-word atomic operations.
1143 * config/rs6000/sync.md (load_lockedti): Insure that the address
1144 is a proper indexed or indirect address for the lqarx instruction.
1145 On little endian systems, swap the hi/lo registers after the lqarx
1147 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1148 insure the address is valid for the lqarx instruction.
1149 (store_conditionalti): Insure that the address is a proper indexed
1150 or indirect address for the stqcrx. instruction. On little endian
1151 systems, swap the hi/lo registers before doing the stqcrx.
1153 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1154 insure the address is valid for the stqcrx. instruction.
1156 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1157 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1158 type of quad memory support is available.
1160 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1163 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1164 there is a danger of looping.
1166 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1168 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1169 force flag_ira_loop_pressure if set via command line.
1171 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1173 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1174 (ashr_simd): New builtin handling DI mode.
1175 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1176 (aarch64_sshr_simddi): New match pattern.
1177 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1178 (vshrd_n_s64): Likewise.
1179 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1181 2014-01-23 Nick Clifton <nickc@redhat.com>
1183 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1184 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1185 favour of mcu specific scripts.
1186 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1189 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1190 Alex Velenko <Alex.Velenko@arm.com>
1192 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1193 (vaddv_s16): Likewise.
1194 (vaddv_s32): Likewise.
1195 (vaddv_u8): Likewise.
1196 (vaddv_u16): Likewise.
1197 (vaddv_u32): Likewise.
1198 (vaddvq_s8): Likewise.
1199 (vaddvq_s16): Likewise.
1200 (vaddvq_s32): Likewise.
1201 (vaddvq_s64): Likewise.
1202 (vaddvq_u8): Likewise.
1203 (vaddvq_u16): Likewise.
1204 (vaddvq_u32): Likewise.
1205 (vaddvq_u64): Likewise.
1206 (vaddv_f32): Likewise.
1207 (vaddvq_f32): Likewise.
1208 (vaddvq_f64): Likewise.
1209 (vmaxv_f32): Likewise.
1210 (vmaxv_s8): Likewise.
1211 (vmaxv_s16): Likewise.
1212 (vmaxv_s32): Likewise.
1213 (vmaxv_u8): Likewise.
1214 (vmaxv_u16): Likewise.
1215 (vmaxv_u32): Likewise.
1216 (vmaxvq_f32): Likewise.
1217 (vmaxvq_f64): Likewise.
1218 (vmaxvq_s8): Likewise.
1219 (vmaxvq_s16): Likewise.
1220 (vmaxvq_s32): Likewise.
1221 (vmaxvq_u8): Likewise.
1222 (vmaxvq_u16): Likewise.
1223 (vmaxvq_u32): Likewise.
1224 (vmaxnmv_f32): Likewise.
1225 (vmaxnmvq_f32): Likewise.
1226 (vmaxnmvq_f64): Likewise.
1227 (vminv_f32): Likewise.
1228 (vminv_s8): Likewise.
1229 (vminv_s16): Likewise.
1230 (vminv_s32): Likewise.
1231 (vminv_u8): Likewise.
1232 (vminv_u16): Likewise.
1233 (vminv_u32): Likewise.
1234 (vminvq_f32): Likewise.
1235 (vminvq_f64): Likewise.
1236 (vminvq_s8): Likewise.
1237 (vminvq_s16): Likewise.
1238 (vminvq_s32): Likewise.
1239 (vminvq_u8): Likewise.
1240 (vminvq_u16): Likewise.
1241 (vminvq_u32): Likewise.
1242 (vminnmv_f32): Likewise.
1243 (vminnmvq_f32): Likewise.
1244 (vminnmvq_f64): Likewise.
1246 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1248 * config/aarch64/aarch64-simd.md
1249 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1250 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1251 (*aarch64_mul3_elt<mode>): Likewise.
1252 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1253 (*aarch64_mul3_elt_to_64v2df): Likewise.
1254 (*aarch64_mla_elt<mode>): Likewise.
1255 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1256 (*aarch64_mls_elt<mode>): Likewise.
1257 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1258 (*aarch64_fma4_elt<mode>): Likewise.
1259 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1260 (*aarch64_fma4_elt_to_64v2df): Likewise.
1261 (*aarch64_fnma4_elt<mode>): Likewise.
1262 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1263 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1264 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1265 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1266 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1267 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1268 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1269 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1270 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1272 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1274 * config/aarch64/aarch64-simd.md
1275 (aarch64_be_checked_get_lane<mode>): New define_expand.
1276 * config/aarch64/aarch64-simd-builtins.def
1277 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1278 New builtin definition.
1279 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1280 Use new safe be builtin.
1282 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1284 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1286 (aarch64_be_st1<mode>): Likewise.
1287 (aarch_ld1<VALL:mode>): Define_expand modified.
1288 (aarch_st1<VALL:mode>): Likewise.
1289 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1290 (UNSPEC_ST1): Likewise.
1292 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1294 * config/microblaze/microblaze.md: Add trap insn and attribute
1296 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1298 PR preprocessor/58580
1299 * input.h (location_get_source_line): Take an additional line_size
1301 (void diagnostics_file_cache_fini): Declare new function.
1302 * input.c (struct fcache): New type.
1303 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1304 New static constants.
1305 (diagnostic_file_cache_init, total_lines_num)
1306 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1307 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1308 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1309 (get_next_line, read_next_line, goto_next_line, read_line_num):
1310 New static function definitions.
1311 (diagnostic_file_cache_fini): New function.
1312 (location_get_source_line): Take an additional output line_len
1313 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1315 * diagnostic.c (diagnostic_finish): Call
1316 diagnostic_file_cache_fini.
1317 (adjust_line): Take an additional input parameter for the length
1318 of the line, rather than calculating it with strlen.
1319 (diagnostic_show_locus): Adjust the use of
1320 location_get_source_line and adjust_line with respect to their new
1321 signature. While displaying a line now, do not stop at the first
1322 null byte. Rather, display the zero byte as a space and keep
1323 going until we reach the size of the line.
1324 * Makefile.in: Add vec.o to OBJS-libcommon
1326 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1327 Ilya Tocar <ilya.tocar@intel.com>
1329 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1330 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1331 (__builtin_ia32_kmov16): Ditto.
1332 * config/i386/i386.md (UNSPEC_KMOV): New.
1335 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1337 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1338 (_mm512_storeu_si512): Ditto.
1340 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1343 * rtl.h (get_referenced_operands): Declare.
1344 * recog.c (get_referenced_operands): New function.
1345 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1346 operands have been referenced when recording LO_SUM references.
1348 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1350 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1352 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1354 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1355 Enable for generic and recent AMD targets.
1357 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1359 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1360 ARG_SIZE note when adjustment was eliminated.
1362 2014-01-22 Jeff Law <law@redhat.com>
1364 PR tree-optimization/59597
1365 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1366 in file. Accept new argument REGISTERING and use it to modify
1367 dump output appropriately.
1368 (register_jump_thread): Corresponding changes.
1369 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1370 thread paths involving joiner blocks. Add code to dump cancelled
1371 jump threading paths.
1373 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1375 PR rtl-optimization/59477
1376 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1377 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1379 2014-01-22 Tom Tromey <tromey@redhat.com>
1381 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1383 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1385 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1387 PR rtl-optimization/59896
1388 * lra-constraints.c (process_alt_operands): Check unused note for
1389 matched operands of insn with no output reloads.
1391 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1393 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1394 (mips_move_from_gpr_cost): Likewise.
1396 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1398 PR rtl-optimization/59858
1399 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1400 ira_class_hard_regs_num.
1401 (process_alt_operands): Increase reject for dying matched operand.
1403 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1406 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1407 smaller than size, perform several stores or loads and stores
1408 at dst + count - size to store or copy all of size bytes, rather
1409 than just last modesize bytes.
1411 2014-01-20 DJ Delorie <dj@redhat.com>
1413 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1414 that CLOBBERs are REGs before propogating their values.
1416 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1419 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1420 (cgraph_inline_failed_type): New function.
1421 * cgraph.h (DEFCIFCODE): Add type.
1422 (cgraph_inline_failed_type_t): New enum.
1423 (cgraph_inline_failed_type): New prototype.
1424 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1425 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1426 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1427 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1428 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1429 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1430 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1431 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1432 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1433 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1434 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1435 OPTIMIZATION_MISMATCH.
1436 * tree-inline.c (expand_call_inline): Emit errors during
1437 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1439 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1442 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1443 mode attribute in insn output.
1445 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1447 * output.h (output_constant): Delete.
1448 * varasm.c (output_constant): Make private.
1450 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1452 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1454 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1457 * tree.h (fold_builtin_strcat): New prototype.
1458 * builtins.c (fold_builtin_strcat): No longer static. Add len
1459 argument, if non-NULL, don't call c_strlen. Optimize
1460 directly into __builtin_memcpy instead of __builtin_strcpy.
1461 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1462 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1464 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1466 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1467 for SImode_address_operand operands, having only a REG argument.
1469 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1471 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1472 loader name using mbig-endian.
1473 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1475 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1477 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1481 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1483 * config/aarch64/aarch64-protos.h
1484 (aarch64_cannot_change_mode_class_ptr): Declare.
1485 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1486 aarch64_cannot_change_mode_class_ptr): New.
1487 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1488 backend hook aarch64_cannot_change_mode_class.
1490 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1492 * common/config/aarch64/aarch64-common.c
1493 (aarch64_handle_option): Don't handle any option order logic here.
1494 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1495 selected_cpu, warn on architecture version mismatch.
1496 (aarch64_override_options): Fix parsing order for option strings.
1498 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1499 Iain Sandoe <iain@codesourcery.com>
1502 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1503 warning. Amend comment to reflect current functionality.
1505 2014-01-20 Richard Biener <rguenther@suse.de>
1508 * builtins.c (fold_builtin_strcat): Remove case better handled
1509 by tree-ssa-strlen.c.
1511 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1513 * config/aarch64/aarch64.opt
1514 (mcpu, march, mtune): Make case-insensitive.
1516 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1519 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1520 if operands[1] is a REG or ZERO_EXTEND of a REG.
1522 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1524 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1526 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1528 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1529 long non-pic millicode calls.
1531 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1533 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1535 2014-01-19 Kito Cheng <kito@0xlab.org>
1537 * builtins.c (expand_movstr): Check movstr expand done or fail.
1539 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1540 H.J. Lu <hongjiu.lu@intel.com>
1543 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1544 to DImode for zero-extended addresses.
1546 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1548 PR rtl-optimization/57763
1549 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1550 on the new indirect jump_insn and increment LABEL_NUSES (label).
1552 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1556 * config.gcc (x86_archs): New variable.
1557 (x86_64_archs): Likewise.
1558 (x86_cpus): Likewise.
1559 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1560 --with-arch/--with-cpu= options.
1561 Support --with-arch=/--with-cpu={nehalem,westmere,
1562 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1564 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1566 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1567 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1569 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1571 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1573 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1576 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1577 clear cpp_get_options (parse_in)->warn_unused_macros for
1578 ix86_target_macros_internal with cpp_define.
1580 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1582 * jump.c (delete_related_insns): Keep (use (insn))s.
1583 * reorg.c (redundant_insn): Check for barriers too.
1585 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1587 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1589 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1591 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1592 call to $$dyncall when TARGET_LONG_CALLS is true.
1594 2014-01-17 Jeff Law <law@redhat.com>
1596 * ree.c (combine_set_extension): Temporarily disable test for
1597 changing number of hard registers.
1599 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1602 * ipa-inline-analysis.c (inline_free_summary):
1603 Do not free summary of aliases.
1605 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1608 * gimplify.c (gimplify_expr): Use create_tmp_var
1609 instead of create_tmp_var_raw. If cond doesn't have
1610 integral type, don't add the IFN_ANNOTATE builtin at all.
1612 2014-01-17 Martin Jambor <mjambor@suse.cz>
1615 * ipa-cp.c (prev_edge_clone): New variable.
1616 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1617 Also resize prev_edge_clone vector.
1618 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1619 (ipcp_edge_removal_hook): New function.
1620 (ipcp_driver): Register ipcp_edge_removal_hook.
1622 2014-01-17 Andrew Pinski <apinski@cavium.com>
1623 Steve Ellcey <sellcey@mips.com>
1626 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1629 2014-01-17 Jeff Law <law@redhat.com>
1632 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1633 so that pass_ccp runs first.
1635 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1637 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1638 (ix86_adjust_cost): Use !TARGET_XXX.
1639 (do_reorder_for_imul): Likewise.
1640 (swap_top_of_ready_list): Likewise.
1641 (ix86_sched_reorder): Likewise.
1643 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1645 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1646 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1647 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1648 (intel_memset): New. Duplicate slm_memset.
1649 (intel_cost): New. Duplicate slm_cost.
1650 (m_INTEL): New macro.
1651 (processor_target_table): Add "intel".
1652 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1653 with PROCESSOR_INTEL for "intel".
1654 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1655 PROCESSOR_SILVERMONT.
1656 (ix86_issue_rate): Likewise.
1657 (ix86_adjust_cost): Likewise.
1658 (ia32_multipass_dfa_lookahead): Likewise.
1659 (swap_top_of_ready_list): Likewise.
1660 (ix86_sched_reorder): Likewise.
1661 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1662 instead of TARGET_OPT_AGU.
1663 * config/i386/i386.h (TARGET_INTEL): New.
1664 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1665 (processor_type): Add PROCESSOR_INTEL.
1666 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1667 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1669 2014-01-17 Marek Polacek <polacek@redhat.com>
1672 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1675 2014-01-17 Richard Biener <rguenther@suse.de>
1677 PR tree-optimization/46590
1678 * opts.c (default_options_table): Add entries for
1679 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1680 all enabled at -O1 but not for -Og.
1681 * common.opt (fbranch-count-reg): Remove Init(1).
1682 (fmove-loop-invariants): Likewise.
1683 (ftree-pta): Likewise.
1685 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1687 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1688 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1689 decls to at least the GCC 4.8 used alignments.
1692 * tree-nested.c (convert_nonlocal_reference_stmt,
1693 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1694 of GIMPLE_BIND stmts, adjust associated decls.
1696 2014-01-17 Richard Biener <rguenther@suse.de>
1698 PR tree-optimization/46590
1699 * vec.h (vec<>::bseach): New member function implementing
1700 binary search according to C89 bsearch.
1701 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1702 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1703 bitmap pointer again. Make accesses_in_loop a flat array.
1704 (mem_ref_obstack): New global.
1705 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1706 (mark_ref_stored): Likewise.
1707 (ref_indep_loop_p_2): Likewise.
1708 (set_ref_stored_in_loop): New helper function.
1709 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1710 (memref_free): Adjust.
1711 (record_mem_ref_loc): Simplify.
1712 (gather_mem_refs_stmt): Adjust.
1713 (sort_locs_in_loop_postorder_cmp): New function.
1714 (analyze_memory_references): Sort accesses_in_loop after
1715 loop postorder number.
1716 (find_ref_loc_in_loop_cmp): New function.
1717 (for_all_locs_in_loop): Find relevant cluster of locs in
1718 accesses_in_loop and iterate without recursion.
1719 (execute_sm): Avoid uninit warning.
1720 (struct ref_always_accessed): Simplify.
1721 (ref_always_accessed::operator ()): Likewise.
1722 (ref_always_accessed_p): Likewise.
1723 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1724 loop postorder numbers here.
1725 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1728 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1731 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1732 on decls for which assemble_alias has been called.
1734 2014-01-17 Nick Clifton <nickc@redhat.com>
1736 * config/msp430/msp430.opt: (mcpu): New option.
1737 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1738 (msp430_option_override): Parse target_cpu. If the MCU name
1739 matches a generic string, clear target_mcu.
1740 (msp430_attr): Allow numeric interrupt values up to 63.
1741 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1742 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1744 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1746 * config/msp430/msp430.md (popm): Use %J rather than %I.
1747 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1748 (addhi_cy_i): Use immediate_operand for operand 2.
1749 * doc/invoke.texi: Document -mcpu option.
1751 2014-01-17 Richard Biener <rguenther@suse.de>
1753 PR rtl-optimization/38518
1754 * df.h (df_analyze_loop): Declare.
1755 * df-core.c: Include cfgloop.h.
1756 (df_analyze_1): Split out main part of df_analyze.
1757 (df_analyze): Adjust.
1758 (loop_inverted_post_order_compute): New function.
1759 (loop_post_order_compute): Likewise.
1760 (df_analyze_loop): New function avoiding whole-function
1762 * loop-invariant.c (find_defs): Use df_analyze_loop.
1763 (find_invariants): Adjust.
1764 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1766 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1768 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1769 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1771 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1773 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1774 traversal when removing references.
1776 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
1779 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1781 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1784 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1785 pushing a reload for an autoinc when we had previously reloaded an
1786 inner part of the address.
1788 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1790 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1792 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1793 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1794 when not giving up or versioning for alias only because of
1796 (vect_analyze_data_ref_dependences): Set to true.
1797 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1799 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1800 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1804 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1807 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1808 operand 0 predicate for gathers, use a new pseudo as subtarget.
1810 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1813 * lra-constraints.c (process_alt_operands): Add printing debug info.
1814 Check absence of input/output reloads for matched operands too.
1816 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1818 PR rtl-optimization/59835
1819 * ira.c (ira_init_register_move_cost): Increase cost for
1822 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1824 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1826 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1829 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1830 non-register objects. Use gen_(high/low)part more consistently.
1833 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1836 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1837 endian support, remove tests for WORDS_BIG_ENDIAN.
1838 (p8_mfvsrd_3_<mode>): Likewise.
1839 (reload_gpr_from_vsx<mode>): Likewise.
1840 (reload_gpr_from_vsxsf): Likewise.
1841 (p8_mfvsrd_4_disf): Likewise.
1843 2014-01-16 Richard Biener <rguenther@suse.de>
1845 PR rtl-optimization/46590
1846 * lcm.c (compute_antinout_edge): Use postorder iteration.
1847 (compute_laterin): Use inverted postorder iteration.
1849 2014-01-16 Nick Clifton <nickc@redhat.com>
1852 * varasm.c (output_constant): Return the number of bytes actually
1854 (output_constructor_array_range): Update the field size with the
1855 number of bytes emitted by output_constant.
1856 (output_constructor_regular_field): Likewise. Also do not
1857 complain if the total number of bytes emitted is now greater
1858 than the expected fieldpos.
1859 * output.h (output_constant): Update prototype and descriptive comment.
1861 2014-01-16 Marek Polacek <polacek@redhat.com>
1864 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1865 it is error_mark_node.
1867 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1869 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1870 VALID_AVX256_REG_OR_OI_MODE.
1872 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1874 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1875 current procedure should be profiled.
1877 2014-01-15 Andrew Pinski <apinski@cavium.com>
1879 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1880 of moving from/to the STACK_REG register class.
1882 2014-01-15 Richard Henderson <rth@redhat.com>
1885 * reginfo.c (global_regs_decl): Globalize.
1886 * rtl.h (global_regs_decl): Declare.
1887 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1888 reserved via global_regs.
1890 2014-01-15 Teresa Johnson <tejohnson@google.com>
1892 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1894 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1896 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1897 and vmulosh rather than call gen_vec_widen_smult_*.
1898 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1899 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1900 (vec_widen_smult_even_v16qi): Likewise.
1901 (vec_widen_umult_even_v8hi): Likewise.
1902 (vec_widen_smult_even_v8hi): Likewise.
1903 (vec_widen_umult_odd_v16qi): Likewise.
1904 (vec_widen_smult_odd_v16qi): Likewise.
1905 (vec_widen_umult_odd_v8hi): Likewise.
1906 (vec_widen_smult_odd_v8hi): Likewise.
1907 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1908 vmuloub rather than call gen_vec_widen_umult_*.
1909 (vec_widen_umult_lo_v16qi): Likewise.
1910 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1911 vmulosb rather than call gen_vec_widen_smult_*.
1912 (vec_widen_smult_lo_v16qi): Likewise.
1913 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1914 rather than call gen_vec_widen_umult_*.
1915 (vec_widen_umult_lo_v8hi): Likewise.
1916 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1917 rather than call gen_vec_widen_smult_*.
1918 (vec_widen_smult_lo_v8hi): Likewise.
1920 2014-01-15 Jeff Law <law@redhat.com>
1922 PR tree-optimization/59747
1923 * ree.c (find_and_remove_re): Properly handle case where a second
1924 eliminated extension requires widening a copy created for elimination
1925 of a prior extension.
1926 (combine_set_extension): Ensure that the number of hard regs needed
1927 for a destination register does not change when we widen it.
1929 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1931 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1932 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1933 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1934 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1935 (avr-*-rtems*): Likewise.
1936 (bfin*-rtems*): Likewise.
1937 (moxie-*-rtems*): Likewise.
1938 (h8300-*-rtems*): Likewise.
1939 (i[34567]86-*-rtems*): Likewise.
1940 (lm32-*-rtems*): Likewise.
1941 (m32r-*-rtems*): Likewise.
1942 (m68k-*-rtems*): Likewise.
1943 (microblaze*-*-rtems*): Likewise.
1944 (mips*-*-rtems*): Likewise.
1945 (powerpc-*-rtems*): Likewise.
1946 (sh-*-rtems*): Likewise.
1947 (sparc-*-rtems*): Likewise.
1948 (sparc64-*-rtems*): Likewise.
1949 (v850-*-rtems*): Likewise.
1950 (m32c-*-rtems*): Likewise.
1952 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1954 PR rtl-optimization/59511
1955 * ira.c (ira_init_register_move_cost): Use memory costs for some
1956 cases of register move cost calculations.
1957 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1958 instead of BB frequency.
1959 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1960 * lra-assigns.c (find_hard_regno_for): Ditto.
1962 2014-01-15 Richard Biener <rguenther@suse.de>
1964 PR tree-optimization/59822
1965 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1966 (vectorizable_load): Use it to hoist defs of uses of invariant
1967 loads out of the loop.
1969 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1970 Kugan Vivekanandarajah <kuganv@linaro.org>
1973 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1976 2014-01-15 Richard Biener <rguenther@suse.de>
1978 PR rtl-optimization/59802
1979 * lcm.c (compute_available): Use inverted postorder to seed
1980 the initial worklist.
1982 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1985 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1986 ADDR_REGS for invalid symrefs in non-PIC code.
1988 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1991 * builtins.c (determine_block_size): Initialize *probable_max_size
1992 even if len_rtx is CONST_INT.
1994 2014-01-14 Andrew Pinski <apinski@cavium.com>
1996 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1997 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1998 (cortexa53_tunings): Likewise.
1999 (aarch64_sched_issue_rate): New function.
2000 (TARGET_SCHED_ISSUE_RATE): Define.
2002 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2004 * ira-costs.c (find_costs_and_classes): Add missed
2005 ira_init_register_move_cost_if_necessary.
2007 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2010 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2012 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2015 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2016 to indicate if type is used for function return value. Warn ABI
2017 change if the vector mode isn't available for function return value.
2018 (ix86_function_arg_advance): Pass false to type_natural_mode.
2019 (ix86_function_arg): Likewise.
2020 (ix86_gimplify_va_arg): Likewise.
2021 (function_arg_32): Don't warn ABI change.
2022 (ix86_function_value): Pass true to type_natural_mode.
2023 (ix86_return_in_memory): Likewise.
2024 (ix86_struct_value_rtx): Removed.
2025 (TARGET_STRUCT_VALUE_RTX): Likewise.
2027 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2029 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2030 converting a conditional jump into a conditional return.
2032 2014-01-14 Richard Biener <rguenther@suse.de>
2034 PR tree-optimization/58921
2035 PR tree-optimization/59006
2036 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2037 hoisting invariant stmts.
2038 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2039 invariant loads on the preheader edge if possible.
2041 2014-01-14 Joey Ye <joey.ye@arm.com>
2043 * doc/plugin.texi (Building GCC plugins): Update to C++.
2045 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2047 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2048 (_mm_rcp28_round_ss): Ditto.
2049 (_mm_rsqrt28_round_sd): Ditto.
2050 (_mm_rsqrt28_round_ss): Ditto.
2051 (_mm_rcp28_sd): Ditto.
2052 (_mm_rcp28_ss): Ditto.
2053 (_mm_rsqrt28_sd): Ditto.
2054 (_mm_rsqrt28_ss): Ditto.
2055 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2056 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2057 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2058 (IX86_BUILTIN_RCP28SD): Ditto.
2059 (IX86_BUILTIN_RCP28SS): Ditto.
2060 (IX86_BUILTIN_RSQRT28SD): Ditto.
2061 (IX86_BUILTIN_RSQRT28SS): Ditto.
2062 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2063 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2064 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2065 (ix86_expand_special_args_builtin): Expand new FTYPE.
2066 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2067 (srcp14<mode>): Make insn unary.
2068 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2069 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2070 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2071 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2072 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2073 Fix rounding: make it SAE only.
2074 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2076 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2078 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2079 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2080 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2081 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2082 (round_saeonly_mask_scalar_operand4): Ditto.
2083 (round_saeonly_mask_scalar_op3): Ditto.
2084 (round_saeonly_mask_scalar_op4): Ditto.
2086 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2089 Implement -maltivec=be for vec_insert and vec_extract.
2091 2014-01-10 DJ Delorie <dj@redhat.com>
2093 * config/msp430/msp430.md (call_internal): Don't allow memory
2094 references with SP as the base register.
2095 (call_value_internal): Likewise.
2096 * config/msp430/constraints.md (Yc): New. For memory references
2097 that don't use SP as a base register.
2099 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2100 "an integer without a # prefix"
2101 * config/msp430/msp430.md (epilogue_helper): Use it.
2103 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2106 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2107 AVX512F gather builtins.
2108 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2109 on gather decls with INTEGER_TYPE masktype.
2110 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2111 directly into the builtin rather than hoisting it before loop.
2113 PR tree-optimization/59387
2114 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2115 (scev_const_prop): If folded_casts and type has undefined overflow,
2116 use force_gimple_operand instead of force_gimple_operand_gsi and
2117 for each added stmt if it is assign with
2118 arith_code_with_undefined_signed_overflow, call
2119 rewrite_to_defined_overflow.
2120 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2121 gimple-fold.h instead.
2122 (arith_code_with_undefined_signed_overflow,
2123 rewrite_to_defined_overflow): Moved to ...
2124 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2125 rewrite_to_defined_overflow): ... here. No longer static.
2126 Include gimplify-me.h.
2127 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2128 rewrite_to_defined_overflow): New prototypes.
2130 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2132 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2134 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2136 * builtins.c (get_object_alignment_2): Minor tweak.
2137 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2139 2014-01-13 Christian Bruel <christian.bruel@st.com>
2141 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2142 optimized non constant lengths.
2144 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2147 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2148 load as __atomic_load_N if possible.
2150 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2152 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2154 (rs6000_expand_builtin): Adjust call.
2156 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2159 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2160 * config/rs6000/rs6000.c: Include target-globals.h.
2161 (rs6000_set_current_function): Instead of doing target_reinit
2162 unconditionally, use save_target_globals_default_opts and
2163 restore_target_globals.
2165 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2167 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2168 (rs6000_expand_builtin): Handle mffs and mtfsf.
2169 (rs6000_init_builtins): Define mffs and mtfsf.
2170 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2171 (rs6000_mffs): New pattern.
2172 (rs6000_mtfsf): New pattern.
2174 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2176 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2177 Start narrowing with START. Apply candidate-use pair
2178 and check overall cost in narrowing.
2179 (iv_ca_prune): Pass new argument.
2181 2014-01-10 Jeff Law <law@redhat.com>
2184 * ree.c (combine_reaching_defs): Ensure the defining statement
2185 occurs before the extension when optimizing extensions with
2186 different source and destination hard registers.
2188 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2191 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2192 vtables into the type inheritance graph.
2194 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2196 PR rtl-optimization/59754
2197 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2198 modes in the REGNO != REGNO case.
2200 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2202 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2204 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2206 PR tree-optimization/59745
2207 * tree-predcom.c (tree_predictive_commoning_loop): Call
2208 free_affine_expand_cache if giving up because components is NULL.
2210 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2211 GC in payload of target_globals struct instead of allocating them on
2212 the heap and the larger structs separately using GC.
2213 * target-globals.h (struct target_globals): Make regs, hard_regs,
2214 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2215 of GTY((skip)) and change type to void *.
2216 (reset_target_globals): Cast loads from those fields to corresponding
2219 2014-01-10 Steve Ellcey <sellcey@mips.com>
2222 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2223 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2224 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2226 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2229 * aarch64-modes.def (CC_Zmode): New flags mode.
2230 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2231 represents an equality.
2232 (aarch64_get_condition_code): Handle CC_Zmode.
2233 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2235 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2237 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2238 extraction in good case.
2240 2014-01-10 Richard Biener <rguenther@suse.de>
2242 PR tree-optimization/59374
2243 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2244 checking after SLP discovery. Mark stmts not participating
2245 in any SLP instance properly.
2247 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2249 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2250 when handling a SET rtx.
2252 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2254 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2255 (cortex-a57): Likewise.
2256 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2258 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2260 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2261 non-iwmmxt builtins.
2263 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2267 * ipa-devirt.c record_target_from_binfo): Take as argument
2268 stack of binfos and lookup matching one for virtual inheritance.
2269 (possible_polymorphic_call_targets_1): Update.
2271 2014-01-10 Huacai Chen <chenhc@lemote.com>
2273 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2274 kernel strings for Loongson-2E/2F/3A.
2276 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2279 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2280 is_gimple_call before calling gimple_call_internal_p.
2282 2014-01-09 Steve Ellcey <sellcey@mips.com>
2284 * Makefile.in (TREE_FLOW_H): Remove.
2285 (TREE_SSA_H): Add file names from tree-flow.h.
2286 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2287 * tree.h: Remove tree-flow.h reference.
2288 * hash-table.h: Remove tree-flow.h reference.
2289 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2290 reference with tree-ssa-loop.h.
2292 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2294 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2295 default element-order behavior for -maltivec.
2296 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2297 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2298 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2299 when targeting big endian, at least for now.
2300 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2302 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2305 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2306 var satisfies use_register_for_decl, just take into account type
2307 alignment, rather than decl alignment.
2309 PR tree-optimization/59622
2310 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2311 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2312 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2313 Don't devirtualize for inplace at all. For targets.length () == 1,
2314 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2316 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2318 * config/i386/i386.md (cpu): Remove the unused btver1.
2320 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2322 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2324 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2327 * tree-core.h (struct target_globals): New forward declaration.
2328 (struct tree_target_option): Add globals field.
2329 * tree.h (TREE_TARGET_GLOBALS): Define.
2330 (prepare_target_option_nodes_for_pch): New prototype.
2331 * target-globals.h (struct target_globals): Define even if
2333 * tree.c (prepare_target_option_node_for_pch,
2334 prepare_target_option_nodes_for_pch): New functions.
2335 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2336 * config/i386/i386.c: Include target-globals.h.
2337 (ix86_set_current_function): Instead of doing target_reinit
2338 unconditionally, use save_target_globals_default_opts and
2339 restore_target_globals.
2341 2014-01-09 Richard Biener <rguenther@suse.de>
2343 PR tree-optimization/59715
2344 * tree-cfg.h (split_critical_edges): Declare.
2345 * tree-cfg.c (split_critical_edges): Export.
2346 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2348 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2350 * cfgexpand.c (expand_stack_vars): Optionally disable
2351 asan stack protection.
2352 (expand_used_vars): Likewise.
2353 (partition_stack_vars): Likewise.
2354 * asan.c (asan_emit_stack_protection): Optionally disable
2355 after return stack usage.
2356 (instrument_derefs): Optionally disable memory access instrumentation.
2357 (instrument_builtin_call): Likewise.
2358 (instrument_strlen_call): Likewise.
2359 (asan_protect_global): Optionally disable global variables protection.
2360 * doc/invoke.texi: Added doc for new options.
2361 * params.def: Added new options.
2362 * params.h: Likewise.
2364 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2366 PR rtl-optimization/59724
2367 * ifcvt.c (cond_exec_process_if_block): Don't call
2368 flow_find_head_matching_sequence with 0 longest_match.
2369 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2370 non-active insns if !stop_after.
2371 (try_head_merge_bb): Revert 2014-01-07 changes.
2373 2014-01-08 Jeff Law <law@redhat.com>
2375 * ree.c (get_sub_rtx): New function, extracted from...
2376 (merge_def_and_ext): Here.
2377 (combine_reaching_defs): Use get_sub_rtx.
2379 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2381 * cgraph.h (varpool_variable_node): Do not choke on null node.
2383 2014-01-08 Catherine Moore <clm@codesourcery.com>
2385 * config/mips/mips.md (simple_return): Attempt to use JRC
2387 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2389 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2391 PR rtl-optimization/59137
2392 * reorg.c (steal_delay_list_from_target): Call update_block for
2394 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2396 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2398 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2399 two duplicate entries.
2401 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2404 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2406 * config/mips/mips.c (mips_truncated_op_cost): New function.
2407 (mips_rtx_costs): Adjust test for BADDU.
2408 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2410 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2412 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2413 (*baddu_si): ...this new pattern.
2415 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2418 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2420 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2423 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2425 (expand_expr, expand_normal): Adjust.
2426 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2427 inner_reference_p. Use inner_reference_p to expand inner references.
2428 (store_expr): Adjust.
2429 * cfgexpand.c (expand_call_stmt): Adjust.
2431 2014-01-08 Rong Xu <xur@google.com>
2433 * gcov-io.c (gcov_var): Move from gcov-io.h.
2434 (gcov_position): Ditto.
2435 (gcov_is_error): Ditto.
2436 (gcov_rewrite): Ditto.
2437 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2438 only part to libgcc/libgcov.h.
2440 2014-01-08 Marek Polacek <polacek@redhat.com>
2443 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2445 2014-01-08 Marek Polacek <polacek@redhat.com>
2448 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2450 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2452 PR rtl-optimization/59649
2453 * stor-layout.c (get_mode_bounds): For BImode return
2454 0 and STORE_FLAG_VALUE.
2456 2014-01-08 Richard Biener <rguenther@suse.de>
2459 * gimple.h (is_gimple_builtin_call): Remove.
2460 (gimple_builtin_call_types_compatible_p): New.
2461 (gimple_call_builtin_p): New overload.
2462 * gimple.c (is_gimple_builtin_call): Remove.
2463 (validate_call): Rename to ...
2464 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2466 (validate_type): New static function.
2467 (gimple_call_builtin_p): New overload and adjust.
2468 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2469 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2470 (gimple_fold_stmt_to_constant_1): Likewise.
2471 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2473 2014-01-08 Richard Biener <rguenther@suse.de>
2476 * gimplify.c (gimplify_expr): Gimplify register-register type
2477 VIEW_CONVERT_EXPRs to separate stmts.
2479 2014-01-07 Jeff Law <law@redhat.com>
2482 * ree.c (combine_set_extension): Handle case where source
2483 and destination registers in an extension insn are different.
2484 (combine_reaching_defs): Allow source and destination registers
2485 in extension to be different under limited circumstances.
2486 (add_removable_extension): Remove restriction that the
2487 source and destination registers in the extension are the same.
2488 (find_and_remove_re): Emit a copy from the extension's
2489 destination to its source after the defining insn if
2490 the source and destination registers are different.
2493 * ifcvt.c (merge_if_block): If we are merging a block with more than
2494 one successor with a block with no successors, remove any BARRIER
2495 after the second block.
2497 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2499 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2501 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2504 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2505 for 14-bit register offsets when INT14_OK_STRICT is false.
2507 2014-01-07 Roland Stigge <stigge@antcom.de>
2508 Michael Meissner <meissner@linux.vnet.ibm.com>
2511 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2512 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2514 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2516 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2519 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2521 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2522 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2523 rtx is const0_rtx or not.
2525 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2528 * target-globals.c (save_target_globals): Remove this_fn_optab
2530 * toplev.c: Include optabs.h.
2531 (target_reinit): Temporarily restore the global options if another
2532 set of options are in force.
2534 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2536 PR rtl-optimization/58668
2537 * cfgcleanup.c (flow_find_cross_jump): Don't count
2538 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2539 to determine what is counted.
2540 (flow_find_head_matching_sequence): Use active_insn_p to determine
2542 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2544 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2545 determine what is counted.
2547 PR tree-optimization/59643
2548 * tree-predcom.c (split_data_refs_to_components): If one dr is
2549 read and one write, determine_offset fails and the write isn't
2550 in the bad component, just put the read into the bad component.
2552 2014-01-07 Mike Stump <mikestump@comcast.net>
2553 Jakub Jelinek <jakub@redhat.com>
2556 * tree-core.h (struct tree_optimization_option): Change optabs
2557 type from unsigned char * to void *.
2558 * optabs.c (init_tree_optimization_optabs): Adjust
2559 TREE_OPTIMIZATION_OPTABS initialization.
2561 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2564 * config/i386/i386.h (struct machine_function): Add
2565 no_drap_save_restore field.
2566 * config/i386/i386.c (ix86_save_reg): Use
2567 !cfun->machine->no_drap_save_restore instead of
2568 crtl->stack_realign_needed.
2569 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2570 this function clears frame_pointer_needed. Set
2571 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2572 and DRAP reg is needed.
2574 2014-01-06 Marek Polacek <polacek@redhat.com>
2577 * doc/implement-c.texi: Mention that other integer types are
2578 permitted as bit-field types in strictly conforming mode.
2580 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2582 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2585 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2587 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2589 2014-01-06 Martin Jambor <mjambor@suse.cz>
2592 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2594 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2596 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2600 * var-tracking.c (add_stores): Preserve the value of the source even if
2601 we don't record the store.
2603 2014-01-06 Terry Guo <terry.guo@arm.com>
2605 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2607 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2610 * config/darwin.c (darwin_function_section): Adjust return values to
2611 correspond to optimisation changes made in r206070.
2613 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2615 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2616 from prefetch_block tune setting.
2617 (nocona_cost): Correct size of prefetch block to 64.
2619 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2621 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2622 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2623 used to save the static chain register in the computation of the offset
2624 from which the FP registers need to be restored.
2626 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2628 PR tree-optimization/59519
2629 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2630 ICE if get_current_def (current_new_name) is already non-NULL, as long
2631 as it is a phi result of some other phi in *new_exit_bb that has
2634 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2635 or vmovdqu* for misaligned_operand.
2636 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2637 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2638 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2639 aligned_mem for AVX512F masked aligned load and store builtins and for
2642 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2644 PR tree-optimization/59651
2645 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2646 Address range for negative step should be added by TYPE_SIZE_UNIT.
2648 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2650 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2651 overlapping registers also for registers other than the stack pointer.
2653 2014-01-03 Marek Polacek <polacek@redhat.com>
2656 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2659 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2662 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2665 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2666 (push splitter): Use <P:MODE_SIZE> instead of
2667 GET_MODE_SIZE (<P:MODE>mode).
2668 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2669 (mov -1, reg peephole2): Likewise.
2670 * config/i386/sse.md (*mov<mode>_internal,
2671 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2672 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2673 *<code><mode>3, *andnot<mode>3<mask_name>,
2674 <mask_codefor><code><mode>3<mask_name>): Likewise.
2675 * config/i386/subst.md (mask_mode512bit_condition,
2676 sd_mask_mode512bit_condition): Likewise.
2678 2014-01-02 Xinliang David Li <davidxl@google.com>
2680 PR tree-optimization/59303
2681 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2682 (dump_predicates): Better output format.
2683 (pred_equal_p): New function.
2684 (is_neq_relop_p): Ditto.
2685 (is_neq_zero_form_p): Ditto.
2686 (pred_expr_equal_p): Ditto.
2687 (pred_neg_p): Ditto.
2688 (simplify_pred): Ditto.
2689 (simplify_preds_2): Ditto.
2690 (simplify_preds_3): Ditto.
2691 (simplify_preds_4): Ditto.
2692 (simplify_preds): Ditto.
2694 (push_to_worklist): Ditto.
2695 (get_pred_info_from_cmp): Ditto.
2696 (is_degenerated_phi): Ditto.
2697 (normalize_one_pred_1): Ditto.
2698 (normalize_one_pred): Ditto.
2699 (normalize_one_pred_chain): Ditto.
2700 (normalize_preds): Ditto.
2701 (normalize_cond_1): Remove function.
2702 (normalize_cond): Ditto.
2703 (is_gcond_subset_of): Ditto.
2704 (is_subset_of_any): Ditto.
2705 (is_or_set_subset_of): Ditto.
2706 (is_and_set_subset_of): Ditto.
2707 (is_norm_cond_subset_of): Ditto.
2708 (pred_chain_length_cmp): Ditto.
2709 (convert_control_dep_chain_into_preds): Type change.
2710 (find_predicates): Ditto.
2711 (find_def_preds): Ditto.
2712 (destroy_predicates_vecs): Ditto.
2713 (find_matching_predicates_in_rest_chains): Ditto.
2714 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2715 (is_pred_expr_subset): Ditto.
2716 (is_pred_chain_subset_of): Ditto.
2717 (is_included_in): Ditto.
2718 (is_superset_of): Ditto.
2720 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2722 Update copyright years.
2724 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2726 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2727 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2728 config/arc/arc.md, config/arc/arc.opt,
2729 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2730 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2731 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2732 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2733 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2734 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2735 vtable-verify.c, vtable-verify.h: Use the standard form for the
2738 2014-01-02 Tobias Burnus <burnus@net-b.de>
2740 * gcc.c (process_command): Update copyright notice dates.
2741 * gcov-dump.c: Ditto.
2743 * doc/cpp.texi: Bump @copying's copyright year.
2744 * doc/cppinternals.texi: Ditto.
2745 * doc/gcc.texi: Ditto.
2746 * doc/gccint.texi: Ditto.
2747 * doc/gcov.texi: Ditto.
2748 * doc/install.texi: Ditto.
2749 * doc/invoke.texi: Ditto.
2751 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2753 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2755 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2757 * config/i386/sse.md (*mov<mode>_internal): Guard
2758 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2760 PR rtl-optimization/59647
2761 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2762 new_rtx into UNSIGNED_FLOAT rtxes.
2764 Copyright (C) 2014 Free Software Foundation, Inc.
2766 Copying and distribution of this file, with or without modification,
2767 are permitted in any medium without royalty provided the copyright
2768 notice and this notice are preserved.