1 2014-02-02 Jan Hubicka <jh@suse.cz>
3 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
4 vtable_pointer_value_to_binfo): New functions.
5 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
6 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
8 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
10 * config/nios2/nios2.md (load_got_register): Initialize GOT
11 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
12 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
14 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
16 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
17 preserverd by passthrough, do not propagate the type.
19 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
21 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
22 (mips_atomic_assign_expand_fenv): New function.
23 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
25 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
27 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
28 (__builtin_mips_set_fcsr): Likewise.
29 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
31 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
32 (mips16_expand_set_fcsr): Likewise.
33 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
34 (mips16_set_fcsr_stub): Likewise.
35 (mips16_get_fcsr_one_only_stub): New class.
36 (mips16_set_fcsr_one_only_stub): Likewise.
37 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
38 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
39 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
40 (hard_float): New availability predicate.
41 (mips_builtins): Add get_fcsr and set_fcsr.
42 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
43 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
44 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
45 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
46 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
49 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
51 * config/mips/mips.c (mips_one_only_stub): New class.
52 (mips_need_mips16_rdhwr_p): Replace with...
53 (mips16_rdhwr_stub): ...this new variable.
54 (mips16_stub_call_address): New function.
55 (mips16_rdhwr_one_only_stub): New class.
56 (mips_expand_thread_pointer): Use mips16_stub_call_address.
57 (mips_output_mips16_rdhwr): Delete.
58 (mips_finish_stub): New function.
59 (mips_code_end): Use it to handle rdhwr stubs.
61 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
64 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
65 when calculating size of integer atomic types.
67 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
69 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
71 2014-02-01 Jakub Jelinek <jakub@redhat.com>
73 PR tree-optimization/60003
74 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
75 * profile.c (branch_prob): Use gimple_call_builtin_p
76 to check for BUILT_IN_SETJMP_RECEIVER.
77 * tree-inline.c (copy_bb): Call notice_special_calls.
79 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
82 * lra-constraints.c (process_alt_operands): Update reload_sum only
85 2014-01-31 Richard Henderson <rth@redhat.com>
88 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
89 until after else_eh is processed.
91 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
93 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
94 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
95 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
96 in smmintrin.h, remove them.
97 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
98 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
99 * config/i386/i386.md (ROUND_SAE): Fix value.
100 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
101 (const48_operand): New.
102 * config/i386/subst.md (round), (round_expand): Use
103 const_4_or_8_to_11_operand.
104 (round_saeonly), (round_saeonly_expand): Use const48_operand.
106 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
108 * config/i386/constraints.md (Yk): Swap meaning with k.
109 * config/i386/i386.md (movhi_internal): Change Yk to k.
110 (movqi_internal): Ditto.
111 (*k<logic><mode>): Ditto.
114 (kandn<mode>): Ditto.
115 (*<code>hi_1): Ditto.
116 (*<code>qi_1): Ditto.
117 (kxnor<mode>): Ditto.
121 (*one_cmplhi2_1): Ditto.
122 (*one_cmplqi2_1): Ditto.
123 * config/i386/sse.md (): Change k to Yk.
124 (avx512f_load<mode>_mask): Ditto.
125 (avx512f_blendm<mode>): Ditto.
126 (avx512f_store<mode>_mask): Ditto.
127 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
128 (avx512f_storedqu<mode>_mask): Ditto.
129 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
131 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
132 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
133 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
134 (avx512f_maskcmp<mode>3): Ditto.
135 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
136 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
137 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
138 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
139 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
140 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
141 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
142 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
143 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
144 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
145 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
146 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
147 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
148 (vec_extract_lo_<mode>_maskm): Ditto.
149 (vec_extract_hi_<mode>_maskm): Ditto.
150 (avx512f_vternlog<mode>_mask): Ditto.
151 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
152 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
153 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
154 (avx512f_<code>v8div16qi2_mask): Ditto.
155 (avx512f_<code>v8div16qi2_mask_store): Ditto.
156 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
157 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
158 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
159 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
160 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
161 (*avx512pf_gatherpf<mode>df_mask): Ditto.
162 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
163 (*avx512pf_scatterpf<mode>df_mask): Ditto.
164 (avx512cd_maskb_vec_dupv8di): Ditto.
165 (avx512cd_maskw_vec_dupv16si): Ditto.
166 (avx512f_vpermi2var<mode>3_maskz): Ditto.
167 (avx512f_vpermi2var<mode>3_mask): Ditto.
168 (avx512f_vpermi2var<mode>3_mask): Ditto.
169 (avx512f_vpermt2var<mode>3_maskz): Ditto.
170 (*avx512f_gathersi<mode>): Ditto.
171 (*avx512f_gathersi<mode>_2): Ditto.
172 (*avx512f_gatherdi<mode>): Ditto.
173 (*avx512f_gatherdi<mode>_2): Ditto.
174 (*avx512f_scattersi<mode>): Ditto.
175 (*avx512f_scatterdi<mode>): Ditto.
176 (avx512f_compress<mode>_mask): Ditto.
177 (avx512f_compressstore<mode>_mask): Ditto.
178 (avx512f_expand<mode>_mask): Ditto.
179 * config/i386/subst.md (mask): Change k to Yk.
180 (mask_scalar_merge): Ditto.
183 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
185 * doc/extend.texi (Vector Extensions): Document ?: in C++.
187 2014-01-31 Richard Biener <rguenther@suse.de>
190 * builtins.c (fold_builtin_memory_op): Make sure to not
191 use a floating-point mode or a boolean or enumeral type for
194 2014-01-30 DJ Delorie <dj@redhat.com>
196 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
197 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
198 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
199 whenever main() has an epilogue.
201 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
203 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
204 unused variable "field".
205 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
206 (vsx_mergeh_<mode>): Likewise.
207 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
208 (altivec_vmrghh): Likewise.
209 (altivec_vmrghw): Likewise.
210 (altivec_vmrglb): Likewise.
211 (altivec_vmrglh): Likewise.
212 (altivec_vmrglw): Likewise.
213 (altivec_vspltb): Add missing uses.
214 (altivec_vsplth): Likewise.
215 (altivec_vspltw): Likewise.
216 (altivec_vspltsf): Likewise.
218 2014-01-30 Jakub Jelinek <jakub@redhat.com>
221 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
222 frame related instructions.
224 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
226 PR rtl-optimization/59959
227 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
228 any reload of register whose subreg is invalid.
230 2014-01-30 Jakub Jelinek <jakub@redhat.com>
232 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
233 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
234 Add missing return type - void.
236 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
238 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
239 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
240 remove element index adjustment for endian (now handled in vsx.md
242 (altivec_expand_vec_perm_const): Use
243 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
244 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
245 (vsx_xxspltw_<mode>): Adjust element index for little endian.
246 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
247 define_expand and a new define_insn *altivec_vspltb_internal;
248 adjust for -maltivec=be on a little endian target.
249 (altivec_vspltb_direct): New.
250 (altivec_vsplth): Divide into a define_expand and a new
251 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
252 little endian target.
253 (altivec_vsplth_direct): New.
254 (altivec_vspltw): Divide into a define_expand and a new
255 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
256 little endian target.
257 (altivec_vspltw_direct): New.
258 (altivec_vspltsf): Divide into a define_expand and a new
259 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
260 a little endian target.
262 2014-01-30 Richard Biener <rguenther@suse.de>
264 PR tree-optimization/59993
265 * tree-ssa-forwprop.c (associate_pointerplus): Check we
266 can propagate form the earlier stmt and avoid the transform
267 when the intermediate result is needed.
269 2014-01-30 Alangi Derick <alangiderick@gmail.com>
271 * README.Portability: Fix typo.
273 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
275 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
276 comparison_operator with ordered_comparison_operator.
278 2014-01-30 Nick Clifton <nickc@redhat.com>
280 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
281 Rename to mn10300_store_multiple_regs.
282 * config/mn10300/mn10300.c: Likewise.
283 * config/mn10300/mn10300.md (store_movm): Fix typo: call
285 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
286 Call mn10300_store_multiple_regs.
288 2014-01-30 Nick Clifton <nickc@redhat.com>
289 DJ Delorie <dj@redhat.com>
291 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
292 %fp 2 to keep registers after it properly word-aligned.
293 (rl78_alloc_physical_registers_umul): Handle the case where both
294 input operands are the same.
296 2014-01-30 Richard Biener <rguenther@suse.de>
298 PR tree-optimization/59903
299 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
302 2014-01-30 Jason Merrill <jason@redhat.com>
305 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
308 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
310 2014-01-30 Richard Biener <rguenther@suse.de>
312 PR tree-optimization/59951
313 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
315 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
318 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
319 SFmode to DFmode case.
321 2014-01-29 DJ Delorie <dj@redhat.com>
323 * config/msp430/msp430.opt (-minrt): New.
324 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
326 (ENDFILE_SPEC): Likewise.
328 2014-01-29 Jan Hubicka <jh@suse.cz>
330 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
331 (estimate_function_body_sizes): Use it.
333 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
336 * dwarf2out.c (is_cxx_auto): New.
337 (is_base_type): Use it.
338 (gen_type_die_with_usage): Likewise.
340 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
342 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
343 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
344 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
345 -maltivec=be with LE targets.
346 (vsx_mergeh_<mode>): Likewise.
347 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
348 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
349 (altivec_vmrghb): Replace with define_expand and new
350 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
351 (altivec_vmrghb_direct): New define_insn.
352 (altivec_vmrghh): Replace with define_expand and new
353 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
354 (altivec_vmrghh_direct): New define_insn.
355 (altivec_vmrghw): Replace with define_expand and new
356 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
357 (altivec_vmrghw_direct): New define_insn.
358 (*altivec_vmrghsf): Adjust for endianness.
359 (altivec_vmrglb): Replace with define_expand and new
360 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
361 (altivec_vmrglb_direct): New define_insn.
362 (altivec_vmrglh): Replace with define_expand and new
363 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
364 (altivec_vmrglh_direct): New define_insn.
365 (altivec_vmrglw): Replace with define_expand and new
366 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
367 (altivec_vmrglw_direct): New define_insn.
368 (*altivec_vmrglsf): Adjust for endianness.
369 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
370 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
371 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
372 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
373 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
374 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
375 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
376 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
378 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
380 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
381 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
384 2014-01-29 Richard Biener <rguenther@suse.de>
386 PR tree-optimization/58742
387 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
388 associate_pointerplus_align.
389 (associate_pointerplus_diff): New function.
390 (associate_pointerplus): Likewise. Call associate_pointerplus_align
391 and associate_pointerplus_diff.
393 2014-01-29 Richard Biener <rguenther@suse.de>
395 * lto-streamer.h (LTO_major_version): Bump to 3.
396 (LTO_minor_version): Reset to 0.
398 2014-01-29 Renlin Li <Renlin.Li@arm.com>
400 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
401 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
402 (arm_file_start): Generate correct asm header for armv7ve.
403 * config/arm/bpabi.h: Add multilib support for armv7ve.
404 * config/arm/driver-arm.c: Change the architectures of cortex-a7
405 and cortex-a15 to armv7ve.
406 * config/arm/t-aprofile: Add multilib support for armv7ve.
407 * doc/invoke.texi: Document -march=armv7ve.
409 2014-01-29 Richard Biener <rguenther@suse.de>
411 PR tree-optimization/58742
412 * tree-ssa-forwprop.c (associate_plusminus): Return true
413 if we changed sth, defer EH cleanup to ...
414 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
415 (simplify_mult): New function.
417 2014-01-29 Jakub Jelinek <jakub@redhat.com>
420 PR tree-optimization/59920
421 * tree.c (build_common_builtin_nodes): Remove
422 __builtin_setjmp_dispatcher initialization.
423 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
424 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
425 instead of gsi_after_labels + manually skipping debug stmts.
426 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
427 ignore bbs with IFN_ABNORMAL_DISPATCHER.
428 * tree-inline.c (copy_edges_for_bb): Remove
429 can_make_abnormal_goto argument, instead add abnormal_goto_dest
430 argument. Ignore computed_goto_p stmts. Don't call
431 make_abnormal_goto_edges. If a call might need abnormal edges
432 for non-local gotos, see if it already has an edge to
433 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
434 with true argument, don't do anything then, otherwise add
435 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
436 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
438 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
439 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
440 (lower_stmt): Don't set data->calls_builtin_setjmp.
441 (lower_builtin_setjmp): Adjust comment.
442 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
443 * tree-cfg.c (found_computed_goto): Remove.
444 (factor_computed_gotos): Remove.
445 (make_goto_expr_edges): Return bool, true for computed gotos.
446 Don't call make_abnormal_goto_edges.
447 (build_gimple_cfg): Don't set found_computed_goto, don't call
448 factor_computed_gotos.
449 (computed_goto_p): No longer static.
450 (make_blocks): Don't set found_computed_goto.
451 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
452 (make_edges): If make_goto_expr_edges returns true, push bb
453 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
454 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
455 vector. Record mapping between bbs and OpenMP regions if there
456 are any, adjust make_gimple_omp_edges caller. Call
457 handle_abnormal_edges.
458 (make_abnormal_goto_edges): Remove.
459 * tree-cfg.h (make_abnormal_goto_edges): Remove.
460 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
461 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
462 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
463 * internal-fn.def (ABNORMAL_DISPATCHER): New.
464 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
465 filling *region also set *region_idx to (*region)->entry->index.
468 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
469 For REGs set ORIGINAL_REGNO.
471 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
473 * doc/md.texi: Mention that a target shouldn't implement
474 vec_widen_(s|u)mul_even/odd pair if it is less efficient
477 2014-01-29 Jakub Jelinek <jakub@redhat.com>
479 PR tree-optimization/59594
480 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
481 a copy of the datarefs vector rather than the vector itself.
483 2014-01-28 Jason Merrill <jason@redhat.com>
486 * dwarf2out.c (auto_die): New static.
487 (gen_type_die_with_usage): Handle C++1y 'auto'.
488 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
491 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
494 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
495 (SPEC_X32): Likewise.
497 * config/i386/i386.c (ix86_option_override_internal): Turn off
498 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
500 (x86_file_start): Output .code16gcc for TARGET_16BIT.
501 * config/i386/i386.h (TARGET_16BIT): New macro.
502 (TARGET_16BIT_P): Likewise.
503 * config/i386/i386.opt: Add m16.
504 * doc/invoke.texi: Document -m16.
506 2014-01-28 Jakub Jelinek <jakub@redhat.com>
508 PR preprocessor/59935
509 * input.c (location_get_source_line): Bail out on when line number
510 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
512 2014-01-28 Richard Biener <rguenther@suse.de>
514 PR tree-optimization/58742
515 * tree-ssa-forwprop.c (associate_plusminus): Handle
516 pointer subtraction of the form (T)(P + A) - (T)P.
518 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
520 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
523 2014-01-28 Richard Biener <rguenther@suse.de>
526 2014-01-28 Richard Biener <rguenther@suse.de>
528 PR rtl-optimization/45364
529 PR rtl-optimization/59890
530 * var-tracking.c (local_get_addr_clear_given_value): Handle
531 already cleared slot.
532 (val_reset): Handle not allocated local_get_addr_cache.
533 (vt_find_locations): Use post-order on the inverted CFG.
535 2014-01-28 Richard Biener <rguenther@suse.de>
537 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
539 2014-01-28 Richard Biener <rguenther@suse.de>
541 PR rtl-optimization/45364
542 PR rtl-optimization/59890
543 * var-tracking.c (local_get_addr_clear_given_value): Handle
544 already cleared slot.
545 (val_reset): Handle not allocated local_get_addr_cache.
546 (vt_find_locations): Use post-order on the inverted CFG.
548 2014-01-28 Alan Modra <amodra@gmail.com>
550 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
551 * configure.ac <recursive call for build != host>: Define
552 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
553 and LD_FOR_BUILD too.
554 * configure: Regenerate.
556 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
558 * config/i386/i386.c (get_builtin_code_for_version): Separate
559 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
560 Broadwell from Haswell.
562 2014-01-27 Steve Ellcey <sellcey@mips.com>
564 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
565 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
566 * config/mips/mips.c (mips_option_override): Change setting
568 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
569 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
570 Change from Mask to Var.
572 2014-01-27 Jeff Law <law@redhat.com>
574 * ipa-inline.c (inline_small_functions): Fix typo.
576 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
578 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
579 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
580 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
581 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
582 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
583 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
584 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
585 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
586 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
587 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
588 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
589 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
590 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
591 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
592 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
593 (_mm512_storeu_epi64): Ditto.
594 (_mm512_cmpge_epi32_mask): Ditto.
595 (_mm512_cmpge_epu32_mask): Ditto.
596 (_mm512_cmpge_epi64_mask): Ditto.
597 (_mm512_cmpge_epu64_mask): Ditto.
598 (_mm512_cmple_epi32_mask): Ditto.
599 (_mm512_cmple_epu32_mask): Ditto.
600 (_mm512_cmple_epi64_mask): Ditto.
601 (_mm512_cmple_epu64_mask): Ditto.
602 (_mm512_cmplt_epi32_mask): Ditto.
603 (_mm512_cmplt_epu32_mask): Ditto.
604 (_mm512_cmplt_epi64_mask): Ditto.
605 (_mm512_cmplt_epu64_mask): Ditto.
606 (_mm512_cmpneq_epi32_mask): Ditto.
607 (_mm512_cmpneq_epu32_mask): Ditto.
608 (_mm512_cmpneq_epi64_mask): Ditto.
609 (_mm512_cmpneq_epu64_mask): Ditto.
610 (_mm512_expand_pd): Ditto.
611 (_mm512_expand_ps): Ditto.
612 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
613 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
614 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
615 * config/i386/i386.c (ix86_builtins): Add
616 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
617 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
618 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
619 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
620 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
621 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
622 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
623 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
624 IX86_BUILTIN_PMOVUSQW512_MEM.
625 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
626 __builtin_ia32_pmovsqd512mem_mask,
627 __builtin_ia32_pmovqd512mem_mask,
628 __builtin_ia32_pmovusqw512mem_mask,
629 __builtin_ia32_pmovsqw512mem_mask,
630 __builtin_ia32_pmovqw512mem_mask,
631 __builtin_ia32_pmovusdw512mem_mask,
632 __builtin_ia32_pmovsdw512mem_mask,
633 __builtin_ia32_pmovdw512mem_mask,
634 __builtin_ia32_pmovqb512mem_mask,
635 __builtin_ia32_pmovusqb512mem_mask,
636 __builtin_ia32_pmovsqb512mem_mask,
637 __builtin_ia32_pmovusdb512mem_mask,
638 __builtin_ia32_pmovsdb512mem_mask,
639 __builtin_ia32_pmovdb512mem_mask.
640 (bdesc_args): Add __builtin_ia32_expanddf512,
641 __builtin_ia32_expandsf512.
642 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
643 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
644 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
645 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
646 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
647 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
648 (avx512f_<code>v8div16qi2_mask_store): This.
649 (avx512f_expand<mode>): New.
651 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
653 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
655 (_mm512_mask_prefetch_i64gather_pd): Ditto.
656 (_mm512_prefetch_i32scatter_pd): Ditto.
657 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
658 (_mm512_prefetch_i64scatter_pd): Ditto.
659 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
660 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
661 (_mm512_mask_prefetch_i64gather_ps): Ditto.
662 (_mm512_prefetch_i32scatter_ps): Ditto.
663 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
664 (_mm512_prefetch_i64scatter_ps): Ditto.
665 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
666 * config/i386/i386-builtin-types.def: Define
667 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
668 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
669 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
670 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
671 IX86_BUILTIN_SCATTERPFQPD.
672 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
673 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
674 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
675 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
676 __builtin_ia32_scatterpfqps.
677 (ix86_expand_builtin): Expand new built-ins.
678 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
679 fix memory access data type.
680 (*avx512pf_gatherpf<mode>_mask): Ditto.
681 (*avx512pf_gatherpf<mode>): Ditto.
682 (avx512pf_scatterpf<mode>): Ditto.
683 (*avx512pf_scatterpf<mode>_mask): Ditto.
684 (*avx512pf_scatterpf<mode>): Ditto.
685 (GATHER_SCATTER_SF_MEM_MODE): New.
686 (avx512pf_gatherpf<mode>df): Ditto.
687 (*avx512pf_gatherpf<mode>df_mask): Ditto.
688 (*avx512pf_scatterpf<mode>df): Ditto.
690 2014-01-27 Jakub Jelinek <jakub@redhat.com>
693 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
694 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
697 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
699 * common/config/arm/arm-common.c
700 (arm_rewrite_mcpu): Handle multiple names.
702 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
704 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
706 * gimple-builder.h (create_gimple_tmp): Delete.
708 2014-01-27 Christian Bruel <christian.bruel@st.com>
710 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
713 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
715 * config/pa/pa.md (call): Generate indirect long calls to non-local
716 functions when outputing 32-bit code.
717 (call_value): Likewise except for special call to buggy powf function.
719 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
720 portable runtime and PIC indirect calls.
721 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
722 and PIC call sequences. Use ldo instead of blr to set return register
723 in PIC call sequence.
725 2014-01-25 Walter Lee <walt@tilera.com>
727 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
728 avoid clobbering a live register.
730 2014-01-25 Walter Lee <walt@tilera.com>
732 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
733 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
734 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
735 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
737 2014-01-25 Walter Lee <walt@tilera.com>
739 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
740 arguments on even registers.
741 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
743 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
744 (BIGGEST_ALIGNMENT): Ditto.
745 (BIGGEST_FIELD_ALIGNMENT): Ditto.
747 2014-01-25 Walter Lee <walt@tilera.com>
749 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
750 insns before bundling.
751 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
753 2014-01-25 Walter Lee <walt@tilera.com>
755 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
756 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
757 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
759 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
761 * config/mips/constraints.md (kl): Delete.
762 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
763 define expands, using...
764 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
765 instructions for MIPS16.
766 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
767 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
769 2014-01-25 Walter Lee <walt@tilera.com>
771 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
775 2014-01-25 Walter Lee <walt@tilera.com>
777 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
778 (TARGET_EXPAND_TO_RTL_HOOK): Define.
780 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
782 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
785 2014-01-25 Jakub Jelinek <jakub@redhat.com>
787 * print-rtl.c (in_call_function_usage): New var.
788 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
789 EXPR_LIST mode as mode and not as reg note name.
792 * cfgloopmanip.c (copy_loop_info): If
793 loop->warned_aggressive_loop_optimizations, make sure
794 the flag is set in target loop too.
796 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
798 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
800 * builtins.def: Likewise.
801 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
802 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
803 * ira.c (ira_setup_eliminable_regset): Likewise.
804 * omp-low.c (gate_expand_omp): Likewise.
805 (execute_lower_omp): Likewise.
806 (diagnose_sb_0): Likewise.
807 (gate_diagnose_omp_blocks): Likewise.
808 (simd_clone_clauses_extract): Likewise.
811 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
813 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
814 correction for little endian...
815 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
818 2014-01-24 Jeff Law <law@redhat.com>
820 PR tree-optimization/59919
821 * tree-vrp.c (find_assert_locations_1): Do not register asserts
822 for non-returning calls.
824 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
826 * common/config/aarch64/aarch64-common.c
827 (aarch64_rewrite_mcpu): Handle multiple names.
828 * config/aarch64/aarch64.h
829 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
831 2014-01-24 Dodji Seketeli <dodji@redhat.com>
833 * input.c (add_file_to_cache_tab): Handle the case where fopen
836 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
839 * config/i386/i386.md (pushsf splitter): Get stack adjustment
840 from push operand if code of push isn't PRE_DEC.
842 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
845 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
846 -mquad-memory-atomic. Update -mquad-memory documentation to say
847 it is only used for non-atomic loads/stores.
849 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
850 -mquad-memory or -mquad-memory-atomic switches.
852 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
853 -mquad-memory-atomic to ISA 2.07 support.
855 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
856 to separate support of normal quad word memory operations (ldq, stq)
857 from the atomic quad word memory operations.
859 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
860 support to separate non-atomic quad word operations from atomic
861 quad word operations. Disable non-atomic quad word operations in
862 little endian mode so that we don't have to swap words after the
863 load and before the store.
864 (quad_load_store_p): Add comment about atomic quad word support.
865 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
866 options printed with -mdebug=reg.
868 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
869 -mquad-memory-atomic as the test for whether we have quad word
871 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
872 or -mp8-vector are used, allow byte/half-word atomic operations.
874 * config/rs6000/sync.md (load_lockedti): Insure that the address
875 is a proper indexed or indirect address for the lqarx instruction.
876 On little endian systems, swap the hi/lo registers after the lqarx
878 (load_lockedpti): Use indexed_or_indirect_operand predicate to
879 insure the address is valid for the lqarx instruction.
880 (store_conditionalti): Insure that the address is a proper indexed
881 or indirect address for the stqcrx. instruction. On little endian
882 systems, swap the hi/lo registers before doing the stqcrx.
884 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
885 insure the address is valid for the stqcrx. instruction.
887 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
888 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
889 type of quad memory support is available.
891 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
894 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
895 there is a danger of looping.
897 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
899 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
900 force flag_ira_loop_pressure if set via command line.
902 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
904 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
905 (ashr_simd): New builtin handling DI mode.
906 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
907 (aarch64_sshr_simddi): New match pattern.
908 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
909 (vshrd_n_s64): Likewise.
910 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
912 2014-01-23 Nick Clifton <nickc@redhat.com>
914 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
915 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
916 favour of mcu specific scripts.
917 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
920 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
921 Alex Velenko <Alex.Velenko@arm.com>
923 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
924 (vaddv_s16): Likewise.
925 (vaddv_s32): Likewise.
926 (vaddv_u8): Likewise.
927 (vaddv_u16): Likewise.
928 (vaddv_u32): Likewise.
929 (vaddvq_s8): Likewise.
930 (vaddvq_s16): Likewise.
931 (vaddvq_s32): Likewise.
932 (vaddvq_s64): Likewise.
933 (vaddvq_u8): Likewise.
934 (vaddvq_u16): Likewise.
935 (vaddvq_u32): Likewise.
936 (vaddvq_u64): Likewise.
937 (vaddv_f32): Likewise.
938 (vaddvq_f32): Likewise.
939 (vaddvq_f64): Likewise.
940 (vmaxv_f32): Likewise.
941 (vmaxv_s8): Likewise.
942 (vmaxv_s16): Likewise.
943 (vmaxv_s32): Likewise.
944 (vmaxv_u8): Likewise.
945 (vmaxv_u16): Likewise.
946 (vmaxv_u32): Likewise.
947 (vmaxvq_f32): Likewise.
948 (vmaxvq_f64): Likewise.
949 (vmaxvq_s8): Likewise.
950 (vmaxvq_s16): Likewise.
951 (vmaxvq_s32): Likewise.
952 (vmaxvq_u8): Likewise.
953 (vmaxvq_u16): Likewise.
954 (vmaxvq_u32): Likewise.
955 (vmaxnmv_f32): Likewise.
956 (vmaxnmvq_f32): Likewise.
957 (vmaxnmvq_f64): Likewise.
958 (vminv_f32): Likewise.
959 (vminv_s8): Likewise.
960 (vminv_s16): Likewise.
961 (vminv_s32): Likewise.
962 (vminv_u8): Likewise.
963 (vminv_u16): Likewise.
964 (vminv_u32): Likewise.
965 (vminvq_f32): Likewise.
966 (vminvq_f64): Likewise.
967 (vminvq_s8): Likewise.
968 (vminvq_s16): Likewise.
969 (vminvq_s32): Likewise.
970 (vminvq_u8): Likewise.
971 (vminvq_u16): Likewise.
972 (vminvq_u32): Likewise.
973 (vminnmv_f32): Likewise.
974 (vminnmvq_f32): Likewise.
975 (vminnmvq_f64): Likewise.
977 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
979 * config/aarch64/aarch64-simd.md
980 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
981 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
982 (*aarch64_mul3_elt<mode>): Likewise.
983 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
984 (*aarch64_mul3_elt_to_64v2df): Likewise.
985 (*aarch64_mla_elt<mode>): Likewise.
986 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
987 (*aarch64_mls_elt<mode>): Likewise.
988 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
989 (*aarch64_fma4_elt<mode>): Likewise.
990 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
991 (*aarch64_fma4_elt_to_64v2df): Likewise.
992 (*aarch64_fnma4_elt<mode>): Likewise.
993 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
994 (*aarch64_fnma4_elt_to_64v2df): Likewise.
995 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
996 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
997 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
998 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
999 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1000 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1001 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1003 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1005 * config/aarch64/aarch64-simd.md
1006 (aarch64_be_checked_get_lane<mode>): New define_expand.
1007 * config/aarch64/aarch64-simd-builtins.def
1008 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1009 New builtin definition.
1010 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1011 Use new safe be builtin.
1013 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1015 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1017 (aarch64_be_st1<mode>): Likewise.
1018 (aarch_ld1<VALL:mode>): Define_expand modified.
1019 (aarch_st1<VALL:mode>): Likewise.
1020 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1021 (UNSPEC_ST1): Likewise.
1023 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1025 * config/microblaze/microblaze.md: Add trap insn and attribute
1027 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1029 PR preprocessor/58580
1030 * input.h (location_get_source_line): Take an additional line_size
1032 (void diagnostics_file_cache_fini): Declare new function.
1033 * input.c (struct fcache): New type.
1034 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1035 New static constants.
1036 (diagnostic_file_cache_init, total_lines_num)
1037 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1038 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1039 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1040 (get_next_line, read_next_line, goto_next_line, read_line_num):
1041 New static function definitions.
1042 (diagnostic_file_cache_fini): New function.
1043 (location_get_source_line): Take an additional output line_len
1044 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1046 * diagnostic.c (diagnostic_finish): Call
1047 diagnostic_file_cache_fini.
1048 (adjust_line): Take an additional input parameter for the length
1049 of the line, rather than calculating it with strlen.
1050 (diagnostic_show_locus): Adjust the use of
1051 location_get_source_line and adjust_line with respect to their new
1052 signature. While displaying a line now, do not stop at the first
1053 null byte. Rather, display the zero byte as a space and keep
1054 going until we reach the size of the line.
1055 * Makefile.in: Add vec.o to OBJS-libcommon
1057 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1058 Ilya Tocar <ilya.tocar@intel.com>
1060 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1061 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1062 (__builtin_ia32_kmov16): Ditto.
1063 * config/i386/i386.md (UNSPEC_KMOV): New.
1066 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1068 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1069 (_mm512_storeu_si512): Ditto.
1071 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1074 * rtl.h (get_referenced_operands): Declare.
1075 * recog.c (get_referenced_operands): New function.
1076 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1077 operands have been referenced when recording LO_SUM references.
1079 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1081 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1083 2014-01-22 Jan Hubicka <jh@suse.cz>
1085 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1086 Enable for generic and recent AMD targets.
1088 2014-01-22 Jan Hubicka <jh@suse.cz>
1090 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1091 ARG_SIZE note when adjustment was eliminated.
1093 2014-01-22 Jeff Law <law@redhat.com>
1095 PR tree-optimization/59597
1096 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1097 in file. Accept new argument REGISTERING and use it to modify
1098 dump output appropriately.
1099 (register_jump_thread): Corresponding changes.
1100 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1101 thread paths involving joiner blocks. Add code to dump cancelled
1102 jump threading paths.
1104 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1106 PR rtl-optimization/59477
1107 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1108 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1110 2014-01-22 Tom Tromey <tromey@redhat.com>
1112 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1114 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1116 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1118 PR rtl-optimization/59896
1119 * lra-constraints.c (process_alt_operands): Check unused note for
1120 matched operands of insn with no output reloads.
1122 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1124 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1125 (mips_move_from_gpr_cost): Likewise.
1127 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1129 PR rtl-optimization/59858
1130 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1131 ira_class_hard_regs_num.
1132 (process_alt_operands): Increase reject for dying matched operand.
1134 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1137 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1138 smaller than size, perform several stores or loads and stores
1139 at dst + count - size to store or copy all of size bytes, rather
1140 than just last modesize bytes.
1142 2014-01-20 DJ Delorie <dj@redhat.com>
1144 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1145 that CLOBBERs are REGs before propogating their values.
1147 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1150 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1151 (cgraph_inline_failed_type): New function.
1152 * cgraph.h (DEFCIFCODE): Add type.
1153 (cgraph_inline_failed_type_t): New enum.
1154 (cgraph_inline_failed_type): New prototype.
1155 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1156 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1157 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1158 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1159 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1160 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1161 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1162 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1163 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1164 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1165 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1166 OPTIMIZATION_MISMATCH.
1167 * tree-inline.c (expand_call_inline): Emit errors during
1168 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1170 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1173 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1174 mode attribute in insn output.
1176 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1178 * output.h (output_constant): Delete.
1179 * varasm.c (output_constant): Make private.
1181 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1183 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1185 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1188 * tree.h (fold_builtin_strcat): New prototype.
1189 * builtins.c (fold_builtin_strcat): No longer static. Add len
1190 argument, if non-NULL, don't call c_strlen. Optimize
1191 directly into __builtin_memcpy instead of __builtin_strcpy.
1192 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1193 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1195 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1197 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1198 for SImode_address_operand operands, having only a REG argument.
1200 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1202 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1203 loader name using mbig-endian.
1204 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1206 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1208 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1212 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1214 * config/aarch64/aarch64-protos.h
1215 (aarch64_cannot_change_mode_class_ptr): Declare.
1216 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1217 aarch64_cannot_change_mode_class_ptr): New.
1218 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1219 backend hook aarch64_cannot_change_mode_class.
1221 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1223 * common/config/aarch64/aarch64-common.c
1224 (aarch64_handle_option): Don't handle any option order logic here.
1225 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1226 selected_cpu, warn on architecture version mismatch.
1227 (aarch64_override_options): Fix parsing order for option strings.
1229 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1230 Iain Sandoe <iain@codesourcery.com>
1233 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1234 warning. Amend comment to reflect current functionality.
1236 2014-01-20 Richard Biener <rguenther@suse.de>
1239 * builtins.c (fold_builtin_strcat): Remove case better handled
1240 by tree-ssa-strlen.c.
1242 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1244 * config/aarch64/aarch64.opt
1245 (mcpu, march, mtune): Make case-insensitive.
1247 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1250 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1251 if operands[1] is a REG or ZERO_EXTEND of a REG.
1253 2014-01-19 Jan Hubicka <jh@suse.cz>
1255 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1257 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1259 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1260 long non-pic millicode calls.
1262 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1264 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1266 2014-01-19 Kito Cheng <kito@0xlab.org>
1268 * builtins.c (expand_movstr): Check movstr expand done or fail.
1270 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1271 H.J. Lu <hongjiu.lu@intel.com>
1274 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1275 to DImode for zero-extended addresses.
1277 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1279 PR rtl-optimization/57763
1280 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1281 on the new indirect jump_insn and increment LABEL_NUSES (label).
1283 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1287 * config.gcc (x86_archs): New variable.
1288 (x86_64_archs): Likewise.
1289 (x86_cpus): Likewise.
1290 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1291 --with-arch/--with-cpu= options.
1292 Support --with-arch=/--with-cpu={nehalem,westmere,
1293 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1295 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1297 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1298 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1300 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1302 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1304 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1307 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1308 clear cpp_get_options (parse_in)->warn_unused_macros for
1309 ix86_target_macros_internal with cpp_define.
1311 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1313 * jump.c (delete_related_insns): Keep (use (insn))s.
1314 * reorg.c (redundant_insn): Check for barriers too.
1316 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1318 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1320 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1322 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1323 call to $$dyncall when TARGET_LONG_CALLS is true.
1325 2014-01-17 Jeff Law <law@redhat.com>
1327 * ree.c (combine_set_extension): Temporarily disable test for
1328 changing number of hard registers.
1330 2014-01-17 Jan Hubicka <jh@suse.cz>
1333 * ipa-inline-analysis.c (inline_free_summary):
1334 Do not free summary of aliases.
1336 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1339 * gimplify.c (gimplify_expr): Use create_tmp_var
1340 instead of create_tmp_var_raw. If cond doesn't have
1341 integral type, don't add the IFN_ANNOTATE builtin at all.
1343 2014-01-17 Martin Jambor <mjambor@suse.cz>
1346 * ipa-cp.c (prev_edge_clone): New variable.
1347 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1348 Also resize prev_edge_clone vector.
1349 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1350 (ipcp_edge_removal_hook): New function.
1351 (ipcp_driver): Register ipcp_edge_removal_hook.
1353 2014-01-17 Andrew Pinski <apinski@cavium.com>
1354 Steve Ellcey <sellcey@mips.com>
1357 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1360 2014-01-17 Jeff Law <law@redhat.com>
1363 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1364 so that pass_ccp runs first.
1366 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1368 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1369 (ix86_adjust_cost): Use !TARGET_XXX.
1370 (do_reorder_for_imul): Likewise.
1371 (swap_top_of_ready_list): Likewise.
1372 (ix86_sched_reorder): Likewise.
1374 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1376 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1377 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1378 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1379 (intel_memset): New. Duplicate slm_memset.
1380 (intel_cost): New. Duplicate slm_cost.
1381 (m_INTEL): New macro.
1382 (processor_target_table): Add "intel".
1383 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1384 with PROCESSOR_INTEL for "intel".
1385 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1386 PROCESSOR_SILVERMONT.
1387 (ix86_issue_rate): Likewise.
1388 (ix86_adjust_cost): Likewise.
1389 (ia32_multipass_dfa_lookahead): Likewise.
1390 (swap_top_of_ready_list): Likewise.
1391 (ix86_sched_reorder): Likewise.
1392 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1393 instead of TARGET_OPT_AGU.
1394 * config/i386/i386.h (TARGET_INTEL): New.
1395 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1396 (processor_type): Add PROCESSOR_INTEL.
1397 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1398 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1400 2014-01-17 Marek Polacek <polacek@redhat.com>
1403 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1406 2014-01-17 Richard Biener <rguenther@suse.de>
1408 PR tree-optimization/46590
1409 * opts.c (default_options_table): Add entries for
1410 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1411 all enabled at -O1 but not for -Og.
1412 * common.opt (fbranch-count-reg): Remove Init(1).
1413 (fmove-loop-invariants): Likewise.
1414 (ftree-pta): Likewise.
1416 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1418 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1419 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1420 decls to at least the GCC 4.8 used alignments.
1423 * tree-nested.c (convert_nonlocal_reference_stmt,
1424 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1425 of GIMPLE_BIND stmts, adjust associated decls.
1427 2014-01-17 Richard Biener <rguenther@suse.de>
1429 PR tree-optimization/46590
1430 * vec.h (vec<>::bseach): New member function implementing
1431 binary search according to C89 bsearch.
1432 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1433 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1434 bitmap pointer again. Make accesses_in_loop a flat array.
1435 (mem_ref_obstack): New global.
1436 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1437 (mark_ref_stored): Likewise.
1438 (ref_indep_loop_p_2): Likewise.
1439 (set_ref_stored_in_loop): New helper function.
1440 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1441 (memref_free): Adjust.
1442 (record_mem_ref_loc): Simplify.
1443 (gather_mem_refs_stmt): Adjust.
1444 (sort_locs_in_loop_postorder_cmp): New function.
1445 (analyze_memory_references): Sort accesses_in_loop after
1446 loop postorder number.
1447 (find_ref_loc_in_loop_cmp): New function.
1448 (for_all_locs_in_loop): Find relevant cluster of locs in
1449 accesses_in_loop and iterate without recursion.
1450 (execute_sm): Avoid uninit warning.
1451 (struct ref_always_accessed): Simplify.
1452 (ref_always_accessed::operator ()): Likewise.
1453 (ref_always_accessed_p): Likewise.
1454 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1455 loop postorder numbers here.
1456 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1459 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1462 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1463 on decls for which assemble_alias has been called.
1465 2014-01-17 Nick Clifton <nickc@redhat.com>
1467 * config/msp430/msp430.opt: (mcpu): New option.
1468 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1469 (msp430_option_override): Parse target_cpu. If the MCU name
1470 matches a generic string, clear target_mcu.
1471 (msp430_attr): Allow numeric interrupt values up to 63.
1472 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1473 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1475 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1477 * config/msp430/msp430.md (popm): Use %J rather than %I.
1478 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1479 (addhi_cy_i): Use immediate_operand for operand 2.
1480 * doc/invoke.texi: Document -mcpu option.
1482 2014-01-17 Richard Biener <rguenther@suse.de>
1484 PR rtl-optimization/38518
1485 * df.h (df_analyze_loop): Declare.
1486 * df-core.c: Include cfgloop.h.
1487 (df_analyze_1): Split out main part of df_analyze.
1488 (df_analyze): Adjust.
1489 (loop_inverted_post_order_compute): New function.
1490 (loop_post_order_compute): Likewise.
1491 (df_analyze_loop): New function avoiding whole-function
1493 * loop-invariant.c (find_defs): Use df_analyze_loop.
1494 (find_invariants): Adjust.
1495 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1497 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1499 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1500 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1502 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1504 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1505 traversal when removing references.
1507 2014-01-16 Jan Hubicka <jh@suse.cz>
1510 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1512 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1515 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1516 pushing a reload for an autoinc when we had previously reloaded an
1517 inner part of the address.
1519 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1521 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1523 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1524 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1525 when not giving up or versioning for alias only because of
1527 (vect_analyze_data_ref_dependences): Set to true.
1528 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1530 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1531 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1535 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1538 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1539 operand 0 predicate for gathers, use a new pseudo as subtarget.
1541 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1544 * lra-constraints.c (process_alt_operands): Add printing debug info.
1545 Check absence of input/output reloads for matched operands too.
1547 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1549 PR rtl-optimization/59835
1550 * ira.c (ira_init_register_move_cost): Increase cost for
1553 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1555 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1557 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1560 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1561 non-register objects. Use gen_(high/low)part more consistently.
1564 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1567 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1568 endian support, remove tests for WORDS_BIG_ENDIAN.
1569 (p8_mfvsrd_3_<mode>): Likewise.
1570 (reload_gpr_from_vsx<mode>): Likewise.
1571 (reload_gpr_from_vsxsf): Likewise.
1572 (p8_mfvsrd_4_disf): Likewise.
1574 2014-01-16 Richard Biener <rguenther@suse.de>
1576 PR rtl-optimization/46590
1577 * lcm.c (compute_antinout_edge): Use postorder iteration.
1578 (compute_laterin): Use inverted postorder iteration.
1580 2014-01-16 Nick Clifton <nickc@redhat.com>
1583 * varasm.c (output_constant): Return the number of bytes actually
1585 (output_constructor_array_range): Update the field size with the
1586 number of bytes emitted by output_constant.
1587 (output_constructor_regular_field): Likewise. Also do not
1588 complain if the total number of bytes emitted is now greater
1589 than the expected fieldpos.
1590 * output.h (output_constant): Update prototype and descriptive comment.
1592 2014-01-16 Marek Polacek <polacek@redhat.com>
1595 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1596 it is error_mark_node.
1598 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1600 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1601 VALID_AVX256_REG_OR_OI_MODE.
1603 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1605 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1606 current procedure should be profiled.
1608 2014-01-15 Andrew Pinski <apinski@cavium.com>
1610 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1611 of moving from/to the STACK_REG register class.
1613 2014-01-15 Richard Henderson <rth@redhat.com>
1616 * reginfo.c (global_regs_decl): Globalize.
1617 * rtl.h (global_regs_decl): Declare.
1618 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1619 reserved via global_regs.
1621 2014-01-15 Teresa Johnson <tejohnson@google.com>
1623 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1625 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1627 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1628 and vmulosh rather than call gen_vec_widen_smult_*.
1629 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1630 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1631 (vec_widen_smult_even_v16qi): Likewise.
1632 (vec_widen_umult_even_v8hi): Likewise.
1633 (vec_widen_smult_even_v8hi): Likewise.
1634 (vec_widen_umult_odd_v16qi): Likewise.
1635 (vec_widen_smult_odd_v16qi): Likewise.
1636 (vec_widen_umult_odd_v8hi): Likewise.
1637 (vec_widen_smult_odd_v8hi): Likewise.
1638 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1639 vmuloub rather than call gen_vec_widen_umult_*.
1640 (vec_widen_umult_lo_v16qi): Likewise.
1641 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1642 vmulosb rather than call gen_vec_widen_smult_*.
1643 (vec_widen_smult_lo_v16qi): Likewise.
1644 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1645 rather than call gen_vec_widen_umult_*.
1646 (vec_widen_umult_lo_v8hi): Likewise.
1647 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1648 rather than call gen_vec_widen_smult_*.
1649 (vec_widen_smult_lo_v8hi): Likewise.
1651 2014-01-15 Jeff Law <law@redhat.com>
1653 PR tree-optimization/59747
1654 * ree.c (find_and_remove_re): Properly handle case where a second
1655 eliminated extension requires widening a copy created for elimination
1656 of a prior extension.
1657 (combine_set_extension): Ensure that the number of hard regs needed
1658 for a destination register does not change when we widen it.
1660 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1662 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1663 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1664 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1665 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1666 (avr-*-rtems*): Likewise.
1667 (bfin*-rtems*): Likewise.
1668 (moxie-*-rtems*): Likewise.
1669 (h8300-*-rtems*): Likewise.
1670 (i[34567]86-*-rtems*): Likewise.
1671 (lm32-*-rtems*): Likewise.
1672 (m32r-*-rtems*): Likewise.
1673 (m68k-*-rtems*): Likewise.
1674 (microblaze*-*-rtems*): Likewise.
1675 (mips*-*-rtems*): Likewise.
1676 (powerpc-*-rtems*): Likewise.
1677 (sh-*-rtems*): Likewise.
1678 (sparc-*-rtems*): Likewise.
1679 (sparc64-*-rtems*): Likewise.
1680 (v850-*-rtems*): Likewise.
1681 (m32c-*-rtems*): Likewise.
1683 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1685 PR rtl-optimization/59511
1686 * ira.c (ira_init_register_move_cost): Use memory costs for some
1687 cases of register move cost calculations.
1688 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1689 instead of BB frequency.
1690 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1691 * lra-assigns.c (find_hard_regno_for): Ditto.
1693 2014-01-15 Richard Biener <rguenther@suse.de>
1695 PR tree-optimization/59822
1696 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1697 (vectorizable_load): Use it to hoist defs of uses of invariant
1698 loads out of the loop.
1700 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1701 Kugan Vivekanandarajah <kuganv@linaro.org>
1704 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1707 2014-01-15 Richard Biener <rguenther@suse.de>
1709 PR rtl-optimization/59802
1710 * lcm.c (compute_available): Use inverted postorder to seed
1711 the initial worklist.
1713 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1716 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1717 ADDR_REGS for invalid symrefs in non-PIC code.
1719 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1722 * builtins.c (determine_block_size): Initialize *probable_max_size
1723 even if len_rtx is CONST_INT.
1725 2014-01-14 Andrew Pinski <apinski@cavium.com>
1727 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1728 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1729 (cortexa53_tunings): Likewise.
1730 (aarch64_sched_issue_rate): New function.
1731 (TARGET_SCHED_ISSUE_RATE): Define.
1733 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1735 * ira-costs.c (find_costs_and_classes): Add missed
1736 ira_init_register_move_cost_if_necessary.
1738 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1741 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1743 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1746 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1747 to indicate if type is used for function return value. Warn ABI
1748 change if the vector mode isn't available for function return value.
1749 (ix86_function_arg_advance): Pass false to type_natural_mode.
1750 (ix86_function_arg): Likewise.
1751 (ix86_gimplify_va_arg): Likewise.
1752 (function_arg_32): Don't warn ABI change.
1753 (ix86_function_value): Pass true to type_natural_mode.
1754 (ix86_return_in_memory): Likewise.
1755 (ix86_struct_value_rtx): Removed.
1756 (TARGET_STRUCT_VALUE_RTX): Likewise.
1758 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1760 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1761 converting a conditional jump into a conditional return.
1763 2014-01-14 Richard Biener <rguenther@suse.de>
1765 PR tree-optimization/58921
1766 PR tree-optimization/59006
1767 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1768 hoisting invariant stmts.
1769 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1770 invariant loads on the preheader edge if possible.
1772 2014-01-14 Joey Ye <joey.ye@arm.com>
1774 * doc/plugin.texi (Building GCC plugins): Update to C++.
1776 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1778 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1779 (_mm_rcp28_round_ss): Ditto.
1780 (_mm_rsqrt28_round_sd): Ditto.
1781 (_mm_rsqrt28_round_ss): Ditto.
1782 (_mm_rcp28_sd): Ditto.
1783 (_mm_rcp28_ss): Ditto.
1784 (_mm_rsqrt28_sd): Ditto.
1785 (_mm_rsqrt28_ss): Ditto.
1786 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1787 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1788 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1789 (IX86_BUILTIN_RCP28SD): Ditto.
1790 (IX86_BUILTIN_RCP28SS): Ditto.
1791 (IX86_BUILTIN_RSQRT28SD): Ditto.
1792 (IX86_BUILTIN_RSQRT28SS): Ditto.
1793 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1794 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1795 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1796 (ix86_expand_special_args_builtin): Expand new FTYPE.
1797 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1798 (srcp14<mode>): Make insn unary.
1799 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1800 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1801 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1802 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1803 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1804 Fix rounding: make it SAE only.
1805 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1807 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1809 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1810 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1811 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1812 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1813 (round_saeonly_mask_scalar_operand4): Ditto.
1814 (round_saeonly_mask_scalar_op3): Ditto.
1815 (round_saeonly_mask_scalar_op4): Ditto.
1817 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1819 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1820 Implement -maltivec=be for vec_insert and vec_extract.
1822 2014-01-10 DJ Delorie <dj@redhat.com>
1824 * config/msp430/msp430.md (call_internal): Don't allow memory
1825 references with SP as the base register.
1826 (call_value_internal): Likewise.
1827 * config/msp430/constraints.md (Yc): New. For memory references
1828 that don't use SP as a base register.
1830 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1831 "an integer without a # prefix"
1832 * config/msp430/msp430.md (epilogue_helper): Use it.
1834 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1837 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1838 AVX512F gather builtins.
1839 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1840 on gather decls with INTEGER_TYPE masktype.
1841 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1842 directly into the builtin rather than hoisting it before loop.
1844 PR tree-optimization/59387
1845 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1846 (scev_const_prop): If folded_casts and type has undefined overflow,
1847 use force_gimple_operand instead of force_gimple_operand_gsi and
1848 for each added stmt if it is assign with
1849 arith_code_with_undefined_signed_overflow, call
1850 rewrite_to_defined_overflow.
1851 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1852 gimple-fold.h instead.
1853 (arith_code_with_undefined_signed_overflow,
1854 rewrite_to_defined_overflow): Moved to ...
1855 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1856 rewrite_to_defined_overflow): ... here. No longer static.
1857 Include gimplify-me.h.
1858 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1859 rewrite_to_defined_overflow): New prototypes.
1861 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1863 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1865 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1867 * builtins.c (get_object_alignment_2): Minor tweak.
1868 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1870 2014-01-13 Christian Bruel <christian.bruel@st.com>
1872 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1873 optimized non constant lengths.
1875 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1878 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1879 load as __atomic_load_N if possible.
1881 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1883 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1885 (rs6000_expand_builtin): Adjust call.
1887 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1890 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1891 * config/rs6000/rs6000.c: Include target-globals.h.
1892 (rs6000_set_current_function): Instead of doing target_reinit
1893 unconditionally, use save_target_globals_default_opts and
1894 restore_target_globals.
1896 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1898 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1899 (rs6000_expand_builtin): Handle mffs and mtfsf.
1900 (rs6000_init_builtins): Define mffs and mtfsf.
1901 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1902 (rs6000_mffs): New pattern.
1903 (rs6000_mtfsf): New pattern.
1905 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1907 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1908 Start narrowing with START. Apply candidate-use pair
1909 and check overall cost in narrowing.
1910 (iv_ca_prune): Pass new argument.
1912 2014-01-10 Jeff Law <law@redhat.com>
1915 * ree.c (combine_reaching_defs): Ensure the defining statement
1916 occurs before the extension when optimizing extensions with
1917 different source and destination hard registers.
1919 2014-01-10 Jan Hubicka <jh@suse.cz>
1922 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1923 vtables into the type inheritance graph.
1925 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1927 PR rtl-optimization/59754
1928 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1929 modes in the REGNO != REGNO case.
1931 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1933 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1935 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1937 PR tree-optimization/59745
1938 * tree-predcom.c (tree_predictive_commoning_loop): Call
1939 free_affine_expand_cache if giving up because components is NULL.
1941 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1942 GC in payload of target_globals struct instead of allocating them on
1943 the heap and the larger structs separately using GC.
1944 * target-globals.h (struct target_globals): Make regs, hard_regs,
1945 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1946 of GTY((skip)) and change type to void *.
1947 (reset_target_globals): Cast loads from those fields to corresponding
1950 2014-01-10 Steve Ellcey <sellcey@mips.com>
1953 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1954 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1955 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1957 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1960 * aarch64-modes.def (CC_Zmode): New flags mode.
1961 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1962 represents an equality.
1963 (aarch64_get_condition_code): Handle CC_Zmode.
1964 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1966 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1968 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1969 extraction in good case.
1971 2014-01-10 Richard Biener <rguenther@suse.de>
1973 PR tree-optimization/59374
1974 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1975 checking after SLP discovery. Mark stmts not participating
1976 in any SLP instance properly.
1978 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1980 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1981 when handling a SET rtx.
1983 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1985 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1986 (cortex-a57): Likewise.
1987 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1989 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1991 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1992 non-iwmmxt builtins.
1994 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1998 * ipa-devirt.c record_target_from_binfo): Take as argument
1999 stack of binfos and lookup matching one for virtual inheritance.
2000 (possible_polymorphic_call_targets_1): Update.
2002 2014-01-10 Huacai Chen <chenhc@lemote.com>
2004 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2005 kernel strings for Loongson-2E/2F/3A.
2007 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2010 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2011 is_gimple_call before calling gimple_call_internal_p.
2013 2014-01-09 Steve Ellcey <sellcey@mips.com>
2015 * Makefile.in (TREE_FLOW_H): Remove.
2016 (TREE_SSA_H): Add file names from tree-flow.h.
2017 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2018 * tree.h: Remove tree-flow.h reference.
2019 * hash-table.h: Remove tree-flow.h reference.
2020 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2021 reference with tree-ssa-loop.h.
2023 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2025 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2026 default element-order behavior for -maltivec.
2027 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2028 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2029 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2030 when targeting big endian, at least for now.
2031 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2033 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2036 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2037 var satisfies use_register_for_decl, just take into account type
2038 alignment, rather than decl alignment.
2040 PR tree-optimization/59622
2041 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2042 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2043 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2044 Don't devirtualize for inplace at all. For targets.length () == 1,
2045 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2047 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2049 * config/i386/i386.md (cpu): Remove the unused btver1.
2051 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2053 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2055 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2058 * tree-core.h (struct target_globals): New forward declaration.
2059 (struct tree_target_option): Add globals field.
2060 * tree.h (TREE_TARGET_GLOBALS): Define.
2061 (prepare_target_option_nodes_for_pch): New prototype.
2062 * target-globals.h (struct target_globals): Define even if
2064 * tree.c (prepare_target_option_node_for_pch,
2065 prepare_target_option_nodes_for_pch): New functions.
2066 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2067 * config/i386/i386.c: Include target-globals.h.
2068 (ix86_set_current_function): Instead of doing target_reinit
2069 unconditionally, use save_target_globals_default_opts and
2070 restore_target_globals.
2072 2014-01-09 Richard Biener <rguenther@suse.de>
2074 PR tree-optimization/59715
2075 * tree-cfg.h (split_critical_edges): Declare.
2076 * tree-cfg.c (split_critical_edges): Export.
2077 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2079 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2081 * cfgexpand.c (expand_stack_vars): Optionally disable
2082 asan stack protection.
2083 (expand_used_vars): Likewise.
2084 (partition_stack_vars): Likewise.
2085 * asan.c (asan_emit_stack_protection): Optionally disable
2086 after return stack usage.
2087 (instrument_derefs): Optionally disable memory access instrumentation.
2088 (instrument_builtin_call): Likewise.
2089 (instrument_strlen_call): Likewise.
2090 (asan_protect_global): Optionally disable global variables protection.
2091 * doc/invoke.texi: Added doc for new options.
2092 * params.def: Added new options.
2093 * params.h: Likewise.
2095 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2097 PR rtl-optimization/59724
2098 * ifcvt.c (cond_exec_process_if_block): Don't call
2099 flow_find_head_matching_sequence with 0 longest_match.
2100 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2101 non-active insns if !stop_after.
2102 (try_head_merge_bb): Revert 2014-01-07 changes.
2104 2014-01-08 Jeff Law <law@redhat.com>
2106 * ree.c (get_sub_rtx): New function, extracted from...
2107 (merge_def_and_ext): Here.
2108 (combine_reaching_defs): Use get_sub_rtx.
2110 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2112 * cgraph.h (varpool_variable_node): Do not choke on null node.
2114 2014-01-08 Catherine Moore <clm@codesourcery.com>
2116 * config/mips/mips.md (simple_return): Attempt to use JRC
2118 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2120 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2122 PR rtl-optimization/59137
2123 * reorg.c (steal_delay_list_from_target): Call update_block for
2125 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2127 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2129 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2130 two duplicate entries.
2132 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2135 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2137 * config/mips/mips.c (mips_truncated_op_cost): New function.
2138 (mips_rtx_costs): Adjust test for BADDU.
2139 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2141 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2143 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2144 (*baddu_si): ...this new pattern.
2146 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2149 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2151 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2154 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2156 (expand_expr, expand_normal): Adjust.
2157 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2158 inner_reference_p. Use inner_reference_p to expand inner references.
2159 (store_expr): Adjust.
2160 * cfgexpand.c (expand_call_stmt): Adjust.
2162 2014-01-08 Rong Xu <xur@google.com>
2164 * gcov-io.c (gcov_var): Move from gcov-io.h.
2165 (gcov_position): Ditto.
2166 (gcov_is_error): Ditto.
2167 (gcov_rewrite): Ditto.
2168 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2169 only part to libgcc/libgcov.h.
2171 2014-01-08 Marek Polacek <polacek@redhat.com>
2174 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2176 2014-01-08 Marek Polacek <polacek@redhat.com>
2179 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2181 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2183 PR rtl-optimization/59649
2184 * stor-layout.c (get_mode_bounds): For BImode return
2185 0 and STORE_FLAG_VALUE.
2187 2014-01-08 Richard Biener <rguenther@suse.de>
2190 * gimple.h (is_gimple_builtin_call): Remove.
2191 (gimple_builtin_call_types_compatible_p): New.
2192 (gimple_call_builtin_p): New overload.
2193 * gimple.c (is_gimple_builtin_call): Remove.
2194 (validate_call): Rename to ...
2195 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2197 (validate_type): New static function.
2198 (gimple_call_builtin_p): New overload and adjust.
2199 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2200 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2201 (gimple_fold_stmt_to_constant_1): Likewise.
2202 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2204 2014-01-08 Richard Biener <rguenther@suse.de>
2207 * gimplify.c (gimplify_expr): Gimplify register-register type
2208 VIEW_CONVERT_EXPRs to separate stmts.
2210 2014-01-07 Jeff Law <law@redhat.com>
2213 * ree.c (combine_set_extension): Handle case where source
2214 and destination registers in an extension insn are different.
2215 (combine_reaching_defs): Allow source and destination registers
2216 in extension to be different under limited circumstances.
2217 (add_removable_extension): Remove restriction that the
2218 source and destination registers in the extension are the same.
2219 (find_and_remove_re): Emit a copy from the extension's
2220 destination to its source after the defining insn if
2221 the source and destination registers are different.
2224 * ifcvt.c (merge_if_block): If we are merging a block with more than
2225 one successor with a block with no successors, remove any BARRIER
2226 after the second block.
2228 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2230 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2232 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2235 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2236 for 14-bit register offsets when INT14_OK_STRICT is false.
2238 2014-01-07 Roland Stigge <stigge@antcom.de>
2239 Michael Meissner <meissner@linux.vnet.ibm.com>
2242 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2243 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2245 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2247 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2250 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2252 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2253 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2254 rtx is const0_rtx or not.
2256 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2259 * target-globals.c (save_target_globals): Remove this_fn_optab
2261 * toplev.c: Include optabs.h.
2262 (target_reinit): Temporarily restore the global options if another
2263 set of options are in force.
2265 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2267 PR rtl-optimization/58668
2268 * cfgcleanup.c (flow_find_cross_jump): Don't count
2269 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2270 to determine what is counted.
2271 (flow_find_head_matching_sequence): Use active_insn_p to determine
2273 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2275 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2276 determine what is counted.
2278 PR tree-optimization/59643
2279 * tree-predcom.c (split_data_refs_to_components): If one dr is
2280 read and one write, determine_offset fails and the write isn't
2281 in the bad component, just put the read into the bad component.
2283 2014-01-07 Mike Stump <mikestump@comcast.net>
2284 Jakub Jelinek <jakub@redhat.com>
2287 * tree-core.h (struct tree_optimization_option): Change optabs
2288 type from unsigned char * to void *.
2289 * optabs.c (init_tree_optimization_optabs): Adjust
2290 TREE_OPTIMIZATION_OPTABS initialization.
2292 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2295 * config/i386/i386.h (struct machine_function): Add
2296 no_drap_save_restore field.
2297 * config/i386/i386.c (ix86_save_reg): Use
2298 !cfun->machine->no_drap_save_restore instead of
2299 crtl->stack_realign_needed.
2300 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2301 this function clears frame_pointer_needed. Set
2302 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2303 and DRAP reg is needed.
2305 2014-01-06 Marek Polacek <polacek@redhat.com>
2308 * doc/implement-c.texi: Mention that other integer types are
2309 permitted as bit-field types in strictly conforming mode.
2311 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2313 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2316 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2318 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2320 2014-01-06 Martin Jambor <mjambor@suse.cz>
2323 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2325 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2327 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2331 * var-tracking.c (add_stores): Preserve the value of the source even if
2332 we don't record the store.
2334 2014-01-06 Terry Guo <terry.guo@arm.com>
2336 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2338 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2341 * config/darwin.c (darwin_function_section): Adjust return values to
2342 correspond to optimisation changes made in r206070.
2344 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2346 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2347 from prefetch_block tune setting.
2348 (nocona_cost): Correct size of prefetch block to 64.
2350 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2352 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2353 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2354 used to save the static chain register in the computation of the offset
2355 from which the FP registers need to be restored.
2357 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2359 PR tree-optimization/59519
2360 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2361 ICE if get_current_def (current_new_name) is already non-NULL, as long
2362 as it is a phi result of some other phi in *new_exit_bb that has
2365 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2366 or vmovdqu* for misaligned_operand.
2367 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2368 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2369 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2370 aligned_mem for AVX512F masked aligned load and store builtins and for
2373 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2375 PR tree-optimization/59651
2376 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2377 Address range for negative step should be added by TYPE_SIZE_UNIT.
2379 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2381 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2382 overlapping registers also for registers other than the stack pointer.
2384 2014-01-03 Marek Polacek <polacek@redhat.com>
2387 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2390 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2393 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2396 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2397 (push splitter): Use <P:MODE_SIZE> instead of
2398 GET_MODE_SIZE (<P:MODE>mode).
2399 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2400 (mov -1, reg peephole2): Likewise.
2401 * config/i386/sse.md (*mov<mode>_internal,
2402 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2403 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2404 *<code><mode>3, *andnot<mode>3<mask_name>,
2405 <mask_codefor><code><mode>3<mask_name>): Likewise.
2406 * config/i386/subst.md (mask_mode512bit_condition,
2407 sd_mask_mode512bit_condition): Likewise.
2409 2014-01-02 Xinliang David Li <davidxl@google.com>
2411 PR tree-optimization/59303
2412 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2413 (dump_predicates): Better output format.
2414 (pred_equal_p): New function.
2415 (is_neq_relop_p): Ditto.
2416 (is_neq_zero_form_p): Ditto.
2417 (pred_expr_equal_p): Ditto.
2418 (pred_neg_p): Ditto.
2419 (simplify_pred): Ditto.
2420 (simplify_preds_2): Ditto.
2421 (simplify_preds_3): Ditto.
2422 (simplify_preds_4): Ditto.
2423 (simplify_preds): Ditto.
2425 (push_to_worklist): Ditto.
2426 (get_pred_info_from_cmp): Ditto.
2427 (is_degenerated_phi): Ditto.
2428 (normalize_one_pred_1): Ditto.
2429 (normalize_one_pred): Ditto.
2430 (normalize_one_pred_chain): Ditto.
2431 (normalize_preds): Ditto.
2432 (normalize_cond_1): Remove function.
2433 (normalize_cond): Ditto.
2434 (is_gcond_subset_of): Ditto.
2435 (is_subset_of_any): Ditto.
2436 (is_or_set_subset_of): Ditto.
2437 (is_and_set_subset_of): Ditto.
2438 (is_norm_cond_subset_of): Ditto.
2439 (pred_chain_length_cmp): Ditto.
2440 (convert_control_dep_chain_into_preds): Type change.
2441 (find_predicates): Ditto.
2442 (find_def_preds): Ditto.
2443 (destroy_predicates_vecs): Ditto.
2444 (find_matching_predicates_in_rest_chains): Ditto.
2445 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2446 (is_pred_expr_subset): Ditto.
2447 (is_pred_chain_subset_of): Ditto.
2448 (is_included_in): Ditto.
2449 (is_superset_of): Ditto.
2451 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2453 Update copyright years.
2455 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2457 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2458 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2459 config/arc/arc.md, config/arc/arc.opt,
2460 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2461 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2462 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2463 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2464 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2465 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2466 vtable-verify.c, vtable-verify.h: Use the standard form for the
2469 2014-01-02 Tobias Burnus <burnus@net-b.de>
2471 * gcc.c (process_command): Update copyright notice dates.
2472 * gcov-dump.c: Ditto.
2474 * doc/cpp.texi: Bump @copying's copyright year.
2475 * doc/cppinternals.texi: Ditto.
2476 * doc/gcc.texi: Ditto.
2477 * doc/gccint.texi: Ditto.
2478 * doc/gcov.texi: Ditto.
2479 * doc/install.texi: Ditto.
2480 * doc/invoke.texi: Ditto.
2482 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2484 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2486 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2488 * config/i386/sse.md (*mov<mode>_internal): Guard
2489 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2491 PR rtl-optimization/59647
2492 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2493 new_rtx into UNSIGNED_FLOAT rtxes.
2495 Copyright (C) 2014 Free Software Foundation, Inc.
2497 Copying and distribution of this file, with or without modification,
2498 are permitted in any medium without royalty provided the copyright
2499 notice and this notice are preserved.