1 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/60003
4 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5 * profile.c (branch_prob): Use gimple_call_builtin_p
6 to check for BUILT_IN_SETJMP_RECEIVER.
7 * tree-inline.c (copy_bb): Call notice_special_calls.
9 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
12 * lra-constraints.c (process_alt_operands): Update reload_sum only
15 2014-01-31 Richard Henderson <rth@redhat.com>
18 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
19 until after else_eh is processed.
21 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
23 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
24 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
25 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
26 in smmintrin.h, remove them.
27 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
28 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
29 * config/i386/i386.md (ROUND_SAE): Fix value.
30 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
31 (const48_operand): New.
32 * config/i386/subst.md (round), (round_expand): Use
33 const_4_or_8_to_11_operand.
34 (round_saeonly), (round_saeonly_expand): Use const48_operand.
36 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
38 * config/i386/constraints.md (Yk): Swap meaning with k.
39 * config/i386/i386.md (movhi_internal): Change Yk to k.
40 (movqi_internal): Ditto.
41 (*k<logic><mode>): Ditto.
51 (*one_cmplhi2_1): Ditto.
52 (*one_cmplqi2_1): Ditto.
53 * config/i386/sse.md (): Change k to Yk.
54 (avx512f_load<mode>_mask): Ditto.
55 (avx512f_blendm<mode>): Ditto.
56 (avx512f_store<mode>_mask): Ditto.
57 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
58 (avx512f_storedqu<mode>_mask): Ditto.
59 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>): Ditto.
60 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
61 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
62 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
63 (avx512f_maskcmp<mode>3): Ditto.
64 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
65 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
66 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
67 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
68 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
69 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
70 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
71 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
72 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
73 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
74 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
75 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
76 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
77 (vec_extract_lo_<mode>_maskm): Ditto.
78 (vec_extract_hi_<mode>_maskm): Ditto.
79 (avx512f_vternlog<mode>_mask): Ditto.
80 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
81 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
82 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
83 (avx512f_<code>v8div16qi2_mask): Ditto.
84 (avx512f_<code>v8div16qi2_mask_store): Ditto.
85 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
86 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
87 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
88 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
89 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
90 (*avx512pf_gatherpf<mode>df_mask): Ditto.
91 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
92 (*avx512pf_scatterpf<mode>df_mask): Ditto.
93 (avx512cd_maskb_vec_dupv8di): Ditto.
94 (avx512cd_maskw_vec_dupv16si): Ditto.
95 (avx512f_vpermi2var<mode>3_maskz): Ditto.
96 (avx512f_vpermi2var<mode>3_mask): Ditto.
97 (avx512f_vpermi2var<mode>3_mask): Ditto.
98 (avx512f_vpermt2var<mode>3_maskz): Ditto.
99 (*avx512f_gathersi<mode>): Ditto.
100 (*avx512f_gathersi<mode>_2): Ditto.
101 (*avx512f_gatherdi<mode>): Ditto.
102 (*avx512f_gatherdi<mode>_2): Ditto.
103 (*avx512f_scattersi<mode>): Ditto.
104 (*avx512f_scatterdi<mode>): Ditto.
105 (avx512f_compress<mode>_mask): Ditto.
106 (avx512f_compressstore<mode>_mask): Ditto.
107 (avx512f_expand<mode>_mask): Ditto.
108 * config/i386/subst.md (mask): Change k to Yk.
109 (mask_scalar_merge): Ditto.
112 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
114 * doc/extend.texi (Vector Extensions): Document ?: in C++.
116 2014-01-31 Richard Biener <rguenther@suse.de>
119 * builtins.c (fold_builtin_memory_op): Make sure to not
120 use a floating-point mode or a boolean or enumeral type for
123 2014-01-30 DJ Delorie <dj@redhat.com>
125 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
126 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
127 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
128 whenever main() has an epilogue.
130 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
132 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
133 unused variable "field".
134 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
135 (vsx_mergeh_<mode>): Likewise.
136 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
137 (altivec_vmrghh): Likewise.
138 (altivec_vmrghw): Likewise.
139 (altivec_vmrglb): Likewise.
140 (altivec_vmrglh): Likewise.
141 (altivec_vmrglw): Likewise.
142 (altivec_vspltb): Add missing uses.
143 (altivec_vsplth): Likewise.
144 (altivec_vspltw): Likewise.
145 (altivec_vspltsf): Likewise.
147 2014-01-30 Jakub Jelinek <jakub@redhat.com>
150 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
151 frame related instructions.
153 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
155 PR rtl-optimization/59959
156 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
157 any reload of register whose subreg is invalid.
159 2014-01-30 Jakub Jelinek <jakub@redhat.com>
161 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing
163 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
164 Add missing return type - void.
166 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
168 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
169 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
170 remove element index adjustment for endian (now handled in vsx.md
172 (altivec_expand_vec_perm_const): Use
173 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
174 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
175 (vsx_xxspltw_<mode>): Adjust element index for little endian.
176 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
177 define_expand and a new define_insn *altivec_vspltb_internal;
178 adjust for -maltivec=be on a little endian target.
179 (altivec_vspltb_direct): New.
180 (altivec_vsplth): Divide into a define_expand and a new
181 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
182 little endian target.
183 (altivec_vsplth_direct): New.
184 (altivec_vspltw): Divide into a define_expand and a new
185 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
186 little endian target.
187 (altivec_vspltw_direct): New.
188 (altivec_vspltsf): Divide into a define_expand and a new
189 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
190 a little endian target.
192 2014-01-30 Richard Biener <rguenther@suse.de>
194 PR tree-optimization/59993
195 * tree-ssa-forwprop.c (associate_pointerplus): Check we
196 can propagate form the earlier stmt and avoid the transform
197 when the intermediate result is needed.
199 2014-01-30 Alangi Derick <alangiderick@gmail.com>
201 * README.Portability: Fix typo.
203 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
205 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
206 comparison_operator with ordered_comparison_operator.
208 2014-01-30 Nick Clifton <nickc@redhat.com>
210 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
211 Rename to mn10300_store_multiple_regs.
212 * config/mn10300/mn10300.c: Likewise.
213 * config/mn10300/mn10300.md (store_movm): Fix typo: call
215 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
216 Call mn10300_store_multiple_regs.
218 2014-01-30 Nick Clifton <nickc@redhat.com>
219 DJ Delorie <dj@redhat.com>
221 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
222 %fp 2 to keep registers after it properly word-aligned.
223 (rl78_alloc_physical_registers_umul): Handle the case where both
224 input operands are the same.
226 2014-01-30 Richard Biener <rguenther@suse.de>
228 PR tree-optimization/59903
229 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
232 2014-01-30 Jason Merrill <jason@redhat.com>
235 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
238 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
240 2014-01-30 Richard Biener <rguenther@suse.de>
242 PR tree-optimization/59951
243 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
246 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
249 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
250 SFmode to DFmode case.
252 2014-01-29 DJ Delorie <dj@redhat.com>
254 * config/msp430/msp430.opt (-minrt): New.
255 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
257 (ENDFILE_SPEC): Likewise.
259 2014-01-29 Jan Hubicka <jh@suse.cz>
261 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
262 (estimate_function_body_sizes): Use it.
264 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
267 * dwarf2out.c (is_cxx_auto): New.
268 (is_base_type): Use it.
269 (gen_type_die_with_usage): Likewise.
271 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
273 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
274 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
275 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
276 -maltivec=be with LE targets.
277 (vsx_mergeh_<mode>): Likewise.
278 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
280 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
281 (altivec_vmrghb): Replace with define_expand and new
282 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
284 (altivec_vmrghb_direct): New define_insn.
285 (altivec_vmrghh): Replace with define_expand and new
286 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
288 (altivec_vmrghh_direct): New define_insn.
289 (altivec_vmrghw): Replace with define_expand and new
290 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
292 (altivec_vmrghw_direct): New define_insn.
293 (*altivec_vmrghsf): Adjust for endianness.
294 (altivec_vmrglb): Replace with define_expand and new
295 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
297 (altivec_vmrglb_direct): New define_insn.
298 (altivec_vmrglh): Replace with define_expand and new
299 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
301 (altivec_vmrglh_direct): New define_insn.
302 (altivec_vmrglw): Replace with define_expand and new
303 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
305 (altivec_vmrglw_direct): New define_insn.
306 (*altivec_vmrglsf): Adjust for endianness.
307 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
308 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
309 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
310 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
311 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
312 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
313 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
314 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
316 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
318 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
319 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
322 2014-01-29 Richard Biener <rguenther@suse.de>
324 PR tree-optimization/58742
325 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
326 associate_pointerplus_align.
327 (associate_pointerplus_diff): New function.
328 (associate_pointerplus): Likewise. Call associate_pointerplus_align
329 and associate_pointerplus_diff.
331 2014-01-29 Richard Biener <rguenther@suse.de>
333 * lto-streamer.h (LTO_major_version): Bump to 3.
334 (LTO_minor_version): Reset to 0.
336 2014-01-29 Renlin Li <Renlin.Li@arm.com>
338 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
339 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
340 (arm_file_start): Generate correct asm header for armv7ve.
341 * config/arm/bpabi.h: Add multilib support for armv7ve.
342 * config/arm/driver-arm.c: Change the architectures of cortex-a7
343 and cortex-a15 to armv7ve.
344 * config/arm/t-aprofile: Add multilib support for armv7ve.
345 * doc/invoke.texi: Document -march=armv7ve.
347 2014-01-29 Richard Biener <rguenther@suse.de>
349 PR tree-optimization/58742
350 * tree-ssa-forwprop.c (associate_plusminus): Return true
351 if we changed sth, defer EH cleanup to ...
352 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
353 (simplify_mult): New function.
355 2014-01-29 Jakub Jelinek <jakub@redhat.com>
358 PR tree-optimization/59920
359 * tree.c (build_common_builtin_nodes): Remove
360 __builtin_setjmp_dispatcher initialization.
361 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
362 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
363 instead of gsi_after_labels + manually skipping debug stmts.
364 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
365 ignore bbs with IFN_ABNORMAL_DISPATCHER.
366 * tree-inline.c (copy_edges_for_bb): Remove
367 can_make_abnormal_goto argument, instead add abnormal_goto_dest
368 argument. Ignore computed_goto_p stmts. Don't call
369 make_abnormal_goto_edges. If a call might need abnormal edges
370 for non-local gotos, see if it already has an edge to
371 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
372 with true argument, don't do anything then, otherwise add
373 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
374 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
376 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
377 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
378 (lower_stmt): Don't set data->calls_builtin_setjmp.
379 (lower_builtin_setjmp): Adjust comment.
380 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
381 * tree-cfg.c (found_computed_goto): Remove.
382 (factor_computed_gotos): Remove.
383 (make_goto_expr_edges): Return bool, true for computed gotos.
384 Don't call make_abnormal_goto_edges.
385 (build_gimple_cfg): Don't set found_computed_goto, don't call
386 factor_computed_gotos.
387 (computed_goto_p): No longer static.
388 (make_blocks): Don't set found_computed_goto.
389 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
390 (make_edges): If make_goto_expr_edges returns true, push bb
391 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
392 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
393 vector. Record mapping between bbs and OpenMP regions if there
394 are any, adjust make_gimple_omp_edges caller. Call
395 handle_abnormal_edges.
396 (make_abnormal_goto_edges): Remove.
397 * tree-cfg.h (make_abnormal_goto_edges): Remove.
398 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
399 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
400 * builtins.c (expand_builtin): Don't handle
401 BUILT_IN_SETJMP_DISPATCHER.
402 * internal-fn.def (ABNORMAL_DISPATCHER): New.
403 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
404 filling *region also set *region_idx to (*region)->entry->index.
407 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
408 For REGs set ORIGINAL_REGNO.
410 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
412 * doc/md.texi: Mention that a target shouldn't implement
413 vec_widen_(s|u)mul_even/odd pair if it is less efficient
416 2014-01-29 Jakub Jelinek <jakub@redhat.com>
418 PR tree-optimization/59594
419 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
420 a copy of the datarefs vector rather than the vector itself.
422 2014-01-28 Jason Merrill <jason@redhat.com>
425 * dwarf2out.c (auto_die): New static.
426 (gen_type_die_with_usage): Handle C++1y 'auto'.
427 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
430 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
433 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
434 (SPEC_X32): Likewise.
436 * config/i386/i386.c (ix86_option_override_internal): Turn off
437 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
439 (x86_file_start): Output .code16gcc for TARGET_16BIT.
440 * config/i386/i386.h (TARGET_16BIT): New macro.
441 (TARGET_16BIT_P): Likewise.
442 * config/i386/i386.opt: Add m16.
443 * doc/invoke.texi: Document -m16.
445 2014-01-28 Jakub Jelinek <jakub@redhat.com>
447 PR preprocessor/59935
448 * input.c (location_get_source_line): Bail out on when line number
449 is zero, and test the return value of
450 lookup_or_add_file_to_cache_tab.
452 2014-01-28 Richard Biener <rguenther@suse.de>
454 PR tree-optimization/58742
455 * tree-ssa-forwprop.c (associate_plusminus): Handle
456 pointer subtraction of the form (T)(P + A) - (T)P.
458 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
460 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
463 2014-01-28 Richard Biener <rguenther@suse.de>
466 2014-01-28 Richard Biener <rguenther@suse.de>
468 PR rtl-optimization/45364
469 PR rtl-optimization/59890
470 * var-tracking.c (local_get_addr_clear_given_value): Handle
471 already cleared slot.
472 (val_reset): Handle not allocated local_get_addr_cache.
473 (vt_find_locations): Use post-order on the inverted CFG.
475 2014-01-28 Richard Biener <rguenther@suse.de>
477 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
480 2014-01-28 Richard Biener <rguenther@suse.de>
482 PR rtl-optimization/45364
483 PR rtl-optimization/59890
484 * var-tracking.c (local_get_addr_clear_given_value): Handle
485 already cleared slot.
486 (val_reset): Handle not allocated local_get_addr_cache.
487 (vt_find_locations): Use post-order on the inverted CFG.
489 2014-01-28 Alan Modra <amodra@gmail.com>
491 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
492 * configure.ac <recursive call for build != host>: Define
493 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
494 and LD_FOR_BUILD too.
495 * configure: Regenerate.
497 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
499 * config/i386/i386.c (get_builtin_code_for_version): Separate
500 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
501 Broadwell from Haswell.
503 2014-01-27 Steve Ellcey <sellcey@mips.com>
505 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
506 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
507 * config/mips/mips.c (mips_option_override): Change setting
509 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
510 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
511 Change from Mask to Var.
513 2014-01-27 Jeff Law <law@redhat.com>
515 * ipa-inline.c (inline_small_functions): Fix typo.
517 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
519 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
520 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
521 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
522 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
523 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
524 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
525 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
526 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
527 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
528 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
529 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
530 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
531 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
532 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
533 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
534 (_mm512_storeu_epi64): Ditto.
535 (_mm512_cmpge_epi32_mask): Ditto.
536 (_mm512_cmpge_epu32_mask): Ditto.
537 (_mm512_cmpge_epi64_mask): Ditto.
538 (_mm512_cmpge_epu64_mask): Ditto.
539 (_mm512_cmple_epi32_mask): Ditto.
540 (_mm512_cmple_epu32_mask): Ditto.
541 (_mm512_cmple_epi64_mask): Ditto.
542 (_mm512_cmple_epu64_mask): Ditto.
543 (_mm512_cmplt_epi32_mask): Ditto.
544 (_mm512_cmplt_epu32_mask): Ditto.
545 (_mm512_cmplt_epi64_mask): Ditto.
546 (_mm512_cmplt_epu64_mask): Ditto.
547 (_mm512_cmpneq_epi32_mask): Ditto.
548 (_mm512_cmpneq_epu32_mask): Ditto.
549 (_mm512_cmpneq_epi64_mask): Ditto.
550 (_mm512_cmpneq_epu64_mask): Ditto.
551 (_mm512_expand_pd): Ditto.
552 (_mm512_expand_ps): Ditto.
553 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
554 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
555 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
556 * config/i386/i386.c (ix86_builtins): Add
557 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
558 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
559 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
560 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
561 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
562 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
563 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
564 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
565 IX86_BUILTIN_PMOVUSQW512_MEM.
566 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
567 __builtin_ia32_pmovsqd512mem_mask,
568 __builtin_ia32_pmovqd512mem_mask,
569 __builtin_ia32_pmovusqw512mem_mask,
570 __builtin_ia32_pmovsqw512mem_mask,
571 __builtin_ia32_pmovqw512mem_mask,
572 __builtin_ia32_pmovusdw512mem_mask,
573 __builtin_ia32_pmovsdw512mem_mask,
574 __builtin_ia32_pmovdw512mem_mask,
575 __builtin_ia32_pmovqb512mem_mask,
576 __builtin_ia32_pmovusqb512mem_mask,
577 __builtin_ia32_pmovsqb512mem_mask,
578 __builtin_ia32_pmovusdb512mem_mask,
579 __builtin_ia32_pmovsdb512mem_mask,
580 __builtin_ia32_pmovdb512mem_mask.
581 (bdesc_args): Add __builtin_ia32_expanddf512,
582 __builtin_ia32_expandsf512.
583 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
584 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
585 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
586 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
587 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
588 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
589 (avx512f_<code>v8div16qi2_mask_store): This.
590 (avx512f_expand<mode>): New.
592 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
594 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
596 (_mm512_mask_prefetch_i64gather_pd): Ditto.
597 (_mm512_prefetch_i32scatter_pd): Ditto.
598 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
599 (_mm512_prefetch_i64scatter_pd): Ditto.
600 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
601 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
602 (_mm512_mask_prefetch_i64gather_ps): Ditto.
603 (_mm512_prefetch_i32scatter_ps): Ditto.
604 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
605 (_mm512_prefetch_i64scatter_ps): Ditto.
606 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
607 * config/i386/i386-builtin-types.def: Define
608 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
609 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
610 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
611 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
612 IX86_BUILTIN_SCATTERPFQPD.
613 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
614 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
615 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
616 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
617 __builtin_ia32_scatterpfqps.
618 (ix86_expand_builtin): Expand new built-ins.
619 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
620 fix memory access data type.
621 (*avx512pf_gatherpf<mode>_mask): Ditto.
622 (*avx512pf_gatherpf<mode>): Ditto.
623 (avx512pf_scatterpf<mode>): Ditto.
624 (*avx512pf_scatterpf<mode>_mask): Ditto.
625 (*avx512pf_scatterpf<mode>): Ditto.
626 (GATHER_SCATTER_SF_MEM_MODE): New.
627 (avx512pf_gatherpf<mode>df): Ditto.
628 (*avx512pf_gatherpf<mode>df_mask): Ditto.
629 (*avx512pf_scatterpf<mode>df): Ditto.
631 2014-01-27 Jakub Jelinek <jakub@redhat.com>
634 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
635 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
638 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
640 * common/config/arm/arm-common.c
641 (arm_rewrite_mcpu): Handle multiple names.
643 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
645 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
647 * gimple-builder.h (create_gimple_tmp): Delete.
649 2014-01-27 Christian Bruel <christian.bruel@st.com>
651 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
654 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
656 * config/pa/pa.md (call): Generate indirect long calls to non-local
657 functions when outputing 32-bit code.
658 (call_value): Likewise except for special call to buggy powf function.
660 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
661 portable runtime and PIC indirect calls.
662 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
663 and PIC call sequences. Use ldo instead of blr to set return register
664 in PIC call sequence.
666 2014-01-25 Walter Lee <walt@tilera.com>
668 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
669 avoid clobbering a live register.
671 2014-01-25 Walter Lee <walt@tilera.com>
673 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
674 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
675 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
676 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
678 2014-01-25 Walter Lee <walt@tilera.com>
680 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
681 arguments on even registers.
682 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
684 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
685 (BIGGEST_ALIGNMENT): Ditto.
686 (BIGGEST_FIELD_ALIGNMENT): Ditto.
688 2014-01-25 Walter Lee <walt@tilera.com>
690 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
691 insns before bundling.
692 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
694 2014-01-25 Walter Lee <walt@tilera.com>
696 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
697 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
698 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
700 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
702 * config/mips/constraints.md (kl): Delete.
703 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
704 define expands, using...
705 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
706 instructions for MIPS16.
707 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
708 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
710 2014-01-25 Walter Lee <walt@tilera.com>
712 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
716 2014-01-25 Walter Lee <walt@tilera.com>
718 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
719 (TARGET_EXPAND_TO_RTL_HOOK): Define.
721 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
723 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
726 2014-01-25 Jakub Jelinek <jakub@redhat.com>
728 * print-rtl.c (in_call_function_usage): New var.
729 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
730 EXPR_LIST mode as mode and not as reg note name.
733 * cfgloopmanip.c (copy_loop_info): If
734 loop->warned_aggressive_loop_optimizations, make sure
735 the flag is set in target loop too.
737 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
739 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
741 * builtins.def: Likewise.
742 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
743 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
744 * ira.c (ira_setup_eliminable_regset): Likewise.
745 * omp-low.c (gate_expand_omp): Likewise.
746 (execute_lower_omp): Likewise.
747 (diagnose_sb_0): Likewise.
748 (gate_diagnose_omp_blocks): Likewise.
749 (simd_clone_clauses_extract): Likewise.
752 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
754 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
755 correction for little endian...
756 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
759 2014-01-24 Jeff Law <law@redhat.com>
761 PR tree-optimization/59919
762 * tree-vrp.c (find_assert_locations_1): Do not register asserts
763 for non-returning calls.
765 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
767 * common/config/aarch64/aarch64-common.c
768 (aarch64_rewrite_mcpu): Handle multiple names.
769 * config/aarch64/aarch64.h
770 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
772 2014-01-24 Dodji Seketeli <dodji@redhat.com>
774 * input.c (add_file_to_cache_tab): Handle the case where fopen
777 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
780 * config/i386/i386.md (pushsf splitter): Get stack adjustment
781 from push operand if code of push isn't PRE_DEC.
783 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
786 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
787 -mquad-memory-atomic. Update -mquad-memory documentation to say
788 it is only used for non-atomic loads/stores.
790 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
791 -mquad-memory or -mquad-memory-atomic switches.
793 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
794 -mquad-memory-atomic to ISA 2.07 support.
796 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
797 to separate support of normal quad word memory operations (ldq, stq)
798 from the atomic quad word memory operations.
800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
801 support to separate non-atomic quad word operations from atomic
802 quad word operations. Disable non-atomic quad word operations in
803 little endian mode so that we don't have to swap words after the
804 load and before the store.
805 (quad_load_store_p): Add comment about atomic quad word support.
806 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
807 options printed with -mdebug=reg.
809 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
810 -mquad-memory-atomic as the test for whether we have quad word
812 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
813 or -mp8-vector are used, allow byte/half-word atomic operations.
815 * config/rs6000/sync.md (load_lockedti): Insure that the address
816 is a proper indexed or indirect address for the lqarx instruction.
817 On little endian systems, swap the hi/lo registers after the lqarx
819 (load_lockedpti): Use indexed_or_indirect_operand predicate to
820 insure the address is valid for the lqarx instruction.
821 (store_conditionalti): Insure that the address is a proper indexed
822 or indirect address for the stqcrx. instruction. On little endian
823 systems, swap the hi/lo registers before doing the stqcrx.
825 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
826 insure the address is valid for the stqcrx. instruction.
828 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
829 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
830 type of quad memory support is available.
832 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
835 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
836 there is a danger of looping.
838 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
841 force flag_ira_loop_pressure if set via command line.
843 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
845 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
846 (ashr_simd): New builtin handling DI mode.
847 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
848 (aarch64_sshr_simddi): New match pattern.
849 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
850 (vshrd_n_s64): Likewise.
851 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
853 2014-01-23 Nick Clifton <nickc@redhat.com>
855 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
856 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
857 favour of mcu specific scripts.
858 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
861 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
862 Alex Velenko <Alex.Velenko@arm.com>
864 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
865 (vaddv_s16): Likewise.
866 (vaddv_s32): Likewise.
867 (vaddv_u8): Likewise.
868 (vaddv_u16): Likewise.
869 (vaddv_u32): Likewise.
870 (vaddvq_s8): Likewise.
871 (vaddvq_s16): Likewise.
872 (vaddvq_s32): Likewise.
873 (vaddvq_s64): Likewise.
874 (vaddvq_u8): Likewise.
875 (vaddvq_u16): Likewise.
876 (vaddvq_u32): Likewise.
877 (vaddvq_u64): Likewise.
878 (vaddv_f32): Likewise.
879 (vaddvq_f32): Likewise.
880 (vaddvq_f64): Likewise.
881 (vmaxv_f32): Likewise.
882 (vmaxv_s8): Likewise.
883 (vmaxv_s16): Likewise.
884 (vmaxv_s32): Likewise.
885 (vmaxv_u8): Likewise.
886 (vmaxv_u16): Likewise.
887 (vmaxv_u32): Likewise.
888 (vmaxvq_f32): Likewise.
889 (vmaxvq_f64): Likewise.
890 (vmaxvq_s8): Likewise.
891 (vmaxvq_s16): Likewise.
892 (vmaxvq_s32): Likewise.
893 (vmaxvq_u8): Likewise.
894 (vmaxvq_u16): Likewise.
895 (vmaxvq_u32): Likewise.
896 (vmaxnmv_f32): Likewise.
897 (vmaxnmvq_f32): Likewise.
898 (vmaxnmvq_f64): Likewise.
899 (vminv_f32): Likewise.
900 (vminv_s8): Likewise.
901 (vminv_s16): Likewise.
902 (vminv_s32): Likewise.
903 (vminv_u8): Likewise.
904 (vminv_u16): Likewise.
905 (vminv_u32): Likewise.
906 (vminvq_f32): Likewise.
907 (vminvq_f64): Likewise.
908 (vminvq_s8): Likewise.
909 (vminvq_s16): Likewise.
910 (vminvq_s32): Likewise.
911 (vminvq_u8): Likewise.
912 (vminvq_u16): Likewise.
913 (vminvq_u32): Likewise.
914 (vminnmv_f32): Likewise.
915 (vminnmvq_f32): Likewise.
916 (vminnmvq_f64): Likewise.
918 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
920 * config/aarch64/aarch64-simd.md
921 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
922 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
923 (*aarch64_mul3_elt<mode>): Likewise.
924 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
925 (*aarch64_mul3_elt_to_64v2df): Likewise.
926 (*aarch64_mla_elt<mode>): Likewise.
927 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
928 (*aarch64_mls_elt<mode>): Likewise.
929 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
930 (*aarch64_fma4_elt<mode>): Likewise.
931 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
932 (*aarch64_fma4_elt_to_64v2df): Likewise.
933 (*aarch64_fnma4_elt<mode>): Likewise.
934 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
935 (*aarch64_fnma4_elt_to_64v2df): Likewise.
936 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
937 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
938 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
939 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
940 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
941 (aarch64_sqdmull_lane<mode>_internal): Likewise.
942 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
944 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
946 * config/aarch64/aarch64-simd.md
947 (aarch64_be_checked_get_lane<mode>): New define_expand.
948 * config/aarch64/aarch64-simd-builtins.def
949 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
950 New builtin definition.
951 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
952 Use new safe be builtin.
954 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
956 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
958 (aarch64_be_st1<mode>): Likewise.
959 (aarch_ld1<VALL:mode>): Define_expand modified.
960 (aarch_st1<VALL:mode>): Likewise.
961 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
962 (UNSPEC_ST1): Likewise.
964 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
966 * config/microblaze/microblaze.md: Add trap insn and attribute
968 2014-01-23 Dodji Seketeli <dodji@redhat.com>
970 PR preprocessor/58580
971 * input.h (location_get_source_line): Take an additional line_size
973 (void diagnostics_file_cache_fini): Declare new function.
974 * input.c (struct fcache): New type.
975 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
976 New static constants.
977 (diagnostic_file_cache_init, total_lines_num)
978 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
979 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
980 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
981 (get_next_line, read_next_line, goto_next_line, read_line_num):
982 New static function definitions.
983 (diagnostic_file_cache_fini): New function.
984 (location_get_source_line): Take an additional output line_len
985 parameter. Re-write using lookup_or_add_file_to_cache_tab and
987 * diagnostic.c (diagnostic_finish): Call
988 diagnostic_file_cache_fini.
989 (adjust_line): Take an additional input parameter for the length
990 of the line, rather than calculating it with strlen.
991 (diagnostic_show_locus): Adjust the use of
992 location_get_source_line and adjust_line with respect to their new
993 signature. While displaying a line now, do not stop at the first
994 null byte. Rather, display the zero byte as a space and keep
995 going until we reach the size of the line.
996 * Makefile.in: Add vec.o to OBJS-libcommon
998 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
999 Ilya Tocar <ilya.tocar@intel.com>
1001 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1002 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1003 (__builtin_ia32_kmov16): Ditto.
1004 * config/i386/i386.md (UNSPEC_KMOV): New.
1007 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1009 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1010 (_mm512_storeu_si512): Ditto.
1012 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1015 * rtl.h (get_referenced_operands): Declare.
1016 * recog.c (get_referenced_operands): New function.
1017 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1018 operands have been referenced when recording LO_SUM references.
1020 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1022 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1024 2014-01-22 Jan Hubicka <jh@suse.cz>
1026 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1027 Enable for generic and recent AMD targets.
1029 2014-01-22 Jan Hubicka <jh@suse.cz>
1031 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1032 ARG_SIZE note when adjustment was eliminated.
1034 2014-01-22 Jeff Law <law@redhat.com>
1036 PR tree-optimization/59597
1037 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1038 in file. Accept new argument REGISTERING and use it to modify
1039 dump output appropriately.
1040 (register_jump_thread): Corresponding changes.
1041 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1042 thread paths involving joiner blocks. Add code to dump cancelled
1043 jump threading paths.
1045 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1047 PR rtl-optimization/59477
1048 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1049 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1051 2014-01-22 Tom Tromey <tromey@redhat.com>
1053 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1055 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1057 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1059 PR rtl-optimization/59896
1060 * lra-constraints.c (process_alt_operands): Check unused note for
1061 matched operands of insn with no output reloads.
1063 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1065 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1066 (mips_move_from_gpr_cost): Likewise.
1068 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1070 PR rtl-optimization/59858
1071 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1072 ira_class_hard_regs_num.
1073 (process_alt_operands): Increase reject for dying matched operand.
1075 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1078 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1079 smaller than size, perform several stores or loads and stores
1080 at dst + count - size to store or copy all of size bytes, rather
1081 than just last modesize bytes.
1083 2014-01-20 DJ Delorie <dj@redhat.com>
1085 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1086 that CLOBBERs are REGs before propogating their values.
1088 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1091 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1092 (cgraph_inline_failed_type): New function.
1093 * cgraph.h (DEFCIFCODE): Add type.
1094 (cgraph_inline_failed_type_t): New enum.
1095 (cgraph_inline_failed_type): New prototype.
1096 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1097 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1098 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1099 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1100 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1101 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1102 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1103 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1104 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1105 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1106 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1107 OPTIMIZATION_MISMATCH.
1108 * tree-inline.c (expand_call_inline): Emit errors during
1109 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1111 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1114 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1115 mode attribute in insn output.
1117 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1119 * output.h (output_constant): Delete.
1120 * varasm.c (output_constant): Make private.
1122 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1124 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1126 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1129 * tree.h (fold_builtin_strcat): New prototype.
1130 * builtins.c (fold_builtin_strcat): No longer static. Add len
1131 argument, if non-NULL, don't call c_strlen. Optimize
1132 directly into __builtin_memcpy instead of __builtin_strcpy.
1133 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1134 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1136 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1138 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1139 for SImode_address_operand operands, having only a REG argument.
1141 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1143 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1144 loader name using mbig-endian.
1145 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1147 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1149 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1153 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1155 * config/aarch64/aarch64-protos.h
1156 (aarch64_cannot_change_mode_class_ptr): Declare.
1157 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1158 aarch64_cannot_change_mode_class_ptr): New.
1159 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1160 backend hook aarch64_cannot_change_mode_class.
1162 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1164 * common/config/aarch64/aarch64-common.c
1165 (aarch64_handle_option): Don't handle any option order logic here.
1166 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1167 selected_cpu, warn on architecture version mismatch.
1168 (aarch64_override_options): Fix parsing order for option strings.
1170 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1171 Iain Sandoe <iain@codesourcery.com>
1174 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1175 warning. Amend comment to reflect current functionality.
1177 2014-01-20 Richard Biener <rguenther@suse.de>
1180 * builtins.c (fold_builtin_strcat): Remove case better handled
1181 by tree-ssa-strlen.c.
1183 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1185 * config/aarch64/aarch64.opt
1186 (mcpu, march, mtune): Make case-insensitive.
1188 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1191 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1192 if operands[1] is a REG or ZERO_EXTEND of a REG.
1194 2014-01-19 Jan Hubicka <jh@suse.cz>
1196 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1198 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1200 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1201 long non-pic millicode calls.
1203 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1205 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1207 2014-01-19 Kito Cheng <kito@0xlab.org>
1209 * builtins.c (expand_movstr): Check movstr expand done or fail.
1211 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1212 H.J. Lu <hongjiu.lu@intel.com>
1215 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1216 to DImode for zero-extended addresses.
1218 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1220 PR rtl-optimization/57763
1221 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1222 on the new indirect jump_insn and increment LABEL_NUSES (label).
1224 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1228 * config.gcc (x86_archs): New variable.
1229 (x86_64_archs): Likewise.
1230 (x86_cpus): Likewise.
1231 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1232 --with-arch/--with-cpu= options.
1233 Support --with-arch=/--with-cpu={nehalem,westmere,
1234 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1236 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1238 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1239 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1241 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1243 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1245 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1248 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1249 clear cpp_get_options (parse_in)->warn_unused_macros for
1250 ix86_target_macros_internal with cpp_define.
1252 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1254 * jump.c (delete_related_insns): Keep (use (insn))s.
1255 * reorg.c (redundant_insn): Check for barriers too.
1257 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1259 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1261 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1263 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1264 call to $$dyncall when TARGET_LONG_CALLS is true.
1266 2014-01-17 Jeff Law <law@redhat.com>
1268 * ree.c (combine_set_extension): Temporarily disable test for
1269 changing number of hard registers.
1271 2014-01-17 Jan Hubicka <jh@suse.cz>
1274 * ipa-inline-analysis.c (inline_free_summary):
1275 Do not free summary of aliases.
1277 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1280 * gimplify.c (gimplify_expr): Use create_tmp_var
1281 instead of create_tmp_var_raw. If cond doesn't have
1282 integral type, don't add the IFN_ANNOTATE builtin at all.
1284 2014-01-17 Martin Jambor <mjambor@suse.cz>
1287 * ipa-cp.c (prev_edge_clone): New variable.
1288 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1289 Also resize prev_edge_clone vector.
1290 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1291 (ipcp_edge_removal_hook): New function.
1292 (ipcp_driver): Register ipcp_edge_removal_hook.
1294 2014-01-17 Andrew Pinski <apinski@cavium.com>
1295 Steve Ellcey <sellcey@mips.com>
1298 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1301 2014-01-17 Jeff Law <law@redhat.com>
1304 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1305 so that pass_ccp runs first.
1307 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1309 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1310 (ix86_adjust_cost): Use !TARGET_XXX.
1311 (do_reorder_for_imul): Likewise.
1312 (swap_top_of_ready_list): Likewise.
1313 (ix86_sched_reorder): Likewise.
1315 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1317 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1318 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1319 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1320 (intel_memset): New. Duplicate slm_memset.
1321 (intel_cost): New. Duplicate slm_cost.
1322 (m_INTEL): New macro.
1323 (processor_target_table): Add "intel".
1324 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1325 with PROCESSOR_INTEL for "intel".
1326 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1327 PROCESSOR_SILVERMONT.
1328 (ix86_issue_rate): Likewise.
1329 (ix86_adjust_cost): Likewise.
1330 (ia32_multipass_dfa_lookahead): Likewise.
1331 (swap_top_of_ready_list): Likewise.
1332 (ix86_sched_reorder): Likewise.
1333 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1334 instead of TARGET_OPT_AGU.
1335 * config/i386/i386.h (TARGET_INTEL): New.
1336 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1337 (processor_type): Add PROCESSOR_INTEL.
1338 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1339 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1341 2014-01-17 Marek Polacek <polacek@redhat.com>
1344 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1347 2014-01-17 Richard Biener <rguenther@suse.de>
1349 PR tree-optimization/46590
1350 * opts.c (default_options_table): Add entries for
1351 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1352 all enabled at -O1 but not for -Og.
1353 * common.opt (fbranch-count-reg): Remove Init(1).
1354 (fmove-loop-invariants): Likewise.
1355 (ftree-pta): Likewise.
1357 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1359 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1360 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1361 decls to at least the GCC 4.8 used alignments.
1364 * tree-nested.c (convert_nonlocal_reference_stmt,
1365 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1366 of GIMPLE_BIND stmts, adjust associated decls.
1368 2014-01-17 Richard Biener <rguenther@suse.de>
1370 PR tree-optimization/46590
1371 * vec.h (vec<>::bseach): New member function implementing
1372 binary search according to C89 bsearch.
1373 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1374 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1375 bitmap pointer again. Make accesses_in_loop a flat array.
1376 (mem_ref_obstack): New global.
1377 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1378 (mark_ref_stored): Likewise.
1379 (ref_indep_loop_p_2): Likewise.
1380 (set_ref_stored_in_loop): New helper function.
1381 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1382 (memref_free): Adjust.
1383 (record_mem_ref_loc): Simplify.
1384 (gather_mem_refs_stmt): Adjust.
1385 (sort_locs_in_loop_postorder_cmp): New function.
1386 (analyze_memory_references): Sort accesses_in_loop after
1387 loop postorder number.
1388 (find_ref_loc_in_loop_cmp): New function.
1389 (for_all_locs_in_loop): Find relevant cluster of locs in
1390 accesses_in_loop and iterate without recursion.
1391 (execute_sm): Avoid uninit warning.
1392 (struct ref_always_accessed): Simplify.
1393 (ref_always_accessed::operator ()): Likewise.
1394 (ref_always_accessed_p): Likewise.
1395 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1396 loop postorder numbers here.
1397 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1400 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1403 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1404 on decls for which assemble_alias has been called.
1406 2014-01-17 Nick Clifton <nickc@redhat.com>
1408 * config/msp430/msp430.opt: (mcpu): New option.
1409 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1410 (msp430_option_override): Parse target_cpu. If the MCU name
1411 matches a generic string, clear target_mcu.
1412 (msp430_attr): Allow numeric interrupt values up to 63.
1413 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1414 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1416 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1418 * config/msp430/msp430.md (popm): Use %J rather than %I.
1419 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1420 (addhi_cy_i): Use immediate_operand for operand 2.
1421 * doc/invoke.texi: Document -mcpu option.
1423 2014-01-17 Richard Biener <rguenther@suse.de>
1425 PR rtl-optimization/38518
1426 * df.h (df_analyze_loop): Declare.
1427 * df-core.c: Include cfgloop.h.
1428 (df_analyze_1): Split out main part of df_analyze.
1429 (df_analyze): Adjust.
1430 (loop_inverted_post_order_compute): New function.
1431 (loop_post_order_compute): Likewise.
1432 (df_analyze_loop): New function avoiding whole-function
1434 * loop-invariant.c (find_defs): Use df_analyze_loop.
1435 (find_invariants): Adjust.
1436 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1438 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1440 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1441 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1443 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1445 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1446 traversal when removing references.
1448 2014-01-16 Jan Hubicka <jh@suse.cz>
1451 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1453 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1456 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1457 pushing a reload for an autoinc when we had previously reloaded an
1458 inner part of the address.
1460 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1462 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1464 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1465 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1466 when not giving up or versioning for alias only because of
1468 (vect_analyze_data_ref_dependences): Set to true.
1469 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1471 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1472 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1476 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1479 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1480 operand 0 predicate for gathers, use a new pseudo as subtarget.
1482 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1485 * lra-constraints.c (process_alt_operands): Add printing debug info.
1486 Check absence of input/output reloads for matched operands too.
1488 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1490 PR rtl-optimization/59835
1491 * ira.c (ira_init_register_move_cost): Increase cost for
1494 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1496 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1498 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1501 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1502 non-register objects. Use gen_(high/low)part more consistently.
1505 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1508 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1509 endian support, remove tests for WORDS_BIG_ENDIAN.
1510 (p8_mfvsrd_3_<mode>): Likewise.
1511 (reload_gpr_from_vsx<mode>): Likewise.
1512 (reload_gpr_from_vsxsf): Likewise.
1513 (p8_mfvsrd_4_disf): Likewise.
1515 2014-01-16 Richard Biener <rguenther@suse.de>
1517 PR rtl-optimization/46590
1518 * lcm.c (compute_antinout_edge): Use postorder iteration.
1519 (compute_laterin): Use inverted postorder iteration.
1521 2014-01-16 Nick Clifton <nickc@redhat.com>
1524 * varasm.c (output_constant): Return the number of bytes actually
1526 (output_constructor_array_range): Update the field size with the
1527 number of bytes emitted by output_constant.
1528 (output_constructor_regular_field): Likewise. Also do not
1529 complain if the total number of bytes emitted is now greater
1530 than the expected fieldpos.
1531 * output.h (output_constant): Update prototype and descriptive comment.
1533 2014-01-16 Marek Polacek <polacek@redhat.com>
1536 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1537 it is error_mark_node.
1539 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1541 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1542 VALID_AVX256_REG_OR_OI_MODE.
1544 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1546 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1547 current procedure should be profiled.
1549 2014-01-15 Andrew Pinski <apinski@cavium.com>
1551 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1552 of moving from/to the STACK_REG register class.
1554 2014-01-15 Richard Henderson <rth@redhat.com>
1557 * reginfo.c (global_regs_decl): Globalize.
1558 * rtl.h (global_regs_decl): Declare.
1559 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1560 reserved via global_regs.
1562 2014-01-15 Teresa Johnson <tejohnson@google.com>
1564 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1566 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1568 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1569 and vmulosh rather than call gen_vec_widen_smult_*.
1570 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1571 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1572 (vec_widen_smult_even_v16qi): Likewise.
1573 (vec_widen_umult_even_v8hi): Likewise.
1574 (vec_widen_smult_even_v8hi): Likewise.
1575 (vec_widen_umult_odd_v16qi): Likewise.
1576 (vec_widen_smult_odd_v16qi): Likewise.
1577 (vec_widen_umult_odd_v8hi): Likewise.
1578 (vec_widen_smult_odd_v8hi): Likewise.
1579 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1580 vmuloub rather than call gen_vec_widen_umult_*.
1581 (vec_widen_umult_lo_v16qi): Likewise.
1582 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1583 vmulosb rather than call gen_vec_widen_smult_*.
1584 (vec_widen_smult_lo_v16qi): Likewise.
1585 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1586 rather than call gen_vec_widen_umult_*.
1587 (vec_widen_umult_lo_v8hi): Likewise.
1588 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1589 rather than call gen_vec_widen_smult_*.
1590 (vec_widen_smult_lo_v8hi): Likewise.
1592 2014-01-15 Jeff Law <law@redhat.com>
1594 PR tree-optimization/59747
1595 * ree.c (find_and_remove_re): Properly handle case where a second
1596 eliminated extension requires widening a copy created for elimination
1597 of a prior extension.
1598 (combine_set_extension): Ensure that the number of hard regs needed
1599 for a destination register does not change when we widen it.
1601 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1603 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1604 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1605 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1606 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1607 (avr-*-rtems*): Likewise.
1608 (bfin*-rtems*): Likewise.
1609 (moxie-*-rtems*): Likewise.
1610 (h8300-*-rtems*): Likewise.
1611 (i[34567]86-*-rtems*): Likewise.
1612 (lm32-*-rtems*): Likewise.
1613 (m32r-*-rtems*): Likewise.
1614 (m68k-*-rtems*): Likewise.
1615 (microblaze*-*-rtems*): Likewise.
1616 (mips*-*-rtems*): Likewise.
1617 (powerpc-*-rtems*): Likewise.
1618 (sh-*-rtems*): Likewise.
1619 (sparc-*-rtems*): Likewise.
1620 (sparc64-*-rtems*): Likewise.
1621 (v850-*-rtems*): Likewise.
1622 (m32c-*-rtems*): Likewise.
1624 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1626 PR rtl-optimization/59511
1627 * ira.c (ira_init_register_move_cost): Use memory costs for some
1628 cases of register move cost calculations.
1629 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1630 instead of BB frequency.
1631 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1632 * lra-assigns.c (find_hard_regno_for): Ditto.
1634 2014-01-15 Richard Biener <rguenther@suse.de>
1636 PR tree-optimization/59822
1637 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1638 (vectorizable_load): Use it to hoist defs of uses of invariant
1639 loads out of the loop.
1641 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1642 Kugan Vivekanandarajah <kuganv@linaro.org>
1645 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1648 2014-01-15 Richard Biener <rguenther@suse.de>
1650 PR rtl-optimization/59802
1651 * lcm.c (compute_available): Use inverted postorder to seed
1652 the initial worklist.
1654 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1657 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1658 ADDR_REGS for invalid symrefs in non-PIC code.
1660 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1663 * builtins.c (determine_block_size): Initialize *probable_max_size
1664 even if len_rtx is CONST_INT.
1666 2014-01-14 Andrew Pinski <apinski@cavium.com>
1668 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1669 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1670 (cortexa53_tunings): Likewise.
1671 (aarch64_sched_issue_rate): New function.
1672 (TARGET_SCHED_ISSUE_RATE): Define.
1674 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1676 * ira-costs.c (find_costs_and_classes): Add missed
1677 ira_init_register_move_cost_if_necessary.
1679 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1682 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1684 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1687 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1688 to indicate if type is used for function return value. Warn ABI
1689 change if the vector mode isn't available for function return value.
1690 (ix86_function_arg_advance): Pass false to type_natural_mode.
1691 (ix86_function_arg): Likewise.
1692 (ix86_gimplify_va_arg): Likewise.
1693 (function_arg_32): Don't warn ABI change.
1694 (ix86_function_value): Pass true to type_natural_mode.
1695 (ix86_return_in_memory): Likewise.
1696 (ix86_struct_value_rtx): Removed.
1697 (TARGET_STRUCT_VALUE_RTX): Likewise.
1699 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1701 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1702 converting a conditional jump into a conditional return.
1704 2014-01-14 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/58921
1707 PR tree-optimization/59006
1708 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1709 hoisting invariant stmts.
1710 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1711 invariant loads on the preheader edge if possible.
1713 2014-01-14 Joey Ye <joey.ye@arm.com>
1715 * doc/plugin.texi (Building GCC plugins): Update to C++.
1717 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1719 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1720 (_mm_rcp28_round_ss): Ditto.
1721 (_mm_rsqrt28_round_sd): Ditto.
1722 (_mm_rsqrt28_round_ss): Ditto.
1723 (_mm_rcp28_sd): Ditto.
1724 (_mm_rcp28_ss): Ditto.
1725 (_mm_rsqrt28_sd): Ditto.
1726 (_mm_rsqrt28_ss): Ditto.
1727 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1728 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1729 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1730 (IX86_BUILTIN_RCP28SD): Ditto.
1731 (IX86_BUILTIN_RCP28SS): Ditto.
1732 (IX86_BUILTIN_RSQRT28SD): Ditto.
1733 (IX86_BUILTIN_RSQRT28SS): Ditto.
1734 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1735 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1736 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1737 (ix86_expand_special_args_builtin): Expand new FTYPE.
1738 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1739 (srcp14<mode>): Make insn unary.
1740 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1741 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1742 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1743 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1744 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1745 Fix rounding: make it SAE only.
1746 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1748 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1750 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1751 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1752 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1753 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1754 (round_saeonly_mask_scalar_operand4): Ditto.
1755 (round_saeonly_mask_scalar_op3): Ditto.
1756 (round_saeonly_mask_scalar_op4): Ditto.
1758 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1760 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1761 Implement -maltivec=be for vec_insert and vec_extract.
1763 2014-01-10 DJ Delorie <dj@redhat.com>
1765 * config/msp430/msp430.md (call_internal): Don't allow memory
1766 references with SP as the base register.
1767 (call_value_internal): Likewise.
1768 * config/msp430/constraints.md (Yc): New. For memory references
1769 that don't use SP as a base register.
1771 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1772 "an integer without a # prefix"
1773 * config/msp430/msp430.md (epilogue_helper): Use it.
1775 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1778 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1779 AVX512F gather builtins.
1780 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1781 on gather decls with INTEGER_TYPE masktype.
1782 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1783 directly into the builtin rather than hoisting it before loop.
1785 PR tree-optimization/59387
1786 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1787 (scev_const_prop): If folded_casts and type has undefined overflow,
1788 use force_gimple_operand instead of force_gimple_operand_gsi and
1789 for each added stmt if it is assign with
1790 arith_code_with_undefined_signed_overflow, call
1791 rewrite_to_defined_overflow.
1792 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1793 gimple-fold.h instead.
1794 (arith_code_with_undefined_signed_overflow,
1795 rewrite_to_defined_overflow): Moved to ...
1796 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1797 rewrite_to_defined_overflow): ... here. No longer static.
1798 Include gimplify-me.h.
1799 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1800 rewrite_to_defined_overflow): New prototypes.
1802 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1804 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1806 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1808 * builtins.c (get_object_alignment_2): Minor tweak.
1809 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1811 2014-01-13 Christian Bruel <christian.bruel@st.com>
1813 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1814 optimized non constant lengths.
1816 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1819 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1820 load as __atomic_load_N if possible.
1822 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1824 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1826 (rs6000_expand_builtin): Adjust call.
1828 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1831 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1832 * config/rs6000/rs6000.c: Include target-globals.h.
1833 (rs6000_set_current_function): Instead of doing target_reinit
1834 unconditionally, use save_target_globals_default_opts and
1835 restore_target_globals.
1837 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1839 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1840 (rs6000_expand_builtin): Handle mffs and mtfsf.
1841 (rs6000_init_builtins): Define mffs and mtfsf.
1842 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1843 (rs6000_mffs): New pattern.
1844 (rs6000_mtfsf): New pattern.
1846 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1848 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1849 Start narrowing with START. Apply candidate-use pair
1850 and check overall cost in narrowing.
1851 (iv_ca_prune): Pass new argument.
1853 2014-01-10 Jeff Law <law@redhat.com>
1856 * ree.c (combine_reaching_defs): Ensure the defining statement
1857 occurs before the extension when optimizing extensions with
1858 different source and destination hard registers.
1860 2014-01-10 Jan Hubicka <jh@suse.cz>
1863 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1864 vtables into the type inheritance graph.
1866 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1868 PR rtl-optimization/59754
1869 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1870 modes in the REGNO != REGNO case.
1872 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1874 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1876 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1878 PR tree-optimization/59745
1879 * tree-predcom.c (tree_predictive_commoning_loop): Call
1880 free_affine_expand_cache if giving up because components is NULL.
1882 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1883 GC in payload of target_globals struct instead of allocating them on
1884 the heap and the larger structs separately using GC.
1885 * target-globals.h (struct target_globals): Make regs, hard_regs,
1886 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1887 of GTY((skip)) and change type to void *.
1888 (reset_target_globals): Cast loads from those fields to corresponding
1891 2014-01-10 Steve Ellcey <sellcey@mips.com>
1894 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1895 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1896 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1898 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1901 * aarch64-modes.def (CC_Zmode): New flags mode.
1902 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1903 represents an equality.
1904 (aarch64_get_condition_code): Handle CC_Zmode.
1905 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1907 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1909 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1910 extraction in good case.
1912 2014-01-10 Richard Biener <rguenther@suse.de>
1914 PR tree-optimization/59374
1915 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1916 checking after SLP discovery. Mark stmts not participating
1917 in any SLP instance properly.
1919 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1921 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1922 when handling a SET rtx.
1924 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1926 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1927 (cortex-a57): Likewise.
1928 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1930 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1932 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1933 non-iwmmxt builtins.
1935 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1939 * ipa-devirt.c record_target_from_binfo): Take as argument
1940 stack of binfos and lookup matching one for virtual inheritance.
1941 (possible_polymorphic_call_targets_1): Update.
1943 2014-01-10 Huacai Chen <chenhc@lemote.com>
1945 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1946 kernel strings for Loongson-2E/2F/3A.
1948 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1951 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1952 is_gimple_call before calling gimple_call_internal_p.
1954 2014-01-09 Steve Ellcey <sellcey@mips.com>
1956 * Makefile.in (TREE_FLOW_H): Remove.
1957 (TREE_SSA_H): Add file names from tree-flow.h.
1958 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1959 * tree.h: Remove tree-flow.h reference.
1960 * hash-table.h: Remove tree-flow.h reference.
1961 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1962 reference with tree-ssa-loop.h.
1964 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1966 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1967 default element-order behavior for -maltivec.
1968 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1969 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1970 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1971 when targeting big endian, at least for now.
1972 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1974 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1977 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1978 var satisfies use_register_for_decl, just take into account type
1979 alignment, rather than decl alignment.
1981 PR tree-optimization/59622
1982 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1983 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1984 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1985 Don't devirtualize for inplace at all. For targets.length () == 1,
1986 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1988 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1990 * config/i386/i386.md (cpu): Remove the unused btver1.
1992 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1994 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1996 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1999 * tree-core.h (struct target_globals): New forward declaration.
2000 (struct tree_target_option): Add globals field.
2001 * tree.h (TREE_TARGET_GLOBALS): Define.
2002 (prepare_target_option_nodes_for_pch): New prototype.
2003 * target-globals.h (struct target_globals): Define even if
2005 * tree.c (prepare_target_option_node_for_pch,
2006 prepare_target_option_nodes_for_pch): New functions.
2007 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2008 * config/i386/i386.c: Include target-globals.h.
2009 (ix86_set_current_function): Instead of doing target_reinit
2010 unconditionally, use save_target_globals_default_opts and
2011 restore_target_globals.
2013 2014-01-09 Richard Biener <rguenther@suse.de>
2015 PR tree-optimization/59715
2016 * tree-cfg.h (split_critical_edges): Declare.
2017 * tree-cfg.c (split_critical_edges): Export.
2018 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2020 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2022 * cfgexpand.c (expand_stack_vars): Optionally disable
2023 asan stack protection.
2024 (expand_used_vars): Likewise.
2025 (partition_stack_vars): Likewise.
2026 * asan.c (asan_emit_stack_protection): Optionally disable
2027 after return stack usage.
2028 (instrument_derefs): Optionally disable memory access instrumentation.
2029 (instrument_builtin_call): Likewise.
2030 (instrument_strlen_call): Likewise.
2031 (asan_protect_global): Optionally disable global variables protection.
2032 * doc/invoke.texi: Added doc for new options.
2033 * params.def: Added new options.
2034 * params.h: Likewise.
2036 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2038 PR rtl-optimization/59724
2039 * ifcvt.c (cond_exec_process_if_block): Don't call
2040 flow_find_head_matching_sequence with 0 longest_match.
2041 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2042 non-active insns if !stop_after.
2043 (try_head_merge_bb): Revert 2014-01-07 changes.
2045 2014-01-08 Jeff Law <law@redhat.com>
2047 * ree.c (get_sub_rtx): New function, extracted from...
2048 (merge_def_and_ext): Here.
2049 (combine_reaching_defs): Use get_sub_rtx.
2051 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2053 * cgraph.h (varpool_variable_node): Do not choke on null node.
2055 2014-01-08 Catherine Moore <clm@codesourcery.com>
2057 * config/mips/mips.md (simple_return): Attempt to use JRC
2059 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2061 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2063 PR rtl-optimization/59137
2064 * reorg.c (steal_delay_list_from_target): Call update_block for
2066 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2068 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2070 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2071 two duplicate entries.
2073 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2076 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2078 * config/mips/mips.c (mips_truncated_op_cost): New function.
2079 (mips_rtx_costs): Adjust test for BADDU.
2080 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2082 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2084 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2085 (*baddu_si): ...this new pattern.
2087 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2090 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2092 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2095 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2097 (expand_expr, expand_normal): Adjust.
2098 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2099 inner_reference_p. Use inner_reference_p to expand inner references.
2100 (store_expr): Adjust.
2101 * cfgexpand.c (expand_call_stmt): Adjust.
2103 2014-01-08 Rong Xu <xur@google.com>
2105 * gcov-io.c (gcov_var): Move from gcov-io.h.
2106 (gcov_position): Ditto.
2107 (gcov_is_error): Ditto.
2108 (gcov_rewrite): Ditto.
2109 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2110 only part to libgcc/libgcov.h.
2112 2014-01-08 Marek Polacek <polacek@redhat.com>
2115 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2117 2014-01-08 Marek Polacek <polacek@redhat.com>
2120 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2122 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2124 PR rtl-optimization/59649
2125 * stor-layout.c (get_mode_bounds): For BImode return
2126 0 and STORE_FLAG_VALUE.
2128 2014-01-08 Richard Biener <rguenther@suse.de>
2131 * gimple.h (is_gimple_builtin_call): Remove.
2132 (gimple_builtin_call_types_compatible_p): New.
2133 (gimple_call_builtin_p): New overload.
2134 * gimple.c (is_gimple_builtin_call): Remove.
2135 (validate_call): Rename to ...
2136 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2138 (validate_type): New static function.
2139 (gimple_call_builtin_p): New overload and adjust.
2140 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2141 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2142 (gimple_fold_stmt_to_constant_1): Likewise.
2143 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2145 2014-01-08 Richard Biener <rguenther@suse.de>
2148 * gimplify.c (gimplify_expr): Gimplify register-register type
2149 VIEW_CONVERT_EXPRs to separate stmts.
2151 2014-01-07 Jeff Law <law@redhat.com>
2154 * ree.c (combine_set_extension): Handle case where source
2155 and destination registers in an extension insn are different.
2156 (combine_reaching_defs): Allow source and destination registers
2157 in extension to be different under limited circumstances.
2158 (add_removable_extension): Remove restriction that the
2159 source and destination registers in the extension are the same.
2160 (find_and_remove_re): Emit a copy from the extension's
2161 destination to its source after the defining insn if
2162 the source and destination registers are different.
2165 * ifcvt.c (merge_if_block): If we are merging a block with more than
2166 one successor with a block with no successors, remove any BARRIER
2167 after the second block.
2169 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2171 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2173 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2176 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2177 for 14-bit register offsets when INT14_OK_STRICT is false.
2179 2014-01-07 Roland Stigge <stigge@antcom.de>
2180 Michael Meissner <meissner@linux.vnet.ibm.com>
2183 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2184 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2186 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2188 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2191 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2193 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2194 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2195 rtx is const0_rtx or not.
2197 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2200 * target-globals.c (save_target_globals): Remove this_fn_optab
2202 * toplev.c: Include optabs.h.
2203 (target_reinit): Temporarily restore the global options if another
2204 set of options are in force.
2206 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2208 PR rtl-optimization/58668
2209 * cfgcleanup.c (flow_find_cross_jump): Don't count
2210 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2211 to determine what is counted.
2212 (flow_find_head_matching_sequence): Use active_insn_p to determine
2214 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2216 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2217 determine what is counted.
2219 PR tree-optimization/59643
2220 * tree-predcom.c (split_data_refs_to_components): If one dr is
2221 read and one write, determine_offset fails and the write isn't
2222 in the bad component, just put the read into the bad component.
2224 2014-01-07 Mike Stump <mikestump@comcast.net>
2225 Jakub Jelinek <jakub@redhat.com>
2228 * tree-core.h (struct tree_optimization_option): Change optabs
2229 type from unsigned char * to void *.
2230 * optabs.c (init_tree_optimization_optabs): Adjust
2231 TREE_OPTIMIZATION_OPTABS initialization.
2233 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2236 * config/i386/i386.h (struct machine_function): Add
2237 no_drap_save_restore field.
2238 * config/i386/i386.c (ix86_save_reg): Use
2239 !cfun->machine->no_drap_save_restore instead of
2240 crtl->stack_realign_needed.
2241 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2242 this function clears frame_pointer_needed. Set
2243 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2244 and DRAP reg is needed.
2246 2014-01-06 Marek Polacek <polacek@redhat.com>
2249 * doc/implement-c.texi: Mention that other integer types are
2250 permitted as bit-field types in strictly conforming mode.
2252 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2254 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2257 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2259 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2261 2014-01-06 Martin Jambor <mjambor@suse.cz>
2264 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2266 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2268 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2272 * var-tracking.c (add_stores): Preserve the value of the source even if
2273 we don't record the store.
2275 2014-01-06 Terry Guo <terry.guo@arm.com>
2277 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2279 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2282 * config/darwin.c (darwin_function_section): Adjust return values to
2283 correspond to optimisation changes made in r206070.
2285 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2287 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2288 from prefetch_block tune setting.
2289 (nocona_cost): Correct size of prefetch block to 64.
2291 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2293 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2294 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2295 used to save the static chain register in the computation of the offset
2296 from which the FP registers need to be restored.
2298 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2300 PR tree-optimization/59519
2301 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2302 ICE if get_current_def (current_new_name) is already non-NULL, as long
2303 as it is a phi result of some other phi in *new_exit_bb that has
2306 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2307 or vmovdqu* for misaligned_operand.
2308 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2309 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2310 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2311 aligned_mem for AVX512F masked aligned load and store builtins and for
2314 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2316 PR tree-optimization/59651
2317 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2318 Address range for negative step should be added by TYPE_SIZE_UNIT.
2320 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2322 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2323 overlapping registers also for registers other than the stack pointer.
2325 2014-01-03 Marek Polacek <polacek@redhat.com>
2328 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2331 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2334 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2337 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2338 (push splitter): Use <P:MODE_SIZE> instead of
2339 GET_MODE_SIZE (<P:MODE>mode).
2340 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2341 (mov -1, reg peephole2): Likewise.
2342 * config/i386/sse.md (*mov<mode>_internal,
2343 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2344 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2345 *<code><mode>3, *andnot<mode>3<mask_name>,
2346 <mask_codefor><code><mode>3<mask_name>): Likewise.
2347 * config/i386/subst.md (mask_mode512bit_condition,
2348 sd_mask_mode512bit_condition): Likewise.
2350 2014-01-02 Xinliang David Li <davidxl@google.com>
2352 PR tree-optimization/59303
2353 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2354 (dump_predicates): Better output format.
2355 (pred_equal_p): New function.
2356 (is_neq_relop_p): Ditto.
2357 (is_neq_zero_form_p): Ditto.
2358 (pred_expr_equal_p): Ditto.
2359 (pred_neg_p): Ditto.
2360 (simplify_pred): Ditto.
2361 (simplify_preds_2): Ditto.
2362 (simplify_preds_3): Ditto.
2363 (simplify_preds_4): Ditto.
2364 (simplify_preds): Ditto.
2366 (push_to_worklist): Ditto.
2367 (get_pred_info_from_cmp): Ditto.
2368 (is_degenerated_phi): Ditto.
2369 (normalize_one_pred_1): Ditto.
2370 (normalize_one_pred): Ditto.
2371 (normalize_one_pred_chain): Ditto.
2372 (normalize_preds): Ditto.
2373 (normalize_cond_1): Remove function.
2374 (normalize_cond): Ditto.
2375 (is_gcond_subset_of): Ditto.
2376 (is_subset_of_any): Ditto.
2377 (is_or_set_subset_of): Ditto.
2378 (is_and_set_subset_of): Ditto.
2379 (is_norm_cond_subset_of): Ditto.
2380 (pred_chain_length_cmp): Ditto.
2381 (convert_control_dep_chain_into_preds): Type change.
2382 (find_predicates): Ditto.
2383 (find_def_preds): Ditto.
2384 (destroy_predicates_vecs): Ditto.
2385 (find_matching_predicates_in_rest_chains): Ditto.
2386 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2387 (is_pred_expr_subset): Ditto.
2388 (is_pred_chain_subset_of): Ditto.
2389 (is_included_in): Ditto.
2390 (is_superset_of): Ditto.
2392 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2394 Update copyright years.
2396 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2398 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2399 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2400 config/arc/arc.md, config/arc/arc.opt,
2401 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2402 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2403 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2404 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2405 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2406 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2407 vtable-verify.c, vtable-verify.h: Use the standard form for the
2410 2014-01-02 Tobias Burnus <burnus@net-b.de>
2412 * gcc.c (process_command): Update copyright notice dates.
2413 * gcov-dump.c: Ditto.
2415 * doc/cpp.texi: Bump @copying's copyright year.
2416 * doc/cppinternals.texi: Ditto.
2417 * doc/gcc.texi: Ditto.
2418 * doc/gccint.texi: Ditto.
2419 * doc/gcov.texi: Ditto.
2420 * doc/install.texi: Ditto.
2421 * doc/invoke.texi: Ditto.
2423 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2425 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2427 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2429 * config/i386/sse.md (*mov<mode>_internal): Guard
2430 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2432 PR rtl-optimization/59647
2433 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2434 new_rtx into UNSIGNED_FLOAT rtxes.
2436 Copyright (C) 2014 Free Software Foundation, Inc.
2438 Copying and distribution of this file, with or without modification,
2439 are permitted in any medium without royalty provided the copyright
2440 notice and this notice are preserved.