1 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3 * doc/extend.texi (Vector Extensions): Document ?: in C++.
5 2014-01-31 Richard Biener <rguenther@suse.de>
8 * builtins.c (fold_builtin_memory_op): Make sure to not
9 use a floating-point mode or a boolean or enumeral type for
12 2014-01-30 DJ Delorie <dj@redhat.com>
14 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
15 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
16 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
17 whenever main() has an epilogue.
19 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
22 unused variable "field".
23 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
24 (vsx_mergeh_<mode>): Likewise.
25 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
26 (altivec_vmrghh): Likewise.
27 (altivec_vmrghw): Likewise.
28 (altivec_vmrglb): Likewise.
29 (altivec_vmrglh): Likewise.
30 (altivec_vmrglw): Likewise.
31 (altivec_vspltb): Add missing uses.
32 (altivec_vsplth): Likewise.
33 (altivec_vspltw): Likewise.
34 (altivec_vspltsf): Likewise.
36 2014-01-30 Jakub Jelinek <jakub@redhat.com>
39 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
40 frame related instructions.
42 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
44 PR rtl-optimization/59959
45 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
46 any reload of register whose subreg is invalid.
48 2014-01-30 Jakub Jelinek <jakub@redhat.com>
50 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing
52 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
53 Add missing return type - void.
55 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
57 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
58 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
59 remove element index adjustment for endian (now handled in vsx.md
61 (altivec_expand_vec_perm_const): Use
62 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
63 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
64 (vsx_xxspltw_<mode>): Adjust element index for little endian.
65 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
66 define_expand and a new define_insn *altivec_vspltb_internal;
67 adjust for -maltivec=be on a little endian target.
68 (altivec_vspltb_direct): New.
69 (altivec_vsplth): Divide into a define_expand and a new
70 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
72 (altivec_vsplth_direct): New.
73 (altivec_vspltw): Divide into a define_expand and a new
74 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
76 (altivec_vspltw_direct): New.
77 (altivec_vspltsf): Divide into a define_expand and a new
78 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
79 a little endian target.
81 2014-01-30 Richard Biener <rguenther@suse.de>
83 PR tree-optimization/59993
84 * tree-ssa-forwprop.c (associate_pointerplus): Check we
85 can propagate form the earlier stmt and avoid the transform
86 when the intermediate result is needed.
88 2014-01-30 Alangi Derick <alangiderick@gmail.com>
90 * README.Portability: Fix typo.
92 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
94 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
95 comparison_operator with ordered_comparison_operator.
97 2014-01-30 Nick Clifton <nickc@redhat.com>
99 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
100 Rename to mn10300_store_multiple_regs.
101 * config/mn10300/mn10300.c: Likewise.
102 * config/mn10300/mn10300.md (store_movm): Fix typo: call
104 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
105 Call mn10300_store_multiple_regs.
107 2014-01-30 Nick Clifton <nickc@redhat.com>
108 DJ Delorie <dj@redhat.com>
110 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
111 %fp 2 to keep registers after it properly word-aligned.
112 (rl78_alloc_physical_registers_umul): Handle the case where both
113 input operands are the same.
115 2014-01-30 Richard Biener <rguenther@suse.de>
117 PR tree-optimization/59903
118 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
121 2014-01-30 Jason Merrill <jason@redhat.com>
124 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
127 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
129 2014-01-30 Richard Biener <rguenther@suse.de>
131 PR tree-optimization/59951
132 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
135 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
138 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
139 SFmode to DFmode case.
141 2014-01-29 DJ Delorie <dj@redhat.com>
143 * config/msp430/msp430.opt (-minrt): New.
144 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
146 (ENDFILE_SPEC): Likewise.
148 2014-01-29 Jan Hubicka <jh@suse.cz>
150 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
151 (estimate_function_body_sizes): Use it.
153 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
156 * dwarf2out.c (is_cxx_auto): New.
157 (is_base_type): Use it.
158 (gen_type_die_with_usage): Likewise.
160 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
162 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
163 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
164 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
165 -maltivec=be with LE targets.
166 (vsx_mergeh_<mode>): Likewise.
167 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
169 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
170 (altivec_vmrghb): Replace with define_expand and new
171 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
173 (altivec_vmrghb_direct): New define_insn.
174 (altivec_vmrghh): Replace with define_expand and new
175 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
177 (altivec_vmrghh_direct): New define_insn.
178 (altivec_vmrghw): Replace with define_expand and new
179 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
181 (altivec_vmrghw_direct): New define_insn.
182 (*altivec_vmrghsf): Adjust for endianness.
183 (altivec_vmrglb): Replace with define_expand and new
184 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
186 (altivec_vmrglb_direct): New define_insn.
187 (altivec_vmrglh): Replace with define_expand and new
188 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
190 (altivec_vmrglh_direct): New define_insn.
191 (altivec_vmrglw): Replace with define_expand and new
192 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
194 (altivec_vmrglw_direct): New define_insn.
195 (*altivec_vmrglsf): Adjust for endianness.
196 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
197 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
198 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
199 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
200 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
201 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
202 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
203 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
205 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
207 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
208 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
211 2014-01-29 Richard Biener <rguenther@suse.de>
213 PR tree-optimization/58742
214 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
215 associate_pointerplus_align.
216 (associate_pointerplus_diff): New function.
217 (associate_pointerplus): Likewise. Call associate_pointerplus_align
218 and associate_pointerplus_diff.
220 2014-01-29 Richard Biener <rguenther@suse.de>
222 * lto-streamer.h (LTO_major_version): Bump to 3.
223 (LTO_minor_version): Reset to 0.
225 2014-01-29 Renlin Li <Renlin.Li@arm.com>
227 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
228 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
229 (arm_file_start): Generate correct asm header for armv7ve.
230 * config/arm/bpabi.h: Add multilib support for armv7ve.
231 * config/arm/driver-arm.c: Change the architectures of cortex-a7
232 and cortex-a15 to armv7ve.
233 * config/arm/t-aprofile: Add multilib support for armv7ve.
234 * doc/invoke.texi: Document -march=armv7ve.
236 2014-01-29 Richard Biener <rguenther@suse.de>
238 PR tree-optimization/58742
239 * tree-ssa-forwprop.c (associate_plusminus): Return true
240 if we changed sth, defer EH cleanup to ...
241 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
242 (simplify_mult): New function.
244 2014-01-29 Jakub Jelinek <jakub@redhat.com>
247 PR tree-optimization/59920
248 * tree.c (build_common_builtin_nodes): Remove
249 __builtin_setjmp_dispatcher initialization.
250 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
251 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
252 instead of gsi_after_labels + manually skipping debug stmts.
253 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
254 ignore bbs with IFN_ABNORMAL_DISPATCHER.
255 * tree-inline.c (copy_edges_for_bb): Remove
256 can_make_abnormal_goto argument, instead add abnormal_goto_dest
257 argument. Ignore computed_goto_p stmts. Don't call
258 make_abnormal_goto_edges. If a call might need abnormal edges
259 for non-local gotos, see if it already has an edge to
260 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
261 with true argument, don't do anything then, otherwise add
262 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
263 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
265 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
266 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
267 (lower_stmt): Don't set data->calls_builtin_setjmp.
268 (lower_builtin_setjmp): Adjust comment.
269 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
270 * tree-cfg.c (found_computed_goto): Remove.
271 (factor_computed_gotos): Remove.
272 (make_goto_expr_edges): Return bool, true for computed gotos.
273 Don't call make_abnormal_goto_edges.
274 (build_gimple_cfg): Don't set found_computed_goto, don't call
275 factor_computed_gotos.
276 (computed_goto_p): No longer static.
277 (make_blocks): Don't set found_computed_goto.
278 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
279 (make_edges): If make_goto_expr_edges returns true, push bb
280 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
281 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
282 vector. Record mapping between bbs and OpenMP regions if there
283 are any, adjust make_gimple_omp_edges caller. Call
284 handle_abnormal_edges.
285 (make_abnormal_goto_edges): Remove.
286 * tree-cfg.h (make_abnormal_goto_edges): Remove.
287 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
288 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
289 * builtins.c (expand_builtin): Don't handle
290 BUILT_IN_SETJMP_DISPATCHER.
291 * internal-fn.def (ABNORMAL_DISPATCHER): New.
292 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
293 filling *region also set *region_idx to (*region)->entry->index.
296 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
297 For REGs set ORIGINAL_REGNO.
299 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
301 * doc/md.texi: Mention that a target shouldn't implement
302 vec_widen_(s|u)mul_even/odd pair if it is less efficient
305 2014-01-29 Jakub Jelinek <jakub@redhat.com>
307 PR tree-optimization/59594
308 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
309 a copy of the datarefs vector rather than the vector itself.
311 2014-01-28 Jason Merrill <jason@redhat.com>
314 * dwarf2out.c (auto_die): New static.
315 (gen_type_die_with_usage): Handle C++1y 'auto'.
316 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
319 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
322 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
323 (SPEC_X32): Likewise.
325 * config/i386/i386.c (ix86_option_override_internal): Turn off
326 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
328 (x86_file_start): Output .code16gcc for TARGET_16BIT.
329 * config/i386/i386.h (TARGET_16BIT): New macro.
330 (TARGET_16BIT_P): Likewise.
331 * config/i386/i386.opt: Add m16.
332 * doc/invoke.texi: Document -m16.
334 2014-01-28 Jakub Jelinek <jakub@redhat.com>
336 PR preprocessor/59935
337 * input.c (location_get_source_line): Bail out on when line number
338 is zero, and test the return value of
339 lookup_or_add_file_to_cache_tab.
341 2014-01-28 Richard Biener <rguenther@suse.de>
343 PR tree-optimization/58742
344 * tree-ssa-forwprop.c (associate_plusminus): Handle
345 pointer subtraction of the form (T)(P + A) - (T)P.
347 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
349 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
352 2014-01-28 Richard Biener <rguenther@suse.de>
355 2014-01-28 Richard Biener <rguenther@suse.de>
357 PR rtl-optimization/45364
358 PR rtl-optimization/59890
359 * var-tracking.c (local_get_addr_clear_given_value): Handle
360 already cleared slot.
361 (val_reset): Handle not allocated local_get_addr_cache.
362 (vt_find_locations): Use post-order on the inverted CFG.
364 2014-01-28 Richard Biener <rguenther@suse.de>
366 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
369 2014-01-28 Richard Biener <rguenther@suse.de>
371 PR rtl-optimization/45364
372 PR rtl-optimization/59890
373 * var-tracking.c (local_get_addr_clear_given_value): Handle
374 already cleared slot.
375 (val_reset): Handle not allocated local_get_addr_cache.
376 (vt_find_locations): Use post-order on the inverted CFG.
378 2014-01-28 Alan Modra <amodra@gmail.com>
380 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
381 * configure.ac <recursive call for build != host>: Define
382 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
383 and LD_FOR_BUILD too.
384 * configure: Regenerate.
386 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
388 * config/i386/i386.c (get_builtin_code_for_version): Separate
389 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
390 Broadwell from Haswell.
392 2014-01-27 Steve Ellcey <sellcey@mips.com>
394 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
395 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
396 * config/mips/mips.c (mips_option_override): Change setting
398 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
399 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
400 Change from Mask to Var.
402 2014-01-27 Jeff Law <law@redhat.com>
404 * ipa-inline.c (inline_small_functions): Fix typo.
406 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
408 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
409 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
410 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
411 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
412 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
413 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
414 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
415 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
416 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
417 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
418 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
419 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
420 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
421 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
422 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
423 (_mm512_storeu_epi64): Ditto.
424 (_mm512_cmpge_epi32_mask): Ditto.
425 (_mm512_cmpge_epu32_mask): Ditto.
426 (_mm512_cmpge_epi64_mask): Ditto.
427 (_mm512_cmpge_epu64_mask): Ditto.
428 (_mm512_cmple_epi32_mask): Ditto.
429 (_mm512_cmple_epu32_mask): Ditto.
430 (_mm512_cmple_epi64_mask): Ditto.
431 (_mm512_cmple_epu64_mask): Ditto.
432 (_mm512_cmplt_epi32_mask): Ditto.
433 (_mm512_cmplt_epu32_mask): Ditto.
434 (_mm512_cmplt_epi64_mask): Ditto.
435 (_mm512_cmplt_epu64_mask): Ditto.
436 (_mm512_cmpneq_epi32_mask): Ditto.
437 (_mm512_cmpneq_epu32_mask): Ditto.
438 (_mm512_cmpneq_epi64_mask): Ditto.
439 (_mm512_cmpneq_epu64_mask): Ditto.
440 (_mm512_expand_pd): Ditto.
441 (_mm512_expand_ps): Ditto.
442 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
443 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
444 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
445 * config/i386/i386.c (ix86_builtins): Add
446 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
447 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
448 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
449 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
450 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
451 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
452 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
453 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
454 IX86_BUILTIN_PMOVUSQW512_MEM.
455 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
456 __builtin_ia32_pmovsqd512mem_mask,
457 __builtin_ia32_pmovqd512mem_mask,
458 __builtin_ia32_pmovusqw512mem_mask,
459 __builtin_ia32_pmovsqw512mem_mask,
460 __builtin_ia32_pmovqw512mem_mask,
461 __builtin_ia32_pmovusdw512mem_mask,
462 __builtin_ia32_pmovsdw512mem_mask,
463 __builtin_ia32_pmovdw512mem_mask,
464 __builtin_ia32_pmovqb512mem_mask,
465 __builtin_ia32_pmovusqb512mem_mask,
466 __builtin_ia32_pmovsqb512mem_mask,
467 __builtin_ia32_pmovusdb512mem_mask,
468 __builtin_ia32_pmovsdb512mem_mask,
469 __builtin_ia32_pmovdb512mem_mask.
470 (bdesc_args): Add __builtin_ia32_expanddf512,
471 __builtin_ia32_expandsf512.
472 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
473 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
474 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
475 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
476 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
477 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
478 (avx512f_<code>v8div16qi2_mask_store): This.
479 (avx512f_expand<mode>): New.
481 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
483 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
485 (_mm512_mask_prefetch_i64gather_pd): Ditto.
486 (_mm512_prefetch_i32scatter_pd): Ditto.
487 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
488 (_mm512_prefetch_i64scatter_pd): Ditto.
489 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
490 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
491 (_mm512_mask_prefetch_i64gather_ps): Ditto.
492 (_mm512_prefetch_i32scatter_ps): Ditto.
493 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
494 (_mm512_prefetch_i64scatter_ps): Ditto.
495 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
496 * config/i386/i386-builtin-types.def: Define
497 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
498 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
499 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
500 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
501 IX86_BUILTIN_SCATTERPFQPD.
502 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
503 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
504 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
505 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
506 __builtin_ia32_scatterpfqps.
507 (ix86_expand_builtin): Expand new built-ins.
508 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
509 fix memory access data type.
510 (*avx512pf_gatherpf<mode>_mask): Ditto.
511 (*avx512pf_gatherpf<mode>): Ditto.
512 (avx512pf_scatterpf<mode>): Ditto.
513 (*avx512pf_scatterpf<mode>_mask): Ditto.
514 (*avx512pf_scatterpf<mode>): Ditto.
515 (GATHER_SCATTER_SF_MEM_MODE): New.
516 (avx512pf_gatherpf<mode>df): Ditto.
517 (*avx512pf_gatherpf<mode>df_mask): Ditto.
518 (*avx512pf_scatterpf<mode>df): Ditto.
520 2014-01-27 Jakub Jelinek <jakub@redhat.com>
523 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
524 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
527 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
529 * common/config/arm/arm-common.c
530 (arm_rewrite_mcpu): Handle multiple names.
532 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
534 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
536 * gimple-builder.h (create_gimple_tmp): Delete.
538 2014-01-27 Christian Bruel <christian.bruel@st.com>
540 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
543 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
545 * config/pa/pa.md (call): Generate indirect long calls to non-local
546 functions when outputing 32-bit code.
547 (call_value): Likewise except for special call to buggy powf function.
549 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
550 portable runtime and PIC indirect calls.
551 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
552 and PIC call sequences. Use ldo instead of blr to set return register
553 in PIC call sequence.
555 2014-01-25 Walter Lee <walt@tilera.com>
557 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
558 avoid clobbering a live register.
560 2014-01-25 Walter Lee <walt@tilera.com>
562 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
563 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
564 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
565 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
567 2014-01-25 Walter Lee <walt@tilera.com>
569 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
570 arguments on even registers.
571 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
573 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
574 (BIGGEST_ALIGNMENT): Ditto.
575 (BIGGEST_FIELD_ALIGNMENT): Ditto.
577 2014-01-25 Walter Lee <walt@tilera.com>
579 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
580 insns before bundling.
581 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
583 2014-01-25 Walter Lee <walt@tilera.com>
585 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
586 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
587 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
589 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
591 * config/mips/constraints.md (kl): Delete.
592 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
593 define expands, using...
594 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
595 instructions for MIPS16.
596 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
597 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
599 2014-01-25 Walter Lee <walt@tilera.com>
601 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
605 2014-01-25 Walter Lee <walt@tilera.com>
607 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
608 (TARGET_EXPAND_TO_RTL_HOOK): Define.
610 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
612 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
615 2014-01-25 Jakub Jelinek <jakub@redhat.com>
617 * print-rtl.c (in_call_function_usage): New var.
618 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
619 EXPR_LIST mode as mode and not as reg note name.
622 * cfgloopmanip.c (copy_loop_info): If
623 loop->warned_aggressive_loop_optimizations, make sure
624 the flag is set in target loop too.
626 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
628 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
630 * builtins.def: Likewise.
631 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
632 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
633 * ira.c (ira_setup_eliminable_regset): Likewise.
634 * omp-low.c (gate_expand_omp): Likewise.
635 (execute_lower_omp): Likewise.
636 (diagnose_sb_0): Likewise.
637 (gate_diagnose_omp_blocks): Likewise.
638 (simd_clone_clauses_extract): Likewise.
641 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
643 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
644 correction for little endian...
645 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
648 2014-01-24 Jeff Law <law@redhat.com>
650 PR tree-optimization/59919
651 * tree-vrp.c (find_assert_locations_1): Do not register asserts
652 for non-returning calls.
654 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
656 * common/config/aarch64/aarch64-common.c
657 (aarch64_rewrite_mcpu): Handle multiple names.
658 * config/aarch64/aarch64.h
659 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
661 2014-01-24 Dodji Seketeli <dodji@redhat.com>
663 * input.c (add_file_to_cache_tab): Handle the case where fopen
666 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
669 * config/i386/i386.md (pushsf splitter): Get stack adjustment
670 from push operand if code of push isn't PRE_DEC.
672 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
675 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
676 -mquad-memory-atomic. Update -mquad-memory documentation to say
677 it is only used for non-atomic loads/stores.
679 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
680 -mquad-memory or -mquad-memory-atomic switches.
682 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
683 -mquad-memory-atomic to ISA 2.07 support.
685 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
686 to separate support of normal quad word memory operations (ldq, stq)
687 from the atomic quad word memory operations.
689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
690 support to separate non-atomic quad word operations from atomic
691 quad word operations. Disable non-atomic quad word operations in
692 little endian mode so that we don't have to swap words after the
693 load and before the store.
694 (quad_load_store_p): Add comment about atomic quad word support.
695 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
696 options printed with -mdebug=reg.
698 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
699 -mquad-memory-atomic as the test for whether we have quad word
701 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
702 or -mp8-vector are used, allow byte/half-word atomic operations.
704 * config/rs6000/sync.md (load_lockedti): Insure that the address
705 is a proper indexed or indirect address for the lqarx instruction.
706 On little endian systems, swap the hi/lo registers after the lqarx
708 (load_lockedpti): Use indexed_or_indirect_operand predicate to
709 insure the address is valid for the lqarx instruction.
710 (store_conditionalti): Insure that the address is a proper indexed
711 or indirect address for the stqcrx. instruction. On little endian
712 systems, swap the hi/lo registers before doing the stqcrx.
714 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
715 insure the address is valid for the stqcrx. instruction.
717 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
718 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
719 type of quad memory support is available.
721 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
724 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
725 there is a danger of looping.
727 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
730 force flag_ira_loop_pressure if set via command line.
732 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
734 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
735 (ashr_simd): New builtin handling DI mode.
736 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
737 (aarch64_sshr_simddi): New match pattern.
738 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
739 (vshrd_n_s64): Likewise.
740 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
742 2014-01-23 Nick Clifton <nickc@redhat.com>
744 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
745 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
746 favour of mcu specific scripts.
747 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
750 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
751 Alex Velenko <Alex.Velenko@arm.com>
753 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
754 (vaddv_s16): Likewise.
755 (vaddv_s32): Likewise.
756 (vaddv_u8): Likewise.
757 (vaddv_u16): Likewise.
758 (vaddv_u32): Likewise.
759 (vaddvq_s8): Likewise.
760 (vaddvq_s16): Likewise.
761 (vaddvq_s32): Likewise.
762 (vaddvq_s64): Likewise.
763 (vaddvq_u8): Likewise.
764 (vaddvq_u16): Likewise.
765 (vaddvq_u32): Likewise.
766 (vaddvq_u64): Likewise.
767 (vaddv_f32): Likewise.
768 (vaddvq_f32): Likewise.
769 (vaddvq_f64): Likewise.
770 (vmaxv_f32): Likewise.
771 (vmaxv_s8): Likewise.
772 (vmaxv_s16): Likewise.
773 (vmaxv_s32): Likewise.
774 (vmaxv_u8): Likewise.
775 (vmaxv_u16): Likewise.
776 (vmaxv_u32): Likewise.
777 (vmaxvq_f32): Likewise.
778 (vmaxvq_f64): Likewise.
779 (vmaxvq_s8): Likewise.
780 (vmaxvq_s16): Likewise.
781 (vmaxvq_s32): Likewise.
782 (vmaxvq_u8): Likewise.
783 (vmaxvq_u16): Likewise.
784 (vmaxvq_u32): Likewise.
785 (vmaxnmv_f32): Likewise.
786 (vmaxnmvq_f32): Likewise.
787 (vmaxnmvq_f64): Likewise.
788 (vminv_f32): Likewise.
789 (vminv_s8): Likewise.
790 (vminv_s16): Likewise.
791 (vminv_s32): Likewise.
792 (vminv_u8): Likewise.
793 (vminv_u16): Likewise.
794 (vminv_u32): Likewise.
795 (vminvq_f32): Likewise.
796 (vminvq_f64): Likewise.
797 (vminvq_s8): Likewise.
798 (vminvq_s16): Likewise.
799 (vminvq_s32): Likewise.
800 (vminvq_u8): Likewise.
801 (vminvq_u16): Likewise.
802 (vminvq_u32): Likewise.
803 (vminnmv_f32): Likewise.
804 (vminnmvq_f32): Likewise.
805 (vminnmvq_f64): Likewise.
807 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
809 * config/aarch64/aarch64-simd.md
810 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
811 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
812 (*aarch64_mul3_elt<mode>): Likewise.
813 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
814 (*aarch64_mul3_elt_to_64v2df): Likewise.
815 (*aarch64_mla_elt<mode>): Likewise.
816 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
817 (*aarch64_mls_elt<mode>): Likewise.
818 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
819 (*aarch64_fma4_elt<mode>): Likewise.
820 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
821 (*aarch64_fma4_elt_to_64v2df): Likewise.
822 (*aarch64_fnma4_elt<mode>): Likewise.
823 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
824 (*aarch64_fnma4_elt_to_64v2df): Likewise.
825 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
826 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
827 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
828 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
829 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
830 (aarch64_sqdmull_lane<mode>_internal): Likewise.
831 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
833 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
835 * config/aarch64/aarch64-simd.md
836 (aarch64_be_checked_get_lane<mode>): New define_expand.
837 * config/aarch64/aarch64-simd-builtins.def
838 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
839 New builtin definition.
840 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
841 Use new safe be builtin.
843 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
845 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
847 (aarch64_be_st1<mode>): Likewise.
848 (aarch_ld1<VALL:mode>): Define_expand modified.
849 (aarch_st1<VALL:mode>): Likewise.
850 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
851 (UNSPEC_ST1): Likewise.
853 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
855 * config/microblaze/microblaze.md: Add trap insn and attribute
857 2014-01-23 Dodji Seketeli <dodji@redhat.com>
859 PR preprocessor/58580
860 * input.h (location_get_source_line): Take an additional line_size
862 (void diagnostics_file_cache_fini): Declare new function.
863 * input.c (struct fcache): New type.
864 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
865 New static constants.
866 (diagnostic_file_cache_init, total_lines_num)
867 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
868 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
869 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
870 (get_next_line, read_next_line, goto_next_line, read_line_num):
871 New static function definitions.
872 (diagnostic_file_cache_fini): New function.
873 (location_get_source_line): Take an additional output line_len
874 parameter. Re-write using lookup_or_add_file_to_cache_tab and
876 * diagnostic.c (diagnostic_finish): Call
877 diagnostic_file_cache_fini.
878 (adjust_line): Take an additional input parameter for the length
879 of the line, rather than calculating it with strlen.
880 (diagnostic_show_locus): Adjust the use of
881 location_get_source_line and adjust_line with respect to their new
882 signature. While displaying a line now, do not stop at the first
883 null byte. Rather, display the zero byte as a space and keep
884 going until we reach the size of the line.
885 * Makefile.in: Add vec.o to OBJS-libcommon
887 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
888 Ilya Tocar <ilya.tocar@intel.com>
890 * config/i386/avx512fintrin.h (_mm512_kmov): New.
891 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
892 (__builtin_ia32_kmov16): Ditto.
893 * config/i386/i386.md (UNSPEC_KMOV): New.
896 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
898 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
899 (_mm512_storeu_si512): Ditto.
901 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
904 * rtl.h (get_referenced_operands): Declare.
905 * recog.c (get_referenced_operands): New function.
906 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
907 operands have been referenced when recording LO_SUM references.
909 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
911 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
913 2014-01-22 Jan Hubicka <jh@suse.cz>
915 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
916 Enable for generic and recent AMD targets.
918 2014-01-22 Jan Hubicka <jh@suse.cz>
920 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
921 ARG_SIZE note when adjustment was eliminated.
923 2014-01-22 Jeff Law <law@redhat.com>
925 PR tree-optimization/59597
926 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
927 in file. Accept new argument REGISTERING and use it to modify
928 dump output appropriately.
929 (register_jump_thread): Corresponding changes.
930 (mark_threaded_blocks): Reinstate code to cancel unprofitable
931 thread paths involving joiner blocks. Add code to dump cancelled
932 jump threading paths.
934 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
936 PR rtl-optimization/59477
937 * lra-constraints.c (inherit_in_ebb): Process call for living hard
938 regs. Update reloads_num and potential_reload_hard_regs for all insns.
940 2014-01-22 Tom Tromey <tromey@redhat.com>
942 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
944 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
946 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
948 PR rtl-optimization/59896
949 * lra-constraints.c (process_alt_operands): Check unused note for
950 matched operands of insn with no output reloads.
952 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
954 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
955 (mips_move_from_gpr_cost): Likewise.
957 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
959 PR rtl-optimization/59858
960 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
961 ira_class_hard_regs_num.
962 (process_alt_operands): Increase reject for dying matched operand.
964 2014-01-21 Jakub Jelinek <jakub@redhat.com>
967 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
968 smaller than size, perform several stores or loads and stores
969 at dst + count - size to store or copy all of size bytes, rather
970 than just last modesize bytes.
972 2014-01-20 DJ Delorie <dj@redhat.com>
974 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
975 that CLOBBERs are REGs before propogating their values.
977 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
980 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
981 (cgraph_inline_failed_type): New function.
982 * cgraph.h (DEFCIFCODE): Add type.
983 (cgraph_inline_failed_type_t): New enum.
984 (cgraph_inline_failed_type): New prototype.
985 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
986 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
987 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
988 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
989 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
990 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
991 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
992 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
993 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
994 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
995 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
996 OPTIMIZATION_MISMATCH.
997 * tree-inline.c (expand_call_inline): Emit errors during
998 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1000 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1003 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1004 mode attribute in insn output.
1006 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1008 * output.h (output_constant): Delete.
1009 * varasm.c (output_constant): Make private.
1011 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1013 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1015 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1018 * tree.h (fold_builtin_strcat): New prototype.
1019 * builtins.c (fold_builtin_strcat): No longer static. Add len
1020 argument, if non-NULL, don't call c_strlen. Optimize
1021 directly into __builtin_memcpy instead of __builtin_strcpy.
1022 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1023 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1025 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1027 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1028 for SImode_address_operand operands, having only a REG argument.
1030 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1032 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1033 loader name using mbig-endian.
1034 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1036 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1038 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1042 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1044 * config/aarch64/aarch64-protos.h
1045 (aarch64_cannot_change_mode_class_ptr): Declare.
1046 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1047 aarch64_cannot_change_mode_class_ptr): New.
1048 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1049 backend hook aarch64_cannot_change_mode_class.
1051 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1053 * common/config/aarch64/aarch64-common.c
1054 (aarch64_handle_option): Don't handle any option order logic here.
1055 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1056 selected_cpu, warn on architecture version mismatch.
1057 (aarch64_override_options): Fix parsing order for option strings.
1059 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1060 Iain Sandoe <iain@codesourcery.com>
1063 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1064 warning. Amend comment to reflect current functionality.
1066 2014-01-20 Richard Biener <rguenther@suse.de>
1069 * builtins.c (fold_builtin_strcat): Remove case better handled
1070 by tree-ssa-strlen.c.
1072 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1074 * config/aarch64/aarch64.opt
1075 (mcpu, march, mtune): Make case-insensitive.
1077 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1080 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1081 if operands[1] is a REG or ZERO_EXTEND of a REG.
1083 2014-01-19 Jan Hubicka <jh@suse.cz>
1085 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1087 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1089 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1090 long non-pic millicode calls.
1092 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1094 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1096 2014-01-19 Kito Cheng <kito@0xlab.org>
1098 * builtins.c (expand_movstr): Check movstr expand done or fail.
1100 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1101 H.J. Lu <hongjiu.lu@intel.com>
1104 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1105 to DImode for zero-extended addresses.
1107 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1109 PR rtl-optimization/57763
1110 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1111 on the new indirect jump_insn and increment LABEL_NUSES (label).
1113 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1117 * config.gcc (x86_archs): New variable.
1118 (x86_64_archs): Likewise.
1119 (x86_cpus): Likewise.
1120 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1121 --with-arch/--with-cpu= options.
1122 Support --with-arch=/--with-cpu={nehalem,westmere,
1123 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1125 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1127 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1128 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1130 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1132 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1134 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1137 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1138 clear cpp_get_options (parse_in)->warn_unused_macros for
1139 ix86_target_macros_internal with cpp_define.
1141 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1143 * jump.c (delete_related_insns): Keep (use (insn))s.
1144 * reorg.c (redundant_insn): Check for barriers too.
1146 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1148 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1150 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1152 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1153 call to $$dyncall when TARGET_LONG_CALLS is true.
1155 2014-01-17 Jeff Law <law@redhat.com>
1157 * ree.c (combine_set_extension): Temporarily disable test for
1158 changing number of hard registers.
1160 2014-01-17 Jan Hubicka <jh@suse.cz>
1163 * ipa-inline-analysis.c (inline_free_summary):
1164 Do not free summary of aliases.
1166 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1169 * gimplify.c (gimplify_expr): Use create_tmp_var
1170 instead of create_tmp_var_raw. If cond doesn't have
1171 integral type, don't add the IFN_ANNOTATE builtin at all.
1173 2014-01-17 Martin Jambor <mjambor@suse.cz>
1176 * ipa-cp.c (prev_edge_clone): New variable.
1177 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1178 Also resize prev_edge_clone vector.
1179 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1180 (ipcp_edge_removal_hook): New function.
1181 (ipcp_driver): Register ipcp_edge_removal_hook.
1183 2014-01-17 Andrew Pinski <apinski@cavium.com>
1184 Steve Ellcey <sellcey@mips.com>
1187 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1190 2014-01-17 Jeff Law <law@redhat.com>
1193 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1194 so that pass_ccp runs first.
1196 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1198 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1199 (ix86_adjust_cost): Use !TARGET_XXX.
1200 (do_reorder_for_imul): Likewise.
1201 (swap_top_of_ready_list): Likewise.
1202 (ix86_sched_reorder): Likewise.
1204 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1206 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1207 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1208 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1209 (intel_memset): New. Duplicate slm_memset.
1210 (intel_cost): New. Duplicate slm_cost.
1211 (m_INTEL): New macro.
1212 (processor_target_table): Add "intel".
1213 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1214 with PROCESSOR_INTEL for "intel".
1215 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1216 PROCESSOR_SILVERMONT.
1217 (ix86_issue_rate): Likewise.
1218 (ix86_adjust_cost): Likewise.
1219 (ia32_multipass_dfa_lookahead): Likewise.
1220 (swap_top_of_ready_list): Likewise.
1221 (ix86_sched_reorder): Likewise.
1222 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1223 instead of TARGET_OPT_AGU.
1224 * config/i386/i386.h (TARGET_INTEL): New.
1225 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1226 (processor_type): Add PROCESSOR_INTEL.
1227 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1228 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1230 2014-01-17 Marek Polacek <polacek@redhat.com>
1233 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1236 2014-01-17 Richard Biener <rguenther@suse.de>
1238 PR tree-optimization/46590
1239 * opts.c (default_options_table): Add entries for
1240 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1241 all enabled at -O1 but not for -Og.
1242 * common.opt (fbranch-count-reg): Remove Init(1).
1243 (fmove-loop-invariants): Likewise.
1244 (ftree-pta): Likewise.
1246 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1248 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1249 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1250 decls to at least the GCC 4.8 used alignments.
1253 * tree-nested.c (convert_nonlocal_reference_stmt,
1254 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1255 of GIMPLE_BIND stmts, adjust associated decls.
1257 2014-01-17 Richard Biener <rguenther@suse.de>
1259 PR tree-optimization/46590
1260 * vec.h (vec<>::bseach): New member function implementing
1261 binary search according to C89 bsearch.
1262 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1263 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1264 bitmap pointer again. Make accesses_in_loop a flat array.
1265 (mem_ref_obstack): New global.
1266 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1267 (mark_ref_stored): Likewise.
1268 (ref_indep_loop_p_2): Likewise.
1269 (set_ref_stored_in_loop): New helper function.
1270 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1271 (memref_free): Adjust.
1272 (record_mem_ref_loc): Simplify.
1273 (gather_mem_refs_stmt): Adjust.
1274 (sort_locs_in_loop_postorder_cmp): New function.
1275 (analyze_memory_references): Sort accesses_in_loop after
1276 loop postorder number.
1277 (find_ref_loc_in_loop_cmp): New function.
1278 (for_all_locs_in_loop): Find relevant cluster of locs in
1279 accesses_in_loop and iterate without recursion.
1280 (execute_sm): Avoid uninit warning.
1281 (struct ref_always_accessed): Simplify.
1282 (ref_always_accessed::operator ()): Likewise.
1283 (ref_always_accessed_p): Likewise.
1284 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1285 loop postorder numbers here.
1286 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1289 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1292 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1293 on decls for which assemble_alias has been called.
1295 2014-01-17 Nick Clifton <nickc@redhat.com>
1297 * config/msp430/msp430.opt: (mcpu): New option.
1298 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1299 (msp430_option_override): Parse target_cpu. If the MCU name
1300 matches a generic string, clear target_mcu.
1301 (msp430_attr): Allow numeric interrupt values up to 63.
1302 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1303 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1305 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1307 * config/msp430/msp430.md (popm): Use %J rather than %I.
1308 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1309 (addhi_cy_i): Use immediate_operand for operand 2.
1310 * doc/invoke.texi: Document -mcpu option.
1312 2014-01-17 Richard Biener <rguenther@suse.de>
1314 PR rtl-optimization/38518
1315 * df.h (df_analyze_loop): Declare.
1316 * df-core.c: Include cfgloop.h.
1317 (df_analyze_1): Split out main part of df_analyze.
1318 (df_analyze): Adjust.
1319 (loop_inverted_post_order_compute): New function.
1320 (loop_post_order_compute): Likewise.
1321 (df_analyze_loop): New function avoiding whole-function
1323 * loop-invariant.c (find_defs): Use df_analyze_loop.
1324 (find_invariants): Adjust.
1325 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1327 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1329 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1330 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1332 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1334 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1335 traversal when removing references.
1337 2014-01-16 Jan Hubicka <jh@suse.cz>
1340 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1342 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1345 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1346 pushing a reload for an autoinc when we had previously reloaded an
1347 inner part of the address.
1349 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1351 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1353 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1354 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1355 when not giving up or versioning for alias only because of
1357 (vect_analyze_data_ref_dependences): Set to true.
1358 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1360 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1361 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1365 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1368 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1369 operand 0 predicate for gathers, use a new pseudo as subtarget.
1371 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1374 * lra-constraints.c (process_alt_operands): Add printing debug info.
1375 Check absence of input/output reloads for matched operands too.
1377 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1379 PR rtl-optimization/59835
1380 * ira.c (ira_init_register_move_cost): Increase cost for
1383 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1385 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1387 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1390 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1391 non-register objects. Use gen_(high/low)part more consistently.
1394 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1397 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1398 endian support, remove tests for WORDS_BIG_ENDIAN.
1399 (p8_mfvsrd_3_<mode>): Likewise.
1400 (reload_gpr_from_vsx<mode>): Likewise.
1401 (reload_gpr_from_vsxsf): Likewise.
1402 (p8_mfvsrd_4_disf): Likewise.
1404 2014-01-16 Richard Biener <rguenther@suse.de>
1406 PR rtl-optimization/46590
1407 * lcm.c (compute_antinout_edge): Use postorder iteration.
1408 (compute_laterin): Use inverted postorder iteration.
1410 2014-01-16 Nick Clifton <nickc@redhat.com>
1413 * varasm.c (output_constant): Return the number of bytes actually
1415 (output_constructor_array_range): Update the field size with the
1416 number of bytes emitted by output_constant.
1417 (output_constructor_regular_field): Likewise. Also do not
1418 complain if the total number of bytes emitted is now greater
1419 than the expected fieldpos.
1420 * output.h (output_constant): Update prototype and descriptive comment.
1422 2014-01-16 Marek Polacek <polacek@redhat.com>
1425 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1426 it is error_mark_node.
1428 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1430 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1431 VALID_AVX256_REG_OR_OI_MODE.
1433 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1435 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1436 current procedure should be profiled.
1438 2014-01-15 Andrew Pinski <apinski@cavium.com>
1440 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1441 of moving from/to the STACK_REG register class.
1443 2014-01-15 Richard Henderson <rth@redhat.com>
1446 * reginfo.c (global_regs_decl): Globalize.
1447 * rtl.h (global_regs_decl): Declare.
1448 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1449 reserved via global_regs.
1451 2014-01-15 Teresa Johnson <tejohnson@google.com>
1453 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1455 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1457 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1458 and vmulosh rather than call gen_vec_widen_smult_*.
1459 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1460 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1461 (vec_widen_smult_even_v16qi): Likewise.
1462 (vec_widen_umult_even_v8hi): Likewise.
1463 (vec_widen_smult_even_v8hi): Likewise.
1464 (vec_widen_umult_odd_v16qi): Likewise.
1465 (vec_widen_smult_odd_v16qi): Likewise.
1466 (vec_widen_umult_odd_v8hi): Likewise.
1467 (vec_widen_smult_odd_v8hi): Likewise.
1468 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1469 vmuloub rather than call gen_vec_widen_umult_*.
1470 (vec_widen_umult_lo_v16qi): Likewise.
1471 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1472 vmulosb rather than call gen_vec_widen_smult_*.
1473 (vec_widen_smult_lo_v16qi): Likewise.
1474 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1475 rather than call gen_vec_widen_umult_*.
1476 (vec_widen_umult_lo_v8hi): Likewise.
1477 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1478 rather than call gen_vec_widen_smult_*.
1479 (vec_widen_smult_lo_v8hi): Likewise.
1481 2014-01-15 Jeff Law <law@redhat.com>
1483 PR tree-optimization/59747
1484 * ree.c (find_and_remove_re): Properly handle case where a second
1485 eliminated extension requires widening a copy created for elimination
1486 of a prior extension.
1487 (combine_set_extension): Ensure that the number of hard regs needed
1488 for a destination register does not change when we widen it.
1490 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1492 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1493 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1494 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1495 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1496 (avr-*-rtems*): Likewise.
1497 (bfin*-rtems*): Likewise.
1498 (moxie-*-rtems*): Likewise.
1499 (h8300-*-rtems*): Likewise.
1500 (i[34567]86-*-rtems*): Likewise.
1501 (lm32-*-rtems*): Likewise.
1502 (m32r-*-rtems*): Likewise.
1503 (m68k-*-rtems*): Likewise.
1504 (microblaze*-*-rtems*): Likewise.
1505 (mips*-*-rtems*): Likewise.
1506 (powerpc-*-rtems*): Likewise.
1507 (sh-*-rtems*): Likewise.
1508 (sparc-*-rtems*): Likewise.
1509 (sparc64-*-rtems*): Likewise.
1510 (v850-*-rtems*): Likewise.
1511 (m32c-*-rtems*): Likewise.
1513 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1515 PR rtl-optimization/59511
1516 * ira.c (ira_init_register_move_cost): Use memory costs for some
1517 cases of register move cost calculations.
1518 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1519 instead of BB frequency.
1520 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1521 * lra-assigns.c (find_hard_regno_for): Ditto.
1523 2014-01-15 Richard Biener <rguenther@suse.de>
1525 PR tree-optimization/59822
1526 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1527 (vectorizable_load): Use it to hoist defs of uses of invariant
1528 loads out of the loop.
1530 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1531 Kugan Vivekanandarajah <kuganv@linaro.org>
1534 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1537 2014-01-15 Richard Biener <rguenther@suse.de>
1539 PR rtl-optimization/59802
1540 * lcm.c (compute_available): Use inverted postorder to seed
1541 the initial worklist.
1543 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1546 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1547 ADDR_REGS for invalid symrefs in non-PIC code.
1549 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1552 * builtins.c (determine_block_size): Initialize *probable_max_size
1553 even if len_rtx is CONST_INT.
1555 2014-01-14 Andrew Pinski <apinski@cavium.com>
1557 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1558 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1559 (cortexa53_tunings): Likewise.
1560 (aarch64_sched_issue_rate): New function.
1561 (TARGET_SCHED_ISSUE_RATE): Define.
1563 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1565 * ira-costs.c (find_costs_and_classes): Add missed
1566 ira_init_register_move_cost_if_necessary.
1568 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1571 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1573 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1576 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1577 to indicate if type is used for function return value. Warn ABI
1578 change if the vector mode isn't available for function return value.
1579 (ix86_function_arg_advance): Pass false to type_natural_mode.
1580 (ix86_function_arg): Likewise.
1581 (ix86_gimplify_va_arg): Likewise.
1582 (function_arg_32): Don't warn ABI change.
1583 (ix86_function_value): Pass true to type_natural_mode.
1584 (ix86_return_in_memory): Likewise.
1585 (ix86_struct_value_rtx): Removed.
1586 (TARGET_STRUCT_VALUE_RTX): Likewise.
1588 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1590 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1591 converting a conditional jump into a conditional return.
1593 2014-01-14 Richard Biener <rguenther@suse.de>
1595 PR tree-optimization/58921
1596 PR tree-optimization/59006
1597 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1598 hoisting invariant stmts.
1599 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1600 invariant loads on the preheader edge if possible.
1602 2014-01-14 Joey Ye <joey.ye@arm.com>
1604 * doc/plugin.texi (Building GCC plugins): Update to C++.
1606 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1608 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1609 (_mm_rcp28_round_ss): Ditto.
1610 (_mm_rsqrt28_round_sd): Ditto.
1611 (_mm_rsqrt28_round_ss): Ditto.
1612 (_mm_rcp28_sd): Ditto.
1613 (_mm_rcp28_ss): Ditto.
1614 (_mm_rsqrt28_sd): Ditto.
1615 (_mm_rsqrt28_ss): Ditto.
1616 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1617 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1618 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1619 (IX86_BUILTIN_RCP28SD): Ditto.
1620 (IX86_BUILTIN_RCP28SS): Ditto.
1621 (IX86_BUILTIN_RSQRT28SD): Ditto.
1622 (IX86_BUILTIN_RSQRT28SS): Ditto.
1623 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1624 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1625 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1626 (ix86_expand_special_args_builtin): Expand new FTYPE.
1627 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1628 (srcp14<mode>): Make insn unary.
1629 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1630 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1631 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1632 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1633 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1634 Fix rounding: make it SAE only.
1635 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1637 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1639 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1640 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1641 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1642 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1643 (round_saeonly_mask_scalar_operand4): Ditto.
1644 (round_saeonly_mask_scalar_op3): Ditto.
1645 (round_saeonly_mask_scalar_op4): Ditto.
1647 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1649 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1650 Implement -maltivec=be for vec_insert and vec_extract.
1652 2014-01-10 DJ Delorie <dj@redhat.com>
1654 * config/msp430/msp430.md (call_internal): Don't allow memory
1655 references with SP as the base register.
1656 (call_value_internal): Likewise.
1657 * config/msp430/constraints.md (Yc): New. For memory references
1658 that don't use SP as a base register.
1660 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1661 "an integer without a # prefix"
1662 * config/msp430/msp430.md (epilogue_helper): Use it.
1664 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1667 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1668 AVX512F gather builtins.
1669 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1670 on gather decls with INTEGER_TYPE masktype.
1671 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1672 directly into the builtin rather than hoisting it before loop.
1674 PR tree-optimization/59387
1675 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1676 (scev_const_prop): If folded_casts and type has undefined overflow,
1677 use force_gimple_operand instead of force_gimple_operand_gsi and
1678 for each added stmt if it is assign with
1679 arith_code_with_undefined_signed_overflow, call
1680 rewrite_to_defined_overflow.
1681 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1682 gimple-fold.h instead.
1683 (arith_code_with_undefined_signed_overflow,
1684 rewrite_to_defined_overflow): Moved to ...
1685 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1686 rewrite_to_defined_overflow): ... here. No longer static.
1687 Include gimplify-me.h.
1688 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1689 rewrite_to_defined_overflow): New prototypes.
1691 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1693 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1695 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1697 * builtins.c (get_object_alignment_2): Minor tweak.
1698 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1700 2014-01-13 Christian Bruel <christian.bruel@st.com>
1702 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1703 optimized non constant lengths.
1705 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1708 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1709 load as __atomic_load_N if possible.
1711 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1713 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1715 (rs6000_expand_builtin): Adjust call.
1717 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1720 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1721 * config/rs6000/rs6000.c: Include target-globals.h.
1722 (rs6000_set_current_function): Instead of doing target_reinit
1723 unconditionally, use save_target_globals_default_opts and
1724 restore_target_globals.
1726 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1728 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1729 (rs6000_expand_builtin): Handle mffs and mtfsf.
1730 (rs6000_init_builtins): Define mffs and mtfsf.
1731 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1732 (rs6000_mffs): New pattern.
1733 (rs6000_mtfsf): New pattern.
1735 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1737 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1738 Start narrowing with START. Apply candidate-use pair
1739 and check overall cost in narrowing.
1740 (iv_ca_prune): Pass new argument.
1742 2014-01-10 Jeff Law <law@redhat.com>
1745 * ree.c (combine_reaching_defs): Ensure the defining statement
1746 occurs before the extension when optimizing extensions with
1747 different source and destination hard registers.
1749 2014-01-10 Jan Hubicka <jh@suse.cz>
1752 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1753 vtables into the type inheritance graph.
1755 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1757 PR rtl-optimization/59754
1758 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1759 modes in the REGNO != REGNO case.
1761 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1763 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1765 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1767 PR tree-optimization/59745
1768 * tree-predcom.c (tree_predictive_commoning_loop): Call
1769 free_affine_expand_cache if giving up because components is NULL.
1771 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1772 GC in payload of target_globals struct instead of allocating them on
1773 the heap and the larger structs separately using GC.
1774 * target-globals.h (struct target_globals): Make regs, hard_regs,
1775 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1776 of GTY((skip)) and change type to void *.
1777 (reset_target_globals): Cast loads from those fields to corresponding
1780 2014-01-10 Steve Ellcey <sellcey@mips.com>
1783 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1784 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1785 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1787 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1790 * aarch64-modes.def (CC_Zmode): New flags mode.
1791 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1792 represents an equality.
1793 (aarch64_get_condition_code): Handle CC_Zmode.
1794 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1796 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1798 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1799 extraction in good case.
1801 2014-01-10 Richard Biener <rguenther@suse.de>
1803 PR tree-optimization/59374
1804 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1805 checking after SLP discovery. Mark stmts not participating
1806 in any SLP instance properly.
1808 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1810 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1811 when handling a SET rtx.
1813 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1815 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1816 (cortex-a57): Likewise.
1817 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1819 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1821 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1822 non-iwmmxt builtins.
1824 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1828 * ipa-devirt.c record_target_from_binfo): Take as argument
1829 stack of binfos and lookup matching one for virtual inheritance.
1830 (possible_polymorphic_call_targets_1): Update.
1832 2014-01-10 Huacai Chen <chenhc@lemote.com>
1834 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1835 kernel strings for Loongson-2E/2F/3A.
1837 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1840 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1841 is_gimple_call before calling gimple_call_internal_p.
1843 2014-01-09 Steve Ellcey <sellcey@mips.com>
1845 * Makefile.in (TREE_FLOW_H): Remove.
1846 (TREE_SSA_H): Add file names from tree-flow.h.
1847 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1848 * tree.h: Remove tree-flow.h reference.
1849 * hash-table.h: Remove tree-flow.h reference.
1850 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1851 reference with tree-ssa-loop.h.
1853 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1855 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1856 default element-order behavior for -maltivec.
1857 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1858 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1859 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1860 when targeting big endian, at least for now.
1861 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1863 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1866 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1867 var satisfies use_register_for_decl, just take into account type
1868 alignment, rather than decl alignment.
1870 PR tree-optimization/59622
1871 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1872 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1873 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1874 Don't devirtualize for inplace at all. For targets.length () == 1,
1875 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1877 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1879 * config/i386/i386.md (cpu): Remove the unused btver1.
1881 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1883 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1885 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1888 * tree-core.h (struct target_globals): New forward declaration.
1889 (struct tree_target_option): Add globals field.
1890 * tree.h (TREE_TARGET_GLOBALS): Define.
1891 (prepare_target_option_nodes_for_pch): New prototype.
1892 * target-globals.h (struct target_globals): Define even if
1894 * tree.c (prepare_target_option_node_for_pch,
1895 prepare_target_option_nodes_for_pch): New functions.
1896 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1897 * config/i386/i386.c: Include target-globals.h.
1898 (ix86_set_current_function): Instead of doing target_reinit
1899 unconditionally, use save_target_globals_default_opts and
1900 restore_target_globals.
1902 2014-01-09 Richard Biener <rguenther@suse.de>
1904 PR tree-optimization/59715
1905 * tree-cfg.h (split_critical_edges): Declare.
1906 * tree-cfg.c (split_critical_edges): Export.
1907 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1909 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1911 * cfgexpand.c (expand_stack_vars): Optionally disable
1912 asan stack protection.
1913 (expand_used_vars): Likewise.
1914 (partition_stack_vars): Likewise.
1915 * asan.c (asan_emit_stack_protection): Optionally disable
1916 after return stack usage.
1917 (instrument_derefs): Optionally disable memory access instrumentation.
1918 (instrument_builtin_call): Likewise.
1919 (instrument_strlen_call): Likewise.
1920 (asan_protect_global): Optionally disable global variables protection.
1921 * doc/invoke.texi: Added doc for new options.
1922 * params.def: Added new options.
1923 * params.h: Likewise.
1925 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1927 PR rtl-optimization/59724
1928 * ifcvt.c (cond_exec_process_if_block): Don't call
1929 flow_find_head_matching_sequence with 0 longest_match.
1930 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1931 non-active insns if !stop_after.
1932 (try_head_merge_bb): Revert 2014-01-07 changes.
1934 2014-01-08 Jeff Law <law@redhat.com>
1936 * ree.c (get_sub_rtx): New function, extracted from...
1937 (merge_def_and_ext): Here.
1938 (combine_reaching_defs): Use get_sub_rtx.
1940 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1942 * cgraph.h (varpool_variable_node): Do not choke on null node.
1944 2014-01-08 Catherine Moore <clm@codesourcery.com>
1946 * config/mips/mips.md (simple_return): Attempt to use JRC
1948 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1950 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1952 PR rtl-optimization/59137
1953 * reorg.c (steal_delay_list_from_target): Call update_block for
1955 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1957 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1959 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1960 two duplicate entries.
1962 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1965 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1967 * config/mips/mips.c (mips_truncated_op_cost): New function.
1968 (mips_rtx_costs): Adjust test for BADDU.
1969 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1971 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1973 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1974 (*baddu_si): ...this new pattern.
1976 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1979 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1981 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1984 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1986 (expand_expr, expand_normal): Adjust.
1987 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1988 inner_reference_p. Use inner_reference_p to expand inner references.
1989 (store_expr): Adjust.
1990 * cfgexpand.c (expand_call_stmt): Adjust.
1992 2014-01-08 Rong Xu <xur@google.com>
1994 * gcov-io.c (gcov_var): Move from gcov-io.h.
1995 (gcov_position): Ditto.
1996 (gcov_is_error): Ditto.
1997 (gcov_rewrite): Ditto.
1998 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1999 only part to libgcc/libgcov.h.
2001 2014-01-08 Marek Polacek <polacek@redhat.com>
2004 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2006 2014-01-08 Marek Polacek <polacek@redhat.com>
2009 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2011 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2013 PR rtl-optimization/59649
2014 * stor-layout.c (get_mode_bounds): For BImode return
2015 0 and STORE_FLAG_VALUE.
2017 2014-01-08 Richard Biener <rguenther@suse.de>
2020 * gimple.h (is_gimple_builtin_call): Remove.
2021 (gimple_builtin_call_types_compatible_p): New.
2022 (gimple_call_builtin_p): New overload.
2023 * gimple.c (is_gimple_builtin_call): Remove.
2024 (validate_call): Rename to ...
2025 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2027 (validate_type): New static function.
2028 (gimple_call_builtin_p): New overload and adjust.
2029 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2030 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2031 (gimple_fold_stmt_to_constant_1): Likewise.
2032 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2034 2014-01-08 Richard Biener <rguenther@suse.de>
2037 * gimplify.c (gimplify_expr): Gimplify register-register type
2038 VIEW_CONVERT_EXPRs to separate stmts.
2040 2014-01-07 Jeff Law <law@redhat.com>
2043 * ree.c (combine_set_extension): Handle case where source
2044 and destination registers in an extension insn are different.
2045 (combine_reaching_defs): Allow source and destination registers
2046 in extension to be different under limited circumstances.
2047 (add_removable_extension): Remove restriction that the
2048 source and destination registers in the extension are the same.
2049 (find_and_remove_re): Emit a copy from the extension's
2050 destination to its source after the defining insn if
2051 the source and destination registers are different.
2054 * ifcvt.c (merge_if_block): If we are merging a block with more than
2055 one successor with a block with no successors, remove any BARRIER
2056 after the second block.
2058 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2060 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2062 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2065 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2066 for 14-bit register offsets when INT14_OK_STRICT is false.
2068 2014-01-07 Roland Stigge <stigge@antcom.de>
2069 Michael Meissner <meissner@linux.vnet.ibm.com>
2072 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2073 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2075 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2077 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2080 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2082 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2083 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2084 rtx is const0_rtx or not.
2086 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2089 * target-globals.c (save_target_globals): Remove this_fn_optab
2091 * toplev.c: Include optabs.h.
2092 (target_reinit): Temporarily restore the global options if another
2093 set of options are in force.
2095 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2097 PR rtl-optimization/58668
2098 * cfgcleanup.c (flow_find_cross_jump): Don't count
2099 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2100 to determine what is counted.
2101 (flow_find_head_matching_sequence): Use active_insn_p to determine
2103 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2105 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2106 determine what is counted.
2108 PR tree-optimization/59643
2109 * tree-predcom.c (split_data_refs_to_components): If one dr is
2110 read and one write, determine_offset fails and the write isn't
2111 in the bad component, just put the read into the bad component.
2113 2014-01-07 Mike Stump <mikestump@comcast.net>
2114 Jakub Jelinek <jakub@redhat.com>
2117 * tree-core.h (struct tree_optimization_option): Change optabs
2118 type from unsigned char * to void *.
2119 * optabs.c (init_tree_optimization_optabs): Adjust
2120 TREE_OPTIMIZATION_OPTABS initialization.
2122 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2125 * config/i386/i386.h (struct machine_function): Add
2126 no_drap_save_restore field.
2127 * config/i386/i386.c (ix86_save_reg): Use
2128 !cfun->machine->no_drap_save_restore instead of
2129 crtl->stack_realign_needed.
2130 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2131 this function clears frame_pointer_needed. Set
2132 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2133 and DRAP reg is needed.
2135 2014-01-06 Marek Polacek <polacek@redhat.com>
2138 * doc/implement-c.texi: Mention that other integer types are
2139 permitted as bit-field types in strictly conforming mode.
2141 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2143 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2146 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2148 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2150 2014-01-06 Martin Jambor <mjambor@suse.cz>
2153 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2155 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2157 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2161 * var-tracking.c (add_stores): Preserve the value of the source even if
2162 we don't record the store.
2164 2014-01-06 Terry Guo <terry.guo@arm.com>
2166 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2168 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2171 * config/darwin.c (darwin_function_section): Adjust return values to
2172 correspond to optimisation changes made in r206070.
2174 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2176 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2177 from prefetch_block tune setting.
2178 (nocona_cost): Correct size of prefetch block to 64.
2180 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2182 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2183 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2184 used to save the static chain register in the computation of the offset
2185 from which the FP registers need to be restored.
2187 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2189 PR tree-optimization/59519
2190 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2191 ICE if get_current_def (current_new_name) is already non-NULL, as long
2192 as it is a phi result of some other phi in *new_exit_bb that has
2195 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2196 or vmovdqu* for misaligned_operand.
2197 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2198 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2199 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2200 aligned_mem for AVX512F masked aligned load and store builtins and for
2203 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2205 PR tree-optimization/59651
2206 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2207 Address range for negative step should be added by TYPE_SIZE_UNIT.
2209 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2211 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2212 overlapping registers also for registers other than the stack pointer.
2214 2014-01-03 Marek Polacek <polacek@redhat.com>
2217 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2220 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2223 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2226 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2227 (push splitter): Use <P:MODE_SIZE> instead of
2228 GET_MODE_SIZE (<P:MODE>mode).
2229 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2230 (mov -1, reg peephole2): Likewise.
2231 * config/i386/sse.md (*mov<mode>_internal,
2232 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2233 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2234 *<code><mode>3, *andnot<mode>3<mask_name>,
2235 <mask_codefor><code><mode>3<mask_name>): Likewise.
2236 * config/i386/subst.md (mask_mode512bit_condition,
2237 sd_mask_mode512bit_condition): Likewise.
2239 2014-01-02 Xinliang David Li <davidxl@google.com>
2241 PR tree-optimization/59303
2242 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2243 (dump_predicates): Better output format.
2244 (pred_equal_p): New function.
2245 (is_neq_relop_p): Ditto.
2246 (is_neq_zero_form_p): Ditto.
2247 (pred_expr_equal_p): Ditto.
2248 (pred_neg_p): Ditto.
2249 (simplify_pred): Ditto.
2250 (simplify_preds_2): Ditto.
2251 (simplify_preds_3): Ditto.
2252 (simplify_preds_4): Ditto.
2253 (simplify_preds): Ditto.
2255 (push_to_worklist): Ditto.
2256 (get_pred_info_from_cmp): Ditto.
2257 (is_degenerated_phi): Ditto.
2258 (normalize_one_pred_1): Ditto.
2259 (normalize_one_pred): Ditto.
2260 (normalize_one_pred_chain): Ditto.
2261 (normalize_preds): Ditto.
2262 (normalize_cond_1): Remove function.
2263 (normalize_cond): Ditto.
2264 (is_gcond_subset_of): Ditto.
2265 (is_subset_of_any): Ditto.
2266 (is_or_set_subset_of): Ditto.
2267 (is_and_set_subset_of): Ditto.
2268 (is_norm_cond_subset_of): Ditto.
2269 (pred_chain_length_cmp): Ditto.
2270 (convert_control_dep_chain_into_preds): Type change.
2271 (find_predicates): Ditto.
2272 (find_def_preds): Ditto.
2273 (destroy_predicates_vecs): Ditto.
2274 (find_matching_predicates_in_rest_chains): Ditto.
2275 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2276 (is_pred_expr_subset): Ditto.
2277 (is_pred_chain_subset_of): Ditto.
2278 (is_included_in): Ditto.
2279 (is_superset_of): Ditto.
2281 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2283 Update copyright years.
2285 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2287 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2288 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2289 config/arc/arc.md, config/arc/arc.opt,
2290 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2291 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2292 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2293 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2294 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2295 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2296 vtable-verify.c, vtable-verify.h: Use the standard form for the
2299 2014-01-02 Tobias Burnus <burnus@net-b.de>
2301 * gcc.c (process_command): Update copyright notice dates.
2302 * gcov-dump.c: Ditto.
2304 * doc/cpp.texi: Bump @copying's copyright year.
2305 * doc/cppinternals.texi: Ditto.
2306 * doc/gcc.texi: Ditto.
2307 * doc/gccint.texi: Ditto.
2308 * doc/gcov.texi: Ditto.
2309 * doc/install.texi: Ditto.
2310 * doc/invoke.texi: Ditto.
2312 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2314 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2316 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2318 * config/i386/sse.md (*mov<mode>_internal): Guard
2319 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2321 PR rtl-optimization/59647
2322 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2323 new_rtx into UNSIGNED_FLOAT rtxes.
2325 Copyright (C) 2014 Free Software Foundation, Inc.
2327 Copying and distribution of this file, with or without modification,
2328 are permitted in any medium without royalty provided the copyright
2329 notice and this notice are preserved.