1 2014-01-30 DJ Delorie <dj@redhat.com>
3 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6 whenever main() has an epilogue.
8 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
11 unused variable "field".
12 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
13 (vsx_mergeh_<mode>): Likewise.
14 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
15 (altivec_vmrghh): Likewise.
16 (altivec_vmrghw): Likewise.
17 (altivec_vmrglb): Likewise.
18 (altivec_vmrglh): Likewise.
19 (altivec_vmrglw): Likewise.
20 (altivec_vspltb): Add missing uses.
21 (altivec_vsplth): Likewise.
22 (altivec_vspltw): Likewise.
23 (altivec_vspltsf): Likewise.
25 2014-01-30 Jakub Jelinek <jakub@redhat.com>
28 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
29 frame related instructions.
31 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
33 PR rtl-optimization/59959
34 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
35 any reload of register whose subreg is invalid.
37 2014-01-30 Jakub Jelinek <jakub@redhat.com>
39 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing
41 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
42 Add missing return type - void.
44 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
46 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
47 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
48 remove element index adjustment for endian (now handled in vsx.md
50 (altivec_expand_vec_perm_const): Use
51 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
52 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
53 (vsx_xxspltw_<mode>): Adjust element index for little endian.
54 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
55 define_expand and a new define_insn *altivec_vspltb_internal;
56 adjust for -maltivec=be on a little endian target.
57 (altivec_vspltb_direct): New.
58 (altivec_vsplth): Divide into a define_expand and a new
59 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
61 (altivec_vsplth_direct): New.
62 (altivec_vspltw): Divide into a define_expand and a new
63 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
65 (altivec_vspltw_direct): New.
66 (altivec_vspltsf): Divide into a define_expand and a new
67 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
68 a little endian target.
70 2014-01-30 Richard Biener <rguenther@suse.de>
72 PR tree-optimization/59993
73 * tree-ssa-forwprop.c (associate_pointerplus): Check we
74 can propagate form the earlier stmt and avoid the transform
75 when the intermediate result is needed.
77 2014-01-30 Alangi Derick <alangiderick@gmail.com>
79 * README.Portability: Fix typo.
81 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
83 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
84 comparison_operator with ordered_comparison_operator.
86 2014-01-30 Nick Clifton <nickc@redhat.com>
88 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
89 Rename to mn10300_store_multiple_regs.
90 * config/mn10300/mn10300.c: Likewise.
91 * config/mn10300/mn10300.md (store_movm): Fix typo: call
93 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
94 Call mn10300_store_multiple_regs.
96 2014-01-30 Nick Clifton <nickc@redhat.com>
97 DJ Delorie <dj@redhat.com>
99 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
100 %fp 2 to keep registers after it properly word-aligned.
101 (rl78_alloc_physical_registers_umul): Handle the case where both
102 input operands are the same.
104 2014-01-30 Richard Biener <rguenther@suse.de>
106 PR tree-optimization/59903
107 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
110 2014-01-30 Jason Merrill <jason@redhat.com>
113 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
116 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
118 2014-01-30 Richard Biener <rguenther@suse.de>
120 PR tree-optimization/59951
121 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
124 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
127 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
128 SFmode to DFmode case.
130 2014-01-29 DJ Delorie <dj@redhat.com>
132 * config/msp430/msp430.opt (-minrt): New.
133 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
135 (ENDFILE_SPEC): Likewise.
137 2014-01-29 Jan Hubicka <jh@suse.cz>
139 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
140 (estimate_function_body_sizes): Use it.
142 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
145 * dwarf2out.c (is_cxx_auto): New.
146 (is_base_type): Use it.
147 (gen_type_die_with_usage): Likewise.
149 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
151 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
152 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
153 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
154 -maltivec=be with LE targets.
155 (vsx_mergeh_<mode>): Likewise.
156 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
158 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
159 (altivec_vmrghb): Replace with define_expand and new
160 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
162 (altivec_vmrghb_direct): New define_insn.
163 (altivec_vmrghh): Replace with define_expand and new
164 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
166 (altivec_vmrghh_direct): New define_insn.
167 (altivec_vmrghw): Replace with define_expand and new
168 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
170 (altivec_vmrghw_direct): New define_insn.
171 (*altivec_vmrghsf): Adjust for endianness.
172 (altivec_vmrglb): Replace with define_expand and new
173 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
175 (altivec_vmrglb_direct): New define_insn.
176 (altivec_vmrglh): Replace with define_expand and new
177 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
179 (altivec_vmrglh_direct): New define_insn.
180 (altivec_vmrglw): Replace with define_expand and new
181 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
183 (altivec_vmrglw_direct): New define_insn.
184 (*altivec_vmrglsf): Adjust for endianness.
185 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
186 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
187 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
188 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
189 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
190 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
191 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
192 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
194 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
196 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
197 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
200 2014-01-29 Richard Biener <rguenther@suse.de>
202 PR tree-optimization/58742
203 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
204 associate_pointerplus_align.
205 (associate_pointerplus_diff): New function.
206 (associate_pointerplus): Likewise. Call associate_pointerplus_align
207 and associate_pointerplus_diff.
209 2014-01-29 Richard Biener <rguenther@suse.de>
211 * lto-streamer.h (LTO_major_version): Bump to 3.
212 (LTO_minor_version): Reset to 0.
214 2014-01-29 Renlin Li <Renlin.Li@arm.com>
216 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
217 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
218 (arm_file_start): Generate correct asm header for armv7ve.
219 * config/arm/bpabi.h: Add multilib support for armv7ve.
220 * config/arm/driver-arm.c: Change the architectures of cortex-a7
221 and cortex-a15 to armv7ve.
222 * config/arm/t-aprofile: Add multilib support for armv7ve.
223 * doc/invoke.texi: Document -march=armv7ve.
225 2014-01-29 Richard Biener <rguenther@suse.de>
227 PR tree-optimization/58742
228 * tree-ssa-forwprop.c (associate_plusminus): Return true
229 if we changed sth, defer EH cleanup to ...
230 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
231 (simplify_mult): New function.
233 2014-01-29 Jakub Jelinek <jakub@redhat.com>
236 PR tree-optimization/59920
237 * tree.c (build_common_builtin_nodes): Remove
238 __builtin_setjmp_dispatcher initialization.
239 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
240 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
241 instead of gsi_after_labels + manually skipping debug stmts.
242 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
243 ignore bbs with IFN_ABNORMAL_DISPATCHER.
244 * tree-inline.c (copy_edges_for_bb): Remove
245 can_make_abnormal_goto argument, instead add abnormal_goto_dest
246 argument. Ignore computed_goto_p stmts. Don't call
247 make_abnormal_goto_edges. If a call might need abnormal edges
248 for non-local gotos, see if it already has an edge to
249 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
250 with true argument, don't do anything then, otherwise add
251 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
252 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
254 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
255 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
256 (lower_stmt): Don't set data->calls_builtin_setjmp.
257 (lower_builtin_setjmp): Adjust comment.
258 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
259 * tree-cfg.c (found_computed_goto): Remove.
260 (factor_computed_gotos): Remove.
261 (make_goto_expr_edges): Return bool, true for computed gotos.
262 Don't call make_abnormal_goto_edges.
263 (build_gimple_cfg): Don't set found_computed_goto, don't call
264 factor_computed_gotos.
265 (computed_goto_p): No longer static.
266 (make_blocks): Don't set found_computed_goto.
267 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
268 (make_edges): If make_goto_expr_edges returns true, push bb
269 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
270 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
271 vector. Record mapping between bbs and OpenMP regions if there
272 are any, adjust make_gimple_omp_edges caller. Call
273 handle_abnormal_edges.
274 (make_abnormal_goto_edges): Remove.
275 * tree-cfg.h (make_abnormal_goto_edges): Remove.
276 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
277 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
278 * builtins.c (expand_builtin): Don't handle
279 BUILT_IN_SETJMP_DISPATCHER.
280 * internal-fn.def (ABNORMAL_DISPATCHER): New.
281 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
282 filling *region also set *region_idx to (*region)->entry->index.
285 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
286 For REGs set ORIGINAL_REGNO.
288 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
290 * doc/md.texi: Mention that a target shouldn't implement
291 vec_widen_(s|u)mul_even/odd pair if it is less efficient
294 2014-01-29 Jakub Jelinek <jakub@redhat.com>
296 PR tree-optimization/59594
297 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
298 a copy of the datarefs vector rather than the vector itself.
300 2014-01-28 Jason Merrill <jason@redhat.com>
303 * dwarf2out.c (auto_die): New static.
304 (gen_type_die_with_usage): Handle C++1y 'auto'.
305 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
308 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
311 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
312 (SPEC_X32): Likewise.
314 * config/i386/i386.c (ix86_option_override_internal): Turn off
315 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
317 (x86_file_start): Output .code16gcc for TARGET_16BIT.
318 * config/i386/i386.h (TARGET_16BIT): New macro.
319 (TARGET_16BIT_P): Likewise.
320 * config/i386/i386.opt: Add m16.
321 * doc/invoke.texi: Document -m16.
323 2014-01-28 Jakub Jelinek <jakub@redhat.com>
325 PR preprocessor/59935
326 * input.c (location_get_source_line): Bail out on when line number
327 is zero, and test the return value of
328 lookup_or_add_file_to_cache_tab.
330 2014-01-28 Richard Biener <rguenther@suse.de>
332 PR tree-optimization/58742
333 * tree-ssa-forwprop.c (associate_plusminus): Handle
334 pointer subtraction of the form (T)(P + A) - (T)P.
336 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
338 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
341 2014-01-28 Richard Biener <rguenther@suse.de>
344 2014-01-28 Richard Biener <rguenther@suse.de>
346 PR rtl-optimization/45364
347 PR rtl-optimization/59890
348 * var-tracking.c (local_get_addr_clear_given_value): Handle
349 already cleared slot.
350 (val_reset): Handle not allocated local_get_addr_cache.
351 (vt_find_locations): Use post-order on the inverted CFG.
353 2014-01-28 Richard Biener <rguenther@suse.de>
355 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
358 2014-01-28 Richard Biener <rguenther@suse.de>
360 PR rtl-optimization/45364
361 PR rtl-optimization/59890
362 * var-tracking.c (local_get_addr_clear_given_value): Handle
363 already cleared slot.
364 (val_reset): Handle not allocated local_get_addr_cache.
365 (vt_find_locations): Use post-order on the inverted CFG.
367 2014-01-28 Alan Modra <amodra@gmail.com>
369 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
370 * configure.ac <recursive call for build != host>: Define
371 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
372 and LD_FOR_BUILD too.
373 * configure: Regenerate.
375 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
377 * config/i386/i386.c (get_builtin_code_for_version): Separate
378 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
379 Broadwell from Haswell.
381 2014-01-27 Steve Ellcey <sellcey@mips.com>
383 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
384 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
385 * config/mips/mips.c (mips_option_override): Change setting
387 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
388 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
389 Change from Mask to Var.
391 2014-01-27 Jeff Law <law@redhat.com>
393 * ipa-inline.c (inline_small_functions): Fix typo.
395 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
397 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
398 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
399 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
400 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
401 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
402 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
403 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
404 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
405 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
406 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
407 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
408 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
409 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
410 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
411 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
412 (_mm512_storeu_epi64): Ditto.
413 (_mm512_cmpge_epi32_mask): Ditto.
414 (_mm512_cmpge_epu32_mask): Ditto.
415 (_mm512_cmpge_epi64_mask): Ditto.
416 (_mm512_cmpge_epu64_mask): Ditto.
417 (_mm512_cmple_epi32_mask): Ditto.
418 (_mm512_cmple_epu32_mask): Ditto.
419 (_mm512_cmple_epi64_mask): Ditto.
420 (_mm512_cmple_epu64_mask): Ditto.
421 (_mm512_cmplt_epi32_mask): Ditto.
422 (_mm512_cmplt_epu32_mask): Ditto.
423 (_mm512_cmplt_epi64_mask): Ditto.
424 (_mm512_cmplt_epu64_mask): Ditto.
425 (_mm512_cmpneq_epi32_mask): Ditto.
426 (_mm512_cmpneq_epu32_mask): Ditto.
427 (_mm512_cmpneq_epi64_mask): Ditto.
428 (_mm512_cmpneq_epu64_mask): Ditto.
429 (_mm512_expand_pd): Ditto.
430 (_mm512_expand_ps): Ditto.
431 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
432 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
433 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
434 * config/i386/i386.c (ix86_builtins): Add
435 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
436 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
437 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
438 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
439 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
440 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
441 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
442 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
443 IX86_BUILTIN_PMOVUSQW512_MEM.
444 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
445 __builtin_ia32_pmovsqd512mem_mask,
446 __builtin_ia32_pmovqd512mem_mask,
447 __builtin_ia32_pmovusqw512mem_mask,
448 __builtin_ia32_pmovsqw512mem_mask,
449 __builtin_ia32_pmovqw512mem_mask,
450 __builtin_ia32_pmovusdw512mem_mask,
451 __builtin_ia32_pmovsdw512mem_mask,
452 __builtin_ia32_pmovdw512mem_mask,
453 __builtin_ia32_pmovqb512mem_mask,
454 __builtin_ia32_pmovusqb512mem_mask,
455 __builtin_ia32_pmovsqb512mem_mask,
456 __builtin_ia32_pmovusdb512mem_mask,
457 __builtin_ia32_pmovsdb512mem_mask,
458 __builtin_ia32_pmovdb512mem_mask.
459 (bdesc_args): Add __builtin_ia32_expanddf512,
460 __builtin_ia32_expandsf512.
461 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
462 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
463 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
464 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
465 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
466 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
467 (avx512f_<code>v8div16qi2_mask_store): This.
468 (avx512f_expand<mode>): New.
470 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
472 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
474 (_mm512_mask_prefetch_i64gather_pd): Ditto.
475 (_mm512_prefetch_i32scatter_pd): Ditto.
476 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
477 (_mm512_prefetch_i64scatter_pd): Ditto.
478 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
479 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
480 (_mm512_mask_prefetch_i64gather_ps): Ditto.
481 (_mm512_prefetch_i32scatter_ps): Ditto.
482 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
483 (_mm512_prefetch_i64scatter_ps): Ditto.
484 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
485 * config/i386/i386-builtin-types.def: Define
486 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
487 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
488 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
489 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
490 IX86_BUILTIN_SCATTERPFQPD.
491 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
492 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
493 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
494 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
495 __builtin_ia32_scatterpfqps.
496 (ix86_expand_builtin): Expand new built-ins.
497 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
498 fix memory access data type.
499 (*avx512pf_gatherpf<mode>_mask): Ditto.
500 (*avx512pf_gatherpf<mode>): Ditto.
501 (avx512pf_scatterpf<mode>): Ditto.
502 (*avx512pf_scatterpf<mode>_mask): Ditto.
503 (*avx512pf_scatterpf<mode>): Ditto.
504 (GATHER_SCATTER_SF_MEM_MODE): New.
505 (avx512pf_gatherpf<mode>df): Ditto.
506 (*avx512pf_gatherpf<mode>df_mask): Ditto.
507 (*avx512pf_scatterpf<mode>df): Ditto.
509 2014-01-27 Jakub Jelinek <jakub@redhat.com>
512 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
513 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
516 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
518 * common/config/arm/arm-common.c
519 (arm_rewrite_mcpu): Handle multiple names.
521 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
523 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
525 * gimple-builder.h (create_gimple_tmp): Delete.
527 2014-01-27 Christian Bruel <christian.bruel@st.com>
529 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
532 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
534 * config/pa/pa.md (call): Generate indirect long calls to non-local
535 functions when outputing 32-bit code.
536 (call_value): Likewise except for special call to buggy powf function.
538 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
539 portable runtime and PIC indirect calls.
540 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
541 and PIC call sequences. Use ldo instead of blr to set return register
542 in PIC call sequence.
544 2014-01-25 Walter Lee <walt@tilera.com>
546 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
547 avoid clobbering a live register.
549 2014-01-25 Walter Lee <walt@tilera.com>
551 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
552 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
553 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
554 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
556 2014-01-25 Walter Lee <walt@tilera.com>
558 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
559 arguments on even registers.
560 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
562 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
563 (BIGGEST_ALIGNMENT): Ditto.
564 (BIGGEST_FIELD_ALIGNMENT): Ditto.
566 2014-01-25 Walter Lee <walt@tilera.com>
568 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
569 insns before bundling.
570 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
572 2014-01-25 Walter Lee <walt@tilera.com>
574 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
575 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
576 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
578 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
580 * config/mips/constraints.md (kl): Delete.
581 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
582 define expands, using...
583 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
584 instructions for MIPS16.
585 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
586 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
588 2014-01-25 Walter Lee <walt@tilera.com>
590 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
594 2014-01-25 Walter Lee <walt@tilera.com>
596 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
597 (TARGET_EXPAND_TO_RTL_HOOK): Define.
599 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
601 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
604 2014-01-25 Jakub Jelinek <jakub@redhat.com>
606 * print-rtl.c (in_call_function_usage): New var.
607 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
608 EXPR_LIST mode as mode and not as reg note name.
611 * cfgloopmanip.c (copy_loop_info): If
612 loop->warned_aggressive_loop_optimizations, make sure
613 the flag is set in target loop too.
615 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
617 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
619 * builtins.def: Likewise.
620 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
621 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
622 * ira.c (ira_setup_eliminable_regset): Likewise.
623 * omp-low.c (gate_expand_omp): Likewise.
624 (execute_lower_omp): Likewise.
625 (diagnose_sb_0): Likewise.
626 (gate_diagnose_omp_blocks): Likewise.
627 (simd_clone_clauses_extract): Likewise.
630 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
632 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
633 correction for little endian...
634 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
637 2014-01-24 Jeff Law <law@redhat.com>
639 PR tree-optimization/59919
640 * tree-vrp.c (find_assert_locations_1): Do not register asserts
641 for non-returning calls.
643 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
645 * common/config/aarch64/aarch64-common.c
646 (aarch64_rewrite_mcpu): Handle multiple names.
647 * config/aarch64/aarch64.h
648 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
650 2014-01-24 Dodji Seketeli <dodji@redhat.com>
652 * input.c (add_file_to_cache_tab): Handle the case where fopen
655 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
658 * config/i386/i386.md (pushsf splitter): Get stack adjustment
659 from push operand if code of push isn't PRE_DEC.
661 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
664 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
665 -mquad-memory-atomic. Update -mquad-memory documentation to say
666 it is only used for non-atomic loads/stores.
668 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
669 -mquad-memory or -mquad-memory-atomic switches.
671 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
672 -mquad-memory-atomic to ISA 2.07 support.
674 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
675 to separate support of normal quad word memory operations (ldq, stq)
676 from the atomic quad word memory operations.
678 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
679 support to separate non-atomic quad word operations from atomic
680 quad word operations. Disable non-atomic quad word operations in
681 little endian mode so that we don't have to swap words after the
682 load and before the store.
683 (quad_load_store_p): Add comment about atomic quad word support.
684 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
685 options printed with -mdebug=reg.
687 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
688 -mquad-memory-atomic as the test for whether we have quad word
690 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
691 or -mp8-vector are used, allow byte/half-word atomic operations.
693 * config/rs6000/sync.md (load_lockedti): Insure that the address
694 is a proper indexed or indirect address for the lqarx instruction.
695 On little endian systems, swap the hi/lo registers after the lqarx
697 (load_lockedpti): Use indexed_or_indirect_operand predicate to
698 insure the address is valid for the lqarx instruction.
699 (store_conditionalti): Insure that the address is a proper indexed
700 or indirect address for the stqcrx. instruction. On little endian
701 systems, swap the hi/lo registers before doing the stqcrx.
703 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
704 insure the address is valid for the stqcrx. instruction.
706 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
707 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
708 type of quad memory support is available.
710 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
713 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
714 there is a danger of looping.
716 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
718 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
719 force flag_ira_loop_pressure if set via command line.
721 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
723 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
724 (ashr_simd): New builtin handling DI mode.
725 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
726 (aarch64_sshr_simddi): New match pattern.
727 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
728 (vshrd_n_s64): Likewise.
729 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
731 2014-01-23 Nick Clifton <nickc@redhat.com>
733 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
734 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
735 favour of mcu specific scripts.
736 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
739 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
740 Alex Velenko <Alex.Velenko@arm.com>
742 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
743 (vaddv_s16): Likewise.
744 (vaddv_s32): Likewise.
745 (vaddv_u8): Likewise.
746 (vaddv_u16): Likewise.
747 (vaddv_u32): Likewise.
748 (vaddvq_s8): Likewise.
749 (vaddvq_s16): Likewise.
750 (vaddvq_s32): Likewise.
751 (vaddvq_s64): Likewise.
752 (vaddvq_u8): Likewise.
753 (vaddvq_u16): Likewise.
754 (vaddvq_u32): Likewise.
755 (vaddvq_u64): Likewise.
756 (vaddv_f32): Likewise.
757 (vaddvq_f32): Likewise.
758 (vaddvq_f64): Likewise.
759 (vmaxv_f32): Likewise.
760 (vmaxv_s8): Likewise.
761 (vmaxv_s16): Likewise.
762 (vmaxv_s32): Likewise.
763 (vmaxv_u8): Likewise.
764 (vmaxv_u16): Likewise.
765 (vmaxv_u32): Likewise.
766 (vmaxvq_f32): Likewise.
767 (vmaxvq_f64): Likewise.
768 (vmaxvq_s8): Likewise.
769 (vmaxvq_s16): Likewise.
770 (vmaxvq_s32): Likewise.
771 (vmaxvq_u8): Likewise.
772 (vmaxvq_u16): Likewise.
773 (vmaxvq_u32): Likewise.
774 (vmaxnmv_f32): Likewise.
775 (vmaxnmvq_f32): Likewise.
776 (vmaxnmvq_f64): Likewise.
777 (vminv_f32): Likewise.
778 (vminv_s8): Likewise.
779 (vminv_s16): Likewise.
780 (vminv_s32): Likewise.
781 (vminv_u8): Likewise.
782 (vminv_u16): Likewise.
783 (vminv_u32): Likewise.
784 (vminvq_f32): Likewise.
785 (vminvq_f64): Likewise.
786 (vminvq_s8): Likewise.
787 (vminvq_s16): Likewise.
788 (vminvq_s32): Likewise.
789 (vminvq_u8): Likewise.
790 (vminvq_u16): Likewise.
791 (vminvq_u32): Likewise.
792 (vminnmv_f32): Likewise.
793 (vminnmvq_f32): Likewise.
794 (vminnmvq_f64): Likewise.
796 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
798 * config/aarch64/aarch64-simd.md
799 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
800 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
801 (*aarch64_mul3_elt<mode>): Likewise.
802 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
803 (*aarch64_mul3_elt_to_64v2df): Likewise.
804 (*aarch64_mla_elt<mode>): Likewise.
805 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
806 (*aarch64_mls_elt<mode>): Likewise.
807 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
808 (*aarch64_fma4_elt<mode>): Likewise.
809 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
810 (*aarch64_fma4_elt_to_64v2df): Likewise.
811 (*aarch64_fnma4_elt<mode>): Likewise.
812 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
813 (*aarch64_fnma4_elt_to_64v2df): Likewise.
814 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
815 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
816 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
817 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
818 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
819 (aarch64_sqdmull_lane<mode>_internal): Likewise.
820 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
822 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
824 * config/aarch64/aarch64-simd.md
825 (aarch64_be_checked_get_lane<mode>): New define_expand.
826 * config/aarch64/aarch64-simd-builtins.def
827 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
828 New builtin definition.
829 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
830 Use new safe be builtin.
832 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
834 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
836 (aarch64_be_st1<mode>): Likewise.
837 (aarch_ld1<VALL:mode>): Define_expand modified.
838 (aarch_st1<VALL:mode>): Likewise.
839 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
840 (UNSPEC_ST1): Likewise.
842 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
844 * config/microblaze/microblaze.md: Add trap insn and attribute
846 2014-01-23 Dodji Seketeli <dodji@redhat.com>
848 PR preprocessor/58580
849 * input.h (location_get_source_line): Take an additional line_size
851 (void diagnostics_file_cache_fini): Declare new function.
852 * input.c (struct fcache): New type.
853 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
854 New static constants.
855 (diagnostic_file_cache_init, total_lines_num)
856 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
857 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
858 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
859 (get_next_line, read_next_line, goto_next_line, read_line_num):
860 New static function definitions.
861 (diagnostic_file_cache_fini): New function.
862 (location_get_source_line): Take an additional output line_len
863 parameter. Re-write using lookup_or_add_file_to_cache_tab and
865 * diagnostic.c (diagnostic_finish): Call
866 diagnostic_file_cache_fini.
867 (adjust_line): Take an additional input parameter for the length
868 of the line, rather than calculating it with strlen.
869 (diagnostic_show_locus): Adjust the use of
870 location_get_source_line and adjust_line with respect to their new
871 signature. While displaying a line now, do not stop at the first
872 null byte. Rather, display the zero byte as a space and keep
873 going until we reach the size of the line.
874 * Makefile.in: Add vec.o to OBJS-libcommon
876 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
877 Ilya Tocar <ilya.tocar@intel.com>
879 * config/i386/avx512fintrin.h (_mm512_kmov): New.
880 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
881 (__builtin_ia32_kmov16): Ditto.
882 * config/i386/i386.md (UNSPEC_KMOV): New.
885 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
887 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
888 (_mm512_storeu_si512): Ditto.
890 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
893 * rtl.h (get_referenced_operands): Declare.
894 * recog.c (get_referenced_operands): New function.
895 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
896 operands have been referenced when recording LO_SUM references.
898 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
900 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
902 2014-01-22 Jan Hubicka <jh@suse.cz>
904 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
905 Enable for generic and recent AMD targets.
907 2014-01-22 Jan Hubicka <jh@suse.cz>
909 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
910 ARG_SIZE note when adjustment was eliminated.
912 2014-01-22 Jeff Law <law@redhat.com>
914 PR tree-optimization/59597
915 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
916 in file. Accept new argument REGISTERING and use it to modify
917 dump output appropriately.
918 (register_jump_thread): Corresponding changes.
919 (mark_threaded_blocks): Reinstate code to cancel unprofitable
920 thread paths involving joiner blocks. Add code to dump cancelled
921 jump threading paths.
923 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
925 PR rtl-optimization/59477
926 * lra-constraints.c (inherit_in_ebb): Process call for living hard
927 regs. Update reloads_num and potential_reload_hard_regs for all insns.
929 2014-01-22 Tom Tromey <tromey@redhat.com>
931 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
933 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
935 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
937 PR rtl-optimization/59896
938 * lra-constraints.c (process_alt_operands): Check unused note for
939 matched operands of insn with no output reloads.
941 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
943 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
944 (mips_move_from_gpr_cost): Likewise.
946 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
948 PR rtl-optimization/59858
949 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
950 ira_class_hard_regs_num.
951 (process_alt_operands): Increase reject for dying matched operand.
953 2014-01-21 Jakub Jelinek <jakub@redhat.com>
956 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
957 smaller than size, perform several stores or loads and stores
958 at dst + count - size to store or copy all of size bytes, rather
959 than just last modesize bytes.
961 2014-01-20 DJ Delorie <dj@redhat.com>
963 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
964 that CLOBBERs are REGs before propogating their values.
966 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
969 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
970 (cgraph_inline_failed_type): New function.
971 * cgraph.h (DEFCIFCODE): Add type.
972 (cgraph_inline_failed_type_t): New enum.
973 (cgraph_inline_failed_type): New prototype.
974 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
975 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
976 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
977 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
978 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
979 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
980 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
981 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
982 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
983 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
984 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
985 OPTIMIZATION_MISMATCH.
986 * tree-inline.c (expand_call_inline): Emit errors during
987 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
989 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
992 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
993 mode attribute in insn output.
995 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
997 * output.h (output_constant): Delete.
998 * varasm.c (output_constant): Make private.
1000 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1002 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1004 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1007 * tree.h (fold_builtin_strcat): New prototype.
1008 * builtins.c (fold_builtin_strcat): No longer static. Add len
1009 argument, if non-NULL, don't call c_strlen. Optimize
1010 directly into __builtin_memcpy instead of __builtin_strcpy.
1011 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1012 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1014 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1016 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1017 for SImode_address_operand operands, having only a REG argument.
1019 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1021 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1022 loader name using mbig-endian.
1023 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1025 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1027 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1031 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1033 * config/aarch64/aarch64-protos.h
1034 (aarch64_cannot_change_mode_class_ptr): Declare.
1035 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1036 aarch64_cannot_change_mode_class_ptr): New.
1037 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1038 backend hook aarch64_cannot_change_mode_class.
1040 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1042 * common/config/aarch64/aarch64-common.c
1043 (aarch64_handle_option): Don't handle any option order logic here.
1044 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1045 selected_cpu, warn on architecture version mismatch.
1046 (aarch64_override_options): Fix parsing order for option strings.
1048 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1049 Iain Sandoe <iain@codesourcery.com>
1052 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1053 warning. Amend comment to reflect current functionality.
1055 2014-01-20 Richard Biener <rguenther@suse.de>
1058 * builtins.c (fold_builtin_strcat): Remove case better handled
1059 by tree-ssa-strlen.c.
1061 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1063 * config/aarch64/aarch64.opt
1064 (mcpu, march, mtune): Make case-insensitive.
1066 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1069 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1070 if operands[1] is a REG or ZERO_EXTEND of a REG.
1072 2014-01-19 Jan Hubicka <jh@suse.cz>
1074 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1076 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1078 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1079 long non-pic millicode calls.
1081 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1083 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1085 2014-01-19 Kito Cheng <kito@0xlab.org>
1087 * builtins.c (expand_movstr): Check movstr expand done or fail.
1089 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1090 H.J. Lu <hongjiu.lu@intel.com>
1093 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1094 to DImode for zero-extended addresses.
1096 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1098 PR rtl-optimization/57763
1099 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1100 on the new indirect jump_insn and increment LABEL_NUSES (label).
1102 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1106 * config.gcc (x86_archs): New variable.
1107 (x86_64_archs): Likewise.
1108 (x86_cpus): Likewise.
1109 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1110 --with-arch/--with-cpu= options.
1111 Support --with-arch=/--with-cpu={nehalem,westmere,
1112 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1114 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1116 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1117 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1119 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1121 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1123 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1126 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1127 clear cpp_get_options (parse_in)->warn_unused_macros for
1128 ix86_target_macros_internal with cpp_define.
1130 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1132 * jump.c (delete_related_insns): Keep (use (insn))s.
1133 * reorg.c (redundant_insn): Check for barriers too.
1135 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1137 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1139 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1141 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1142 call to $$dyncall when TARGET_LONG_CALLS is true.
1144 2014-01-17 Jeff Law <law@redhat.com>
1146 * ree.c (combine_set_extension): Temporarily disable test for
1147 changing number of hard registers.
1149 2014-01-17 Jan Hubicka <jh@suse.cz>
1152 * ipa-inline-analysis.c (inline_free_summary):
1153 Do not free summary of aliases.
1155 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1158 * gimplify.c (gimplify_expr): Use create_tmp_var
1159 instead of create_tmp_var_raw. If cond doesn't have
1160 integral type, don't add the IFN_ANNOTATE builtin at all.
1162 2014-01-17 Martin Jambor <mjambor@suse.cz>
1165 * ipa-cp.c (prev_edge_clone): New variable.
1166 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1167 Also resize prev_edge_clone vector.
1168 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1169 (ipcp_edge_removal_hook): New function.
1170 (ipcp_driver): Register ipcp_edge_removal_hook.
1172 2014-01-17 Andrew Pinski <apinski@cavium.com>
1173 Steve Ellcey <sellcey@mips.com>
1176 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1179 2014-01-17 Jeff Law <law@redhat.com>
1182 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1183 so that pass_ccp runs first.
1185 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1187 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1188 (ix86_adjust_cost): Use !TARGET_XXX.
1189 (do_reorder_for_imul): Likewise.
1190 (swap_top_of_ready_list): Likewise.
1191 (ix86_sched_reorder): Likewise.
1193 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1195 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1196 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1197 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1198 (intel_memset): New. Duplicate slm_memset.
1199 (intel_cost): New. Duplicate slm_cost.
1200 (m_INTEL): New macro.
1201 (processor_target_table): Add "intel".
1202 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1203 with PROCESSOR_INTEL for "intel".
1204 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1205 PROCESSOR_SILVERMONT.
1206 (ix86_issue_rate): Likewise.
1207 (ix86_adjust_cost): Likewise.
1208 (ia32_multipass_dfa_lookahead): Likewise.
1209 (swap_top_of_ready_list): Likewise.
1210 (ix86_sched_reorder): Likewise.
1211 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1212 instead of TARGET_OPT_AGU.
1213 * config/i386/i386.h (TARGET_INTEL): New.
1214 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1215 (processor_type): Add PROCESSOR_INTEL.
1216 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1217 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1219 2014-01-17 Marek Polacek <polacek@redhat.com>
1222 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1225 2014-01-17 Richard Biener <rguenther@suse.de>
1227 PR tree-optimization/46590
1228 * opts.c (default_options_table): Add entries for
1229 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1230 all enabled at -O1 but not for -Og.
1231 * common.opt (fbranch-count-reg): Remove Init(1).
1232 (fmove-loop-invariants): Likewise.
1233 (ftree-pta): Likewise.
1235 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1237 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1238 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1239 decls to at least the GCC 4.8 used alignments.
1242 * tree-nested.c (convert_nonlocal_reference_stmt,
1243 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1244 of GIMPLE_BIND stmts, adjust associated decls.
1246 2014-01-17 Richard Biener <rguenther@suse.de>
1248 PR tree-optimization/46590
1249 * vec.h (vec<>::bseach): New member function implementing
1250 binary search according to C89 bsearch.
1251 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1252 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1253 bitmap pointer again. Make accesses_in_loop a flat array.
1254 (mem_ref_obstack): New global.
1255 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1256 (mark_ref_stored): Likewise.
1257 (ref_indep_loop_p_2): Likewise.
1258 (set_ref_stored_in_loop): New helper function.
1259 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1260 (memref_free): Adjust.
1261 (record_mem_ref_loc): Simplify.
1262 (gather_mem_refs_stmt): Adjust.
1263 (sort_locs_in_loop_postorder_cmp): New function.
1264 (analyze_memory_references): Sort accesses_in_loop after
1265 loop postorder number.
1266 (find_ref_loc_in_loop_cmp): New function.
1267 (for_all_locs_in_loop): Find relevant cluster of locs in
1268 accesses_in_loop and iterate without recursion.
1269 (execute_sm): Avoid uninit warning.
1270 (struct ref_always_accessed): Simplify.
1271 (ref_always_accessed::operator ()): Likewise.
1272 (ref_always_accessed_p): Likewise.
1273 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1274 loop postorder numbers here.
1275 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1278 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1281 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1282 on decls for which assemble_alias has been called.
1284 2014-01-17 Nick Clifton <nickc@redhat.com>
1286 * config/msp430/msp430.opt: (mcpu): New option.
1287 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1288 (msp430_option_override): Parse target_cpu. If the MCU name
1289 matches a generic string, clear target_mcu.
1290 (msp430_attr): Allow numeric interrupt values up to 63.
1291 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1292 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1294 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1296 * config/msp430/msp430.md (popm): Use %J rather than %I.
1297 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1298 (addhi_cy_i): Use immediate_operand for operand 2.
1299 * doc/invoke.texi: Document -mcpu option.
1301 2014-01-17 Richard Biener <rguenther@suse.de>
1303 PR rtl-optimization/38518
1304 * df.h (df_analyze_loop): Declare.
1305 * df-core.c: Include cfgloop.h.
1306 (df_analyze_1): Split out main part of df_analyze.
1307 (df_analyze): Adjust.
1308 (loop_inverted_post_order_compute): New function.
1309 (loop_post_order_compute): Likewise.
1310 (df_analyze_loop): New function avoiding whole-function
1312 * loop-invariant.c (find_defs): Use df_analyze_loop.
1313 (find_invariants): Adjust.
1314 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1316 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1318 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1319 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1321 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1323 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1324 traversal when removing references.
1326 2014-01-16 Jan Hubicka <jh@suse.cz>
1329 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1331 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1334 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1335 pushing a reload for an autoinc when we had previously reloaded an
1336 inner part of the address.
1338 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1340 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1342 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1343 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1344 when not giving up or versioning for alias only because of
1346 (vect_analyze_data_ref_dependences): Set to true.
1347 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1349 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1350 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1354 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1357 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1358 operand 0 predicate for gathers, use a new pseudo as subtarget.
1360 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1363 * lra-constraints.c (process_alt_operands): Add printing debug info.
1364 Check absence of input/output reloads for matched operands too.
1366 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1368 PR rtl-optimization/59835
1369 * ira.c (ira_init_register_move_cost): Increase cost for
1372 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1374 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1376 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1379 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1380 non-register objects. Use gen_(high/low)part more consistently.
1383 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1386 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1387 endian support, remove tests for WORDS_BIG_ENDIAN.
1388 (p8_mfvsrd_3_<mode>): Likewise.
1389 (reload_gpr_from_vsx<mode>): Likewise.
1390 (reload_gpr_from_vsxsf): Likewise.
1391 (p8_mfvsrd_4_disf): Likewise.
1393 2014-01-16 Richard Biener <rguenther@suse.de>
1395 PR rtl-optimization/46590
1396 * lcm.c (compute_antinout_edge): Use postorder iteration.
1397 (compute_laterin): Use inverted postorder iteration.
1399 2014-01-16 Nick Clifton <nickc@redhat.com>
1402 * varasm.c (output_constant): Return the number of bytes actually
1404 (output_constructor_array_range): Update the field size with the
1405 number of bytes emitted by output_constant.
1406 (output_constructor_regular_field): Likewise. Also do not
1407 complain if the total number of bytes emitted is now greater
1408 than the expected fieldpos.
1409 * output.h (output_constant): Update prototype and descriptive comment.
1411 2014-01-16 Marek Polacek <polacek@redhat.com>
1414 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1415 it is error_mark_node.
1417 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1419 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1420 VALID_AVX256_REG_OR_OI_MODE.
1422 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1424 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1425 current procedure should be profiled.
1427 2014-01-15 Andrew Pinski <apinski@cavium.com>
1429 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1430 of moving from/to the STACK_REG register class.
1432 2014-01-15 Richard Henderson <rth@redhat.com>
1435 * reginfo.c (global_regs_decl): Globalize.
1436 * rtl.h (global_regs_decl): Declare.
1437 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1438 reserved via global_regs.
1440 2014-01-15 Teresa Johnson <tejohnson@google.com>
1442 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1444 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1446 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1447 and vmulosh rather than call gen_vec_widen_smult_*.
1448 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1449 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1450 (vec_widen_smult_even_v16qi): Likewise.
1451 (vec_widen_umult_even_v8hi): Likewise.
1452 (vec_widen_smult_even_v8hi): Likewise.
1453 (vec_widen_umult_odd_v16qi): Likewise.
1454 (vec_widen_smult_odd_v16qi): Likewise.
1455 (vec_widen_umult_odd_v8hi): Likewise.
1456 (vec_widen_smult_odd_v8hi): Likewise.
1457 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1458 vmuloub rather than call gen_vec_widen_umult_*.
1459 (vec_widen_umult_lo_v16qi): Likewise.
1460 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1461 vmulosb rather than call gen_vec_widen_smult_*.
1462 (vec_widen_smult_lo_v16qi): Likewise.
1463 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1464 rather than call gen_vec_widen_umult_*.
1465 (vec_widen_umult_lo_v8hi): Likewise.
1466 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1467 rather than call gen_vec_widen_smult_*.
1468 (vec_widen_smult_lo_v8hi): Likewise.
1470 2014-01-15 Jeff Law <law@redhat.com>
1472 PR tree-optimization/59747
1473 * ree.c (find_and_remove_re): Properly handle case where a second
1474 eliminated extension requires widening a copy created for elimination
1475 of a prior extension.
1476 (combine_set_extension): Ensure that the number of hard regs needed
1477 for a destination register does not change when we widen it.
1479 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1481 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1482 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1483 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1484 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1485 (avr-*-rtems*): Likewise.
1486 (bfin*-rtems*): Likewise.
1487 (moxie-*-rtems*): Likewise.
1488 (h8300-*-rtems*): Likewise.
1489 (i[34567]86-*-rtems*): Likewise.
1490 (lm32-*-rtems*): Likewise.
1491 (m32r-*-rtems*): Likewise.
1492 (m68k-*-rtems*): Likewise.
1493 (microblaze*-*-rtems*): Likewise.
1494 (mips*-*-rtems*): Likewise.
1495 (powerpc-*-rtems*): Likewise.
1496 (sh-*-rtems*): Likewise.
1497 (sparc-*-rtems*): Likewise.
1498 (sparc64-*-rtems*): Likewise.
1499 (v850-*-rtems*): Likewise.
1500 (m32c-*-rtems*): Likewise.
1502 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1504 PR rtl-optimization/59511
1505 * ira.c (ira_init_register_move_cost): Use memory costs for some
1506 cases of register move cost calculations.
1507 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1508 instead of BB frequency.
1509 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1510 * lra-assigns.c (find_hard_regno_for): Ditto.
1512 2014-01-15 Richard Biener <rguenther@suse.de>
1514 PR tree-optimization/59822
1515 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1516 (vectorizable_load): Use it to hoist defs of uses of invariant
1517 loads out of the loop.
1519 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1520 Kugan Vivekanandarajah <kuganv@linaro.org>
1523 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1526 2014-01-15 Richard Biener <rguenther@suse.de>
1528 PR rtl-optimization/59802
1529 * lcm.c (compute_available): Use inverted postorder to seed
1530 the initial worklist.
1532 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1535 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1536 ADDR_REGS for invalid symrefs in non-PIC code.
1538 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1541 * builtins.c (determine_block_size): Initialize *probable_max_size
1542 even if len_rtx is CONST_INT.
1544 2014-01-14 Andrew Pinski <apinski@cavium.com>
1546 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1547 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1548 (cortexa53_tunings): Likewise.
1549 (aarch64_sched_issue_rate): New function.
1550 (TARGET_SCHED_ISSUE_RATE): Define.
1552 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1554 * ira-costs.c (find_costs_and_classes): Add missed
1555 ira_init_register_move_cost_if_necessary.
1557 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1560 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1562 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1565 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1566 to indicate if type is used for function return value. Warn ABI
1567 change if the vector mode isn't available for function return value.
1568 (ix86_function_arg_advance): Pass false to type_natural_mode.
1569 (ix86_function_arg): Likewise.
1570 (ix86_gimplify_va_arg): Likewise.
1571 (function_arg_32): Don't warn ABI change.
1572 (ix86_function_value): Pass true to type_natural_mode.
1573 (ix86_return_in_memory): Likewise.
1574 (ix86_struct_value_rtx): Removed.
1575 (TARGET_STRUCT_VALUE_RTX): Likewise.
1577 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1579 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1580 converting a conditional jump into a conditional return.
1582 2014-01-14 Richard Biener <rguenther@suse.de>
1584 PR tree-optimization/58921
1585 PR tree-optimization/59006
1586 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1587 hoisting invariant stmts.
1588 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1589 invariant loads on the preheader edge if possible.
1591 2014-01-14 Joey Ye <joey.ye@arm.com>
1593 * doc/plugin.texi (Building GCC plugins): Update to C++.
1595 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1597 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1598 (_mm_rcp28_round_ss): Ditto.
1599 (_mm_rsqrt28_round_sd): Ditto.
1600 (_mm_rsqrt28_round_ss): Ditto.
1601 (_mm_rcp28_sd): Ditto.
1602 (_mm_rcp28_ss): Ditto.
1603 (_mm_rsqrt28_sd): Ditto.
1604 (_mm_rsqrt28_ss): Ditto.
1605 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1606 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1607 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1608 (IX86_BUILTIN_RCP28SD): Ditto.
1609 (IX86_BUILTIN_RCP28SS): Ditto.
1610 (IX86_BUILTIN_RSQRT28SD): Ditto.
1611 (IX86_BUILTIN_RSQRT28SS): Ditto.
1612 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1613 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1614 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1615 (ix86_expand_special_args_builtin): Expand new FTYPE.
1616 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1617 (srcp14<mode>): Make insn unary.
1618 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1619 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1620 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1621 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1622 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1623 Fix rounding: make it SAE only.
1624 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1626 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1628 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1629 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1630 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1631 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1632 (round_saeonly_mask_scalar_operand4): Ditto.
1633 (round_saeonly_mask_scalar_op3): Ditto.
1634 (round_saeonly_mask_scalar_op4): Ditto.
1636 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1638 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1639 Implement -maltivec=be for vec_insert and vec_extract.
1641 2014-01-10 DJ Delorie <dj@redhat.com>
1643 * config/msp430/msp430.md (call_internal): Don't allow memory
1644 references with SP as the base register.
1645 (call_value_internal): Likewise.
1646 * config/msp430/constraints.md (Yc): New. For memory references
1647 that don't use SP as a base register.
1649 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1650 "an integer without a # prefix"
1651 * config/msp430/msp430.md (epilogue_helper): Use it.
1653 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1656 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1657 AVX512F gather builtins.
1658 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1659 on gather decls with INTEGER_TYPE masktype.
1660 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1661 directly into the builtin rather than hoisting it before loop.
1663 PR tree-optimization/59387
1664 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1665 (scev_const_prop): If folded_casts and type has undefined overflow,
1666 use force_gimple_operand instead of force_gimple_operand_gsi and
1667 for each added stmt if it is assign with
1668 arith_code_with_undefined_signed_overflow, call
1669 rewrite_to_defined_overflow.
1670 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1671 gimple-fold.h instead.
1672 (arith_code_with_undefined_signed_overflow,
1673 rewrite_to_defined_overflow): Moved to ...
1674 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1675 rewrite_to_defined_overflow): ... here. No longer static.
1676 Include gimplify-me.h.
1677 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1678 rewrite_to_defined_overflow): New prototypes.
1680 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1682 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1684 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1686 * builtins.c (get_object_alignment_2): Minor tweak.
1687 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1689 2014-01-13 Christian Bruel <christian.bruel@st.com>
1691 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1692 optimized non constant lengths.
1694 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1697 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1698 load as __atomic_load_N if possible.
1700 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1702 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1704 (rs6000_expand_builtin): Adjust call.
1706 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1709 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1710 * config/rs6000/rs6000.c: Include target-globals.h.
1711 (rs6000_set_current_function): Instead of doing target_reinit
1712 unconditionally, use save_target_globals_default_opts and
1713 restore_target_globals.
1715 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1717 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1718 (rs6000_expand_builtin): Handle mffs and mtfsf.
1719 (rs6000_init_builtins): Define mffs and mtfsf.
1720 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1721 (rs6000_mffs): New pattern.
1722 (rs6000_mtfsf): New pattern.
1724 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1726 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1727 Start narrowing with START. Apply candidate-use pair
1728 and check overall cost in narrowing.
1729 (iv_ca_prune): Pass new argument.
1731 2014-01-10 Jeff Law <law@redhat.com>
1734 * ree.c (combine_reaching_defs): Ensure the defining statement
1735 occurs before the extension when optimizing extensions with
1736 different source and destination hard registers.
1738 2014-01-10 Jan Hubicka <jh@suse.cz>
1741 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1742 vtables into the type inheritance graph.
1744 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1746 PR rtl-optimization/59754
1747 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1748 modes in the REGNO != REGNO case.
1750 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1752 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1754 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1756 PR tree-optimization/59745
1757 * tree-predcom.c (tree_predictive_commoning_loop): Call
1758 free_affine_expand_cache if giving up because components is NULL.
1760 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1761 GC in payload of target_globals struct instead of allocating them on
1762 the heap and the larger structs separately using GC.
1763 * target-globals.h (struct target_globals): Make regs, hard_regs,
1764 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1765 of GTY((skip)) and change type to void *.
1766 (reset_target_globals): Cast loads from those fields to corresponding
1769 2014-01-10 Steve Ellcey <sellcey@mips.com>
1772 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1773 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1774 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1776 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1779 * aarch64-modes.def (CC_Zmode): New flags mode.
1780 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1781 represents an equality.
1782 (aarch64_get_condition_code): Handle CC_Zmode.
1783 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1785 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1787 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1788 extraction in good case.
1790 2014-01-10 Richard Biener <rguenther@suse.de>
1792 PR tree-optimization/59374
1793 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1794 checking after SLP discovery. Mark stmts not participating
1795 in any SLP instance properly.
1797 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1799 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1800 when handling a SET rtx.
1802 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1804 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1805 (cortex-a57): Likewise.
1806 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1808 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1810 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1811 non-iwmmxt builtins.
1813 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1817 * ipa-devirt.c record_target_from_binfo): Take as argument
1818 stack of binfos and lookup matching one for virtual inheritance.
1819 (possible_polymorphic_call_targets_1): Update.
1821 2014-01-10 Huacai Chen <chenhc@lemote.com>
1823 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1824 kernel strings for Loongson-2E/2F/3A.
1826 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1829 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1830 is_gimple_call before calling gimple_call_internal_p.
1832 2014-01-09 Steve Ellcey <sellcey@mips.com>
1834 * Makefile.in (TREE_FLOW_H): Remove.
1835 (TREE_SSA_H): Add file names from tree-flow.h.
1836 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1837 * tree.h: Remove tree-flow.h reference.
1838 * hash-table.h: Remove tree-flow.h reference.
1839 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1840 reference with tree-ssa-loop.h.
1842 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1844 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1845 default element-order behavior for -maltivec.
1846 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1847 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1848 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1849 when targeting big endian, at least for now.
1850 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1852 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1855 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1856 var satisfies use_register_for_decl, just take into account type
1857 alignment, rather than decl alignment.
1859 PR tree-optimization/59622
1860 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1861 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1862 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1863 Don't devirtualize for inplace at all. For targets.length () == 1,
1864 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1866 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1868 * config/i386/i386.md (cpu): Remove the unused btver1.
1870 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1872 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1874 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1877 * tree-core.h (struct target_globals): New forward declaration.
1878 (struct tree_target_option): Add globals field.
1879 * tree.h (TREE_TARGET_GLOBALS): Define.
1880 (prepare_target_option_nodes_for_pch): New prototype.
1881 * target-globals.h (struct target_globals): Define even if
1883 * tree.c (prepare_target_option_node_for_pch,
1884 prepare_target_option_nodes_for_pch): New functions.
1885 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1886 * config/i386/i386.c: Include target-globals.h.
1887 (ix86_set_current_function): Instead of doing target_reinit
1888 unconditionally, use save_target_globals_default_opts and
1889 restore_target_globals.
1891 2014-01-09 Richard Biener <rguenther@suse.de>
1893 PR tree-optimization/59715
1894 * tree-cfg.h (split_critical_edges): Declare.
1895 * tree-cfg.c (split_critical_edges): Export.
1896 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1898 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1900 * cfgexpand.c (expand_stack_vars): Optionally disable
1901 asan stack protection.
1902 (expand_used_vars): Likewise.
1903 (partition_stack_vars): Likewise.
1904 * asan.c (asan_emit_stack_protection): Optionally disable
1905 after return stack usage.
1906 (instrument_derefs): Optionally disable memory access instrumentation.
1907 (instrument_builtin_call): Likewise.
1908 (instrument_strlen_call): Likewise.
1909 (asan_protect_global): Optionally disable global variables protection.
1910 * doc/invoke.texi: Added doc for new options.
1911 * params.def: Added new options.
1912 * params.h: Likewise.
1914 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1916 PR rtl-optimization/59724
1917 * ifcvt.c (cond_exec_process_if_block): Don't call
1918 flow_find_head_matching_sequence with 0 longest_match.
1919 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1920 non-active insns if !stop_after.
1921 (try_head_merge_bb): Revert 2014-01-07 changes.
1923 2014-01-08 Jeff Law <law@redhat.com>
1925 * ree.c (get_sub_rtx): New function, extracted from...
1926 (merge_def_and_ext): Here.
1927 (combine_reaching_defs): Use get_sub_rtx.
1929 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1931 * cgraph.h (varpool_variable_node): Do not choke on null node.
1933 2014-01-08 Catherine Moore <clm@codesourcery.com>
1935 * config/mips/mips.md (simple_return): Attempt to use JRC
1937 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1939 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1941 PR rtl-optimization/59137
1942 * reorg.c (steal_delay_list_from_target): Call update_block for
1944 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1946 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1949 two duplicate entries.
1951 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1954 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1956 * config/mips/mips.c (mips_truncated_op_cost): New function.
1957 (mips_rtx_costs): Adjust test for BADDU.
1958 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1960 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1962 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1963 (*baddu_si): ...this new pattern.
1965 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1968 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1970 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1973 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1975 (expand_expr, expand_normal): Adjust.
1976 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1977 inner_reference_p. Use inner_reference_p to expand inner references.
1978 (store_expr): Adjust.
1979 * cfgexpand.c (expand_call_stmt): Adjust.
1981 2014-01-08 Rong Xu <xur@google.com>
1983 * gcov-io.c (gcov_var): Move from gcov-io.h.
1984 (gcov_position): Ditto.
1985 (gcov_is_error): Ditto.
1986 (gcov_rewrite): Ditto.
1987 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1988 only part to libgcc/libgcov.h.
1990 2014-01-08 Marek Polacek <polacek@redhat.com>
1993 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1995 2014-01-08 Marek Polacek <polacek@redhat.com>
1998 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2000 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2002 PR rtl-optimization/59649
2003 * stor-layout.c (get_mode_bounds): For BImode return
2004 0 and STORE_FLAG_VALUE.
2006 2014-01-08 Richard Biener <rguenther@suse.de>
2009 * gimple.h (is_gimple_builtin_call): Remove.
2010 (gimple_builtin_call_types_compatible_p): New.
2011 (gimple_call_builtin_p): New overload.
2012 * gimple.c (is_gimple_builtin_call): Remove.
2013 (validate_call): Rename to ...
2014 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2016 (validate_type): New static function.
2017 (gimple_call_builtin_p): New overload and adjust.
2018 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2019 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2020 (gimple_fold_stmt_to_constant_1): Likewise.
2021 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2023 2014-01-08 Richard Biener <rguenther@suse.de>
2026 * gimplify.c (gimplify_expr): Gimplify register-register type
2027 VIEW_CONVERT_EXPRs to separate stmts.
2029 2014-01-07 Jeff Law <law@redhat.com>
2032 * ree.c (combine_set_extension): Handle case where source
2033 and destination registers in an extension insn are different.
2034 (combine_reaching_defs): Allow source and destination registers
2035 in extension to be different under limited circumstances.
2036 (add_removable_extension): Remove restriction that the
2037 source and destination registers in the extension are the same.
2038 (find_and_remove_re): Emit a copy from the extension's
2039 destination to its source after the defining insn if
2040 the source and destination registers are different.
2043 * ifcvt.c (merge_if_block): If we are merging a block with more than
2044 one successor with a block with no successors, remove any BARRIER
2045 after the second block.
2047 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2049 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2051 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2054 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2055 for 14-bit register offsets when INT14_OK_STRICT is false.
2057 2014-01-07 Roland Stigge <stigge@antcom.de>
2058 Michael Meissner <meissner@linux.vnet.ibm.com>
2061 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2062 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2064 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2066 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2069 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2071 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2072 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2073 rtx is const0_rtx or not.
2075 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2078 * target-globals.c (save_target_globals): Remove this_fn_optab
2080 * toplev.c: Include optabs.h.
2081 (target_reinit): Temporarily restore the global options if another
2082 set of options are in force.
2084 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2086 PR rtl-optimization/58668
2087 * cfgcleanup.c (flow_find_cross_jump): Don't count
2088 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2089 to determine what is counted.
2090 (flow_find_head_matching_sequence): Use active_insn_p to determine
2092 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2094 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2095 determine what is counted.
2097 PR tree-optimization/59643
2098 * tree-predcom.c (split_data_refs_to_components): If one dr is
2099 read and one write, determine_offset fails and the write isn't
2100 in the bad component, just put the read into the bad component.
2102 2014-01-07 Mike Stump <mikestump@comcast.net>
2103 Jakub Jelinek <jakub@redhat.com>
2106 * tree-core.h (struct tree_optimization_option): Change optabs
2107 type from unsigned char * to void *.
2108 * optabs.c (init_tree_optimization_optabs): Adjust
2109 TREE_OPTIMIZATION_OPTABS initialization.
2111 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2114 * config/i386/i386.h (struct machine_function): Add
2115 no_drap_save_restore field.
2116 * config/i386/i386.c (ix86_save_reg): Use
2117 !cfun->machine->no_drap_save_restore instead of
2118 crtl->stack_realign_needed.
2119 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2120 this function clears frame_pointer_needed. Set
2121 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2122 and DRAP reg is needed.
2124 2014-01-06 Marek Polacek <polacek@redhat.com>
2127 * doc/implement-c.texi: Mention that other integer types are
2128 permitted as bit-field types in strictly conforming mode.
2130 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2132 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2135 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2137 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2139 2014-01-06 Martin Jambor <mjambor@suse.cz>
2142 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2144 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2146 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2150 * var-tracking.c (add_stores): Preserve the value of the source even if
2151 we don't record the store.
2153 2014-01-06 Terry Guo <terry.guo@arm.com>
2155 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2157 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2160 * config/darwin.c (darwin_function_section): Adjust return values to
2161 correspond to optimisation changes made in r206070.
2163 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2165 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2166 from prefetch_block tune setting.
2167 (nocona_cost): Correct size of prefetch block to 64.
2169 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2171 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2172 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2173 used to save the static chain register in the computation of the offset
2174 from which the FP registers need to be restored.
2176 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2178 PR tree-optimization/59519
2179 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2180 ICE if get_current_def (current_new_name) is already non-NULL, as long
2181 as it is a phi result of some other phi in *new_exit_bb that has
2184 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2185 or vmovdqu* for misaligned_operand.
2186 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2187 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2188 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2189 aligned_mem for AVX512F masked aligned load and store builtins and for
2192 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2194 PR tree-optimization/59651
2195 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2196 Address range for negative step should be added by TYPE_SIZE_UNIT.
2198 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2200 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2201 overlapping registers also for registers other than the stack pointer.
2203 2014-01-03 Marek Polacek <polacek@redhat.com>
2206 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2209 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2212 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2215 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2216 (push splitter): Use <P:MODE_SIZE> instead of
2217 GET_MODE_SIZE (<P:MODE>mode).
2218 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2219 (mov -1, reg peephole2): Likewise.
2220 * config/i386/sse.md (*mov<mode>_internal,
2221 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2222 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2223 *<code><mode>3, *andnot<mode>3<mask_name>,
2224 <mask_codefor><code><mode>3<mask_name>): Likewise.
2225 * config/i386/subst.md (mask_mode512bit_condition,
2226 sd_mask_mode512bit_condition): Likewise.
2228 2014-01-02 Xinliang David Li <davidxl@google.com>
2230 PR tree-optimization/59303
2231 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2232 (dump_predicates): Better output format.
2233 (pred_equal_p): New function.
2234 (is_neq_relop_p): Ditto.
2235 (is_neq_zero_form_p): Ditto.
2236 (pred_expr_equal_p): Ditto.
2237 (pred_neg_p): Ditto.
2238 (simplify_pred): Ditto.
2239 (simplify_preds_2): Ditto.
2240 (simplify_preds_3): Ditto.
2241 (simplify_preds_4): Ditto.
2242 (simplify_preds): Ditto.
2244 (push_to_worklist): Ditto.
2245 (get_pred_info_from_cmp): Ditto.
2246 (is_degenerated_phi): Ditto.
2247 (normalize_one_pred_1): Ditto.
2248 (normalize_one_pred): Ditto.
2249 (normalize_one_pred_chain): Ditto.
2250 (normalize_preds): Ditto.
2251 (normalize_cond_1): Remove function.
2252 (normalize_cond): Ditto.
2253 (is_gcond_subset_of): Ditto.
2254 (is_subset_of_any): Ditto.
2255 (is_or_set_subset_of): Ditto.
2256 (is_and_set_subset_of): Ditto.
2257 (is_norm_cond_subset_of): Ditto.
2258 (pred_chain_length_cmp): Ditto.
2259 (convert_control_dep_chain_into_preds): Type change.
2260 (find_predicates): Ditto.
2261 (find_def_preds): Ditto.
2262 (destroy_predicates_vecs): Ditto.
2263 (find_matching_predicates_in_rest_chains): Ditto.
2264 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2265 (is_pred_expr_subset): Ditto.
2266 (is_pred_chain_subset_of): Ditto.
2267 (is_included_in): Ditto.
2268 (is_superset_of): Ditto.
2270 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2272 Update copyright years.
2274 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2276 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2277 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2278 config/arc/arc.md, config/arc/arc.opt,
2279 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2280 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2281 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2282 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2283 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2284 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2285 vtable-verify.c, vtable-verify.h: Use the standard form for the
2288 2014-01-02 Tobias Burnus <burnus@net-b.de>
2290 * gcc.c (process_command): Update copyright notice dates.
2291 * gcov-dump.c: Ditto.
2293 * doc/cpp.texi: Bump @copying's copyright year.
2294 * doc/cppinternals.texi: Ditto.
2295 * doc/gcc.texi: Ditto.
2296 * doc/gccint.texi: Ditto.
2297 * doc/gcov.texi: Ditto.
2298 * doc/install.texi: Ditto.
2299 * doc/invoke.texi: Ditto.
2301 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2303 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2305 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2307 * config/i386/sse.md (*mov<mode>_internal): Guard
2308 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2310 PR rtl-optimization/59647
2311 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2312 new_rtx into UNSIGNED_FLOAT rtxes.
2314 Copyright (C) 2014 Free Software Foundation, Inc.
2316 Copying and distribution of this file, with or without modification,
2317 are permitted in any medium without royalty provided the copyright
2318 notice and this notice are preserved.