1 2014-01-30 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/59993
4 * tree-ssa-forwprop.c (associate_pointerplus): Check we
5 can propagate form the earlier stmt and avoid the transform
6 when the intermediate result is needed.
8 2014-01-30 Alangi Derick <alangiderick@gmail.com>
10 * README.Portability: Fix typo.
12 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
14 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
15 comparison_operator with ordered_comparison_operator.
17 2014-01-30 Nick Clifton <nickc@redhat.com>
19 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
20 Rename to mn10300_store_multiple_regs.
21 * config/mn10300/mn10300.c: Likewise.
22 * config/mn10300/mn10300.md (store_movm): Fix typo: call
24 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
25 Call mn10300_store_multiple_regs.
27 2014-01-30 Nick Clifton <nickc@redhat.com>
28 DJ Delorie <dj@redhat.com>
30 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
31 %fp 2 to keep registers after it properly word-aligned.
32 (rl78_alloc_physical_registers_umul): Handle the case where both
33 input operands are the same.
35 2014-01-30 Richard Biener <rguenther@suse.de>
37 PR tree-optimization/59903
38 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
41 2014-01-30 Jason Merrill <jason@redhat.com>
44 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
47 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
49 2014-01-30 Richard Biener <rguenther@suse.de>
51 PR tree-optimization/59951
52 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
55 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
58 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
59 SFmode to DFmode case.
61 2014-01-29 DJ Delorie <dj@redhat.com>
63 * config/msp430/msp430.opt (-minrt): New.
64 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
66 (ENDFILE_SPEC): Likewise.
68 2014-01-29 Jan Hubicka <jh@suse.cz>
70 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
71 (estimate_function_body_sizes): Use it.
73 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
76 * dwarf2out.c (is_cxx_auto): New.
77 (is_base_type): Use it.
78 (gen_type_die_with_usage): Likewise.
80 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
82 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
83 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
84 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
85 -maltivec=be with LE targets.
86 (vsx_mergeh_<mode>): Likewise.
87 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
89 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
90 (altivec_vmrghb): Replace with define_expand and new
91 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
93 (altivec_vmrghb_direct): New define_insn.
94 (altivec_vmrghh): Replace with define_expand and new
95 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
97 (altivec_vmrghh_direct): New define_insn.
98 (altivec_vmrghw): Replace with define_expand and new
99 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
101 (altivec_vmrghw_direct): New define_insn.
102 (*altivec_vmrghsf): Adjust for endianness.
103 (altivec_vmrglb): Replace with define_expand and new
104 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
106 (altivec_vmrglb_direct): New define_insn.
107 (altivec_vmrglh): Replace with define_expand and new
108 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
110 (altivec_vmrglh_direct): New define_insn.
111 (altivec_vmrglw): Replace with define_expand and new
112 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
114 (altivec_vmrglw_direct): New define_insn.
115 (*altivec_vmrglsf): Adjust for endianness.
116 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
117 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
118 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
119 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
120 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
121 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
122 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
123 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
125 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
127 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
128 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
131 2014-01-29 Richard Biener <rguenther@suse.de>
133 PR tree-optimization/58742
134 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
135 associate_pointerplus_align.
136 (associate_pointerplus_diff): New function.
137 (associate_pointerplus): Likewise. Call associate_pointerplus_align
138 and associate_pointerplus_diff.
140 2014-01-29 Richard Biener <rguenther@suse.de>
142 * lto-streamer.h (LTO_major_version): Bump to 3.
143 (LTO_minor_version): Reset to 0.
145 2014-01-29 Renlin Li <Renlin.Li@arm.com>
147 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
148 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
149 (arm_file_start): Generate correct asm header for armv7ve.
150 * config/arm/bpabi.h: Add multilib support for armv7ve.
151 * config/arm/driver-arm.c: Change the architectures of cortex-a7
152 and cortex-a15 to armv7ve.
153 * config/arm/t-aprofile: Add multilib support for armv7ve.
154 * doc/invoke.texi: Document -march=armv7ve.
156 2014-01-29 Richard Biener <rguenther@suse.de>
158 PR tree-optimization/58742
159 * tree-ssa-forwprop.c (associate_plusminus): Return true
160 if we changed sth, defer EH cleanup to ...
161 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
162 (simplify_mult): New function.
164 2014-01-29 Jakub Jelinek <jakub@redhat.com>
167 PR tree-optimization/59920
168 * tree.c (build_common_builtin_nodes): Remove
169 __builtin_setjmp_dispatcher initialization.
170 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
171 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
172 instead of gsi_after_labels + manually skipping debug stmts.
173 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
174 ignore bbs with IFN_ABNORMAL_DISPATCHER.
175 * tree-inline.c (copy_edges_for_bb): Remove
176 can_make_abnormal_goto argument, instead add abnormal_goto_dest
177 argument. Ignore computed_goto_p stmts. Don't call
178 make_abnormal_goto_edges. If a call might need abnormal edges
179 for non-local gotos, see if it already has an edge to
180 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
181 with true argument, don't do anything then, otherwise add
182 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
183 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
185 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
186 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
187 (lower_stmt): Don't set data->calls_builtin_setjmp.
188 (lower_builtin_setjmp): Adjust comment.
189 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
190 * tree-cfg.c (found_computed_goto): Remove.
191 (factor_computed_gotos): Remove.
192 (make_goto_expr_edges): Return bool, true for computed gotos.
193 Don't call make_abnormal_goto_edges.
194 (build_gimple_cfg): Don't set found_computed_goto, don't call
195 factor_computed_gotos.
196 (computed_goto_p): No longer static.
197 (make_blocks): Don't set found_computed_goto.
198 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
199 (make_edges): If make_goto_expr_edges returns true, push bb
200 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
201 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
202 vector. Record mapping between bbs and OpenMP regions if there
203 are any, adjust make_gimple_omp_edges caller. Call
204 handle_abnormal_edges.
205 (make_abnormal_goto_edges): Remove.
206 * tree-cfg.h (make_abnormal_goto_edges): Remove.
207 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
208 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
209 * builtins.c (expand_builtin): Don't handle
210 BUILT_IN_SETJMP_DISPATCHER.
211 * internal-fn.def (ABNORMAL_DISPATCHER): New.
212 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
213 filling *region also set *region_idx to (*region)->entry->index.
216 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
217 For REGs set ORIGINAL_REGNO.
219 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
221 * doc/md.texi: Mention that a target shouldn't implement
222 vec_widen_(s|u)mul_even/odd pair if it is less efficient
225 2014-01-29 Jakub Jelinek <jakub@redhat.com>
227 PR tree-optimization/59594
228 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
229 a copy of the datarefs vector rather than the vector itself.
231 2014-01-28 Jason Merrill <jason@redhat.com>
234 * dwarf2out.c (auto_die): New static.
235 (gen_type_die_with_usage): Handle C++1y 'auto'.
236 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
239 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
242 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
243 (SPEC_X32): Likewise.
245 * config/i386/i386.c (ix86_option_override_internal): Turn off
246 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
248 (x86_file_start): Output .code16gcc for TARGET_16BIT.
249 * config/i386/i386.h (TARGET_16BIT): New macro.
250 (TARGET_16BIT_P): Likewise.
251 * config/i386/i386.opt: Add m16.
252 * doc/invoke.texi: Document -m16.
254 2014-01-28 Jakub Jelinek <jakub@redhat.com>
256 PR preprocessor/59935
257 * input.c (location_get_source_line): Bail out on when line number
258 is zero, and test the return value of
259 lookup_or_add_file_to_cache_tab.
261 2014-01-28 Richard Biener <rguenther@suse.de>
263 PR tree-optimization/58742
264 * tree-ssa-forwprop.c (associate_plusminus): Handle
265 pointer subtraction of the form (T)(P + A) - (T)P.
267 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
269 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
272 2014-01-28 Richard Biener <rguenther@suse.de>
275 2014-01-28 Richard Biener <rguenther@suse.de>
277 PR rtl-optimization/45364
278 PR rtl-optimization/59890
279 * var-tracking.c (local_get_addr_clear_given_value): Handle
280 already cleared slot.
281 (val_reset): Handle not allocated local_get_addr_cache.
282 (vt_find_locations): Use post-order on the inverted CFG.
284 2014-01-28 Richard Biener <rguenther@suse.de>
286 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
289 2014-01-28 Richard Biener <rguenther@suse.de>
291 PR rtl-optimization/45364
292 PR rtl-optimization/59890
293 * var-tracking.c (local_get_addr_clear_given_value): Handle
294 already cleared slot.
295 (val_reset): Handle not allocated local_get_addr_cache.
296 (vt_find_locations): Use post-order on the inverted CFG.
298 2014-01-28 Alan Modra <amodra@gmail.com>
300 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
301 * configure.ac <recursive call for build != host>: Define
302 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
303 and LD_FOR_BUILD too.
304 * configure: Regenerate.
306 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
308 * config/i386/i386.c (get_builtin_code_for_version): Separate
309 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
310 Broadwell from Haswell.
312 2014-01-27 Steve Ellcey <sellcey@mips.com>
314 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
315 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
316 * config/mips/mips.c (mips_option_override): Change setting
318 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
319 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
320 Change from Mask to Var.
322 2014-01-27 Jeff Law <law@redhat.com>
324 * ipa-inline.c (inline_small_functions): Fix typo.
326 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
328 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
329 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
330 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
331 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
332 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
333 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
334 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
335 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
336 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
337 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
338 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
339 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
340 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
341 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
342 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
343 (_mm512_storeu_epi64): Ditto.
344 (_mm512_cmpge_epi32_mask): Ditto.
345 (_mm512_cmpge_epu32_mask): Ditto.
346 (_mm512_cmpge_epi64_mask): Ditto.
347 (_mm512_cmpge_epu64_mask): Ditto.
348 (_mm512_cmple_epi32_mask): Ditto.
349 (_mm512_cmple_epu32_mask): Ditto.
350 (_mm512_cmple_epi64_mask): Ditto.
351 (_mm512_cmple_epu64_mask): Ditto.
352 (_mm512_cmplt_epi32_mask): Ditto.
353 (_mm512_cmplt_epu32_mask): Ditto.
354 (_mm512_cmplt_epi64_mask): Ditto.
355 (_mm512_cmplt_epu64_mask): Ditto.
356 (_mm512_cmpneq_epi32_mask): Ditto.
357 (_mm512_cmpneq_epu32_mask): Ditto.
358 (_mm512_cmpneq_epi64_mask): Ditto.
359 (_mm512_cmpneq_epu64_mask): Ditto.
360 (_mm512_expand_pd): Ditto.
361 (_mm512_expand_ps): Ditto.
362 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
363 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
364 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
365 * config/i386/i386.c (ix86_builtins): Add
366 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
367 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
368 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
369 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
370 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
371 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
372 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
373 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
374 IX86_BUILTIN_PMOVUSQW512_MEM.
375 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
376 __builtin_ia32_pmovsqd512mem_mask,
377 __builtin_ia32_pmovqd512mem_mask,
378 __builtin_ia32_pmovusqw512mem_mask,
379 __builtin_ia32_pmovsqw512mem_mask,
380 __builtin_ia32_pmovqw512mem_mask,
381 __builtin_ia32_pmovusdw512mem_mask,
382 __builtin_ia32_pmovsdw512mem_mask,
383 __builtin_ia32_pmovdw512mem_mask,
384 __builtin_ia32_pmovqb512mem_mask,
385 __builtin_ia32_pmovusqb512mem_mask,
386 __builtin_ia32_pmovsqb512mem_mask,
387 __builtin_ia32_pmovusdb512mem_mask,
388 __builtin_ia32_pmovsdb512mem_mask,
389 __builtin_ia32_pmovdb512mem_mask.
390 (bdesc_args): Add __builtin_ia32_expanddf512,
391 __builtin_ia32_expandsf512.
392 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
393 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
394 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
395 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
396 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
397 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
398 (avx512f_<code>v8div16qi2_mask_store): This.
399 (avx512f_expand<mode>): New.
401 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
403 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
405 (_mm512_mask_prefetch_i64gather_pd): Ditto.
406 (_mm512_prefetch_i32scatter_pd): Ditto.
407 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
408 (_mm512_prefetch_i64scatter_pd): Ditto.
409 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
410 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
411 (_mm512_mask_prefetch_i64gather_ps): Ditto.
412 (_mm512_prefetch_i32scatter_ps): Ditto.
413 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
414 (_mm512_prefetch_i64scatter_ps): Ditto.
415 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
416 * config/i386/i386-builtin-types.def: Define
417 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
418 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
419 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
420 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
421 IX86_BUILTIN_SCATTERPFQPD.
422 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
423 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
424 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
425 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
426 __builtin_ia32_scatterpfqps.
427 (ix86_expand_builtin): Expand new built-ins.
428 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
429 fix memory access data type.
430 (*avx512pf_gatherpf<mode>_mask): Ditto.
431 (*avx512pf_gatherpf<mode>): Ditto.
432 (avx512pf_scatterpf<mode>): Ditto.
433 (*avx512pf_scatterpf<mode>_mask): Ditto.
434 (*avx512pf_scatterpf<mode>): Ditto.
435 (GATHER_SCATTER_SF_MEM_MODE): New.
436 (avx512pf_gatherpf<mode>df): Ditto.
437 (*avx512pf_gatherpf<mode>df_mask): Ditto.
438 (*avx512pf_scatterpf<mode>df): Ditto.
440 2014-01-27 Jakub Jelinek <jakub@redhat.com>
443 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
444 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
447 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
449 * common/config/arm/arm-common.c
450 (arm_rewrite_mcpu): Handle multiple names.
452 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
454 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
456 * gimple-builder.h (create_gimple_tmp): Delete.
458 2014-01-27 Christian Bruel <christian.bruel@st.com>
460 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
463 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
465 * config/pa/pa.md (call): Generate indirect long calls to non-local
466 functions when outputing 32-bit code.
467 (call_value): Likewise except for special call to buggy powf function.
469 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
470 portable runtime and PIC indirect calls.
471 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
472 and PIC call sequences. Use ldo instead of blr to set return register
473 in PIC call sequence.
475 2014-01-25 Walter Lee <walt@tilera.com>
477 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
478 avoid clobbering a live register.
480 2014-01-25 Walter Lee <walt@tilera.com>
482 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
483 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
484 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
485 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
487 2014-01-25 Walter Lee <walt@tilera.com>
489 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
490 arguments on even registers.
491 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
493 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
494 (BIGGEST_ALIGNMENT): Ditto.
495 (BIGGEST_FIELD_ALIGNMENT): Ditto.
497 2014-01-25 Walter Lee <walt@tilera.com>
499 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
500 insns before bundling.
501 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
503 2014-01-25 Walter Lee <walt@tilera.com>
505 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
506 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
507 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
509 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
511 * config/mips/constraints.md (kl): Delete.
512 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
513 define expands, using...
514 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
515 instructions for MIPS16.
516 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
517 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
519 2014-01-25 Walter Lee <walt@tilera.com>
521 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
525 2014-01-25 Walter Lee <walt@tilera.com>
527 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
528 (TARGET_EXPAND_TO_RTL_HOOK): Define.
530 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
532 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
535 2014-01-25 Jakub Jelinek <jakub@redhat.com>
537 * print-rtl.c (in_call_function_usage): New var.
538 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
539 EXPR_LIST mode as mode and not as reg note name.
542 * cfgloopmanip.c (copy_loop_info): If
543 loop->warned_aggressive_loop_optimizations, make sure
544 the flag is set in target loop too.
546 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
548 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
550 * builtins.def: Likewise.
551 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
552 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
553 * ira.c (ira_setup_eliminable_regset): Likewise.
554 * omp-low.c (gate_expand_omp): Likewise.
555 (execute_lower_omp): Likewise.
556 (diagnose_sb_0): Likewise.
557 (gate_diagnose_omp_blocks): Likewise.
558 (simd_clone_clauses_extract): Likewise.
561 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
563 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
564 correction for little endian...
565 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
568 2014-01-24 Jeff Law <law@redhat.com>
570 PR tree-optimization/59919
571 * tree-vrp.c (find_assert_locations_1): Do not register asserts
572 for non-returning calls.
574 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
576 * common/config/aarch64/aarch64-common.c
577 (aarch64_rewrite_mcpu): Handle multiple names.
578 * config/aarch64/aarch64.h
579 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
581 2014-01-24 Dodji Seketeli <dodji@redhat.com>
583 * input.c (add_file_to_cache_tab): Handle the case where fopen
586 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
589 * config/i386/i386.md (pushsf splitter): Get stack adjustment
590 from push operand if code of push isn't PRE_DEC.
592 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
595 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
596 -mquad-memory-atomic. Update -mquad-memory documentation to say
597 it is only used for non-atomic loads/stores.
599 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
600 -mquad-memory or -mquad-memory-atomic switches.
602 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
603 -mquad-memory-atomic to ISA 2.07 support.
605 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
606 to separate support of normal quad word memory operations (ldq, stq)
607 from the atomic quad word memory operations.
609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
610 support to separate non-atomic quad word operations from atomic
611 quad word operations. Disable non-atomic quad word operations in
612 little endian mode so that we don't have to swap words after the
613 load and before the store.
614 (quad_load_store_p): Add comment about atomic quad word support.
615 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
616 options printed with -mdebug=reg.
618 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
619 -mquad-memory-atomic as the test for whether we have quad word
621 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
622 or -mp8-vector are used, allow byte/half-word atomic operations.
624 * config/rs6000/sync.md (load_lockedti): Insure that the address
625 is a proper indexed or indirect address for the lqarx instruction.
626 On little endian systems, swap the hi/lo registers after the lqarx
628 (load_lockedpti): Use indexed_or_indirect_operand predicate to
629 insure the address is valid for the lqarx instruction.
630 (store_conditionalti): Insure that the address is a proper indexed
631 or indirect address for the stqcrx. instruction. On little endian
632 systems, swap the hi/lo registers before doing the stqcrx.
634 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
635 insure the address is valid for the stqcrx. instruction.
637 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
638 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
639 type of quad memory support is available.
641 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
644 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
645 there is a danger of looping.
647 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
649 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
650 force flag_ira_loop_pressure if set via command line.
652 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
654 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
655 (ashr_simd): New builtin handling DI mode.
656 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
657 (aarch64_sshr_simddi): New match pattern.
658 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
659 (vshrd_n_s64): Likewise.
660 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
662 2014-01-23 Nick Clifton <nickc@redhat.com>
664 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
665 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
666 favour of mcu specific scripts.
667 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
670 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
671 Alex Velenko <Alex.Velenko@arm.com>
673 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
674 (vaddv_s16): Likewise.
675 (vaddv_s32): Likewise.
676 (vaddv_u8): Likewise.
677 (vaddv_u16): Likewise.
678 (vaddv_u32): Likewise.
679 (vaddvq_s8): Likewise.
680 (vaddvq_s16): Likewise.
681 (vaddvq_s32): Likewise.
682 (vaddvq_s64): Likewise.
683 (vaddvq_u8): Likewise.
684 (vaddvq_u16): Likewise.
685 (vaddvq_u32): Likewise.
686 (vaddvq_u64): Likewise.
687 (vaddv_f32): Likewise.
688 (vaddvq_f32): Likewise.
689 (vaddvq_f64): Likewise.
690 (vmaxv_f32): Likewise.
691 (vmaxv_s8): Likewise.
692 (vmaxv_s16): Likewise.
693 (vmaxv_s32): Likewise.
694 (vmaxv_u8): Likewise.
695 (vmaxv_u16): Likewise.
696 (vmaxv_u32): Likewise.
697 (vmaxvq_f32): Likewise.
698 (vmaxvq_f64): Likewise.
699 (vmaxvq_s8): Likewise.
700 (vmaxvq_s16): Likewise.
701 (vmaxvq_s32): Likewise.
702 (vmaxvq_u8): Likewise.
703 (vmaxvq_u16): Likewise.
704 (vmaxvq_u32): Likewise.
705 (vmaxnmv_f32): Likewise.
706 (vmaxnmvq_f32): Likewise.
707 (vmaxnmvq_f64): Likewise.
708 (vminv_f32): Likewise.
709 (vminv_s8): Likewise.
710 (vminv_s16): Likewise.
711 (vminv_s32): Likewise.
712 (vminv_u8): Likewise.
713 (vminv_u16): Likewise.
714 (vminv_u32): Likewise.
715 (vminvq_f32): Likewise.
716 (vminvq_f64): Likewise.
717 (vminvq_s8): Likewise.
718 (vminvq_s16): Likewise.
719 (vminvq_s32): Likewise.
720 (vminvq_u8): Likewise.
721 (vminvq_u16): Likewise.
722 (vminvq_u32): Likewise.
723 (vminnmv_f32): Likewise.
724 (vminnmvq_f32): Likewise.
725 (vminnmvq_f64): Likewise.
727 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
729 * config/aarch64/aarch64-simd.md
730 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
731 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
732 (*aarch64_mul3_elt<mode>): Likewise.
733 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
734 (*aarch64_mul3_elt_to_64v2df): Likewise.
735 (*aarch64_mla_elt<mode>): Likewise.
736 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
737 (*aarch64_mls_elt<mode>): Likewise.
738 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
739 (*aarch64_fma4_elt<mode>): Likewise.
740 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
741 (*aarch64_fma4_elt_to_64v2df): Likewise.
742 (*aarch64_fnma4_elt<mode>): Likewise.
743 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
744 (*aarch64_fnma4_elt_to_64v2df): Likewise.
745 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
746 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
747 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
748 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
749 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
750 (aarch64_sqdmull_lane<mode>_internal): Likewise.
751 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
753 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
755 * config/aarch64/aarch64-simd.md
756 (aarch64_be_checked_get_lane<mode>): New define_expand.
757 * config/aarch64/aarch64-simd-builtins.def
758 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
759 New builtin definition.
760 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
761 Use new safe be builtin.
763 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
765 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
767 (aarch64_be_st1<mode>): Likewise.
768 (aarch_ld1<VALL:mode>): Define_expand modified.
769 (aarch_st1<VALL:mode>): Likewise.
770 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
771 (UNSPEC_ST1): Likewise.
773 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
775 * config/microblaze/microblaze.md: Add trap insn and attribute
777 2014-01-23 Dodji Seketeli <dodji@redhat.com>
779 PR preprocessor/58580
780 * input.h (location_get_source_line): Take an additional line_size
782 (void diagnostics_file_cache_fini): Declare new function.
783 * input.c (struct fcache): New type.
784 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
785 New static constants.
786 (diagnostic_file_cache_init, total_lines_num)
787 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
788 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
789 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
790 (get_next_line, read_next_line, goto_next_line, read_line_num):
791 New static function definitions.
792 (diagnostic_file_cache_fini): New function.
793 (location_get_source_line): Take an additional output line_len
794 parameter. Re-write using lookup_or_add_file_to_cache_tab and
796 * diagnostic.c (diagnostic_finish): Call
797 diagnostic_file_cache_fini.
798 (adjust_line): Take an additional input parameter for the length
799 of the line, rather than calculating it with strlen.
800 (diagnostic_show_locus): Adjust the use of
801 location_get_source_line and adjust_line with respect to their new
802 signature. While displaying a line now, do not stop at the first
803 null byte. Rather, display the zero byte as a space and keep
804 going until we reach the size of the line.
805 * Makefile.in: Add vec.o to OBJS-libcommon
807 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
808 Ilya Tocar <ilya.tocar@intel.com>
810 * config/i386/avx512fintrin.h (_mm512_kmov): New.
811 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
812 (__builtin_ia32_kmov16): Ditto.
813 * config/i386/i386.md (UNSPEC_KMOV): New.
816 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
818 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
819 (_mm512_storeu_si512): Ditto.
821 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
824 * rtl.h (get_referenced_operands): Declare.
825 * recog.c (get_referenced_operands): New function.
826 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
827 operands have been referenced when recording LO_SUM references.
829 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
831 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
833 2014-01-22 Jan Hubicka <jh@suse.cz>
835 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
836 Enable for generic and recent AMD targets.
838 2014-01-22 Jan Hubicka <jh@suse.cz>
840 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
841 ARG_SIZE note when adjustment was eliminated.
843 2014-01-22 Jeff Law <law@redhat.com>
845 PR tree-optimization/59597
846 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
847 in file. Accept new argument REGISTERING and use it to modify
848 dump output appropriately.
849 (register_jump_thread): Corresponding changes.
850 (mark_threaded_blocks): Reinstate code to cancel unprofitable
851 thread paths involving joiner blocks. Add code to dump cancelled
852 jump threading paths.
854 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
856 PR rtl-optimization/59477
857 * lra-constraints.c (inherit_in_ebb): Process call for living hard
858 regs. Update reloads_num and potential_reload_hard_regs for all insns.
860 2014-01-22 Tom Tromey <tromey@redhat.com>
862 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
864 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
866 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
868 PR rtl-optimization/59896
869 * lra-constraints.c (process_alt_operands): Check unused note for
870 matched operands of insn with no output reloads.
872 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
874 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
875 (mips_move_from_gpr_cost): Likewise.
877 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
879 PR rtl-optimization/59858
880 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
881 ira_class_hard_regs_num.
882 (process_alt_operands): Increase reject for dying matched operand.
884 2014-01-21 Jakub Jelinek <jakub@redhat.com>
887 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
888 smaller than size, perform several stores or loads and stores
889 at dst + count - size to store or copy all of size bytes, rather
890 than just last modesize bytes.
892 2014-01-20 DJ Delorie <dj@redhat.com>
894 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
895 that CLOBBERs are REGs before propogating their values.
897 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
900 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
901 (cgraph_inline_failed_type): New function.
902 * cgraph.h (DEFCIFCODE): Add type.
903 (cgraph_inline_failed_type_t): New enum.
904 (cgraph_inline_failed_type): New prototype.
905 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
906 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
907 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
908 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
909 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
910 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
911 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
912 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
913 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
914 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
915 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
916 OPTIMIZATION_MISMATCH.
917 * tree-inline.c (expand_call_inline): Emit errors during
918 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
920 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
923 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
924 mode attribute in insn output.
926 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
928 * output.h (output_constant): Delete.
929 * varasm.c (output_constant): Make private.
931 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
933 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
935 2014-01-20 Jakub Jelinek <jakub@redhat.com>
938 * tree.h (fold_builtin_strcat): New prototype.
939 * builtins.c (fold_builtin_strcat): No longer static. Add len
940 argument, if non-NULL, don't call c_strlen. Optimize
941 directly into __builtin_memcpy instead of __builtin_strcpy.
942 (fold_builtin_2): Adjust fold_builtin_strcat caller.
943 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
945 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
947 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
948 for SImode_address_operand operands, having only a REG argument.
950 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
952 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
953 loader name using mbig-endian.
954 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
956 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
958 * doc/invoke.texi (-march): Clarify documentation for AArch64.
962 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
964 * config/aarch64/aarch64-protos.h
965 (aarch64_cannot_change_mode_class_ptr): Declare.
966 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
967 aarch64_cannot_change_mode_class_ptr): New.
968 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
969 backend hook aarch64_cannot_change_mode_class.
971 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
973 * common/config/aarch64/aarch64-common.c
974 (aarch64_handle_option): Don't handle any option order logic here.
975 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
976 selected_cpu, warn on architecture version mismatch.
977 (aarch64_override_options): Fix parsing order for option strings.
979 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
980 Iain Sandoe <iain@codesourcery.com>
983 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
984 warning. Amend comment to reflect current functionality.
986 2014-01-20 Richard Biener <rguenther@suse.de>
989 * builtins.c (fold_builtin_strcat): Remove case better handled
990 by tree-ssa-strlen.c.
992 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
994 * config/aarch64/aarch64.opt
995 (mcpu, march, mtune): Make case-insensitive.
997 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1000 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1001 if operands[1] is a REG or ZERO_EXTEND of a REG.
1003 2014-01-19 Jan Hubicka <jh@suse.cz>
1005 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1007 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1009 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1010 long non-pic millicode calls.
1012 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1014 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1016 2014-01-19 Kito Cheng <kito@0xlab.org>
1018 * builtins.c (expand_movstr): Check movstr expand done or fail.
1020 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1021 H.J. Lu <hongjiu.lu@intel.com>
1024 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1025 to DImode for zero-extended addresses.
1027 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1029 PR rtl-optimization/57763
1030 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1031 on the new indirect jump_insn and increment LABEL_NUSES (label).
1033 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1037 * config.gcc (x86_archs): New variable.
1038 (x86_64_archs): Likewise.
1039 (x86_cpus): Likewise.
1040 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1041 --with-arch/--with-cpu= options.
1042 Support --with-arch=/--with-cpu={nehalem,westmere,
1043 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1045 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1047 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1048 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1050 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1052 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1054 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1057 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1058 clear cpp_get_options (parse_in)->warn_unused_macros for
1059 ix86_target_macros_internal with cpp_define.
1061 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1063 * jump.c (delete_related_insns): Keep (use (insn))s.
1064 * reorg.c (redundant_insn): Check for barriers too.
1066 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1068 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1070 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1072 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1073 call to $$dyncall when TARGET_LONG_CALLS is true.
1075 2014-01-17 Jeff Law <law@redhat.com>
1077 * ree.c (combine_set_extension): Temporarily disable test for
1078 changing number of hard registers.
1080 2014-01-17 Jan Hubicka <jh@suse.cz>
1083 * ipa-inline-analysis.c (inline_free_summary):
1084 Do not free summary of aliases.
1086 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1089 * gimplify.c (gimplify_expr): Use create_tmp_var
1090 instead of create_tmp_var_raw. If cond doesn't have
1091 integral type, don't add the IFN_ANNOTATE builtin at all.
1093 2014-01-17 Martin Jambor <mjambor@suse.cz>
1096 * ipa-cp.c (prev_edge_clone): New variable.
1097 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1098 Also resize prev_edge_clone vector.
1099 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1100 (ipcp_edge_removal_hook): New function.
1101 (ipcp_driver): Register ipcp_edge_removal_hook.
1103 2014-01-17 Andrew Pinski <apinski@cavium.com>
1104 Steve Ellcey <sellcey@mips.com>
1107 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1110 2014-01-17 Jeff Law <law@redhat.com>
1113 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1114 so that pass_ccp runs first.
1116 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1118 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1119 (ix86_adjust_cost): Use !TARGET_XXX.
1120 (do_reorder_for_imul): Likewise.
1121 (swap_top_of_ready_list): Likewise.
1122 (ix86_sched_reorder): Likewise.
1124 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1126 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1127 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1128 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1129 (intel_memset): New. Duplicate slm_memset.
1130 (intel_cost): New. Duplicate slm_cost.
1131 (m_INTEL): New macro.
1132 (processor_target_table): Add "intel".
1133 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1134 with PROCESSOR_INTEL for "intel".
1135 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1136 PROCESSOR_SILVERMONT.
1137 (ix86_issue_rate): Likewise.
1138 (ix86_adjust_cost): Likewise.
1139 (ia32_multipass_dfa_lookahead): Likewise.
1140 (swap_top_of_ready_list): Likewise.
1141 (ix86_sched_reorder): Likewise.
1142 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1143 instead of TARGET_OPT_AGU.
1144 * config/i386/i386.h (TARGET_INTEL): New.
1145 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1146 (processor_type): Add PROCESSOR_INTEL.
1147 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1148 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1150 2014-01-17 Marek Polacek <polacek@redhat.com>
1153 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1156 2014-01-17 Richard Biener <rguenther@suse.de>
1158 PR tree-optimization/46590
1159 * opts.c (default_options_table): Add entries for
1160 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1161 all enabled at -O1 but not for -Og.
1162 * common.opt (fbranch-count-reg): Remove Init(1).
1163 (fmove-loop-invariants): Likewise.
1164 (ftree-pta): Likewise.
1166 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1168 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1169 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1170 decls to at least the GCC 4.8 used alignments.
1173 * tree-nested.c (convert_nonlocal_reference_stmt,
1174 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1175 of GIMPLE_BIND stmts, adjust associated decls.
1177 2014-01-17 Richard Biener <rguenther@suse.de>
1179 PR tree-optimization/46590
1180 * vec.h (vec<>::bseach): New member function implementing
1181 binary search according to C89 bsearch.
1182 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1183 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1184 bitmap pointer again. Make accesses_in_loop a flat array.
1185 (mem_ref_obstack): New global.
1186 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1187 (mark_ref_stored): Likewise.
1188 (ref_indep_loop_p_2): Likewise.
1189 (set_ref_stored_in_loop): New helper function.
1190 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1191 (memref_free): Adjust.
1192 (record_mem_ref_loc): Simplify.
1193 (gather_mem_refs_stmt): Adjust.
1194 (sort_locs_in_loop_postorder_cmp): New function.
1195 (analyze_memory_references): Sort accesses_in_loop after
1196 loop postorder number.
1197 (find_ref_loc_in_loop_cmp): New function.
1198 (for_all_locs_in_loop): Find relevant cluster of locs in
1199 accesses_in_loop and iterate without recursion.
1200 (execute_sm): Avoid uninit warning.
1201 (struct ref_always_accessed): Simplify.
1202 (ref_always_accessed::operator ()): Likewise.
1203 (ref_always_accessed_p): Likewise.
1204 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1205 loop postorder numbers here.
1206 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1209 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1212 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1213 on decls for which assemble_alias has been called.
1215 2014-01-17 Nick Clifton <nickc@redhat.com>
1217 * config/msp430/msp430.opt: (mcpu): New option.
1218 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1219 (msp430_option_override): Parse target_cpu. If the MCU name
1220 matches a generic string, clear target_mcu.
1221 (msp430_attr): Allow numeric interrupt values up to 63.
1222 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1223 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1225 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1227 * config/msp430/msp430.md (popm): Use %J rather than %I.
1228 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1229 (addhi_cy_i): Use immediate_operand for operand 2.
1230 * doc/invoke.texi: Document -mcpu option.
1232 2014-01-17 Richard Biener <rguenther@suse.de>
1234 PR rtl-optimization/38518
1235 * df.h (df_analyze_loop): Declare.
1236 * df-core.c: Include cfgloop.h.
1237 (df_analyze_1): Split out main part of df_analyze.
1238 (df_analyze): Adjust.
1239 (loop_inverted_post_order_compute): New function.
1240 (loop_post_order_compute): Likewise.
1241 (df_analyze_loop): New function avoiding whole-function
1243 * loop-invariant.c (find_defs): Use df_analyze_loop.
1244 (find_invariants): Adjust.
1245 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1247 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1249 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1250 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1252 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1254 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1255 traversal when removing references.
1257 2014-01-16 Jan Hubicka <jh@suse.cz>
1260 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1262 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1265 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1266 pushing a reload for an autoinc when we had previously reloaded an
1267 inner part of the address.
1269 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1271 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1273 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1274 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1275 when not giving up or versioning for alias only because of
1277 (vect_analyze_data_ref_dependences): Set to true.
1278 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1280 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1281 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1285 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1288 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1289 operand 0 predicate for gathers, use a new pseudo as subtarget.
1291 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1294 * lra-constraints.c (process_alt_operands): Add printing debug info.
1295 Check absence of input/output reloads for matched operands too.
1297 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1299 PR rtl-optimization/59835
1300 * ira.c (ira_init_register_move_cost): Increase cost for
1303 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1305 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1307 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1310 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1311 non-register objects. Use gen_(high/low)part more consistently.
1314 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1317 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1318 endian support, remove tests for WORDS_BIG_ENDIAN.
1319 (p8_mfvsrd_3_<mode>): Likewise.
1320 (reload_gpr_from_vsx<mode>): Likewise.
1321 (reload_gpr_from_vsxsf): Likewise.
1322 (p8_mfvsrd_4_disf): Likewise.
1324 2014-01-16 Richard Biener <rguenther@suse.de>
1326 PR rtl-optimization/46590
1327 * lcm.c (compute_antinout_edge): Use postorder iteration.
1328 (compute_laterin): Use inverted postorder iteration.
1330 2014-01-16 Nick Clifton <nickc@redhat.com>
1333 * varasm.c (output_constant): Return the number of bytes actually
1335 (output_constructor_array_range): Update the field size with the
1336 number of bytes emitted by output_constant.
1337 (output_constructor_regular_field): Likewise. Also do not
1338 complain if the total number of bytes emitted is now greater
1339 than the expected fieldpos.
1340 * output.h (output_constant): Update prototype and descriptive comment.
1342 2014-01-16 Marek Polacek <polacek@redhat.com>
1345 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1346 it is error_mark_node.
1348 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1350 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1351 VALID_AVX256_REG_OR_OI_MODE.
1353 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1355 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1356 current procedure should be profiled.
1358 2014-01-15 Andrew Pinski <apinski@cavium.com>
1360 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1361 of moving from/to the STACK_REG register class.
1363 2014-01-15 Richard Henderson <rth@redhat.com>
1366 * reginfo.c (global_regs_decl): Globalize.
1367 * rtl.h (global_regs_decl): Declare.
1368 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1369 reserved via global_regs.
1371 2014-01-15 Teresa Johnson <tejohnson@google.com>
1373 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1375 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1377 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1378 and vmulosh rather than call gen_vec_widen_smult_*.
1379 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1380 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1381 (vec_widen_smult_even_v16qi): Likewise.
1382 (vec_widen_umult_even_v8hi): Likewise.
1383 (vec_widen_smult_even_v8hi): Likewise.
1384 (vec_widen_umult_odd_v16qi): Likewise.
1385 (vec_widen_smult_odd_v16qi): Likewise.
1386 (vec_widen_umult_odd_v8hi): Likewise.
1387 (vec_widen_smult_odd_v8hi): Likewise.
1388 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1389 vmuloub rather than call gen_vec_widen_umult_*.
1390 (vec_widen_umult_lo_v16qi): Likewise.
1391 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1392 vmulosb rather than call gen_vec_widen_smult_*.
1393 (vec_widen_smult_lo_v16qi): Likewise.
1394 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1395 rather than call gen_vec_widen_umult_*.
1396 (vec_widen_umult_lo_v8hi): Likewise.
1397 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1398 rather than call gen_vec_widen_smult_*.
1399 (vec_widen_smult_lo_v8hi): Likewise.
1401 2014-01-15 Jeff Law <law@redhat.com>
1403 PR tree-optimization/59747
1404 * ree.c (find_and_remove_re): Properly handle case where a second
1405 eliminated extension requires widening a copy created for elimination
1406 of a prior extension.
1407 (combine_set_extension): Ensure that the number of hard regs needed
1408 for a destination register does not change when we widen it.
1410 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1412 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1413 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1414 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1415 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1416 (avr-*-rtems*): Likewise.
1417 (bfin*-rtems*): Likewise.
1418 (moxie-*-rtems*): Likewise.
1419 (h8300-*-rtems*): Likewise.
1420 (i[34567]86-*-rtems*): Likewise.
1421 (lm32-*-rtems*): Likewise.
1422 (m32r-*-rtems*): Likewise.
1423 (m68k-*-rtems*): Likewise.
1424 (microblaze*-*-rtems*): Likewise.
1425 (mips*-*-rtems*): Likewise.
1426 (powerpc-*-rtems*): Likewise.
1427 (sh-*-rtems*): Likewise.
1428 (sparc-*-rtems*): Likewise.
1429 (sparc64-*-rtems*): Likewise.
1430 (v850-*-rtems*): Likewise.
1431 (m32c-*-rtems*): Likewise.
1433 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1435 PR rtl-optimization/59511
1436 * ira.c (ira_init_register_move_cost): Use memory costs for some
1437 cases of register move cost calculations.
1438 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1439 instead of BB frequency.
1440 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1441 * lra-assigns.c (find_hard_regno_for): Ditto.
1443 2014-01-15 Richard Biener <rguenther@suse.de>
1445 PR tree-optimization/59822
1446 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1447 (vectorizable_load): Use it to hoist defs of uses of invariant
1448 loads out of the loop.
1450 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1451 Kugan Vivekanandarajah <kuganv@linaro.org>
1454 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1457 2014-01-15 Richard Biener <rguenther@suse.de>
1459 PR rtl-optimization/59802
1460 * lcm.c (compute_available): Use inverted postorder to seed
1461 the initial worklist.
1463 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1466 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1467 ADDR_REGS for invalid symrefs in non-PIC code.
1469 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1472 * builtins.c (determine_block_size): Initialize *probable_max_size
1473 even if len_rtx is CONST_INT.
1475 2014-01-14 Andrew Pinski <apinski@cavium.com>
1477 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1478 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1479 (cortexa53_tunings): Likewise.
1480 (aarch64_sched_issue_rate): New function.
1481 (TARGET_SCHED_ISSUE_RATE): Define.
1483 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1485 * ira-costs.c (find_costs_and_classes): Add missed
1486 ira_init_register_move_cost_if_necessary.
1488 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1491 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1493 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1496 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1497 to indicate if type is used for function return value. Warn ABI
1498 change if the vector mode isn't available for function return value.
1499 (ix86_function_arg_advance): Pass false to type_natural_mode.
1500 (ix86_function_arg): Likewise.
1501 (ix86_gimplify_va_arg): Likewise.
1502 (function_arg_32): Don't warn ABI change.
1503 (ix86_function_value): Pass true to type_natural_mode.
1504 (ix86_return_in_memory): Likewise.
1505 (ix86_struct_value_rtx): Removed.
1506 (TARGET_STRUCT_VALUE_RTX): Likewise.
1508 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1510 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1511 converting a conditional jump into a conditional return.
1513 2014-01-14 Richard Biener <rguenther@suse.de>
1515 PR tree-optimization/58921
1516 PR tree-optimization/59006
1517 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1518 hoisting invariant stmts.
1519 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1520 invariant loads on the preheader edge if possible.
1522 2014-01-14 Joey Ye <joey.ye@arm.com>
1524 * doc/plugin.texi (Building GCC plugins): Update to C++.
1526 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1528 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1529 (_mm_rcp28_round_ss): Ditto.
1530 (_mm_rsqrt28_round_sd): Ditto.
1531 (_mm_rsqrt28_round_ss): Ditto.
1532 (_mm_rcp28_sd): Ditto.
1533 (_mm_rcp28_ss): Ditto.
1534 (_mm_rsqrt28_sd): Ditto.
1535 (_mm_rsqrt28_ss): Ditto.
1536 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1537 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1538 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1539 (IX86_BUILTIN_RCP28SD): Ditto.
1540 (IX86_BUILTIN_RCP28SS): Ditto.
1541 (IX86_BUILTIN_RSQRT28SD): Ditto.
1542 (IX86_BUILTIN_RSQRT28SS): Ditto.
1543 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1544 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1545 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1546 (ix86_expand_special_args_builtin): Expand new FTYPE.
1547 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1548 (srcp14<mode>): Make insn unary.
1549 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1550 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1551 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1552 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1553 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1554 Fix rounding: make it SAE only.
1555 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1557 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1559 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1560 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1561 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1562 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1563 (round_saeonly_mask_scalar_operand4): Ditto.
1564 (round_saeonly_mask_scalar_op3): Ditto.
1565 (round_saeonly_mask_scalar_op4): Ditto.
1567 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1569 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1570 Implement -maltivec=be for vec_insert and vec_extract.
1572 2014-01-10 DJ Delorie <dj@redhat.com>
1574 * config/msp430/msp430.md (call_internal): Don't allow memory
1575 references with SP as the base register.
1576 (call_value_internal): Likewise.
1577 * config/msp430/constraints.md (Yc): New. For memory references
1578 that don't use SP as a base register.
1580 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1581 "an integer without a # prefix"
1582 * config/msp430/msp430.md (epilogue_helper): Use it.
1584 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1587 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1588 AVX512F gather builtins.
1589 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1590 on gather decls with INTEGER_TYPE masktype.
1591 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1592 directly into the builtin rather than hoisting it before loop.
1594 PR tree-optimization/59387
1595 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1596 (scev_const_prop): If folded_casts and type has undefined overflow,
1597 use force_gimple_operand instead of force_gimple_operand_gsi and
1598 for each added stmt if it is assign with
1599 arith_code_with_undefined_signed_overflow, call
1600 rewrite_to_defined_overflow.
1601 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1602 gimple-fold.h instead.
1603 (arith_code_with_undefined_signed_overflow,
1604 rewrite_to_defined_overflow): Moved to ...
1605 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1606 rewrite_to_defined_overflow): ... here. No longer static.
1607 Include gimplify-me.h.
1608 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1609 rewrite_to_defined_overflow): New prototypes.
1611 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1613 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1615 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1617 * builtins.c (get_object_alignment_2): Minor tweak.
1618 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1620 2014-01-13 Christian Bruel <christian.bruel@st.com>
1622 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1623 optimized non constant lengths.
1625 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1628 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1629 load as __atomic_load_N if possible.
1631 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1633 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1635 (rs6000_expand_builtin): Adjust call.
1637 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1640 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1641 * config/rs6000/rs6000.c: Include target-globals.h.
1642 (rs6000_set_current_function): Instead of doing target_reinit
1643 unconditionally, use save_target_globals_default_opts and
1644 restore_target_globals.
1646 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1648 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1649 (rs6000_expand_builtin): Handle mffs and mtfsf.
1650 (rs6000_init_builtins): Define mffs and mtfsf.
1651 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1652 (rs6000_mffs): New pattern.
1653 (rs6000_mtfsf): New pattern.
1655 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1657 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1658 Start narrowing with START. Apply candidate-use pair
1659 and check overall cost in narrowing.
1660 (iv_ca_prune): Pass new argument.
1662 2014-01-10 Jeff Law <law@redhat.com>
1665 * ree.c (combine_reaching_defs): Ensure the defining statement
1666 occurs before the extension when optimizing extensions with
1667 different source and destination hard registers.
1669 2014-01-10 Jan Hubicka <jh@suse.cz>
1672 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1673 vtables into the type inheritance graph.
1675 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1677 PR rtl-optimization/59754
1678 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1679 modes in the REGNO != REGNO case.
1681 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1683 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1685 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1687 PR tree-optimization/59745
1688 * tree-predcom.c (tree_predictive_commoning_loop): Call
1689 free_affine_expand_cache if giving up because components is NULL.
1691 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1692 GC in payload of target_globals struct instead of allocating them on
1693 the heap and the larger structs separately using GC.
1694 * target-globals.h (struct target_globals): Make regs, hard_regs,
1695 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1696 of GTY((skip)) and change type to void *.
1697 (reset_target_globals): Cast loads from those fields to corresponding
1700 2014-01-10 Steve Ellcey <sellcey@mips.com>
1703 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1704 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1705 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1707 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1710 * aarch64-modes.def (CC_Zmode): New flags mode.
1711 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1712 represents an equality.
1713 (aarch64_get_condition_code): Handle CC_Zmode.
1714 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1716 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1718 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1719 extraction in good case.
1721 2014-01-10 Richard Biener <rguenther@suse.de>
1723 PR tree-optimization/59374
1724 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1725 checking after SLP discovery. Mark stmts not participating
1726 in any SLP instance properly.
1728 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1730 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1731 when handling a SET rtx.
1733 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1735 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1736 (cortex-a57): Likewise.
1737 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1739 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1741 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1742 non-iwmmxt builtins.
1744 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1748 * ipa-devirt.c record_target_from_binfo): Take as argument
1749 stack of binfos and lookup matching one for virtual inheritance.
1750 (possible_polymorphic_call_targets_1): Update.
1752 2014-01-10 Huacai Chen <chenhc@lemote.com>
1754 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1755 kernel strings for Loongson-2E/2F/3A.
1757 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1760 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1761 is_gimple_call before calling gimple_call_internal_p.
1763 2014-01-09 Steve Ellcey <sellcey@mips.com>
1765 * Makefile.in (TREE_FLOW_H): Remove.
1766 (TREE_SSA_H): Add file names from tree-flow.h.
1767 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1768 * tree.h: Remove tree-flow.h reference.
1769 * hash-table.h: Remove tree-flow.h reference.
1770 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1771 reference with tree-ssa-loop.h.
1773 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1775 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1776 default element-order behavior for -maltivec.
1777 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1778 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1779 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1780 when targeting big endian, at least for now.
1781 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1783 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1786 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1787 var satisfies use_register_for_decl, just take into account type
1788 alignment, rather than decl alignment.
1790 PR tree-optimization/59622
1791 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1792 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1793 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1794 Don't devirtualize for inplace at all. For targets.length () == 1,
1795 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1797 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1799 * config/i386/i386.md (cpu): Remove the unused btver1.
1801 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1803 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1805 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1808 * tree-core.h (struct target_globals): New forward declaration.
1809 (struct tree_target_option): Add globals field.
1810 * tree.h (TREE_TARGET_GLOBALS): Define.
1811 (prepare_target_option_nodes_for_pch): New prototype.
1812 * target-globals.h (struct target_globals): Define even if
1814 * tree.c (prepare_target_option_node_for_pch,
1815 prepare_target_option_nodes_for_pch): New functions.
1816 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1817 * config/i386/i386.c: Include target-globals.h.
1818 (ix86_set_current_function): Instead of doing target_reinit
1819 unconditionally, use save_target_globals_default_opts and
1820 restore_target_globals.
1822 2014-01-09 Richard Biener <rguenther@suse.de>
1824 PR tree-optimization/59715
1825 * tree-cfg.h (split_critical_edges): Declare.
1826 * tree-cfg.c (split_critical_edges): Export.
1827 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1829 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1831 * cfgexpand.c (expand_stack_vars): Optionally disable
1832 asan stack protection.
1833 (expand_used_vars): Likewise.
1834 (partition_stack_vars): Likewise.
1835 * asan.c (asan_emit_stack_protection): Optionally disable
1836 after return stack usage.
1837 (instrument_derefs): Optionally disable memory access instrumentation.
1838 (instrument_builtin_call): Likewise.
1839 (instrument_strlen_call): Likewise.
1840 (asan_protect_global): Optionally disable global variables protection.
1841 * doc/invoke.texi: Added doc for new options.
1842 * params.def: Added new options.
1843 * params.h: Likewise.
1845 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1847 PR rtl-optimization/59724
1848 * ifcvt.c (cond_exec_process_if_block): Don't call
1849 flow_find_head_matching_sequence with 0 longest_match.
1850 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1851 non-active insns if !stop_after.
1852 (try_head_merge_bb): Revert 2014-01-07 changes.
1854 2014-01-08 Jeff Law <law@redhat.com>
1856 * ree.c (get_sub_rtx): New function, extracted from...
1857 (merge_def_and_ext): Here.
1858 (combine_reaching_defs): Use get_sub_rtx.
1860 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1862 * cgraph.h (varpool_variable_node): Do not choke on null node.
1864 2014-01-08 Catherine Moore <clm@codesourcery.com>
1866 * config/mips/mips.md (simple_return): Attempt to use JRC
1868 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1870 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1872 PR rtl-optimization/59137
1873 * reorg.c (steal_delay_list_from_target): Call update_block for
1875 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1877 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1879 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1880 two duplicate entries.
1882 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1885 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1887 * config/mips/mips.c (mips_truncated_op_cost): New function.
1888 (mips_rtx_costs): Adjust test for BADDU.
1889 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1891 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1893 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1894 (*baddu_si): ...this new pattern.
1896 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1899 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1901 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1904 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1906 (expand_expr, expand_normal): Adjust.
1907 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1908 inner_reference_p. Use inner_reference_p to expand inner references.
1909 (store_expr): Adjust.
1910 * cfgexpand.c (expand_call_stmt): Adjust.
1912 2014-01-08 Rong Xu <xur@google.com>
1914 * gcov-io.c (gcov_var): Move from gcov-io.h.
1915 (gcov_position): Ditto.
1916 (gcov_is_error): Ditto.
1917 (gcov_rewrite): Ditto.
1918 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1919 only part to libgcc/libgcov.h.
1921 2014-01-08 Marek Polacek <polacek@redhat.com>
1924 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1926 2014-01-08 Marek Polacek <polacek@redhat.com>
1929 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1931 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1933 PR rtl-optimization/59649
1934 * stor-layout.c (get_mode_bounds): For BImode return
1935 0 and STORE_FLAG_VALUE.
1937 2014-01-08 Richard Biener <rguenther@suse.de>
1940 * gimple.h (is_gimple_builtin_call): Remove.
1941 (gimple_builtin_call_types_compatible_p): New.
1942 (gimple_call_builtin_p): New overload.
1943 * gimple.c (is_gimple_builtin_call): Remove.
1944 (validate_call): Rename to ...
1945 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1947 (validate_type): New static function.
1948 (gimple_call_builtin_p): New overload and adjust.
1949 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1950 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1951 (gimple_fold_stmt_to_constant_1): Likewise.
1952 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1954 2014-01-08 Richard Biener <rguenther@suse.de>
1957 * gimplify.c (gimplify_expr): Gimplify register-register type
1958 VIEW_CONVERT_EXPRs to separate stmts.
1960 2014-01-07 Jeff Law <law@redhat.com>
1963 * ree.c (combine_set_extension): Handle case where source
1964 and destination registers in an extension insn are different.
1965 (combine_reaching_defs): Allow source and destination registers
1966 in extension to be different under limited circumstances.
1967 (add_removable_extension): Remove restriction that the
1968 source and destination registers in the extension are the same.
1969 (find_and_remove_re): Emit a copy from the extension's
1970 destination to its source after the defining insn if
1971 the source and destination registers are different.
1974 * ifcvt.c (merge_if_block): If we are merging a block with more than
1975 one successor with a block with no successors, remove any BARRIER
1976 after the second block.
1978 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1980 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1982 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1985 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1986 for 14-bit register offsets when INT14_OK_STRICT is false.
1988 2014-01-07 Roland Stigge <stigge@antcom.de>
1989 Michael Meissner <meissner@linux.vnet.ibm.com>
1992 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1993 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1995 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1997 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2000 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2002 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2003 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2004 rtx is const0_rtx or not.
2006 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2009 * target-globals.c (save_target_globals): Remove this_fn_optab
2011 * toplev.c: Include optabs.h.
2012 (target_reinit): Temporarily restore the global options if another
2013 set of options are in force.
2015 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2017 PR rtl-optimization/58668
2018 * cfgcleanup.c (flow_find_cross_jump): Don't count
2019 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2020 to determine what is counted.
2021 (flow_find_head_matching_sequence): Use active_insn_p to determine
2023 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2025 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2026 determine what is counted.
2028 PR tree-optimization/59643
2029 * tree-predcom.c (split_data_refs_to_components): If one dr is
2030 read and one write, determine_offset fails and the write isn't
2031 in the bad component, just put the read into the bad component.
2033 2014-01-07 Mike Stump <mikestump@comcast.net>
2034 Jakub Jelinek <jakub@redhat.com>
2037 * tree-core.h (struct tree_optimization_option): Change optabs
2038 type from unsigned char * to void *.
2039 * optabs.c (init_tree_optimization_optabs): Adjust
2040 TREE_OPTIMIZATION_OPTABS initialization.
2042 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2045 * config/i386/i386.h (struct machine_function): Add
2046 no_drap_save_restore field.
2047 * config/i386/i386.c (ix86_save_reg): Use
2048 !cfun->machine->no_drap_save_restore instead of
2049 crtl->stack_realign_needed.
2050 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2051 this function clears frame_pointer_needed. Set
2052 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2053 and DRAP reg is needed.
2055 2014-01-06 Marek Polacek <polacek@redhat.com>
2058 * doc/implement-c.texi: Mention that other integer types are
2059 permitted as bit-field types in strictly conforming mode.
2061 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2063 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2066 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2068 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2070 2014-01-06 Martin Jambor <mjambor@suse.cz>
2073 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2075 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2077 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2081 * var-tracking.c (add_stores): Preserve the value of the source even if
2082 we don't record the store.
2084 2014-01-06 Terry Guo <terry.guo@arm.com>
2086 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2088 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2091 * config/darwin.c (darwin_function_section): Adjust return values to
2092 correspond to optimisation changes made in r206070.
2094 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2096 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2097 from prefetch_block tune setting.
2098 (nocona_cost): Correct size of prefetch block to 64.
2100 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2102 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2103 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2104 used to save the static chain register in the computation of the offset
2105 from which the FP registers need to be restored.
2107 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2109 PR tree-optimization/59519
2110 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2111 ICE if get_current_def (current_new_name) is already non-NULL, as long
2112 as it is a phi result of some other phi in *new_exit_bb that has
2115 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2116 or vmovdqu* for misaligned_operand.
2117 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2118 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2119 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2120 aligned_mem for AVX512F masked aligned load and store builtins and for
2123 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2125 PR tree-optimization/59651
2126 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2127 Address range for negative step should be added by TYPE_SIZE_UNIT.
2129 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2131 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2132 overlapping registers also for registers other than the stack pointer.
2134 2014-01-03 Marek Polacek <polacek@redhat.com>
2137 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2140 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2143 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2146 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2147 (push splitter): Use <P:MODE_SIZE> instead of
2148 GET_MODE_SIZE (<P:MODE>mode).
2149 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2150 (mov -1, reg peephole2): Likewise.
2151 * config/i386/sse.md (*mov<mode>_internal,
2152 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2153 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2154 *<code><mode>3, *andnot<mode>3<mask_name>,
2155 <mask_codefor><code><mode>3<mask_name>): Likewise.
2156 * config/i386/subst.md (mask_mode512bit_condition,
2157 sd_mask_mode512bit_condition): Likewise.
2159 2014-01-02 Xinliang David Li <davidxl@google.com>
2161 PR tree-optimization/59303
2162 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2163 (dump_predicates): Better output format.
2164 (pred_equal_p): New function.
2165 (is_neq_relop_p): Ditto.
2166 (is_neq_zero_form_p): Ditto.
2167 (pred_expr_equal_p): Ditto.
2168 (pred_neg_p): Ditto.
2169 (simplify_pred): Ditto.
2170 (simplify_preds_2): Ditto.
2171 (simplify_preds_3): Ditto.
2172 (simplify_preds_4): Ditto.
2173 (simplify_preds): Ditto.
2175 (push_to_worklist): Ditto.
2176 (get_pred_info_from_cmp): Ditto.
2177 (is_degenerated_phi): Ditto.
2178 (normalize_one_pred_1): Ditto.
2179 (normalize_one_pred): Ditto.
2180 (normalize_one_pred_chain): Ditto.
2181 (normalize_preds): Ditto.
2182 (normalize_cond_1): Remove function.
2183 (normalize_cond): Ditto.
2184 (is_gcond_subset_of): Ditto.
2185 (is_subset_of_any): Ditto.
2186 (is_or_set_subset_of): Ditto.
2187 (is_and_set_subset_of): Ditto.
2188 (is_norm_cond_subset_of): Ditto.
2189 (pred_chain_length_cmp): Ditto.
2190 (convert_control_dep_chain_into_preds): Type change.
2191 (find_predicates): Ditto.
2192 (find_def_preds): Ditto.
2193 (destroy_predicates_vecs): Ditto.
2194 (find_matching_predicates_in_rest_chains): Ditto.
2195 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2196 (is_pred_expr_subset): Ditto.
2197 (is_pred_chain_subset_of): Ditto.
2198 (is_included_in): Ditto.
2199 (is_superset_of): Ditto.
2201 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2203 Update copyright years.
2205 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2207 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2208 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2209 config/arc/arc.md, config/arc/arc.opt,
2210 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2211 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2212 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2213 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2214 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2215 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2216 vtable-verify.c, vtable-verify.h: Use the standard form for the
2219 2014-01-02 Tobias Burnus <burnus@net-b.de>
2221 * gcc.c (process_command): Update copyright notice dates.
2222 * gcov-dump.c: Ditto.
2224 * doc/cpp.texi: Bump @copying's copyright year.
2225 * doc/cppinternals.texi: Ditto.
2226 * doc/gcc.texi: Ditto.
2227 * doc/gccint.texi: Ditto.
2228 * doc/gcov.texi: Ditto.
2229 * doc/install.texi: Ditto.
2230 * doc/invoke.texi: Ditto.
2232 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2234 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2236 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2238 * config/i386/sse.md (*mov<mode>_internal): Guard
2239 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2241 PR rtl-optimization/59647
2242 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2243 new_rtx into UNSIGNED_FLOAT rtxes.
2245 Copyright (C) 2014 Free Software Foundation, Inc.
2247 Copying and distribution of this file, with or without modification,
2248 are permitted in any medium without royalty provided the copyright
2249 notice and this notice are preserved.