1 2014-01-30 Nick Clifton <nickc@redhat.com>
3 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4 Rename to mn10300_store_multiple_regs.
5 * config/mn10300/mn10300.c: Likewise.
6 * config/mn10300/mn10300.md (store_movm): Fix typo: call
8 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
9 Call mn10300_store_multiple_regs.
11 2014-01-30 Nick Clifton <nickc@redhat.com>
12 DJ Delorie <dj@redhat.com>
14 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
15 %fp 2 to keep registers after it properly word-aligned.
16 (rl78_alloc_physical_registers_umul): Handle the case where both
17 input operands are the same.
19 2014-01-30 Richard Biener <rguenther@suse.de>
21 PR tree-optimization/59903
22 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
25 2014-01-30 Jason Merrill <jason@redhat.com>
28 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
31 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
33 2014-01-30 Richard Biener <rguenther@suse.de>
35 PR tree-optimization/59951
36 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
39 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
42 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
43 SFmode to DFmode case.
45 2014-01-29 DJ Delorie <dj@redhat.com>
47 * config/msp430/msp430.opt (-minrt): New.
48 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
50 (ENDFILE_SPEC): Likewise.
52 2014-01-29 Jan Hubicka <jh@suse.cz>
54 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
55 (estimate_function_body_sizes): Use it.
57 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
60 * dwarf2out.c (is_cxx_auto): New.
61 (is_base_type): Use it.
62 (gen_type_die_with_usage): Likewise.
64 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
66 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
67 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
68 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
69 -maltivec=be with LE targets.
70 (vsx_mergeh_<mode>): Likewise.
71 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
73 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
74 (altivec_vmrghb): Replace with define_expand and new
75 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
77 (altivec_vmrghb_direct): New define_insn.
78 (altivec_vmrghh): Replace with define_expand and new
79 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
81 (altivec_vmrghh_direct): New define_insn.
82 (altivec_vmrghw): Replace with define_expand and new
83 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
85 (altivec_vmrghw_direct): New define_insn.
86 (*altivec_vmrghsf): Adjust for endianness.
87 (altivec_vmrglb): Replace with define_expand and new
88 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
90 (altivec_vmrglb_direct): New define_insn.
91 (altivec_vmrglh): Replace with define_expand and new
92 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
94 (altivec_vmrglh_direct): New define_insn.
95 (altivec_vmrglw): Replace with define_expand and new
96 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
98 (altivec_vmrglw_direct): New define_insn.
99 (*altivec_vmrglsf): Adjust for endianness.
100 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
101 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
102 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
103 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
104 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
105 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
106 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
107 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
109 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
111 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
112 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
115 2014-01-29 Richard Biener <rguenther@suse.de>
117 PR tree-optimization/58742
118 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
119 associate_pointerplus_align.
120 (associate_pointerplus_diff): New function.
121 (associate_pointerplus): Likewise. Call associate_pointerplus_align
122 and associate_pointerplus_diff.
124 2014-01-29 Richard Biener <rguenther@suse.de>
126 * lto-streamer.h (LTO_major_version): Bump to 3.
127 (LTO_minor_version): Reset to 0.
129 2014-01-29 Renlin Li <Renlin.Li@arm.com>
131 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
132 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
133 (arm_file_start): Generate correct asm header for armv7ve.
134 * config/arm/bpabi.h: Add multilib support for armv7ve.
135 * config/arm/driver-arm.c: Change the architectures of cortex-a7
136 and cortex-a15 to armv7ve.
137 * config/arm/t-aprofile: Add multilib support for armv7ve.
138 * doc/invoke.texi: Document -march=armv7ve.
140 2014-01-29 Richard Biener <rguenther@suse.de>
142 PR tree-optimization/58742
143 * tree-ssa-forwprop.c (associate_plusminus): Return true
144 if we changed sth, defer EH cleanup to ...
145 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
146 (simplify_mult): New function.
148 2014-01-29 Jakub Jelinek <jakub@redhat.com>
151 PR tree-optimization/59920
152 * tree.c (build_common_builtin_nodes): Remove
153 __builtin_setjmp_dispatcher initialization.
154 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
155 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
156 instead of gsi_after_labels + manually skipping debug stmts.
157 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
158 ignore bbs with IFN_ABNORMAL_DISPATCHER.
159 * tree-inline.c (copy_edges_for_bb): Remove
160 can_make_abnormal_goto argument, instead add abnormal_goto_dest
161 argument. Ignore computed_goto_p stmts. Don't call
162 make_abnormal_goto_edges. If a call might need abnormal edges
163 for non-local gotos, see if it already has an edge to
164 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
165 with true argument, don't do anything then, otherwise add
166 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
167 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
169 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
170 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
171 (lower_stmt): Don't set data->calls_builtin_setjmp.
172 (lower_builtin_setjmp): Adjust comment.
173 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
174 * tree-cfg.c (found_computed_goto): Remove.
175 (factor_computed_gotos): Remove.
176 (make_goto_expr_edges): Return bool, true for computed gotos.
177 Don't call make_abnormal_goto_edges.
178 (build_gimple_cfg): Don't set found_computed_goto, don't call
179 factor_computed_gotos.
180 (computed_goto_p): No longer static.
181 (make_blocks): Don't set found_computed_goto.
182 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
183 (make_edges): If make_goto_expr_edges returns true, push bb
184 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
185 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
186 vector. Record mapping between bbs and OpenMP regions if there
187 are any, adjust make_gimple_omp_edges caller. Call
188 handle_abnormal_edges.
189 (make_abnormal_goto_edges): Remove.
190 * tree-cfg.h (make_abnormal_goto_edges): Remove.
191 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
192 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
193 * builtins.c (expand_builtin): Don't handle
194 BUILT_IN_SETJMP_DISPATCHER.
195 * internal-fn.def (ABNORMAL_DISPATCHER): New.
196 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
197 filling *region also set *region_idx to (*region)->entry->index.
200 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
201 For REGs set ORIGINAL_REGNO.
203 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
205 * doc/md.texi: Mention that a target shouldn't implement
206 vec_widen_(s|u)mul_even/odd pair if it is less efficient
209 2014-01-29 Jakub Jelinek <jakub@redhat.com>
211 PR tree-optimization/59594
212 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
213 a copy of the datarefs vector rather than the vector itself.
215 2014-01-28 Jason Merrill <jason@redhat.com>
218 * dwarf2out.c (auto_die): New static.
219 (gen_type_die_with_usage): Handle C++1y 'auto'.
220 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
223 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
226 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
227 (SPEC_X32): Likewise.
229 * config/i386/i386.c (ix86_option_override_internal): Turn off
230 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
232 (x86_file_start): Output .code16gcc for TARGET_16BIT.
233 * config/i386/i386.h (TARGET_16BIT): New macro.
234 (TARGET_16BIT_P): Likewise.
235 * config/i386/i386.opt: Add m16.
236 * doc/invoke.texi: Document -m16.
238 2014-01-28 Jakub Jelinek <jakub@redhat.com>
240 PR preprocessor/59935
241 * input.c (location_get_source_line): Bail out on when line number
242 is zero, and test the return value of
243 lookup_or_add_file_to_cache_tab.
245 2014-01-28 Richard Biener <rguenther@suse.de>
247 PR tree-optimization/58742
248 * tree-ssa-forwprop.c (associate_plusminus): Handle
249 pointer subtraction of the form (T)(P + A) - (T)P.
251 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
253 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
256 2014-01-28 Richard Biener <rguenther@suse.de>
259 2014-01-28 Richard Biener <rguenther@suse.de>
261 PR rtl-optimization/45364
262 PR rtl-optimization/59890
263 * var-tracking.c (local_get_addr_clear_given_value): Handle
264 already cleared slot.
265 (val_reset): Handle not allocated local_get_addr_cache.
266 (vt_find_locations): Use post-order on the inverted CFG.
268 2014-01-28 Richard Biener <rguenther@suse.de>
270 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
273 2014-01-28 Richard Biener <rguenther@suse.de>
275 PR rtl-optimization/45364
276 PR rtl-optimization/59890
277 * var-tracking.c (local_get_addr_clear_given_value): Handle
278 already cleared slot.
279 (val_reset): Handle not allocated local_get_addr_cache.
280 (vt_find_locations): Use post-order on the inverted CFG.
282 2014-01-28 Alan Modra <amodra@gmail.com>
284 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
285 * configure.ac <recursive call for build != host>: Define
286 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
287 and LD_FOR_BUILD too.
288 * configure: Regenerate.
290 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
292 * config/i386/i386.c (get_builtin_code_for_version): Separate
293 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
294 Broadwell from Haswell.
296 2014-01-27 Steve Ellcey <sellcey@mips.com>
298 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
299 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
300 * config/mips/mips.c (mips_option_override): Change setting
302 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
303 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
304 Change from Mask to Var.
306 2014-01-27 Jeff Law <law@redhat.com>
308 * ipa-inline.c (inline_small_functions): Fix typo.
310 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
312 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
313 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
314 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
315 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
316 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
317 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
318 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
319 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
320 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
321 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
322 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
323 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
324 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
325 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
326 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
327 (_mm512_storeu_epi64): Ditto.
328 (_mm512_cmpge_epi32_mask): Ditto.
329 (_mm512_cmpge_epu32_mask): Ditto.
330 (_mm512_cmpge_epi64_mask): Ditto.
331 (_mm512_cmpge_epu64_mask): Ditto.
332 (_mm512_cmple_epi32_mask): Ditto.
333 (_mm512_cmple_epu32_mask): Ditto.
334 (_mm512_cmple_epi64_mask): Ditto.
335 (_mm512_cmple_epu64_mask): Ditto.
336 (_mm512_cmplt_epi32_mask): Ditto.
337 (_mm512_cmplt_epu32_mask): Ditto.
338 (_mm512_cmplt_epi64_mask): Ditto.
339 (_mm512_cmplt_epu64_mask): Ditto.
340 (_mm512_cmpneq_epi32_mask): Ditto.
341 (_mm512_cmpneq_epu32_mask): Ditto.
342 (_mm512_cmpneq_epi64_mask): Ditto.
343 (_mm512_cmpneq_epu64_mask): Ditto.
344 (_mm512_expand_pd): Ditto.
345 (_mm512_expand_ps): Ditto.
346 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
347 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
348 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
349 * config/i386/i386.c (ix86_builtins): Add
350 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
351 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
352 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
353 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
354 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
355 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
356 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
357 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
358 IX86_BUILTIN_PMOVUSQW512_MEM.
359 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
360 __builtin_ia32_pmovsqd512mem_mask,
361 __builtin_ia32_pmovqd512mem_mask,
362 __builtin_ia32_pmovusqw512mem_mask,
363 __builtin_ia32_pmovsqw512mem_mask,
364 __builtin_ia32_pmovqw512mem_mask,
365 __builtin_ia32_pmovusdw512mem_mask,
366 __builtin_ia32_pmovsdw512mem_mask,
367 __builtin_ia32_pmovdw512mem_mask,
368 __builtin_ia32_pmovqb512mem_mask,
369 __builtin_ia32_pmovusqb512mem_mask,
370 __builtin_ia32_pmovsqb512mem_mask,
371 __builtin_ia32_pmovusdb512mem_mask,
372 __builtin_ia32_pmovsdb512mem_mask,
373 __builtin_ia32_pmovdb512mem_mask.
374 (bdesc_args): Add __builtin_ia32_expanddf512,
375 __builtin_ia32_expandsf512.
376 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
377 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
378 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
379 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
380 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
381 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
382 (avx512f_<code>v8div16qi2_mask_store): This.
383 (avx512f_expand<mode>): New.
385 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
387 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
389 (_mm512_mask_prefetch_i64gather_pd): Ditto.
390 (_mm512_prefetch_i32scatter_pd): Ditto.
391 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
392 (_mm512_prefetch_i64scatter_pd): Ditto.
393 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
394 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
395 (_mm512_mask_prefetch_i64gather_ps): Ditto.
396 (_mm512_prefetch_i32scatter_ps): Ditto.
397 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
398 (_mm512_prefetch_i64scatter_ps): Ditto.
399 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
400 * config/i386/i386-builtin-types.def: Define
401 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
402 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
403 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
404 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
405 IX86_BUILTIN_SCATTERPFQPD.
406 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
407 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
408 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
409 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
410 __builtin_ia32_scatterpfqps.
411 (ix86_expand_builtin): Expand new built-ins.
412 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
413 fix memory access data type.
414 (*avx512pf_gatherpf<mode>_mask): Ditto.
415 (*avx512pf_gatherpf<mode>): Ditto.
416 (avx512pf_scatterpf<mode>): Ditto.
417 (*avx512pf_scatterpf<mode>_mask): Ditto.
418 (*avx512pf_scatterpf<mode>): Ditto.
419 (GATHER_SCATTER_SF_MEM_MODE): New.
420 (avx512pf_gatherpf<mode>df): Ditto.
421 (*avx512pf_gatherpf<mode>df_mask): Ditto.
422 (*avx512pf_scatterpf<mode>df): Ditto.
424 2014-01-27 Jakub Jelinek <jakub@redhat.com>
427 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
428 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
431 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
433 * common/config/arm/arm-common.c
434 (arm_rewrite_mcpu): Handle multiple names.
436 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
438 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
440 * gimple-builder.h (create_gimple_tmp): Delete.
442 2014-01-27 Christian Bruel <christian.bruel@st.com>
444 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
447 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
449 * config/pa/pa.md (call): Generate indirect long calls to non-local
450 functions when outputing 32-bit code.
451 (call_value): Likewise except for special call to buggy powf function.
453 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
454 portable runtime and PIC indirect calls.
455 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
456 and PIC call sequences. Use ldo instead of blr to set return register
457 in PIC call sequence.
459 2014-01-25 Walter Lee <walt@tilera.com>
461 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
462 avoid clobbering a live register.
464 2014-01-25 Walter Lee <walt@tilera.com>
466 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
467 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
468 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
469 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
471 2014-01-25 Walter Lee <walt@tilera.com>
473 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
474 arguments on even registers.
475 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
477 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
478 (BIGGEST_ALIGNMENT): Ditto.
479 (BIGGEST_FIELD_ALIGNMENT): Ditto.
481 2014-01-25 Walter Lee <walt@tilera.com>
483 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
484 insns before bundling.
485 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
487 2014-01-25 Walter Lee <walt@tilera.com>
489 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
490 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
491 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
493 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
495 * config/mips/constraints.md (kl): Delete.
496 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
497 define expands, using...
498 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
499 instructions for MIPS16.
500 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
501 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
503 2014-01-25 Walter Lee <walt@tilera.com>
505 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
509 2014-01-25 Walter Lee <walt@tilera.com>
511 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
512 (TARGET_EXPAND_TO_RTL_HOOK): Define.
514 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
516 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
519 2014-01-25 Jakub Jelinek <jakub@redhat.com>
521 * print-rtl.c (in_call_function_usage): New var.
522 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
523 EXPR_LIST mode as mode and not as reg note name.
526 * cfgloopmanip.c (copy_loop_info): If
527 loop->warned_aggressive_loop_optimizations, make sure
528 the flag is set in target loop too.
530 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
532 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
534 * builtins.def: Likewise.
535 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
536 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
537 * ira.c (ira_setup_eliminable_regset): Likewise.
538 * omp-low.c (gate_expand_omp): Likewise.
539 (execute_lower_omp): Likewise.
540 (diagnose_sb_0): Likewise.
541 (gate_diagnose_omp_blocks): Likewise.
542 (simd_clone_clauses_extract): Likewise.
545 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
547 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
548 correction for little endian...
549 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
552 2014-01-24 Jeff Law <law@redhat.com>
554 PR tree-optimization/59919
555 * tree-vrp.c (find_assert_locations_1): Do not register asserts
556 for non-returning calls.
558 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
560 * common/config/aarch64/aarch64-common.c
561 (aarch64_rewrite_mcpu): Handle multiple names.
562 * config/aarch64/aarch64.h
563 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
565 2014-01-24 Dodji Seketeli <dodji@redhat.com>
567 * input.c (add_file_to_cache_tab): Handle the case where fopen
570 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
573 * config/i386/i386.md (pushsf splitter): Get stack adjustment
574 from push operand if code of push isn't PRE_DEC.
576 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
579 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
580 -mquad-memory-atomic. Update -mquad-memory documentation to say
581 it is only used for non-atomic loads/stores.
583 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
584 -mquad-memory or -mquad-memory-atomic switches.
586 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
587 -mquad-memory-atomic to ISA 2.07 support.
589 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
590 to separate support of normal quad word memory operations (ldq, stq)
591 from the atomic quad word memory operations.
593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
594 support to separate non-atomic quad word operations from atomic
595 quad word operations. Disable non-atomic quad word operations in
596 little endian mode so that we don't have to swap words after the
597 load and before the store.
598 (quad_load_store_p): Add comment about atomic quad word support.
599 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
600 options printed with -mdebug=reg.
602 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
603 -mquad-memory-atomic as the test for whether we have quad word
605 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
606 or -mp8-vector are used, allow byte/half-word atomic operations.
608 * config/rs6000/sync.md (load_lockedti): Insure that the address
609 is a proper indexed or indirect address for the lqarx instruction.
610 On little endian systems, swap the hi/lo registers after the lqarx
612 (load_lockedpti): Use indexed_or_indirect_operand predicate to
613 insure the address is valid for the lqarx instruction.
614 (store_conditionalti): Insure that the address is a proper indexed
615 or indirect address for the stqcrx. instruction. On little endian
616 systems, swap the hi/lo registers before doing the stqcrx.
618 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
619 insure the address is valid for the stqcrx. instruction.
621 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
622 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
623 type of quad memory support is available.
625 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
628 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
629 there is a danger of looping.
631 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
634 force flag_ira_loop_pressure if set via command line.
636 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
638 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
639 (ashr_simd): New builtin handling DI mode.
640 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
641 (aarch64_sshr_simddi): New match pattern.
642 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
643 (vshrd_n_s64): Likewise.
644 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
646 2014-01-23 Nick Clifton <nickc@redhat.com>
648 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
649 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
650 favour of mcu specific scripts.
651 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
654 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
655 Alex Velenko <Alex.Velenko@arm.com>
657 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
658 (vaddv_s16): Likewise.
659 (vaddv_s32): Likewise.
660 (vaddv_u8): Likewise.
661 (vaddv_u16): Likewise.
662 (vaddv_u32): Likewise.
663 (vaddvq_s8): Likewise.
664 (vaddvq_s16): Likewise.
665 (vaddvq_s32): Likewise.
666 (vaddvq_s64): Likewise.
667 (vaddvq_u8): Likewise.
668 (vaddvq_u16): Likewise.
669 (vaddvq_u32): Likewise.
670 (vaddvq_u64): Likewise.
671 (vaddv_f32): Likewise.
672 (vaddvq_f32): Likewise.
673 (vaddvq_f64): Likewise.
674 (vmaxv_f32): Likewise.
675 (vmaxv_s8): Likewise.
676 (vmaxv_s16): Likewise.
677 (vmaxv_s32): Likewise.
678 (vmaxv_u8): Likewise.
679 (vmaxv_u16): Likewise.
680 (vmaxv_u32): Likewise.
681 (vmaxvq_f32): Likewise.
682 (vmaxvq_f64): Likewise.
683 (vmaxvq_s8): Likewise.
684 (vmaxvq_s16): Likewise.
685 (vmaxvq_s32): Likewise.
686 (vmaxvq_u8): Likewise.
687 (vmaxvq_u16): Likewise.
688 (vmaxvq_u32): Likewise.
689 (vmaxnmv_f32): Likewise.
690 (vmaxnmvq_f32): Likewise.
691 (vmaxnmvq_f64): Likewise.
692 (vminv_f32): Likewise.
693 (vminv_s8): Likewise.
694 (vminv_s16): Likewise.
695 (vminv_s32): Likewise.
696 (vminv_u8): Likewise.
697 (vminv_u16): Likewise.
698 (vminv_u32): Likewise.
699 (vminvq_f32): Likewise.
700 (vminvq_f64): Likewise.
701 (vminvq_s8): Likewise.
702 (vminvq_s16): Likewise.
703 (vminvq_s32): Likewise.
704 (vminvq_u8): Likewise.
705 (vminvq_u16): Likewise.
706 (vminvq_u32): Likewise.
707 (vminnmv_f32): Likewise.
708 (vminnmvq_f32): Likewise.
709 (vminnmvq_f64): Likewise.
711 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
713 * config/aarch64/aarch64-simd.md
714 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
715 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
716 (*aarch64_mul3_elt<mode>): Likewise.
717 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
718 (*aarch64_mul3_elt_to_64v2df): Likewise.
719 (*aarch64_mla_elt<mode>): Likewise.
720 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
721 (*aarch64_mls_elt<mode>): Likewise.
722 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
723 (*aarch64_fma4_elt<mode>): Likewise.
724 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
725 (*aarch64_fma4_elt_to_64v2df): Likewise.
726 (*aarch64_fnma4_elt<mode>): Likewise.
727 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
728 (*aarch64_fnma4_elt_to_64v2df): Likewise.
729 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
730 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
731 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
732 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
733 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
734 (aarch64_sqdmull_lane<mode>_internal): Likewise.
735 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
737 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
739 * config/aarch64/aarch64-simd.md
740 (aarch64_be_checked_get_lane<mode>): New define_expand.
741 * config/aarch64/aarch64-simd-builtins.def
742 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
743 New builtin definition.
744 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
745 Use new safe be builtin.
747 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
749 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
751 (aarch64_be_st1<mode>): Likewise.
752 (aarch_ld1<VALL:mode>): Define_expand modified.
753 (aarch_st1<VALL:mode>): Likewise.
754 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
755 (UNSPEC_ST1): Likewise.
757 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
759 * config/microblaze/microblaze.md: Add trap insn and attribute
761 2014-01-23 Dodji Seketeli <dodji@redhat.com>
763 PR preprocessor/58580
764 * input.h (location_get_source_line): Take an additional line_size
766 (void diagnostics_file_cache_fini): Declare new function.
767 * input.c (struct fcache): New type.
768 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
769 New static constants.
770 (diagnostic_file_cache_init, total_lines_num)
771 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
772 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
773 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
774 (get_next_line, read_next_line, goto_next_line, read_line_num):
775 New static function definitions.
776 (diagnostic_file_cache_fini): New function.
777 (location_get_source_line): Take an additional output line_len
778 parameter. Re-write using lookup_or_add_file_to_cache_tab and
780 * diagnostic.c (diagnostic_finish): Call
781 diagnostic_file_cache_fini.
782 (adjust_line): Take an additional input parameter for the length
783 of the line, rather than calculating it with strlen.
784 (diagnostic_show_locus): Adjust the use of
785 location_get_source_line and adjust_line with respect to their new
786 signature. While displaying a line now, do not stop at the first
787 null byte. Rather, display the zero byte as a space and keep
788 going until we reach the size of the line.
789 * Makefile.in: Add vec.o to OBJS-libcommon
791 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
792 Ilya Tocar <ilya.tocar@intel.com>
794 * config/i386/avx512fintrin.h (_mm512_kmov): New.
795 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
796 (__builtin_ia32_kmov16): Ditto.
797 * config/i386/i386.md (UNSPEC_KMOV): New.
800 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
802 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
803 (_mm512_storeu_si512): Ditto.
805 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
808 * rtl.h (get_referenced_operands): Declare.
809 * recog.c (get_referenced_operands): New function.
810 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
811 operands have been referenced when recording LO_SUM references.
813 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
815 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
817 2014-01-22 Jan Hubicka <jh@suse.cz>
819 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
820 Enable for generic and recent AMD targets.
822 2014-01-22 Jan Hubicka <jh@suse.cz>
824 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
825 ARG_SIZE note when adjustment was eliminated.
827 2014-01-22 Jeff Law <law@redhat.com>
829 PR tree-optimization/59597
830 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
831 in file. Accept new argument REGISTERING and use it to modify
832 dump output appropriately.
833 (register_jump_thread): Corresponding changes.
834 (mark_threaded_blocks): Reinstate code to cancel unprofitable
835 thread paths involving joiner blocks. Add code to dump cancelled
836 jump threading paths.
838 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
840 PR rtl-optimization/59477
841 * lra-constraints.c (inherit_in_ebb): Process call for living hard
842 regs. Update reloads_num and potential_reload_hard_regs for all insns.
844 2014-01-22 Tom Tromey <tromey@redhat.com>
846 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
848 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
850 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
852 PR rtl-optimization/59896
853 * lra-constraints.c (process_alt_operands): Check unused note for
854 matched operands of insn with no output reloads.
856 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
858 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
859 (mips_move_from_gpr_cost): Likewise.
861 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
863 PR rtl-optimization/59858
864 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
865 ira_class_hard_regs_num.
866 (process_alt_operands): Increase reject for dying matched operand.
868 2014-01-21 Jakub Jelinek <jakub@redhat.com>
871 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
872 smaller than size, perform several stores or loads and stores
873 at dst + count - size to store or copy all of size bytes, rather
874 than just last modesize bytes.
876 2014-01-20 DJ Delorie <dj@redhat.com>
878 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
879 that CLOBBERs are REGs before propogating their values.
881 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
884 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
885 (cgraph_inline_failed_type): New function.
886 * cgraph.h (DEFCIFCODE): Add type.
887 (cgraph_inline_failed_type_t): New enum.
888 (cgraph_inline_failed_type): New prototype.
889 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
890 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
891 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
892 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
893 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
894 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
895 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
896 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
897 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
898 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
899 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
900 OPTIMIZATION_MISMATCH.
901 * tree-inline.c (expand_call_inline): Emit errors during
902 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
904 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
907 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
908 mode attribute in insn output.
910 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
912 * output.h (output_constant): Delete.
913 * varasm.c (output_constant): Make private.
915 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
917 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
919 2014-01-20 Jakub Jelinek <jakub@redhat.com>
922 * tree.h (fold_builtin_strcat): New prototype.
923 * builtins.c (fold_builtin_strcat): No longer static. Add len
924 argument, if non-NULL, don't call c_strlen. Optimize
925 directly into __builtin_memcpy instead of __builtin_strcpy.
926 (fold_builtin_2): Adjust fold_builtin_strcat caller.
927 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
929 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
931 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
932 for SImode_address_operand operands, having only a REG argument.
934 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
936 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
937 loader name using mbig-endian.
938 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
940 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
942 * doc/invoke.texi (-march): Clarify documentation for AArch64.
946 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
948 * config/aarch64/aarch64-protos.h
949 (aarch64_cannot_change_mode_class_ptr): Declare.
950 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
951 aarch64_cannot_change_mode_class_ptr): New.
952 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
953 backend hook aarch64_cannot_change_mode_class.
955 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
957 * common/config/aarch64/aarch64-common.c
958 (aarch64_handle_option): Don't handle any option order logic here.
959 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
960 selected_cpu, warn on architecture version mismatch.
961 (aarch64_override_options): Fix parsing order for option strings.
963 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
964 Iain Sandoe <iain@codesourcery.com>
967 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
968 warning. Amend comment to reflect current functionality.
970 2014-01-20 Richard Biener <rguenther@suse.de>
973 * builtins.c (fold_builtin_strcat): Remove case better handled
974 by tree-ssa-strlen.c.
976 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
978 * config/aarch64/aarch64.opt
979 (mcpu, march, mtune): Make case-insensitive.
981 2014-01-20 Jakub Jelinek <jakub@redhat.com>
984 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
985 if operands[1] is a REG or ZERO_EXTEND of a REG.
987 2014-01-19 Jan Hubicka <jh@suse.cz>
989 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
991 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
993 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
994 long non-pic millicode calls.
996 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
998 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1000 2014-01-19 Kito Cheng <kito@0xlab.org>
1002 * builtins.c (expand_movstr): Check movstr expand done or fail.
1004 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1005 H.J. Lu <hongjiu.lu@intel.com>
1008 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1009 to DImode for zero-extended addresses.
1011 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1013 PR rtl-optimization/57763
1014 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1015 on the new indirect jump_insn and increment LABEL_NUSES (label).
1017 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1021 * config.gcc (x86_archs): New variable.
1022 (x86_64_archs): Likewise.
1023 (x86_cpus): Likewise.
1024 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1025 --with-arch/--with-cpu= options.
1026 Support --with-arch=/--with-cpu={nehalem,westmere,
1027 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1029 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1031 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1032 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1034 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1036 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1038 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1041 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1042 clear cpp_get_options (parse_in)->warn_unused_macros for
1043 ix86_target_macros_internal with cpp_define.
1045 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1047 * jump.c (delete_related_insns): Keep (use (insn))s.
1048 * reorg.c (redundant_insn): Check for barriers too.
1050 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1052 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1054 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1056 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1057 call to $$dyncall when TARGET_LONG_CALLS is true.
1059 2014-01-17 Jeff Law <law@redhat.com>
1061 * ree.c (combine_set_extension): Temporarily disable test for
1062 changing number of hard registers.
1064 2014-01-17 Jan Hubicka <jh@suse.cz>
1067 * ipa-inline-analysis.c (inline_free_summary):
1068 Do not free summary of aliases.
1070 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1073 * gimplify.c (gimplify_expr): Use create_tmp_var
1074 instead of create_tmp_var_raw. If cond doesn't have
1075 integral type, don't add the IFN_ANNOTATE builtin at all.
1077 2014-01-17 Martin Jambor <mjambor@suse.cz>
1080 * ipa-cp.c (prev_edge_clone): New variable.
1081 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1082 Also resize prev_edge_clone vector.
1083 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1084 (ipcp_edge_removal_hook): New function.
1085 (ipcp_driver): Register ipcp_edge_removal_hook.
1087 2014-01-17 Andrew Pinski <apinski@cavium.com>
1088 Steve Ellcey <sellcey@mips.com>
1091 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1094 2014-01-17 Jeff Law <law@redhat.com>
1097 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1098 so that pass_ccp runs first.
1100 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1102 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1103 (ix86_adjust_cost): Use !TARGET_XXX.
1104 (do_reorder_for_imul): Likewise.
1105 (swap_top_of_ready_list): Likewise.
1106 (ix86_sched_reorder): Likewise.
1108 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1110 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1111 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1112 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1113 (intel_memset): New. Duplicate slm_memset.
1114 (intel_cost): New. Duplicate slm_cost.
1115 (m_INTEL): New macro.
1116 (processor_target_table): Add "intel".
1117 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1118 with PROCESSOR_INTEL for "intel".
1119 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1120 PROCESSOR_SILVERMONT.
1121 (ix86_issue_rate): Likewise.
1122 (ix86_adjust_cost): Likewise.
1123 (ia32_multipass_dfa_lookahead): Likewise.
1124 (swap_top_of_ready_list): Likewise.
1125 (ix86_sched_reorder): Likewise.
1126 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1127 instead of TARGET_OPT_AGU.
1128 * config/i386/i386.h (TARGET_INTEL): New.
1129 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1130 (processor_type): Add PROCESSOR_INTEL.
1131 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1132 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1134 2014-01-17 Marek Polacek <polacek@redhat.com>
1137 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1140 2014-01-17 Richard Biener <rguenther@suse.de>
1142 PR tree-optimization/46590
1143 * opts.c (default_options_table): Add entries for
1144 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1145 all enabled at -O1 but not for -Og.
1146 * common.opt (fbranch-count-reg): Remove Init(1).
1147 (fmove-loop-invariants): Likewise.
1148 (ftree-pta): Likewise.
1150 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1152 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1153 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1154 decls to at least the GCC 4.8 used alignments.
1157 * tree-nested.c (convert_nonlocal_reference_stmt,
1158 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1159 of GIMPLE_BIND stmts, adjust associated decls.
1161 2014-01-17 Richard Biener <rguenther@suse.de>
1163 PR tree-optimization/46590
1164 * vec.h (vec<>::bseach): New member function implementing
1165 binary search according to C89 bsearch.
1166 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1167 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1168 bitmap pointer again. Make accesses_in_loop a flat array.
1169 (mem_ref_obstack): New global.
1170 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1171 (mark_ref_stored): Likewise.
1172 (ref_indep_loop_p_2): Likewise.
1173 (set_ref_stored_in_loop): New helper function.
1174 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1175 (memref_free): Adjust.
1176 (record_mem_ref_loc): Simplify.
1177 (gather_mem_refs_stmt): Adjust.
1178 (sort_locs_in_loop_postorder_cmp): New function.
1179 (analyze_memory_references): Sort accesses_in_loop after
1180 loop postorder number.
1181 (find_ref_loc_in_loop_cmp): New function.
1182 (for_all_locs_in_loop): Find relevant cluster of locs in
1183 accesses_in_loop and iterate without recursion.
1184 (execute_sm): Avoid uninit warning.
1185 (struct ref_always_accessed): Simplify.
1186 (ref_always_accessed::operator ()): Likewise.
1187 (ref_always_accessed_p): Likewise.
1188 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1189 loop postorder numbers here.
1190 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1193 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1196 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1197 on decls for which assemble_alias has been called.
1199 2014-01-17 Nick Clifton <nickc@redhat.com>
1201 * config/msp430/msp430.opt: (mcpu): New option.
1202 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1203 (msp430_option_override): Parse target_cpu. If the MCU name
1204 matches a generic string, clear target_mcu.
1205 (msp430_attr): Allow numeric interrupt values up to 63.
1206 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1207 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1209 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1211 * config/msp430/msp430.md (popm): Use %J rather than %I.
1212 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1213 (addhi_cy_i): Use immediate_operand for operand 2.
1214 * doc/invoke.texi: Document -mcpu option.
1216 2014-01-17 Richard Biener <rguenther@suse.de>
1218 PR rtl-optimization/38518
1219 * df.h (df_analyze_loop): Declare.
1220 * df-core.c: Include cfgloop.h.
1221 (df_analyze_1): Split out main part of df_analyze.
1222 (df_analyze): Adjust.
1223 (loop_inverted_post_order_compute): New function.
1224 (loop_post_order_compute): Likewise.
1225 (df_analyze_loop): New function avoiding whole-function
1227 * loop-invariant.c (find_defs): Use df_analyze_loop.
1228 (find_invariants): Adjust.
1229 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1231 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1233 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1234 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1236 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1238 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1239 traversal when removing references.
1241 2014-01-16 Jan Hubicka <jh@suse.cz>
1244 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1246 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1249 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1250 pushing a reload for an autoinc when we had previously reloaded an
1251 inner part of the address.
1253 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1255 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1257 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1258 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1259 when not giving up or versioning for alias only because of
1261 (vect_analyze_data_ref_dependences): Set to true.
1262 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1264 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1265 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1269 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1272 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1273 operand 0 predicate for gathers, use a new pseudo as subtarget.
1275 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1278 * lra-constraints.c (process_alt_operands): Add printing debug info.
1279 Check absence of input/output reloads for matched operands too.
1281 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1283 PR rtl-optimization/59835
1284 * ira.c (ira_init_register_move_cost): Increase cost for
1287 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1289 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1291 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1294 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1295 non-register objects. Use gen_(high/low)part more consistently.
1298 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1301 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1302 endian support, remove tests for WORDS_BIG_ENDIAN.
1303 (p8_mfvsrd_3_<mode>): Likewise.
1304 (reload_gpr_from_vsx<mode>): Likewise.
1305 (reload_gpr_from_vsxsf): Likewise.
1306 (p8_mfvsrd_4_disf): Likewise.
1308 2014-01-16 Richard Biener <rguenther@suse.de>
1310 PR rtl-optimization/46590
1311 * lcm.c (compute_antinout_edge): Use postorder iteration.
1312 (compute_laterin): Use inverted postorder iteration.
1314 2014-01-16 Nick Clifton <nickc@redhat.com>
1317 * varasm.c (output_constant): Return the number of bytes actually
1319 (output_constructor_array_range): Update the field size with the
1320 number of bytes emitted by output_constant.
1321 (output_constructor_regular_field): Likewise. Also do not
1322 complain if the total number of bytes emitted is now greater
1323 than the expected fieldpos.
1324 * output.h (output_constant): Update prototype and descriptive comment.
1326 2014-01-16 Marek Polacek <polacek@redhat.com>
1329 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1330 it is error_mark_node.
1332 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1334 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1335 VALID_AVX256_REG_OR_OI_MODE.
1337 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1339 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1340 current procedure should be profiled.
1342 2014-01-15 Andrew Pinski <apinski@cavium.com>
1344 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1345 of moving from/to the STACK_REG register class.
1347 2014-01-15 Richard Henderson <rth@redhat.com>
1350 * reginfo.c (global_regs_decl): Globalize.
1351 * rtl.h (global_regs_decl): Declare.
1352 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1353 reserved via global_regs.
1355 2014-01-15 Teresa Johnson <tejohnson@google.com>
1357 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1359 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1361 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1362 and vmulosh rather than call gen_vec_widen_smult_*.
1363 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1364 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1365 (vec_widen_smult_even_v16qi): Likewise.
1366 (vec_widen_umult_even_v8hi): Likewise.
1367 (vec_widen_smult_even_v8hi): Likewise.
1368 (vec_widen_umult_odd_v16qi): Likewise.
1369 (vec_widen_smult_odd_v16qi): Likewise.
1370 (vec_widen_umult_odd_v8hi): Likewise.
1371 (vec_widen_smult_odd_v8hi): Likewise.
1372 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1373 vmuloub rather than call gen_vec_widen_umult_*.
1374 (vec_widen_umult_lo_v16qi): Likewise.
1375 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1376 vmulosb rather than call gen_vec_widen_smult_*.
1377 (vec_widen_smult_lo_v16qi): Likewise.
1378 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1379 rather than call gen_vec_widen_umult_*.
1380 (vec_widen_umult_lo_v8hi): Likewise.
1381 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1382 rather than call gen_vec_widen_smult_*.
1383 (vec_widen_smult_lo_v8hi): Likewise.
1385 2014-01-15 Jeff Law <law@redhat.com>
1387 PR tree-optimization/59747
1388 * ree.c (find_and_remove_re): Properly handle case where a second
1389 eliminated extension requires widening a copy created for elimination
1390 of a prior extension.
1391 (combine_set_extension): Ensure that the number of hard regs needed
1392 for a destination register does not change when we widen it.
1394 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1396 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1397 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1398 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1399 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1400 (avr-*-rtems*): Likewise.
1401 (bfin*-rtems*): Likewise.
1402 (moxie-*-rtems*): Likewise.
1403 (h8300-*-rtems*): Likewise.
1404 (i[34567]86-*-rtems*): Likewise.
1405 (lm32-*-rtems*): Likewise.
1406 (m32r-*-rtems*): Likewise.
1407 (m68k-*-rtems*): Likewise.
1408 (microblaze*-*-rtems*): Likewise.
1409 (mips*-*-rtems*): Likewise.
1410 (powerpc-*-rtems*): Likewise.
1411 (sh-*-rtems*): Likewise.
1412 (sparc-*-rtems*): Likewise.
1413 (sparc64-*-rtems*): Likewise.
1414 (v850-*-rtems*): Likewise.
1415 (m32c-*-rtems*): Likewise.
1417 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1419 PR rtl-optimization/59511
1420 * ira.c (ira_init_register_move_cost): Use memory costs for some
1421 cases of register move cost calculations.
1422 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1423 instead of BB frequency.
1424 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1425 * lra-assigns.c (find_hard_regno_for): Ditto.
1427 2014-01-15 Richard Biener <rguenther@suse.de>
1429 PR tree-optimization/59822
1430 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1431 (vectorizable_load): Use it to hoist defs of uses of invariant
1432 loads out of the loop.
1434 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1435 Kugan Vivekanandarajah <kuganv@linaro.org>
1438 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1441 2014-01-15 Richard Biener <rguenther@suse.de>
1443 PR rtl-optimization/59802
1444 * lcm.c (compute_available): Use inverted postorder to seed
1445 the initial worklist.
1447 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1450 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1451 ADDR_REGS for invalid symrefs in non-PIC code.
1453 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1456 * builtins.c (determine_block_size): Initialize *probable_max_size
1457 even if len_rtx is CONST_INT.
1459 2014-01-14 Andrew Pinski <apinski@cavium.com>
1461 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1462 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1463 (cortexa53_tunings): Likewise.
1464 (aarch64_sched_issue_rate): New function.
1465 (TARGET_SCHED_ISSUE_RATE): Define.
1467 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1469 * ira-costs.c (find_costs_and_classes): Add missed
1470 ira_init_register_move_cost_if_necessary.
1472 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1475 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1477 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1480 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1481 to indicate if type is used for function return value. Warn ABI
1482 change if the vector mode isn't available for function return value.
1483 (ix86_function_arg_advance): Pass false to type_natural_mode.
1484 (ix86_function_arg): Likewise.
1485 (ix86_gimplify_va_arg): Likewise.
1486 (function_arg_32): Don't warn ABI change.
1487 (ix86_function_value): Pass true to type_natural_mode.
1488 (ix86_return_in_memory): Likewise.
1489 (ix86_struct_value_rtx): Removed.
1490 (TARGET_STRUCT_VALUE_RTX): Likewise.
1492 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1494 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1495 converting a conditional jump into a conditional return.
1497 2014-01-14 Richard Biener <rguenther@suse.de>
1499 PR tree-optimization/58921
1500 PR tree-optimization/59006
1501 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1502 hoisting invariant stmts.
1503 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1504 invariant loads on the preheader edge if possible.
1506 2014-01-14 Joey Ye <joey.ye@arm.com>
1508 * doc/plugin.texi (Building GCC plugins): Update to C++.
1510 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1512 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1513 (_mm_rcp28_round_ss): Ditto.
1514 (_mm_rsqrt28_round_sd): Ditto.
1515 (_mm_rsqrt28_round_ss): Ditto.
1516 (_mm_rcp28_sd): Ditto.
1517 (_mm_rcp28_ss): Ditto.
1518 (_mm_rsqrt28_sd): Ditto.
1519 (_mm_rsqrt28_ss): Ditto.
1520 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1521 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1522 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1523 (IX86_BUILTIN_RCP28SD): Ditto.
1524 (IX86_BUILTIN_RCP28SS): Ditto.
1525 (IX86_BUILTIN_RSQRT28SD): Ditto.
1526 (IX86_BUILTIN_RSQRT28SS): Ditto.
1527 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1528 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1529 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1530 (ix86_expand_special_args_builtin): Expand new FTYPE.
1531 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1532 (srcp14<mode>): Make insn unary.
1533 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1534 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1535 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1536 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1537 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1538 Fix rounding: make it SAE only.
1539 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1541 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1543 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1544 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1545 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1546 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1547 (round_saeonly_mask_scalar_operand4): Ditto.
1548 (round_saeonly_mask_scalar_op3): Ditto.
1549 (round_saeonly_mask_scalar_op4): Ditto.
1551 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1553 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1554 Implement -maltivec=be for vec_insert and vec_extract.
1556 2014-01-10 DJ Delorie <dj@redhat.com>
1558 * config/msp430/msp430.md (call_internal): Don't allow memory
1559 references with SP as the base register.
1560 (call_value_internal): Likewise.
1561 * config/msp430/constraints.md (Yc): New. For memory references
1562 that don't use SP as a base register.
1564 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1565 "an integer without a # prefix"
1566 * config/msp430/msp430.md (epilogue_helper): Use it.
1568 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1571 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1572 AVX512F gather builtins.
1573 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1574 on gather decls with INTEGER_TYPE masktype.
1575 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1576 directly into the builtin rather than hoisting it before loop.
1578 PR tree-optimization/59387
1579 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1580 (scev_const_prop): If folded_casts and type has undefined overflow,
1581 use force_gimple_operand instead of force_gimple_operand_gsi and
1582 for each added stmt if it is assign with
1583 arith_code_with_undefined_signed_overflow, call
1584 rewrite_to_defined_overflow.
1585 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1586 gimple-fold.h instead.
1587 (arith_code_with_undefined_signed_overflow,
1588 rewrite_to_defined_overflow): Moved to ...
1589 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1590 rewrite_to_defined_overflow): ... here. No longer static.
1591 Include gimplify-me.h.
1592 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1593 rewrite_to_defined_overflow): New prototypes.
1595 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1597 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1599 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1601 * builtins.c (get_object_alignment_2): Minor tweak.
1602 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1604 2014-01-13 Christian Bruel <christian.bruel@st.com>
1606 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1607 optimized non constant lengths.
1609 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1612 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1613 load as __atomic_load_N if possible.
1615 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1617 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1619 (rs6000_expand_builtin): Adjust call.
1621 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1624 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1625 * config/rs6000/rs6000.c: Include target-globals.h.
1626 (rs6000_set_current_function): Instead of doing target_reinit
1627 unconditionally, use save_target_globals_default_opts and
1628 restore_target_globals.
1630 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1632 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1633 (rs6000_expand_builtin): Handle mffs and mtfsf.
1634 (rs6000_init_builtins): Define mffs and mtfsf.
1635 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1636 (rs6000_mffs): New pattern.
1637 (rs6000_mtfsf): New pattern.
1639 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1641 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1642 Start narrowing with START. Apply candidate-use pair
1643 and check overall cost in narrowing.
1644 (iv_ca_prune): Pass new argument.
1646 2014-01-10 Jeff Law <law@redhat.com>
1649 * ree.c (combine_reaching_defs): Ensure the defining statement
1650 occurs before the extension when optimizing extensions with
1651 different source and destination hard registers.
1653 2014-01-10 Jan Hubicka <jh@suse.cz>
1656 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1657 vtables into the type inheritance graph.
1659 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1661 PR rtl-optimization/59754
1662 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1663 modes in the REGNO != REGNO case.
1665 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1667 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1669 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1671 PR tree-optimization/59745
1672 * tree-predcom.c (tree_predictive_commoning_loop): Call
1673 free_affine_expand_cache if giving up because components is NULL.
1675 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1676 GC in payload of target_globals struct instead of allocating them on
1677 the heap and the larger structs separately using GC.
1678 * target-globals.h (struct target_globals): Make regs, hard_regs,
1679 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1680 of GTY((skip)) and change type to void *.
1681 (reset_target_globals): Cast loads from those fields to corresponding
1684 2014-01-10 Steve Ellcey <sellcey@mips.com>
1687 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1688 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1689 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1691 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1694 * aarch64-modes.def (CC_Zmode): New flags mode.
1695 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1696 represents an equality.
1697 (aarch64_get_condition_code): Handle CC_Zmode.
1698 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1700 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1702 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1703 extraction in good case.
1705 2014-01-10 Richard Biener <rguenther@suse.de>
1707 PR tree-optimization/59374
1708 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1709 checking after SLP discovery. Mark stmts not participating
1710 in any SLP instance properly.
1712 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1714 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1715 when handling a SET rtx.
1717 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1719 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1720 (cortex-a57): Likewise.
1721 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1723 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1725 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1726 non-iwmmxt builtins.
1728 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1732 * ipa-devirt.c record_target_from_binfo): Take as argument
1733 stack of binfos and lookup matching one for virtual inheritance.
1734 (possible_polymorphic_call_targets_1): Update.
1736 2014-01-10 Huacai Chen <chenhc@lemote.com>
1738 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1739 kernel strings for Loongson-2E/2F/3A.
1741 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1744 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1745 is_gimple_call before calling gimple_call_internal_p.
1747 2014-01-09 Steve Ellcey <sellcey@mips.com>
1749 * Makefile.in (TREE_FLOW_H): Remove.
1750 (TREE_SSA_H): Add file names from tree-flow.h.
1751 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1752 * tree.h: Remove tree-flow.h reference.
1753 * hash-table.h: Remove tree-flow.h reference.
1754 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1755 reference with tree-ssa-loop.h.
1757 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1759 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1760 default element-order behavior for -maltivec.
1761 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1762 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1763 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1764 when targeting big endian, at least for now.
1765 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1767 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1770 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1771 var satisfies use_register_for_decl, just take into account type
1772 alignment, rather than decl alignment.
1774 PR tree-optimization/59622
1775 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1776 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1777 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1778 Don't devirtualize for inplace at all. For targets.length () == 1,
1779 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1781 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1783 * config/i386/i386.md (cpu): Remove the unused btver1.
1785 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1787 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1789 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1792 * tree-core.h (struct target_globals): New forward declaration.
1793 (struct tree_target_option): Add globals field.
1794 * tree.h (TREE_TARGET_GLOBALS): Define.
1795 (prepare_target_option_nodes_for_pch): New prototype.
1796 * target-globals.h (struct target_globals): Define even if
1798 * tree.c (prepare_target_option_node_for_pch,
1799 prepare_target_option_nodes_for_pch): New functions.
1800 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1801 * config/i386/i386.c: Include target-globals.h.
1802 (ix86_set_current_function): Instead of doing target_reinit
1803 unconditionally, use save_target_globals_default_opts and
1804 restore_target_globals.
1806 2014-01-09 Richard Biener <rguenther@suse.de>
1808 PR tree-optimization/59715
1809 * tree-cfg.h (split_critical_edges): Declare.
1810 * tree-cfg.c (split_critical_edges): Export.
1811 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1813 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1815 * cfgexpand.c (expand_stack_vars): Optionally disable
1816 asan stack protection.
1817 (expand_used_vars): Likewise.
1818 (partition_stack_vars): Likewise.
1819 * asan.c (asan_emit_stack_protection): Optionally disable
1820 after return stack usage.
1821 (instrument_derefs): Optionally disable memory access instrumentation.
1822 (instrument_builtin_call): Likewise.
1823 (instrument_strlen_call): Likewise.
1824 (asan_protect_global): Optionally disable global variables protection.
1825 * doc/invoke.texi: Added doc for new options.
1826 * params.def: Added new options.
1827 * params.h: Likewise.
1829 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1831 PR rtl-optimization/59724
1832 * ifcvt.c (cond_exec_process_if_block): Don't call
1833 flow_find_head_matching_sequence with 0 longest_match.
1834 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1835 non-active insns if !stop_after.
1836 (try_head_merge_bb): Revert 2014-01-07 changes.
1838 2014-01-08 Jeff Law <law@redhat.com>
1840 * ree.c (get_sub_rtx): New function, extracted from...
1841 (merge_def_and_ext): Here.
1842 (combine_reaching_defs): Use get_sub_rtx.
1844 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1846 * cgraph.h (varpool_variable_node): Do not choke on null node.
1848 2014-01-08 Catherine Moore <clm@codesourcery.com>
1850 * config/mips/mips.md (simple_return): Attempt to use JRC
1852 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1854 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1856 PR rtl-optimization/59137
1857 * reorg.c (steal_delay_list_from_target): Call update_block for
1859 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1861 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1863 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1864 two duplicate entries.
1866 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1869 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1871 * config/mips/mips.c (mips_truncated_op_cost): New function.
1872 (mips_rtx_costs): Adjust test for BADDU.
1873 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1875 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1877 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1878 (*baddu_si): ...this new pattern.
1880 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1883 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1885 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1888 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1890 (expand_expr, expand_normal): Adjust.
1891 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1892 inner_reference_p. Use inner_reference_p to expand inner references.
1893 (store_expr): Adjust.
1894 * cfgexpand.c (expand_call_stmt): Adjust.
1896 2014-01-08 Rong Xu <xur@google.com>
1898 * gcov-io.c (gcov_var): Move from gcov-io.h.
1899 (gcov_position): Ditto.
1900 (gcov_is_error): Ditto.
1901 (gcov_rewrite): Ditto.
1902 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1903 only part to libgcc/libgcov.h.
1905 2014-01-08 Marek Polacek <polacek@redhat.com>
1908 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1910 2014-01-08 Marek Polacek <polacek@redhat.com>
1913 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1915 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1917 PR rtl-optimization/59649
1918 * stor-layout.c (get_mode_bounds): For BImode return
1919 0 and STORE_FLAG_VALUE.
1921 2014-01-08 Richard Biener <rguenther@suse.de>
1924 * gimple.h (is_gimple_builtin_call): Remove.
1925 (gimple_builtin_call_types_compatible_p): New.
1926 (gimple_call_builtin_p): New overload.
1927 * gimple.c (is_gimple_builtin_call): Remove.
1928 (validate_call): Rename to ...
1929 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1931 (validate_type): New static function.
1932 (gimple_call_builtin_p): New overload and adjust.
1933 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1934 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1935 (gimple_fold_stmt_to_constant_1): Likewise.
1936 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1938 2014-01-08 Richard Biener <rguenther@suse.de>
1941 * gimplify.c (gimplify_expr): Gimplify register-register type
1942 VIEW_CONVERT_EXPRs to separate stmts.
1944 2014-01-07 Jeff Law <law@redhat.com>
1947 * ree.c (combine_set_extension): Handle case where source
1948 and destination registers in an extension insn are different.
1949 (combine_reaching_defs): Allow source and destination registers
1950 in extension to be different under limited circumstances.
1951 (add_removable_extension): Remove restriction that the
1952 source and destination registers in the extension are the same.
1953 (find_and_remove_re): Emit a copy from the extension's
1954 destination to its source after the defining insn if
1955 the source and destination registers are different.
1958 * ifcvt.c (merge_if_block): If we are merging a block with more than
1959 one successor with a block with no successors, remove any BARRIER
1960 after the second block.
1962 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1964 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1966 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1969 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1970 for 14-bit register offsets when INT14_OK_STRICT is false.
1972 2014-01-07 Roland Stigge <stigge@antcom.de>
1973 Michael Meissner <meissner@linux.vnet.ibm.com>
1976 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1977 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1979 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1981 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1984 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1986 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1987 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1988 rtx is const0_rtx or not.
1990 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1993 * target-globals.c (save_target_globals): Remove this_fn_optab
1995 * toplev.c: Include optabs.h.
1996 (target_reinit): Temporarily restore the global options if another
1997 set of options are in force.
1999 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2001 PR rtl-optimization/58668
2002 * cfgcleanup.c (flow_find_cross_jump): Don't count
2003 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2004 to determine what is counted.
2005 (flow_find_head_matching_sequence): Use active_insn_p to determine
2007 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2009 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2010 determine what is counted.
2012 PR tree-optimization/59643
2013 * tree-predcom.c (split_data_refs_to_components): If one dr is
2014 read and one write, determine_offset fails and the write isn't
2015 in the bad component, just put the read into the bad component.
2017 2014-01-07 Mike Stump <mikestump@comcast.net>
2018 Jakub Jelinek <jakub@redhat.com>
2021 * tree-core.h (struct tree_optimization_option): Change optabs
2022 type from unsigned char * to void *.
2023 * optabs.c (init_tree_optimization_optabs): Adjust
2024 TREE_OPTIMIZATION_OPTABS initialization.
2026 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2029 * config/i386/i386.h (struct machine_function): Add
2030 no_drap_save_restore field.
2031 * config/i386/i386.c (ix86_save_reg): Use
2032 !cfun->machine->no_drap_save_restore instead of
2033 crtl->stack_realign_needed.
2034 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2035 this function clears frame_pointer_needed. Set
2036 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2037 and DRAP reg is needed.
2039 2014-01-06 Marek Polacek <polacek@redhat.com>
2042 * doc/implement-c.texi: Mention that other integer types are
2043 permitted as bit-field types in strictly conforming mode.
2045 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2047 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2050 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2052 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2054 2014-01-06 Martin Jambor <mjambor@suse.cz>
2057 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2059 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2061 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2065 * var-tracking.c (add_stores): Preserve the value of the source even if
2066 we don't record the store.
2068 2014-01-06 Terry Guo <terry.guo@arm.com>
2070 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2072 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2075 * config/darwin.c (darwin_function_section): Adjust return values to
2076 correspond to optimisation changes made in r206070.
2078 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2080 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2081 from prefetch_block tune setting.
2082 (nocona_cost): Correct size of prefetch block to 64.
2084 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2086 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2087 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2088 used to save the static chain register in the computation of the offset
2089 from which the FP registers need to be restored.
2091 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2093 PR tree-optimization/59519
2094 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2095 ICE if get_current_def (current_new_name) is already non-NULL, as long
2096 as it is a phi result of some other phi in *new_exit_bb that has
2099 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2100 or vmovdqu* for misaligned_operand.
2101 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2102 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2103 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2104 aligned_mem for AVX512F masked aligned load and store builtins and for
2107 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2109 PR tree-optimization/59651
2110 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2111 Address range for negative step should be added by TYPE_SIZE_UNIT.
2113 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2115 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2116 overlapping registers also for registers other than the stack pointer.
2118 2014-01-03 Marek Polacek <polacek@redhat.com>
2121 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2124 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2127 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2130 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2131 (push splitter): Use <P:MODE_SIZE> instead of
2132 GET_MODE_SIZE (<P:MODE>mode).
2133 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2134 (mov -1, reg peephole2): Likewise.
2135 * config/i386/sse.md (*mov<mode>_internal,
2136 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2137 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2138 *<code><mode>3, *andnot<mode>3<mask_name>,
2139 <mask_codefor><code><mode>3<mask_name>): Likewise.
2140 * config/i386/subst.md (mask_mode512bit_condition,
2141 sd_mask_mode512bit_condition): Likewise.
2143 2014-01-02 Xinliang David Li <davidxl@google.com>
2145 PR tree-optimization/59303
2146 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2147 (dump_predicates): Better output format.
2148 (pred_equal_p): New function.
2149 (is_neq_relop_p): Ditto.
2150 (is_neq_zero_form_p): Ditto.
2151 (pred_expr_equal_p): Ditto.
2152 (pred_neg_p): Ditto.
2153 (simplify_pred): Ditto.
2154 (simplify_preds_2): Ditto.
2155 (simplify_preds_3): Ditto.
2156 (simplify_preds_4): Ditto.
2157 (simplify_preds): Ditto.
2159 (push_to_worklist): Ditto.
2160 (get_pred_info_from_cmp): Ditto.
2161 (is_degenerated_phi): Ditto.
2162 (normalize_one_pred_1): Ditto.
2163 (normalize_one_pred): Ditto.
2164 (normalize_one_pred_chain): Ditto.
2165 (normalize_preds): Ditto.
2166 (normalize_cond_1): Remove function.
2167 (normalize_cond): Ditto.
2168 (is_gcond_subset_of): Ditto.
2169 (is_subset_of_any): Ditto.
2170 (is_or_set_subset_of): Ditto.
2171 (is_and_set_subset_of): Ditto.
2172 (is_norm_cond_subset_of): Ditto.
2173 (pred_chain_length_cmp): Ditto.
2174 (convert_control_dep_chain_into_preds): Type change.
2175 (find_predicates): Ditto.
2176 (find_def_preds): Ditto.
2177 (destroy_predicates_vecs): Ditto.
2178 (find_matching_predicates_in_rest_chains): Ditto.
2179 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2180 (is_pred_expr_subset): Ditto.
2181 (is_pred_chain_subset_of): Ditto.
2182 (is_included_in): Ditto.
2183 (is_superset_of): Ditto.
2185 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2187 Update copyright years.
2189 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2191 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2192 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2193 config/arc/arc.md, config/arc/arc.opt,
2194 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2195 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2196 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2197 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2198 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2199 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2200 vtable-verify.c, vtable-verify.h: Use the standard form for the
2203 2014-01-02 Tobias Burnus <burnus@net-b.de>
2205 * gcc.c (process_command): Update copyright notice dates.
2206 * gcov-dump.c: Ditto.
2208 * doc/cpp.texi: Bump @copying's copyright year.
2209 * doc/cppinternals.texi: Ditto.
2210 * doc/gcc.texi: Ditto.
2211 * doc/gccint.texi: Ditto.
2212 * doc/gcov.texi: Ditto.
2213 * doc/install.texi: Ditto.
2214 * doc/invoke.texi: Ditto.
2216 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2218 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2220 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2222 * config/i386/sse.md (*mov<mode>_internal): Guard
2223 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2225 PR rtl-optimization/59647
2226 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2227 new_rtx into UNSIGNED_FLOAT rtxes.
2229 Copyright (C) 2014 Free Software Foundation, Inc.
2231 Copying and distribution of this file, with or without modification,
2232 are permitted in any medium without royalty provided the copyright
2233 notice and this notice are preserved.