1 2014-01-30 Jason Merrill <jason@redhat.com>
4 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
7 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
9 2014-01-30 Richard Biener <rguenther@suse.de>
11 PR tree-optimization/59951
12 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
15 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
18 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
19 SFmode to DFmode case.
21 2014-01-29 DJ Delorie <dj@redhat.com>
23 * config/msp430/msp430.opt (-minrt): New.
24 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
26 (ENDFILE_SPEC): Likewise.
28 2014-01-29 Jan Hubicka <jh@suse.cz>
30 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
31 (estimate_function_body_sizes): Use it.
33 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
36 * dwarf2out.c (is_cxx_auto): New.
37 (is_base_type): Use it.
38 (gen_type_die_with_usage): Likewise.
40 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
42 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
43 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
44 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
45 -maltivec=be with LE targets.
46 (vsx_mergeh_<mode>): Likewise.
47 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
49 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
50 (altivec_vmrghb): Replace with define_expand and new
51 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
53 (altivec_vmrghb_direct): New define_insn.
54 (altivec_vmrghh): Replace with define_expand and new
55 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
57 (altivec_vmrghh_direct): New define_insn.
58 (altivec_vmrghw): Replace with define_expand and new
59 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
61 (altivec_vmrghw_direct): New define_insn.
62 (*altivec_vmrghsf): Adjust for endianness.
63 (altivec_vmrglb): Replace with define_expand and new
64 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
66 (altivec_vmrglb_direct): New define_insn.
67 (altivec_vmrglh): Replace with define_expand and new
68 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
70 (altivec_vmrglh_direct): New define_insn.
71 (altivec_vmrglw): Replace with define_expand and new
72 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
74 (altivec_vmrglw_direct): New define_insn.
75 (*altivec_vmrglsf): Adjust for endianness.
76 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
77 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
78 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
79 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
80 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
81 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
82 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
83 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
85 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
87 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
88 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
91 2014-01-29 Richard Biener <rguenther@suse.de>
93 PR tree-optimization/58742
94 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
95 associate_pointerplus_align.
96 (associate_pointerplus_diff): New function.
97 (associate_pointerplus): Likewise. Call associate_pointerplus_align
98 and associate_pointerplus_diff.
100 2014-01-29 Richard Biener <rguenther@suse.de>
102 * lto-streamer.h (LTO_major_version): Bump to 3.
103 (LTO_minor_version): Reset to 0.
105 2014-01-29 Renlin Li <Renlin.Li@arm.com>
107 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
108 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
109 (arm_file_start): Generate correct asm header for armv7ve.
110 * config/arm/bpabi.h: Add multilib support for armv7ve.
111 * config/arm/driver-arm.c: Change the architectures of cortex-a7
112 and cortex-a15 to armv7ve.
113 * config/arm/t-aprofile: Add multilib support for armv7ve.
114 * doc/invoke.texi: Document -march=armv7ve.
116 2014-01-29 Richard Biener <rguenther@suse.de>
118 PR tree-optimization/58742
119 * tree-ssa-forwprop.c (associate_plusminus): Return true
120 if we changed sth, defer EH cleanup to ...
121 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
122 (simplify_mult): New function.
124 2014-01-29 Jakub Jelinek <jakub@redhat.com>
127 PR tree-optimization/59920
128 * tree.c (build_common_builtin_nodes): Remove
129 __builtin_setjmp_dispatcher initialization.
130 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
131 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
132 instead of gsi_after_labels + manually skipping debug stmts.
133 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
134 ignore bbs with IFN_ABNORMAL_DISPATCHER.
135 * tree-inline.c (copy_edges_for_bb): Remove
136 can_make_abnormal_goto argument, instead add abnormal_goto_dest
137 argument. Ignore computed_goto_p stmts. Don't call
138 make_abnormal_goto_edges. If a call might need abnormal edges
139 for non-local gotos, see if it already has an edge to
140 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
141 with true argument, don't do anything then, otherwise add
142 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
143 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
145 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
146 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
147 (lower_stmt): Don't set data->calls_builtin_setjmp.
148 (lower_builtin_setjmp): Adjust comment.
149 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
150 * tree-cfg.c (found_computed_goto): Remove.
151 (factor_computed_gotos): Remove.
152 (make_goto_expr_edges): Return bool, true for computed gotos.
153 Don't call make_abnormal_goto_edges.
154 (build_gimple_cfg): Don't set found_computed_goto, don't call
155 factor_computed_gotos.
156 (computed_goto_p): No longer static.
157 (make_blocks): Don't set found_computed_goto.
158 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
159 (make_edges): If make_goto_expr_edges returns true, push bb
160 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
161 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
162 vector. Record mapping between bbs and OpenMP regions if there
163 are any, adjust make_gimple_omp_edges caller. Call
164 handle_abnormal_edges.
165 (make_abnormal_goto_edges): Remove.
166 * tree-cfg.h (make_abnormal_goto_edges): Remove.
167 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
168 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
169 * builtins.c (expand_builtin): Don't handle
170 BUILT_IN_SETJMP_DISPATCHER.
171 * internal-fn.def (ABNORMAL_DISPATCHER): New.
172 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
173 filling *region also set *region_idx to (*region)->entry->index.
176 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
177 For REGs set ORIGINAL_REGNO.
179 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
181 * doc/md.texi: Mention that a target shouldn't implement
182 vec_widen_(s|u)mul_even/odd pair if it is less efficient
185 2014-01-29 Jakub Jelinek <jakub@redhat.com>
187 PR tree-optimization/59594
188 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
189 a copy of the datarefs vector rather than the vector itself.
191 2014-01-28 Jason Merrill <jason@redhat.com>
194 * dwarf2out.c (auto_die): New static.
195 (gen_type_die_with_usage): Handle C++1y 'auto'.
196 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
199 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
202 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
203 (SPEC_X32): Likewise.
205 * config/i386/i386.c (ix86_option_override_internal): Turn off
206 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
208 (x86_file_start): Output .code16gcc for TARGET_16BIT.
209 * config/i386/i386.h (TARGET_16BIT): New macro.
210 (TARGET_16BIT_P): Likewise.
211 * config/i386/i386.opt: Add m16.
212 * doc/invoke.texi: Document -m16.
214 2014-01-28 Jakub Jelinek <jakub@redhat.com>
216 PR preprocessor/59935
217 * input.c (location_get_source_line): Bail out on when line number
218 is zero, and test the return value of
219 lookup_or_add_file_to_cache_tab.
221 2014-01-28 Richard Biener <rguenther@suse.de>
223 PR tree-optimization/58742
224 * tree-ssa-forwprop.c (associate_plusminus): Handle
225 pointer subtraction of the form (T)(P + A) - (T)P.
227 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
229 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
232 2014-01-28 Richard Biener <rguenther@suse.de>
235 2014-01-28 Richard Biener <rguenther@suse.de>
237 PR rtl-optimization/45364
238 PR rtl-optimization/59890
239 * var-tracking.c (local_get_addr_clear_given_value): Handle
240 already cleared slot.
241 (val_reset): Handle not allocated local_get_addr_cache.
242 (vt_find_locations): Use post-order on the inverted CFG.
244 2014-01-28 Richard Biener <rguenther@suse.de>
246 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
249 2014-01-28 Richard Biener <rguenther@suse.de>
251 PR rtl-optimization/45364
252 PR rtl-optimization/59890
253 * var-tracking.c (local_get_addr_clear_given_value): Handle
254 already cleared slot.
255 (val_reset): Handle not allocated local_get_addr_cache.
256 (vt_find_locations): Use post-order on the inverted CFG.
258 2014-01-28 Alan Modra <amodra@gmail.com>
260 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
261 * configure.ac <recursive call for build != host>: Define
262 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
263 and LD_FOR_BUILD too.
264 * configure: Regenerate.
266 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
268 * config/i386/i386.c (get_builtin_code_for_version): Separate
269 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
270 Broadwell from Haswell.
272 2014-01-27 Steve Ellcey <sellcey@mips.com>
274 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
275 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
276 * config/mips/mips.c (mips_option_override): Change setting
278 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
279 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
280 Change from Mask to Var.
282 2014-01-27 Jeff Law <law@redhat.com>
284 * ipa-inline.c (inline_small_functions): Fix typo.
286 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
288 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
289 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
290 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
291 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
292 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
293 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
294 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
295 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
296 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
297 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
298 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
299 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
300 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
301 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
302 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
303 (_mm512_storeu_epi64): Ditto.
304 (_mm512_cmpge_epi32_mask): Ditto.
305 (_mm512_cmpge_epu32_mask): Ditto.
306 (_mm512_cmpge_epi64_mask): Ditto.
307 (_mm512_cmpge_epu64_mask): Ditto.
308 (_mm512_cmple_epi32_mask): Ditto.
309 (_mm512_cmple_epu32_mask): Ditto.
310 (_mm512_cmple_epi64_mask): Ditto.
311 (_mm512_cmple_epu64_mask): Ditto.
312 (_mm512_cmplt_epi32_mask): Ditto.
313 (_mm512_cmplt_epu32_mask): Ditto.
314 (_mm512_cmplt_epi64_mask): Ditto.
315 (_mm512_cmplt_epu64_mask): Ditto.
316 (_mm512_cmpneq_epi32_mask): Ditto.
317 (_mm512_cmpneq_epu32_mask): Ditto.
318 (_mm512_cmpneq_epi64_mask): Ditto.
319 (_mm512_cmpneq_epu64_mask): Ditto.
320 (_mm512_expand_pd): Ditto.
321 (_mm512_expand_ps): Ditto.
322 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
323 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
324 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
325 * config/i386/i386.c (ix86_builtins): Add
326 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
327 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
328 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
329 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
330 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
331 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
332 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
333 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
334 IX86_BUILTIN_PMOVUSQW512_MEM.
335 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
336 __builtin_ia32_pmovsqd512mem_mask,
337 __builtin_ia32_pmovqd512mem_mask,
338 __builtin_ia32_pmovusqw512mem_mask,
339 __builtin_ia32_pmovsqw512mem_mask,
340 __builtin_ia32_pmovqw512mem_mask,
341 __builtin_ia32_pmovusdw512mem_mask,
342 __builtin_ia32_pmovsdw512mem_mask,
343 __builtin_ia32_pmovdw512mem_mask,
344 __builtin_ia32_pmovqb512mem_mask,
345 __builtin_ia32_pmovusqb512mem_mask,
346 __builtin_ia32_pmovsqb512mem_mask,
347 __builtin_ia32_pmovusdb512mem_mask,
348 __builtin_ia32_pmovsdb512mem_mask,
349 __builtin_ia32_pmovdb512mem_mask.
350 (bdesc_args): Add __builtin_ia32_expanddf512,
351 __builtin_ia32_expandsf512.
352 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
353 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
354 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
355 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
356 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
357 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
358 (avx512f_<code>v8div16qi2_mask_store): This.
359 (avx512f_expand<mode>): New.
361 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
363 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
365 (_mm512_mask_prefetch_i64gather_pd): Ditto.
366 (_mm512_prefetch_i32scatter_pd): Ditto.
367 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
368 (_mm512_prefetch_i64scatter_pd): Ditto.
369 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
370 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
371 (_mm512_mask_prefetch_i64gather_ps): Ditto.
372 (_mm512_prefetch_i32scatter_ps): Ditto.
373 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
374 (_mm512_prefetch_i64scatter_ps): Ditto.
375 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
376 * config/i386/i386-builtin-types.def: Define
377 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
378 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
379 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
380 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
381 IX86_BUILTIN_SCATTERPFQPD.
382 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
383 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
384 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
385 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
386 __builtin_ia32_scatterpfqps.
387 (ix86_expand_builtin): Expand new built-ins.
388 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
389 fix memory access data type.
390 (*avx512pf_gatherpf<mode>_mask): Ditto.
391 (*avx512pf_gatherpf<mode>): Ditto.
392 (avx512pf_scatterpf<mode>): Ditto.
393 (*avx512pf_scatterpf<mode>_mask): Ditto.
394 (*avx512pf_scatterpf<mode>): Ditto.
395 (GATHER_SCATTER_SF_MEM_MODE): New.
396 (avx512pf_gatherpf<mode>df): Ditto.
397 (*avx512pf_gatherpf<mode>df_mask): Ditto.
398 (*avx512pf_scatterpf<mode>df): Ditto.
400 2014-01-27 Jakub Jelinek <jakub@redhat.com>
403 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
404 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
407 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
409 * common/config/arm/arm-common.c
410 (arm_rewrite_mcpu): Handle multiple names.
412 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
414 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
416 * gimple-builder.h (create_gimple_tmp): Delete.
418 2014-01-27 Christian Bruel <christian.bruel@st.com>
420 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
423 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
425 * config/pa/pa.md (call): Generate indirect long calls to non-local
426 functions when outputing 32-bit code.
427 (call_value): Likewise except for special call to buggy powf function.
429 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
430 portable runtime and PIC indirect calls.
431 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
432 and PIC call sequences. Use ldo instead of blr to set return register
433 in PIC call sequence.
435 2014-01-25 Walter Lee <walt@tilera.com>
437 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
438 avoid clobbering a live register.
440 2014-01-25 Walter Lee <walt@tilera.com>
442 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
443 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
444 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
445 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
447 2014-01-25 Walter Lee <walt@tilera.com>
449 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
450 arguments on even registers.
451 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
453 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
454 (BIGGEST_ALIGNMENT): Ditto.
455 (BIGGEST_FIELD_ALIGNMENT): Ditto.
457 2014-01-25 Walter Lee <walt@tilera.com>
459 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
460 insns before bundling.
461 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
463 2014-01-25 Walter Lee <walt@tilera.com>
465 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
466 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
467 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
469 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
471 * config/mips/constraints.md (kl): Delete.
472 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
473 define expands, using...
474 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
475 instructions for MIPS16.
476 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
477 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
479 2014-01-25 Walter Lee <walt@tilera.com>
481 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
485 2014-01-25 Walter Lee <walt@tilera.com>
487 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
488 (TARGET_EXPAND_TO_RTL_HOOK): Define.
490 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
492 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
495 2014-01-25 Jakub Jelinek <jakub@redhat.com>
497 * print-rtl.c (in_call_function_usage): New var.
498 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
499 EXPR_LIST mode as mode and not as reg note name.
502 * cfgloopmanip.c (copy_loop_info): If
503 loop->warned_aggressive_loop_optimizations, make sure
504 the flag is set in target loop too.
506 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
508 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
510 * builtins.def: Likewise.
511 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
512 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
513 * ira.c (ira_setup_eliminable_regset): Likewise.
514 * omp-low.c (gate_expand_omp): Likewise.
515 (execute_lower_omp): Likewise.
516 (diagnose_sb_0): Likewise.
517 (gate_diagnose_omp_blocks): Likewise.
518 (simd_clone_clauses_extract): Likewise.
521 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
523 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
524 correction for little endian...
525 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
528 2014-01-24 Jeff Law <law@redhat.com>
530 PR tree-optimization/59919
531 * tree-vrp.c (find_assert_locations_1): Do not register asserts
532 for non-returning calls.
534 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
536 * common/config/aarch64/aarch64-common.c
537 (aarch64_rewrite_mcpu): Handle multiple names.
538 * config/aarch64/aarch64.h
539 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
541 2014-01-24 Dodji Seketeli <dodji@redhat.com>
543 * input.c (add_file_to_cache_tab): Handle the case where fopen
546 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
549 * config/i386/i386.md (pushsf splitter): Get stack adjustment
550 from push operand if code of push isn't PRE_DEC.
552 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
555 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
556 -mquad-memory-atomic. Update -mquad-memory documentation to say
557 it is only used for non-atomic loads/stores.
559 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
560 -mquad-memory or -mquad-memory-atomic switches.
562 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
563 -mquad-memory-atomic to ISA 2.07 support.
565 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
566 to separate support of normal quad word memory operations (ldq, stq)
567 from the atomic quad word memory operations.
569 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
570 support to separate non-atomic quad word operations from atomic
571 quad word operations. Disable non-atomic quad word operations in
572 little endian mode so that we don't have to swap words after the
573 load and before the store.
574 (quad_load_store_p): Add comment about atomic quad word support.
575 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
576 options printed with -mdebug=reg.
578 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
579 -mquad-memory-atomic as the test for whether we have quad word
581 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
582 or -mp8-vector are used, allow byte/half-word atomic operations.
584 * config/rs6000/sync.md (load_lockedti): Insure that the address
585 is a proper indexed or indirect address for the lqarx instruction.
586 On little endian systems, swap the hi/lo registers after the lqarx
588 (load_lockedpti): Use indexed_or_indirect_operand predicate to
589 insure the address is valid for the lqarx instruction.
590 (store_conditionalti): Insure that the address is a proper indexed
591 or indirect address for the stqcrx. instruction. On little endian
592 systems, swap the hi/lo registers before doing the stqcrx.
594 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
595 insure the address is valid for the stqcrx. instruction.
597 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
598 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
599 type of quad memory support is available.
601 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
604 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
605 there is a danger of looping.
607 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
610 force flag_ira_loop_pressure if set via command line.
612 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
614 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
615 (ashr_simd): New builtin handling DI mode.
616 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
617 (aarch64_sshr_simddi): New match pattern.
618 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
619 (vshrd_n_s64): Likewise.
620 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
622 2014-01-23 Nick Clifton <nickc@redhat.com>
624 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
625 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
626 favour of mcu specific scripts.
627 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
630 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
631 Alex Velenko <Alex.Velenko@arm.com>
633 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
634 (vaddv_s16): Likewise.
635 (vaddv_s32): Likewise.
636 (vaddv_u8): Likewise.
637 (vaddv_u16): Likewise.
638 (vaddv_u32): Likewise.
639 (vaddvq_s8): Likewise.
640 (vaddvq_s16): Likewise.
641 (vaddvq_s32): Likewise.
642 (vaddvq_s64): Likewise.
643 (vaddvq_u8): Likewise.
644 (vaddvq_u16): Likewise.
645 (vaddvq_u32): Likewise.
646 (vaddvq_u64): Likewise.
647 (vaddv_f32): Likewise.
648 (vaddvq_f32): Likewise.
649 (vaddvq_f64): Likewise.
650 (vmaxv_f32): Likewise.
651 (vmaxv_s8): Likewise.
652 (vmaxv_s16): Likewise.
653 (vmaxv_s32): Likewise.
654 (vmaxv_u8): Likewise.
655 (vmaxv_u16): Likewise.
656 (vmaxv_u32): Likewise.
657 (vmaxvq_f32): Likewise.
658 (vmaxvq_f64): Likewise.
659 (vmaxvq_s8): Likewise.
660 (vmaxvq_s16): Likewise.
661 (vmaxvq_s32): Likewise.
662 (vmaxvq_u8): Likewise.
663 (vmaxvq_u16): Likewise.
664 (vmaxvq_u32): Likewise.
665 (vmaxnmv_f32): Likewise.
666 (vmaxnmvq_f32): Likewise.
667 (vmaxnmvq_f64): Likewise.
668 (vminv_f32): Likewise.
669 (vminv_s8): Likewise.
670 (vminv_s16): Likewise.
671 (vminv_s32): Likewise.
672 (vminv_u8): Likewise.
673 (vminv_u16): Likewise.
674 (vminv_u32): Likewise.
675 (vminvq_f32): Likewise.
676 (vminvq_f64): Likewise.
677 (vminvq_s8): Likewise.
678 (vminvq_s16): Likewise.
679 (vminvq_s32): Likewise.
680 (vminvq_u8): Likewise.
681 (vminvq_u16): Likewise.
682 (vminvq_u32): Likewise.
683 (vminnmv_f32): Likewise.
684 (vminnmvq_f32): Likewise.
685 (vminnmvq_f64): Likewise.
687 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
689 * config/aarch64/aarch64-simd.md
690 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
691 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
692 (*aarch64_mul3_elt<mode>): Likewise.
693 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
694 (*aarch64_mul3_elt_to_64v2df): Likewise.
695 (*aarch64_mla_elt<mode>): Likewise.
696 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
697 (*aarch64_mls_elt<mode>): Likewise.
698 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
699 (*aarch64_fma4_elt<mode>): Likewise.
700 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
701 (*aarch64_fma4_elt_to_64v2df): Likewise.
702 (*aarch64_fnma4_elt<mode>): Likewise.
703 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
704 (*aarch64_fnma4_elt_to_64v2df): Likewise.
705 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
706 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
707 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
708 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
709 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
710 (aarch64_sqdmull_lane<mode>_internal): Likewise.
711 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
713 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
715 * config/aarch64/aarch64-simd.md
716 (aarch64_be_checked_get_lane<mode>): New define_expand.
717 * config/aarch64/aarch64-simd-builtins.def
718 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
719 New builtin definition.
720 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
721 Use new safe be builtin.
723 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
725 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
727 (aarch64_be_st1<mode>): Likewise.
728 (aarch_ld1<VALL:mode>): Define_expand modified.
729 (aarch_st1<VALL:mode>): Likewise.
730 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
731 (UNSPEC_ST1): Likewise.
733 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
735 * config/microblaze/microblaze.md: Add trap insn and attribute
737 2014-01-23 Dodji Seketeli <dodji@redhat.com>
739 PR preprocessor/58580
740 * input.h (location_get_source_line): Take an additional line_size
742 (void diagnostics_file_cache_fini): Declare new function.
743 * input.c (struct fcache): New type.
744 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
745 New static constants.
746 (diagnostic_file_cache_init, total_lines_num)
747 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
748 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
749 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
750 (get_next_line, read_next_line, goto_next_line, read_line_num):
751 New static function definitions.
752 (diagnostic_file_cache_fini): New function.
753 (location_get_source_line): Take an additional output line_len
754 parameter. Re-write using lookup_or_add_file_to_cache_tab and
756 * diagnostic.c (diagnostic_finish): Call
757 diagnostic_file_cache_fini.
758 (adjust_line): Take an additional input parameter for the length
759 of the line, rather than calculating it with strlen.
760 (diagnostic_show_locus): Adjust the use of
761 location_get_source_line and adjust_line with respect to their new
762 signature. While displaying a line now, do not stop at the first
763 null byte. Rather, display the zero byte as a space and keep
764 going until we reach the size of the line.
765 * Makefile.in: Add vec.o to OBJS-libcommon
767 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
768 Ilya Tocar <ilya.tocar@intel.com>
770 * config/i386/avx512fintrin.h (_mm512_kmov): New.
771 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
772 (__builtin_ia32_kmov16): Ditto.
773 * config/i386/i386.md (UNSPEC_KMOV): New.
776 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
778 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
779 (_mm512_storeu_si512): Ditto.
781 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
784 * rtl.h (get_referenced_operands): Declare.
785 * recog.c (get_referenced_operands): New function.
786 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
787 operands have been referenced when recording LO_SUM references.
789 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
791 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
793 2014-01-22 Jan Hubicka <jh@suse.cz>
795 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
796 Enable for generic and recent AMD targets.
798 2014-01-22 Jan Hubicka <jh@suse.cz>
800 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
801 ARG_SIZE note when adjustment was eliminated.
803 2014-01-22 Jeff Law <law@redhat.com>
805 PR tree-optimization/59597
806 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
807 in file. Accept new argument REGISTERING and use it to modify
808 dump output appropriately.
809 (register_jump_thread): Corresponding changes.
810 (mark_threaded_blocks): Reinstate code to cancel unprofitable
811 thread paths involving joiner blocks. Add code to dump cancelled
812 jump threading paths.
814 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
816 PR rtl-optimization/59477
817 * lra-constraints.c (inherit_in_ebb): Process call for living hard
818 regs. Update reloads_num and potential_reload_hard_regs for all insns.
820 2014-01-22 Tom Tromey <tromey@redhat.com>
822 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
824 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
826 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
828 PR rtl-optimization/59896
829 * lra-constraints.c (process_alt_operands): Check unused note for
830 matched operands of insn with no output reloads.
832 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
834 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
835 (mips_move_from_gpr_cost): Likewise.
837 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
839 PR rtl-optimization/59858
840 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
841 ira_class_hard_regs_num.
842 (process_alt_operands): Increase reject for dying matched operand.
844 2014-01-21 Jakub Jelinek <jakub@redhat.com>
847 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
848 smaller than size, perform several stores or loads and stores
849 at dst + count - size to store or copy all of size bytes, rather
850 than just last modesize bytes.
852 2014-01-20 DJ Delorie <dj@redhat.com>
854 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
855 that CLOBBERs are REGs before propogating their values.
857 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
860 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
861 (cgraph_inline_failed_type): New function.
862 * cgraph.h (DEFCIFCODE): Add type.
863 (cgraph_inline_failed_type_t): New enum.
864 (cgraph_inline_failed_type): New prototype.
865 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
866 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
867 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
868 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
869 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
870 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
871 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
872 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
873 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
874 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
875 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
876 OPTIMIZATION_MISMATCH.
877 * tree-inline.c (expand_call_inline): Emit errors during
878 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
880 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
883 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
884 mode attribute in insn output.
886 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
888 * output.h (output_constant): Delete.
889 * varasm.c (output_constant): Make private.
891 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
893 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
895 2014-01-20 Jakub Jelinek <jakub@redhat.com>
898 * tree.h (fold_builtin_strcat): New prototype.
899 * builtins.c (fold_builtin_strcat): No longer static. Add len
900 argument, if non-NULL, don't call c_strlen. Optimize
901 directly into __builtin_memcpy instead of __builtin_strcpy.
902 (fold_builtin_2): Adjust fold_builtin_strcat caller.
903 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
905 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
907 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
908 for SImode_address_operand operands, having only a REG argument.
910 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
912 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
913 loader name using mbig-endian.
914 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
916 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
918 * doc/invoke.texi (-march): Clarify documentation for AArch64.
922 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
924 * config/aarch64/aarch64-protos.h
925 (aarch64_cannot_change_mode_class_ptr): Declare.
926 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
927 aarch64_cannot_change_mode_class_ptr): New.
928 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
929 backend hook aarch64_cannot_change_mode_class.
931 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
933 * common/config/aarch64/aarch64-common.c
934 (aarch64_handle_option): Don't handle any option order logic here.
935 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
936 selected_cpu, warn on architecture version mismatch.
937 (aarch64_override_options): Fix parsing order for option strings.
939 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
940 Iain Sandoe <iain@codesourcery.com>
943 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
944 warning. Amend comment to reflect current functionality.
946 2014-01-20 Richard Biener <rguenther@suse.de>
949 * builtins.c (fold_builtin_strcat): Remove case better handled
950 by tree-ssa-strlen.c.
952 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
954 * config/aarch64/aarch64.opt
955 (mcpu, march, mtune): Make case-insensitive.
957 2014-01-20 Jakub Jelinek <jakub@redhat.com>
960 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
961 if operands[1] is a REG or ZERO_EXTEND of a REG.
963 2014-01-19 Jan Hubicka <jh@suse.cz>
965 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
967 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
969 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
970 long non-pic millicode calls.
972 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
974 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
976 2014-01-19 Kito Cheng <kito@0xlab.org>
978 * builtins.c (expand_movstr): Check movstr expand done or fail.
980 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
981 H.J. Lu <hongjiu.lu@intel.com>
984 * config/i386/i386.md (*lea<mode>): Zero-extend return register
985 to DImode for zero-extended addresses.
987 2014-01-19 Jakub Jelinek <jakub@redhat.com>
989 PR rtl-optimization/57763
990 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
991 on the new indirect jump_insn and increment LABEL_NUSES (label).
993 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
997 * config.gcc (x86_archs): New variable.
998 (x86_64_archs): Likewise.
999 (x86_cpus): Likewise.
1000 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1001 --with-arch/--with-cpu= options.
1002 Support --with-arch=/--with-cpu={nehalem,westmere,
1003 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1005 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1007 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1008 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1010 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1012 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1014 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1017 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1018 clear cpp_get_options (parse_in)->warn_unused_macros for
1019 ix86_target_macros_internal with cpp_define.
1021 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1023 * jump.c (delete_related_insns): Keep (use (insn))s.
1024 * reorg.c (redundant_insn): Check for barriers too.
1026 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1028 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1030 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1032 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1033 call to $$dyncall when TARGET_LONG_CALLS is true.
1035 2014-01-17 Jeff Law <law@redhat.com>
1037 * ree.c (combine_set_extension): Temporarily disable test for
1038 changing number of hard registers.
1040 2014-01-17 Jan Hubicka <jh@suse.cz>
1043 * ipa-inline-analysis.c (inline_free_summary):
1044 Do not free summary of aliases.
1046 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1049 * gimplify.c (gimplify_expr): Use create_tmp_var
1050 instead of create_tmp_var_raw. If cond doesn't have
1051 integral type, don't add the IFN_ANNOTATE builtin at all.
1053 2014-01-17 Martin Jambor <mjambor@suse.cz>
1056 * ipa-cp.c (prev_edge_clone): New variable.
1057 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1058 Also resize prev_edge_clone vector.
1059 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1060 (ipcp_edge_removal_hook): New function.
1061 (ipcp_driver): Register ipcp_edge_removal_hook.
1063 2014-01-17 Andrew Pinski <apinski@cavium.com>
1064 Steve Ellcey <sellcey@mips.com>
1067 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1070 2014-01-17 Jeff Law <law@redhat.com>
1073 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1074 so that pass_ccp runs first.
1076 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1078 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1079 (ix86_adjust_cost): Use !TARGET_XXX.
1080 (do_reorder_for_imul): Likewise.
1081 (swap_top_of_ready_list): Likewise.
1082 (ix86_sched_reorder): Likewise.
1084 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1086 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1087 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1088 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1089 (intel_memset): New. Duplicate slm_memset.
1090 (intel_cost): New. Duplicate slm_cost.
1091 (m_INTEL): New macro.
1092 (processor_target_table): Add "intel".
1093 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1094 with PROCESSOR_INTEL for "intel".
1095 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1096 PROCESSOR_SILVERMONT.
1097 (ix86_issue_rate): Likewise.
1098 (ix86_adjust_cost): Likewise.
1099 (ia32_multipass_dfa_lookahead): Likewise.
1100 (swap_top_of_ready_list): Likewise.
1101 (ix86_sched_reorder): Likewise.
1102 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1103 instead of TARGET_OPT_AGU.
1104 * config/i386/i386.h (TARGET_INTEL): New.
1105 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1106 (processor_type): Add PROCESSOR_INTEL.
1107 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1108 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1110 2014-01-17 Marek Polacek <polacek@redhat.com>
1113 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1116 2014-01-17 Richard Biener <rguenther@suse.de>
1118 PR tree-optimization/46590
1119 * opts.c (default_options_table): Add entries for
1120 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1121 all enabled at -O1 but not for -Og.
1122 * common.opt (fbranch-count-reg): Remove Init(1).
1123 (fmove-loop-invariants): Likewise.
1124 (ftree-pta): Likewise.
1126 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1128 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1129 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1130 decls to at least the GCC 4.8 used alignments.
1133 * tree-nested.c (convert_nonlocal_reference_stmt,
1134 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1135 of GIMPLE_BIND stmts, adjust associated decls.
1137 2014-01-17 Richard Biener <rguenther@suse.de>
1139 PR tree-optimization/46590
1140 * vec.h (vec<>::bseach): New member function implementing
1141 binary search according to C89 bsearch.
1142 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1143 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1144 bitmap pointer again. Make accesses_in_loop a flat array.
1145 (mem_ref_obstack): New global.
1146 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1147 (mark_ref_stored): Likewise.
1148 (ref_indep_loop_p_2): Likewise.
1149 (set_ref_stored_in_loop): New helper function.
1150 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1151 (memref_free): Adjust.
1152 (record_mem_ref_loc): Simplify.
1153 (gather_mem_refs_stmt): Adjust.
1154 (sort_locs_in_loop_postorder_cmp): New function.
1155 (analyze_memory_references): Sort accesses_in_loop after
1156 loop postorder number.
1157 (find_ref_loc_in_loop_cmp): New function.
1158 (for_all_locs_in_loop): Find relevant cluster of locs in
1159 accesses_in_loop and iterate without recursion.
1160 (execute_sm): Avoid uninit warning.
1161 (struct ref_always_accessed): Simplify.
1162 (ref_always_accessed::operator ()): Likewise.
1163 (ref_always_accessed_p): Likewise.
1164 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1165 loop postorder numbers here.
1166 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1169 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1172 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1173 on decls for which assemble_alias has been called.
1175 2014-01-17 Nick Clifton <nickc@redhat.com>
1177 * config/msp430/msp430.opt: (mcpu): New option.
1178 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1179 (msp430_option_override): Parse target_cpu. If the MCU name
1180 matches a generic string, clear target_mcu.
1181 (msp430_attr): Allow numeric interrupt values up to 63.
1182 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1183 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1185 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1187 * config/msp430/msp430.md (popm): Use %J rather than %I.
1188 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1189 (addhi_cy_i): Use immediate_operand for operand 2.
1190 * doc/invoke.texi: Document -mcpu option.
1192 2014-01-17 Richard Biener <rguenther@suse.de>
1194 PR rtl-optimization/38518
1195 * df.h (df_analyze_loop): Declare.
1196 * df-core.c: Include cfgloop.h.
1197 (df_analyze_1): Split out main part of df_analyze.
1198 (df_analyze): Adjust.
1199 (loop_inverted_post_order_compute): New function.
1200 (loop_post_order_compute): Likewise.
1201 (df_analyze_loop): New function avoiding whole-function
1203 * loop-invariant.c (find_defs): Use df_analyze_loop.
1204 (find_invariants): Adjust.
1205 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1207 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1209 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1210 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1212 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1214 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1215 traversal when removing references.
1217 2014-01-16 Jan Hubicka <jh@suse.cz>
1220 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1222 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1225 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1226 pushing a reload for an autoinc when we had previously reloaded an
1227 inner part of the address.
1229 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1231 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1233 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1234 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1235 when not giving up or versioning for alias only because of
1237 (vect_analyze_data_ref_dependences): Set to true.
1238 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1240 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1241 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1245 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1248 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1249 operand 0 predicate for gathers, use a new pseudo as subtarget.
1251 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1254 * lra-constraints.c (process_alt_operands): Add printing debug info.
1255 Check absence of input/output reloads for matched operands too.
1257 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1259 PR rtl-optimization/59835
1260 * ira.c (ira_init_register_move_cost): Increase cost for
1263 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1265 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1267 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1270 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1271 non-register objects. Use gen_(high/low)part more consistently.
1274 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1277 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1278 endian support, remove tests for WORDS_BIG_ENDIAN.
1279 (p8_mfvsrd_3_<mode>): Likewise.
1280 (reload_gpr_from_vsx<mode>): Likewise.
1281 (reload_gpr_from_vsxsf): Likewise.
1282 (p8_mfvsrd_4_disf): Likewise.
1284 2014-01-16 Richard Biener <rguenther@suse.de>
1286 PR rtl-optimization/46590
1287 * lcm.c (compute_antinout_edge): Use postorder iteration.
1288 (compute_laterin): Use inverted postorder iteration.
1290 2014-01-16 Nick Clifton <nickc@redhat.com>
1293 * varasm.c (output_constant): Return the number of bytes actually
1295 (output_constructor_array_range): Update the field size with the
1296 number of bytes emitted by output_constant.
1297 (output_constructor_regular_field): Likewise. Also do not
1298 complain if the total number of bytes emitted is now greater
1299 than the expected fieldpos.
1300 * output.h (output_constant): Update prototype and descriptive comment.
1302 2014-01-16 Marek Polacek <polacek@redhat.com>
1305 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1306 it is error_mark_node.
1308 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1310 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1311 VALID_AVX256_REG_OR_OI_MODE.
1313 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1315 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1316 current procedure should be profiled.
1318 2014-01-15 Andrew Pinski <apinski@cavium.com>
1320 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1321 of moving from/to the STACK_REG register class.
1323 2014-01-15 Richard Henderson <rth@redhat.com>
1326 * reginfo.c (global_regs_decl): Globalize.
1327 * rtl.h (global_regs_decl): Declare.
1328 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1329 reserved via global_regs.
1331 2014-01-15 Teresa Johnson <tejohnson@google.com>
1333 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1335 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1337 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1338 and vmulosh rather than call gen_vec_widen_smult_*.
1339 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1340 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1341 (vec_widen_smult_even_v16qi): Likewise.
1342 (vec_widen_umult_even_v8hi): Likewise.
1343 (vec_widen_smult_even_v8hi): Likewise.
1344 (vec_widen_umult_odd_v16qi): Likewise.
1345 (vec_widen_smult_odd_v16qi): Likewise.
1346 (vec_widen_umult_odd_v8hi): Likewise.
1347 (vec_widen_smult_odd_v8hi): Likewise.
1348 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1349 vmuloub rather than call gen_vec_widen_umult_*.
1350 (vec_widen_umult_lo_v16qi): Likewise.
1351 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1352 vmulosb rather than call gen_vec_widen_smult_*.
1353 (vec_widen_smult_lo_v16qi): Likewise.
1354 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1355 rather than call gen_vec_widen_umult_*.
1356 (vec_widen_umult_lo_v8hi): Likewise.
1357 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1358 rather than call gen_vec_widen_smult_*.
1359 (vec_widen_smult_lo_v8hi): Likewise.
1361 2014-01-15 Jeff Law <law@redhat.com>
1363 PR tree-optimization/59747
1364 * ree.c (find_and_remove_re): Properly handle case where a second
1365 eliminated extension requires widening a copy created for elimination
1366 of a prior extension.
1367 (combine_set_extension): Ensure that the number of hard regs needed
1368 for a destination register does not change when we widen it.
1370 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1372 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1373 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1374 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1375 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1376 (avr-*-rtems*): Likewise.
1377 (bfin*-rtems*): Likewise.
1378 (moxie-*-rtems*): Likewise.
1379 (h8300-*-rtems*): Likewise.
1380 (i[34567]86-*-rtems*): Likewise.
1381 (lm32-*-rtems*): Likewise.
1382 (m32r-*-rtems*): Likewise.
1383 (m68k-*-rtems*): Likewise.
1384 (microblaze*-*-rtems*): Likewise.
1385 (mips*-*-rtems*): Likewise.
1386 (powerpc-*-rtems*): Likewise.
1387 (sh-*-rtems*): Likewise.
1388 (sparc-*-rtems*): Likewise.
1389 (sparc64-*-rtems*): Likewise.
1390 (v850-*-rtems*): Likewise.
1391 (m32c-*-rtems*): Likewise.
1393 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1395 PR rtl-optimization/59511
1396 * ira.c (ira_init_register_move_cost): Use memory costs for some
1397 cases of register move cost calculations.
1398 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1399 instead of BB frequency.
1400 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1401 * lra-assigns.c (find_hard_regno_for): Ditto.
1403 2014-01-15 Richard Biener <rguenther@suse.de>
1405 PR tree-optimization/59822
1406 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1407 (vectorizable_load): Use it to hoist defs of uses of invariant
1408 loads out of the loop.
1410 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1411 Kugan Vivekanandarajah <kuganv@linaro.org>
1414 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1417 2014-01-15 Richard Biener <rguenther@suse.de>
1419 PR rtl-optimization/59802
1420 * lcm.c (compute_available): Use inverted postorder to seed
1421 the initial worklist.
1423 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1426 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1427 ADDR_REGS for invalid symrefs in non-PIC code.
1429 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1432 * builtins.c (determine_block_size): Initialize *probable_max_size
1433 even if len_rtx is CONST_INT.
1435 2014-01-14 Andrew Pinski <apinski@cavium.com>
1437 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1438 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1439 (cortexa53_tunings): Likewise.
1440 (aarch64_sched_issue_rate): New function.
1441 (TARGET_SCHED_ISSUE_RATE): Define.
1443 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1445 * ira-costs.c (find_costs_and_classes): Add missed
1446 ira_init_register_move_cost_if_necessary.
1448 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1451 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1453 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1456 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1457 to indicate if type is used for function return value. Warn ABI
1458 change if the vector mode isn't available for function return value.
1459 (ix86_function_arg_advance): Pass false to type_natural_mode.
1460 (ix86_function_arg): Likewise.
1461 (ix86_gimplify_va_arg): Likewise.
1462 (function_arg_32): Don't warn ABI change.
1463 (ix86_function_value): Pass true to type_natural_mode.
1464 (ix86_return_in_memory): Likewise.
1465 (ix86_struct_value_rtx): Removed.
1466 (TARGET_STRUCT_VALUE_RTX): Likewise.
1468 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1470 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1471 converting a conditional jump into a conditional return.
1473 2014-01-14 Richard Biener <rguenther@suse.de>
1475 PR tree-optimization/58921
1476 PR tree-optimization/59006
1477 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1478 hoisting invariant stmts.
1479 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1480 invariant loads on the preheader edge if possible.
1482 2014-01-14 Joey Ye <joey.ye@arm.com>
1484 * doc/plugin.texi (Building GCC plugins): Update to C++.
1486 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1488 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1489 (_mm_rcp28_round_ss): Ditto.
1490 (_mm_rsqrt28_round_sd): Ditto.
1491 (_mm_rsqrt28_round_ss): Ditto.
1492 (_mm_rcp28_sd): Ditto.
1493 (_mm_rcp28_ss): Ditto.
1494 (_mm_rsqrt28_sd): Ditto.
1495 (_mm_rsqrt28_ss): Ditto.
1496 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1497 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1498 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1499 (IX86_BUILTIN_RCP28SD): Ditto.
1500 (IX86_BUILTIN_RCP28SS): Ditto.
1501 (IX86_BUILTIN_RSQRT28SD): Ditto.
1502 (IX86_BUILTIN_RSQRT28SS): Ditto.
1503 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1504 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1505 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1506 (ix86_expand_special_args_builtin): Expand new FTYPE.
1507 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1508 (srcp14<mode>): Make insn unary.
1509 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1510 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1511 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1512 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1513 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1514 Fix rounding: make it SAE only.
1515 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1517 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1519 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1520 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1521 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1522 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1523 (round_saeonly_mask_scalar_operand4): Ditto.
1524 (round_saeonly_mask_scalar_op3): Ditto.
1525 (round_saeonly_mask_scalar_op4): Ditto.
1527 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1530 Implement -maltivec=be for vec_insert and vec_extract.
1532 2014-01-10 DJ Delorie <dj@redhat.com>
1534 * config/msp430/msp430.md (call_internal): Don't allow memory
1535 references with SP as the base register.
1536 (call_value_internal): Likewise.
1537 * config/msp430/constraints.md (Yc): New. For memory references
1538 that don't use SP as a base register.
1540 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1541 "an integer without a # prefix"
1542 * config/msp430/msp430.md (epilogue_helper): Use it.
1544 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1547 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1548 AVX512F gather builtins.
1549 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1550 on gather decls with INTEGER_TYPE masktype.
1551 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1552 directly into the builtin rather than hoisting it before loop.
1554 PR tree-optimization/59387
1555 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1556 (scev_const_prop): If folded_casts and type has undefined overflow,
1557 use force_gimple_operand instead of force_gimple_operand_gsi and
1558 for each added stmt if it is assign with
1559 arith_code_with_undefined_signed_overflow, call
1560 rewrite_to_defined_overflow.
1561 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1562 gimple-fold.h instead.
1563 (arith_code_with_undefined_signed_overflow,
1564 rewrite_to_defined_overflow): Moved to ...
1565 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1566 rewrite_to_defined_overflow): ... here. No longer static.
1567 Include gimplify-me.h.
1568 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1569 rewrite_to_defined_overflow): New prototypes.
1571 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1573 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1575 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1577 * builtins.c (get_object_alignment_2): Minor tweak.
1578 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1580 2014-01-13 Christian Bruel <christian.bruel@st.com>
1582 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1583 optimized non constant lengths.
1585 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1588 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1589 load as __atomic_load_N if possible.
1591 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1593 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1595 (rs6000_expand_builtin): Adjust call.
1597 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1600 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1601 * config/rs6000/rs6000.c: Include target-globals.h.
1602 (rs6000_set_current_function): Instead of doing target_reinit
1603 unconditionally, use save_target_globals_default_opts and
1604 restore_target_globals.
1606 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1608 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1609 (rs6000_expand_builtin): Handle mffs and mtfsf.
1610 (rs6000_init_builtins): Define mffs and mtfsf.
1611 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1612 (rs6000_mffs): New pattern.
1613 (rs6000_mtfsf): New pattern.
1615 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1617 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1618 Start narrowing with START. Apply candidate-use pair
1619 and check overall cost in narrowing.
1620 (iv_ca_prune): Pass new argument.
1622 2014-01-10 Jeff Law <law@redhat.com>
1625 * ree.c (combine_reaching_defs): Ensure the defining statement
1626 occurs before the extension when optimizing extensions with
1627 different source and destination hard registers.
1629 2014-01-10 Jan Hubicka <jh@suse.cz>
1632 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1633 vtables into the type inheritance graph.
1635 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1637 PR rtl-optimization/59754
1638 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1639 modes in the REGNO != REGNO case.
1641 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1643 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1645 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1647 PR tree-optimization/59745
1648 * tree-predcom.c (tree_predictive_commoning_loop): Call
1649 free_affine_expand_cache if giving up because components is NULL.
1651 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1652 GC in payload of target_globals struct instead of allocating them on
1653 the heap and the larger structs separately using GC.
1654 * target-globals.h (struct target_globals): Make regs, hard_regs,
1655 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1656 of GTY((skip)) and change type to void *.
1657 (reset_target_globals): Cast loads from those fields to corresponding
1660 2014-01-10 Steve Ellcey <sellcey@mips.com>
1663 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1664 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1665 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1667 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1670 * aarch64-modes.def (CC_Zmode): New flags mode.
1671 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1672 represents an equality.
1673 (aarch64_get_condition_code): Handle CC_Zmode.
1674 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1676 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1678 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1679 extraction in good case.
1681 2014-01-10 Richard Biener <rguenther@suse.de>
1683 PR tree-optimization/59374
1684 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1685 checking after SLP discovery. Mark stmts not participating
1686 in any SLP instance properly.
1688 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1690 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1691 when handling a SET rtx.
1693 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1695 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1696 (cortex-a57): Likewise.
1697 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1699 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1701 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1702 non-iwmmxt builtins.
1704 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1708 * ipa-devirt.c record_target_from_binfo): Take as argument
1709 stack of binfos and lookup matching one for virtual inheritance.
1710 (possible_polymorphic_call_targets_1): Update.
1712 2014-01-10 Huacai Chen <chenhc@lemote.com>
1714 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1715 kernel strings for Loongson-2E/2F/3A.
1717 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1720 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1721 is_gimple_call before calling gimple_call_internal_p.
1723 2014-01-09 Steve Ellcey <sellcey@mips.com>
1725 * Makefile.in (TREE_FLOW_H): Remove.
1726 (TREE_SSA_H): Add file names from tree-flow.h.
1727 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1728 * tree.h: Remove tree-flow.h reference.
1729 * hash-table.h: Remove tree-flow.h reference.
1730 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1731 reference with tree-ssa-loop.h.
1733 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1735 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1736 default element-order behavior for -maltivec.
1737 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1739 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1740 when targeting big endian, at least for now.
1741 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1743 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1746 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1747 var satisfies use_register_for_decl, just take into account type
1748 alignment, rather than decl alignment.
1750 PR tree-optimization/59622
1751 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1752 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1753 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1754 Don't devirtualize for inplace at all. For targets.length () == 1,
1755 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1757 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1759 * config/i386/i386.md (cpu): Remove the unused btver1.
1761 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1763 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1765 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1768 * tree-core.h (struct target_globals): New forward declaration.
1769 (struct tree_target_option): Add globals field.
1770 * tree.h (TREE_TARGET_GLOBALS): Define.
1771 (prepare_target_option_nodes_for_pch): New prototype.
1772 * target-globals.h (struct target_globals): Define even if
1774 * tree.c (prepare_target_option_node_for_pch,
1775 prepare_target_option_nodes_for_pch): New functions.
1776 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1777 * config/i386/i386.c: Include target-globals.h.
1778 (ix86_set_current_function): Instead of doing target_reinit
1779 unconditionally, use save_target_globals_default_opts and
1780 restore_target_globals.
1782 2014-01-09 Richard Biener <rguenther@suse.de>
1784 PR tree-optimization/59715
1785 * tree-cfg.h (split_critical_edges): Declare.
1786 * tree-cfg.c (split_critical_edges): Export.
1787 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1789 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1791 * cfgexpand.c (expand_stack_vars): Optionally disable
1792 asan stack protection.
1793 (expand_used_vars): Likewise.
1794 (partition_stack_vars): Likewise.
1795 * asan.c (asan_emit_stack_protection): Optionally disable
1796 after return stack usage.
1797 (instrument_derefs): Optionally disable memory access instrumentation.
1798 (instrument_builtin_call): Likewise.
1799 (instrument_strlen_call): Likewise.
1800 (asan_protect_global): Optionally disable global variables protection.
1801 * doc/invoke.texi: Added doc for new options.
1802 * params.def: Added new options.
1803 * params.h: Likewise.
1805 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1807 PR rtl-optimization/59724
1808 * ifcvt.c (cond_exec_process_if_block): Don't call
1809 flow_find_head_matching_sequence with 0 longest_match.
1810 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1811 non-active insns if !stop_after.
1812 (try_head_merge_bb): Revert 2014-01-07 changes.
1814 2014-01-08 Jeff Law <law@redhat.com>
1816 * ree.c (get_sub_rtx): New function, extracted from...
1817 (merge_def_and_ext): Here.
1818 (combine_reaching_defs): Use get_sub_rtx.
1820 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1822 * cgraph.h (varpool_variable_node): Do not choke on null node.
1824 2014-01-08 Catherine Moore <clm@codesourcery.com>
1826 * config/mips/mips.md (simple_return): Attempt to use JRC
1828 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1830 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1832 PR rtl-optimization/59137
1833 * reorg.c (steal_delay_list_from_target): Call update_block for
1835 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1837 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1839 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1840 two duplicate entries.
1842 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1845 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1847 * config/mips/mips.c (mips_truncated_op_cost): New function.
1848 (mips_rtx_costs): Adjust test for BADDU.
1849 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1851 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1853 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1854 (*baddu_si): ...this new pattern.
1856 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1859 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1861 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1864 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1866 (expand_expr, expand_normal): Adjust.
1867 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1868 inner_reference_p. Use inner_reference_p to expand inner references.
1869 (store_expr): Adjust.
1870 * cfgexpand.c (expand_call_stmt): Adjust.
1872 2014-01-08 Rong Xu <xur@google.com>
1874 * gcov-io.c (gcov_var): Move from gcov-io.h.
1875 (gcov_position): Ditto.
1876 (gcov_is_error): Ditto.
1877 (gcov_rewrite): Ditto.
1878 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1879 only part to libgcc/libgcov.h.
1881 2014-01-08 Marek Polacek <polacek@redhat.com>
1884 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1886 2014-01-08 Marek Polacek <polacek@redhat.com>
1889 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1891 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1893 PR rtl-optimization/59649
1894 * stor-layout.c (get_mode_bounds): For BImode return
1895 0 and STORE_FLAG_VALUE.
1897 2014-01-08 Richard Biener <rguenther@suse.de>
1900 * gimple.h (is_gimple_builtin_call): Remove.
1901 (gimple_builtin_call_types_compatible_p): New.
1902 (gimple_call_builtin_p): New overload.
1903 * gimple.c (is_gimple_builtin_call): Remove.
1904 (validate_call): Rename to ...
1905 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1907 (validate_type): New static function.
1908 (gimple_call_builtin_p): New overload and adjust.
1909 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1910 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1911 (gimple_fold_stmt_to_constant_1): Likewise.
1912 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1914 2014-01-08 Richard Biener <rguenther@suse.de>
1917 * gimplify.c (gimplify_expr): Gimplify register-register type
1918 VIEW_CONVERT_EXPRs to separate stmts.
1920 2014-01-07 Jeff Law <law@redhat.com>
1923 * ree.c (combine_set_extension): Handle case where source
1924 and destination registers in an extension insn are different.
1925 (combine_reaching_defs): Allow source and destination registers
1926 in extension to be different under limited circumstances.
1927 (add_removable_extension): Remove restriction that the
1928 source and destination registers in the extension are the same.
1929 (find_and_remove_re): Emit a copy from the extension's
1930 destination to its source after the defining insn if
1931 the source and destination registers are different.
1934 * ifcvt.c (merge_if_block): If we are merging a block with more than
1935 one successor with a block with no successors, remove any BARRIER
1936 after the second block.
1938 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1940 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1942 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1945 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1946 for 14-bit register offsets when INT14_OK_STRICT is false.
1948 2014-01-07 Roland Stigge <stigge@antcom.de>
1949 Michael Meissner <meissner@linux.vnet.ibm.com>
1952 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1953 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1955 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1957 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1960 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1962 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1963 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1964 rtx is const0_rtx or not.
1966 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1969 * target-globals.c (save_target_globals): Remove this_fn_optab
1971 * toplev.c: Include optabs.h.
1972 (target_reinit): Temporarily restore the global options if another
1973 set of options are in force.
1975 2014-01-07 Jakub Jelinek <jakub@redhat.com>
1977 PR rtl-optimization/58668
1978 * cfgcleanup.c (flow_find_cross_jump): Don't count
1979 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
1980 to determine what is counted.
1981 (flow_find_head_matching_sequence): Use active_insn_p to determine
1983 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1985 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1986 determine what is counted.
1988 PR tree-optimization/59643
1989 * tree-predcom.c (split_data_refs_to_components): If one dr is
1990 read and one write, determine_offset fails and the write isn't
1991 in the bad component, just put the read into the bad component.
1993 2014-01-07 Mike Stump <mikestump@comcast.net>
1994 Jakub Jelinek <jakub@redhat.com>
1997 * tree-core.h (struct tree_optimization_option): Change optabs
1998 type from unsigned char * to void *.
1999 * optabs.c (init_tree_optimization_optabs): Adjust
2000 TREE_OPTIMIZATION_OPTABS initialization.
2002 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2005 * config/i386/i386.h (struct machine_function): Add
2006 no_drap_save_restore field.
2007 * config/i386/i386.c (ix86_save_reg): Use
2008 !cfun->machine->no_drap_save_restore instead of
2009 crtl->stack_realign_needed.
2010 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2011 this function clears frame_pointer_needed. Set
2012 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2013 and DRAP reg is needed.
2015 2014-01-06 Marek Polacek <polacek@redhat.com>
2018 * doc/implement-c.texi: Mention that other integer types are
2019 permitted as bit-field types in strictly conforming mode.
2021 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2023 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2026 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2028 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2030 2014-01-06 Martin Jambor <mjambor@suse.cz>
2033 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2035 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2037 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2041 * var-tracking.c (add_stores): Preserve the value of the source even if
2042 we don't record the store.
2044 2014-01-06 Terry Guo <terry.guo@arm.com>
2046 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2048 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2051 * config/darwin.c (darwin_function_section): Adjust return values to
2052 correspond to optimisation changes made in r206070.
2054 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2056 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2057 from prefetch_block tune setting.
2058 (nocona_cost): Correct size of prefetch block to 64.
2060 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2062 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2063 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2064 used to save the static chain register in the computation of the offset
2065 from which the FP registers need to be restored.
2067 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2069 PR tree-optimization/59519
2070 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2071 ICE if get_current_def (current_new_name) is already non-NULL, as long
2072 as it is a phi result of some other phi in *new_exit_bb that has
2075 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2076 or vmovdqu* for misaligned_operand.
2077 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2078 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2079 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2080 aligned_mem for AVX512F masked aligned load and store builtins and for
2083 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2085 PR tree-optimization/59651
2086 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2087 Address range for negative step should be added by TYPE_SIZE_UNIT.
2089 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2091 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2092 overlapping registers also for registers other than the stack pointer.
2094 2014-01-03 Marek Polacek <polacek@redhat.com>
2097 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2100 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2103 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2106 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2107 (push splitter): Use <P:MODE_SIZE> instead of
2108 GET_MODE_SIZE (<P:MODE>mode).
2109 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2110 (mov -1, reg peephole2): Likewise.
2111 * config/i386/sse.md (*mov<mode>_internal,
2112 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2113 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2114 *<code><mode>3, *andnot<mode>3<mask_name>,
2115 <mask_codefor><code><mode>3<mask_name>): Likewise.
2116 * config/i386/subst.md (mask_mode512bit_condition,
2117 sd_mask_mode512bit_condition): Likewise.
2119 2014-01-02 Xinliang David Li <davidxl@google.com>
2121 PR tree-optimization/59303
2122 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2123 (dump_predicates): Better output format.
2124 (pred_equal_p): New function.
2125 (is_neq_relop_p): Ditto.
2126 (is_neq_zero_form_p): Ditto.
2127 (pred_expr_equal_p): Ditto.
2128 (pred_neg_p): Ditto.
2129 (simplify_pred): Ditto.
2130 (simplify_preds_2): Ditto.
2131 (simplify_preds_3): Ditto.
2132 (simplify_preds_4): Ditto.
2133 (simplify_preds): Ditto.
2135 (push_to_worklist): Ditto.
2136 (get_pred_info_from_cmp): Ditto.
2137 (is_degenerated_phi): Ditto.
2138 (normalize_one_pred_1): Ditto.
2139 (normalize_one_pred): Ditto.
2140 (normalize_one_pred_chain): Ditto.
2141 (normalize_preds): Ditto.
2142 (normalize_cond_1): Remove function.
2143 (normalize_cond): Ditto.
2144 (is_gcond_subset_of): Ditto.
2145 (is_subset_of_any): Ditto.
2146 (is_or_set_subset_of): Ditto.
2147 (is_and_set_subset_of): Ditto.
2148 (is_norm_cond_subset_of): Ditto.
2149 (pred_chain_length_cmp): Ditto.
2150 (convert_control_dep_chain_into_preds): Type change.
2151 (find_predicates): Ditto.
2152 (find_def_preds): Ditto.
2153 (destroy_predicates_vecs): Ditto.
2154 (find_matching_predicates_in_rest_chains): Ditto.
2155 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2156 (is_pred_expr_subset): Ditto.
2157 (is_pred_chain_subset_of): Ditto.
2158 (is_included_in): Ditto.
2159 (is_superset_of): Ditto.
2161 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2163 Update copyright years.
2165 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2167 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2168 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2169 config/arc/arc.md, config/arc/arc.opt,
2170 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2171 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2172 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2173 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2174 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2175 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2176 vtable-verify.c, vtable-verify.h: Use the standard form for the
2179 2014-01-02 Tobias Burnus <burnus@net-b.de>
2181 * gcc.c (process_command): Update copyright notice dates.
2182 * gcov-dump.c: Ditto.
2184 * doc/cpp.texi: Bump @copying's copyright year.
2185 * doc/cppinternals.texi: Ditto.
2186 * doc/gcc.texi: Ditto.
2187 * doc/gccint.texi: Ditto.
2188 * doc/gcov.texi: Ditto.
2189 * doc/install.texi: Ditto.
2190 * doc/invoke.texi: Ditto.
2192 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2194 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2196 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2198 * config/i386/sse.md (*mov<mode>_internal): Guard
2199 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2201 PR rtl-optimization/59647
2202 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2203 new_rtx into UNSIGNED_FLOAT rtxes.
2205 Copyright (C) 2014 Free Software Foundation, Inc.
2207 Copying and distribution of this file, with or without modification,
2208 are permitted in any medium without royalty provided the copyright
2209 notice and this notice are preserved.