1 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4 * input.c (location_get_source_line): Bail out on when line number
5 is zero, and test the return value of
6 lookup_or_add_file_to_cache_tab.
8 2014-01-28 Richard Biener <rguenther@suse.de>
10 PR tree-optimization/58742
11 * tree-ssa-forwprop.c (associate_plusminus): Handle
12 pointer subtraction of the form (T)(P + A) - (T)P.
14 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
19 2014-01-28 Richard Biener <rguenther@suse.de>
22 2014-01-28 Richard Biener <rguenther@suse.de>
24 PR rtl-optimization/45364
25 PR rtl-optimization/59890
26 * var-tracking.c (local_get_addr_clear_given_value): Handle
28 (val_reset): Handle not allocated local_get_addr_cache.
29 (vt_find_locations): Use post-order on the inverted CFG.
31 2014-01-28 Richard Biener <rguenther@suse.de>
33 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
36 2014-01-28 Richard Biener <rguenther@suse.de>
38 PR rtl-optimization/45364
39 PR rtl-optimization/59890
40 * var-tracking.c (local_get_addr_clear_given_value): Handle
42 (val_reset): Handle not allocated local_get_addr_cache.
43 (vt_find_locations): Use post-order on the inverted CFG.
45 2014-01-28 Alan Modra <amodra@gmail.com>
47 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
48 * configure.ac <recursive call for build != host>: Define
49 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
51 * configure: Regenerate.
53 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
55 * config/i386/i386.c (get_builtin_code_for_version): Separate
56 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
57 Broadwell from Haswell.
59 2014-01-27 Steve Ellcey <sellcey@mips.com>
61 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
62 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
63 * config/mips/mips.c (mips_option_override): Change setting
65 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
66 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
67 Change from Mask to Var.
69 2014-01-27 Jeff Law <law@redhat.com>
71 * ipa-inline.c (inline_small_functions): Fix typo.
73 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
75 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
76 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
77 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
78 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
79 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
80 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
81 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
82 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
83 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
84 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
85 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
86 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
87 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
88 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
89 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
90 (_mm512_storeu_epi64): Ditto.
91 (_mm512_cmpge_epi32_mask): Ditto.
92 (_mm512_cmpge_epu32_mask): Ditto.
93 (_mm512_cmpge_epi64_mask): Ditto.
94 (_mm512_cmpge_epu64_mask): Ditto.
95 (_mm512_cmple_epi32_mask): Ditto.
96 (_mm512_cmple_epu32_mask): Ditto.
97 (_mm512_cmple_epi64_mask): Ditto.
98 (_mm512_cmple_epu64_mask): Ditto.
99 (_mm512_cmplt_epi32_mask): Ditto.
100 (_mm512_cmplt_epu32_mask): Ditto.
101 (_mm512_cmplt_epi64_mask): Ditto.
102 (_mm512_cmplt_epu64_mask): Ditto.
103 (_mm512_cmpneq_epi32_mask): Ditto.
104 (_mm512_cmpneq_epu32_mask): Ditto.
105 (_mm512_cmpneq_epi64_mask): Ditto.
106 (_mm512_cmpneq_epu64_mask): Ditto.
107 (_mm512_expand_pd): Ditto.
108 (_mm512_expand_ps): Ditto.
109 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
110 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
111 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
112 * config/i386/i386.c (ix86_builtins): Add
113 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
114 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
115 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
116 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
117 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
118 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
119 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
120 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
121 IX86_BUILTIN_PMOVUSQW512_MEM.
122 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
123 __builtin_ia32_pmovsqd512mem_mask,
124 __builtin_ia32_pmovqd512mem_mask,
125 __builtin_ia32_pmovusqw512mem_mask,
126 __builtin_ia32_pmovsqw512mem_mask,
127 __builtin_ia32_pmovqw512mem_mask,
128 __builtin_ia32_pmovusdw512mem_mask,
129 __builtin_ia32_pmovsdw512mem_mask,
130 __builtin_ia32_pmovdw512mem_mask,
131 __builtin_ia32_pmovqb512mem_mask,
132 __builtin_ia32_pmovusqb512mem_mask,
133 __builtin_ia32_pmovsqb512mem_mask,
134 __builtin_ia32_pmovusdb512mem_mask,
135 __builtin_ia32_pmovsdb512mem_mask,
136 __builtin_ia32_pmovdb512mem_mask.
137 (bdesc_args): Add __builtin_ia32_expanddf512,
138 __builtin_ia32_expandsf512.
139 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
140 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
141 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
142 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
143 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
144 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
145 (avx512f_<code>v8div16qi2_mask_store): This.
146 (avx512f_expand<mode>): New.
148 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
150 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
152 (_mm512_mask_prefetch_i64gather_pd): Ditto.
153 (_mm512_prefetch_i32scatter_pd): Ditto.
154 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
155 (_mm512_prefetch_i64scatter_pd): Ditto.
156 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
157 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
158 (_mm512_mask_prefetch_i64gather_ps): Ditto.
159 (_mm512_prefetch_i32scatter_ps): Ditto.
160 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
161 (_mm512_prefetch_i64scatter_ps): Ditto.
162 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
163 * config/i386/i386-builtin-types.def: Define
164 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
165 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
166 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
167 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
168 IX86_BUILTIN_SCATTERPFQPD.
169 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
170 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
171 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
172 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
173 __builtin_ia32_scatterpfqps.
174 (ix86_expand_builtin): Expand new built-ins.
175 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
176 fix memory access data type.
177 (*avx512pf_gatherpf<mode>_mask): Ditto.
178 (*avx512pf_gatherpf<mode>): Ditto.
179 (avx512pf_scatterpf<mode>): Ditto.
180 (*avx512pf_scatterpf<mode>_mask): Ditto.
181 (*avx512pf_scatterpf<mode>): Ditto.
182 (GATHER_SCATTER_SF_MEM_MODE): New.
183 (avx512pf_gatherpf<mode>df): Ditto.
184 (*avx512pf_gatherpf<mode>df_mask): Ditto.
185 (*avx512pf_scatterpf<mode>df): Ditto.
187 2014-01-27 Jakub Jelinek <jakub@redhat.com>
190 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
191 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
194 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
196 * common/config/arm/arm-common.c
197 (arm_rewrite_mcpu): Handle multiple names.
199 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
201 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
203 * gimple-builder.h (create_gimple_tmp): Delete.
205 2014-01-27 Christian Bruel <christian.bruel@st.com>
207 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
210 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
212 * config/pa/pa.md (call): Generate indirect long calls to non-local
213 functions when outputing 32-bit code.
214 (call_value): Likewise except for special call to buggy powf function.
216 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
217 portable runtime and PIC indirect calls.
218 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
219 and PIC call sequences. Use ldo instead of blr to set return register
220 in PIC call sequence.
222 2014-01-25 Walter Lee <walt@tilera.com>
224 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
225 avoid clobbering a live register.
227 2014-01-25 Walter Lee <walt@tilera.com>
229 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
230 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
231 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
232 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
234 2014-01-25 Walter Lee <walt@tilera.com>
236 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
237 arguments on even registers.
238 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
240 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
241 (BIGGEST_ALIGNMENT): Ditto.
242 (BIGGEST_FIELD_ALIGNMENT): Ditto.
244 2014-01-25 Walter Lee <walt@tilera.com>
246 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
247 insns before bundling.
248 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
250 2014-01-25 Walter Lee <walt@tilera.com>
252 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
253 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
254 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
256 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
258 * config/mips/constraints.md (kl): Delete.
259 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
260 define expands, using...
261 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
262 instructions for MIPS16.
263 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
264 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
266 2014-01-25 Walter Lee <walt@tilera.com>
268 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
272 2014-01-25 Walter Lee <walt@tilera.com>
274 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
275 (TARGET_EXPAND_TO_RTL_HOOK): Define.
277 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
279 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
282 2014-01-25 Jakub Jelinek <jakub@redhat.com>
284 * print-rtl.c (in_call_function_usage): New var.
285 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
286 EXPR_LIST mode as mode and not as reg note name.
289 * cfgloopmanip.c (copy_loop_info): If
290 loop->warned_aggressive_loop_optimizations, make sure
291 the flag is set in target loop too.
293 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
295 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
297 * builtins.def: Likewise.
298 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
299 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
300 * ira.c (ira_setup_eliminable_regset): Likewise.
301 * omp-low.c (gate_expand_omp): Likewise.
302 (execute_lower_omp): Likewise.
303 (diagnose_sb_0): Likewise.
304 (gate_diagnose_omp_blocks): Likewise.
305 (simd_clone_clauses_extract): Likewise.
308 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
310 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
311 correction for little endian...
312 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
315 2014-01-24 Jeff Law <law@redhat.com>
317 PR tree-optimization/59919
318 * tree-vrp.c (find_assert_locations_1): Do not register asserts
319 for non-returning calls.
321 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
323 * common/config/aarch64/aarch64-common.c
324 (aarch64_rewrite_mcpu): Handle multiple names.
325 * config/aarch64/aarch64.h
326 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
328 2014-01-24 Dodji Seketeli <dodji@redhat.com>
330 * input.c (add_file_to_cache_tab): Handle the case where fopen
333 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
336 * config/i386/i386.md (pushsf splitter): Get stack adjustment
337 from push operand if code of push isn't PRE_DEC.
339 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
342 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
343 -mquad-memory-atomic. Update -mquad-memory documentation to say
344 it is only used for non-atomic loads/stores.
346 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
347 -mquad-memory or -mquad-memory-atomic switches.
349 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
350 -mquad-memory-atomic to ISA 2.07 support.
352 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
353 to separate support of normal quad word memory operations (ldq, stq)
354 from the atomic quad word memory operations.
356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
357 support to separate non-atomic quad word operations from atomic
358 quad word operations. Disable non-atomic quad word operations in
359 little endian mode so that we don't have to swap words after the
360 load and before the store.
361 (quad_load_store_p): Add comment about atomic quad word support.
362 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
363 options printed with -mdebug=reg.
365 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
366 -mquad-memory-atomic as the test for whether we have quad word
368 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
369 or -mp8-vector are used, allow byte/half-word atomic operations.
371 * config/rs6000/sync.md (load_lockedti): Insure that the address
372 is a proper indexed or indirect address for the lqarx instruction.
373 On little endian systems, swap the hi/lo registers after the lqarx
375 (load_lockedpti): Use indexed_or_indirect_operand predicate to
376 insure the address is valid for the lqarx instruction.
377 (store_conditionalti): Insure that the address is a proper indexed
378 or indirect address for the stqcrx. instruction. On little endian
379 systems, swap the hi/lo registers before doing the stqcrx.
381 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
382 insure the address is valid for the stqcrx. instruction.
384 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
385 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
386 type of quad memory support is available.
388 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
391 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
392 there is a danger of looping.
394 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
396 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
397 force flag_ira_loop_pressure if set via command line.
399 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
401 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
402 (ashr_simd): New builtin handling DI mode.
403 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
404 (aarch64_sshr_simddi): New match pattern.
405 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
406 (vshrd_n_s64): Likewise.
407 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
409 2014-01-23 Nick Clifton <nickc@redhat.com>
411 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
412 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
413 favour of mcu specific scripts.
414 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
417 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
418 Alex Velenko <Alex.Velenko@arm.com>
420 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
421 (vaddv_s16): Likewise.
422 (vaddv_s32): Likewise.
423 (vaddv_u8): Likewise.
424 (vaddv_u16): Likewise.
425 (vaddv_u32): Likewise.
426 (vaddvq_s8): Likewise.
427 (vaddvq_s16): Likewise.
428 (vaddvq_s32): Likewise.
429 (vaddvq_s64): Likewise.
430 (vaddvq_u8): Likewise.
431 (vaddvq_u16): Likewise.
432 (vaddvq_u32): Likewise.
433 (vaddvq_u64): Likewise.
434 (vaddv_f32): Likewise.
435 (vaddvq_f32): Likewise.
436 (vaddvq_f64): Likewise.
437 (vmaxv_f32): Likewise.
438 (vmaxv_s8): Likewise.
439 (vmaxv_s16): Likewise.
440 (vmaxv_s32): Likewise.
441 (vmaxv_u8): Likewise.
442 (vmaxv_u16): Likewise.
443 (vmaxv_u32): Likewise.
444 (vmaxvq_f32): Likewise.
445 (vmaxvq_f64): Likewise.
446 (vmaxvq_s8): Likewise.
447 (vmaxvq_s16): Likewise.
448 (vmaxvq_s32): Likewise.
449 (vmaxvq_u8): Likewise.
450 (vmaxvq_u16): Likewise.
451 (vmaxvq_u32): Likewise.
452 (vmaxnmv_f32): Likewise.
453 (vmaxnmvq_f32): Likewise.
454 (vmaxnmvq_f64): Likewise.
455 (vminv_f32): Likewise.
456 (vminv_s8): Likewise.
457 (vminv_s16): Likewise.
458 (vminv_s32): Likewise.
459 (vminv_u8): Likewise.
460 (vminv_u16): Likewise.
461 (vminv_u32): Likewise.
462 (vminvq_f32): Likewise.
463 (vminvq_f64): Likewise.
464 (vminvq_s8): Likewise.
465 (vminvq_s16): Likewise.
466 (vminvq_s32): Likewise.
467 (vminvq_u8): Likewise.
468 (vminvq_u16): Likewise.
469 (vminvq_u32): Likewise.
470 (vminnmv_f32): Likewise.
471 (vminnmvq_f32): Likewise.
472 (vminnmvq_f64): Likewise.
474 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
476 * config/aarch64/aarch64-simd.md
477 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
478 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
479 (*aarch64_mul3_elt<mode>): Likewise.
480 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
481 (*aarch64_mul3_elt_to_64v2df): Likewise.
482 (*aarch64_mla_elt<mode>): Likewise.
483 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
484 (*aarch64_mls_elt<mode>): Likewise.
485 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
486 (*aarch64_fma4_elt<mode>): Likewise.
487 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
488 (*aarch64_fma4_elt_to_64v2df): Likewise.
489 (*aarch64_fnma4_elt<mode>): Likewise.
490 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
491 (*aarch64_fnma4_elt_to_64v2df): Likewise.
492 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
493 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
494 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
495 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
496 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
497 (aarch64_sqdmull_lane<mode>_internal): Likewise.
498 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
500 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
502 * config/aarch64/aarch64-simd.md
503 (aarch64_be_checked_get_lane<mode>): New define_expand.
504 * config/aarch64/aarch64-simd-builtins.def
505 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
506 New builtin definition.
507 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
508 Use new safe be builtin.
510 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
512 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
514 (aarch64_be_st1<mode>): Likewise.
515 (aarch_ld1<VALL:mode>): Define_expand modified.
516 (aarch_st1<VALL:mode>): Likewise.
517 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
518 (UNSPEC_ST1): Likewise.
520 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
522 * config/microblaze/microblaze.md: Add trap insn and attribute
524 2014-01-23 Dodji Seketeli <dodji@redhat.com>
526 PR preprocessor/58580
527 * input.h (location_get_source_line): Take an additional line_size
529 (void diagnostics_file_cache_fini): Declare new function.
530 * input.c (struct fcache): New type.
531 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
532 New static constants.
533 (diagnostic_file_cache_init, total_lines_num)
534 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
535 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
536 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
537 (get_next_line, read_next_line, goto_next_line, read_line_num):
538 New static function definitions.
539 (diagnostic_file_cache_fini): New function.
540 (location_get_source_line): Take an additional output line_len
541 parameter. Re-write using lookup_or_add_file_to_cache_tab and
543 * diagnostic.c (diagnostic_finish): Call
544 diagnostic_file_cache_fini.
545 (adjust_line): Take an additional input parameter for the length
546 of the line, rather than calculating it with strlen.
547 (diagnostic_show_locus): Adjust the use of
548 location_get_source_line and adjust_line with respect to their new
549 signature. While displaying a line now, do not stop at the first
550 null byte. Rather, display the zero byte as a space and keep
551 going until we reach the size of the line.
552 * Makefile.in: Add vec.o to OBJS-libcommon
554 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
555 Ilya Tocar <ilya.tocar@intel.com>
557 * config/i386/avx512fintrin.h (_mm512_kmov): New.
558 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
559 (__builtin_ia32_kmov16): Ditto.
560 * config/i386/i386.md (UNSPEC_KMOV): New.
563 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
565 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
566 (_mm512_storeu_si512): Ditto.
568 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
571 * rtl.h (get_referenced_operands): Declare.
572 * recog.c (get_referenced_operands): New function.
573 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
574 operands have been referenced when recording LO_SUM references.
576 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
578 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
580 2014-01-22 Jan Hubicka <jh@suse.cz>
582 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
583 Enable for generic and recent AMD targets.
585 2014-01-22 Jan Hubicka <jh@suse.cz>
587 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
588 ARG_SIZE note when adjustment was eliminated.
590 2014-01-22 Jeff Law <law@redhat.com>
592 PR tree-optimization/59597
593 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
594 in file. Accept new argument REGISTERING and use it to modify
595 dump output appropriately.
596 (register_jump_thread): Corresponding changes.
597 (mark_threaded_blocks): Reinstate code to cancel unprofitable
598 thread paths involving joiner blocks. Add code to dump cancelled
599 jump threading paths.
601 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
603 PR rtl-optimization/59477
604 * lra-constraints.c (inherit_in_ebb): Process call for living hard
605 regs. Update reloads_num and potential_reload_hard_regs for all insns.
607 2014-01-22 Tom Tromey <tromey@redhat.com>
609 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
611 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
613 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
615 PR rtl-optimization/59896
616 * lra-constraints.c (process_alt_operands): Check unused note for
617 matched operands of insn with no output reloads.
619 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
621 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
622 (mips_move_from_gpr_cost): Likewise.
624 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
626 PR rtl-optimization/59858
627 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
628 ira_class_hard_regs_num.
629 (process_alt_operands): Increase reject for dying matched operand.
631 2014-01-21 Jakub Jelinek <jakub@redhat.com>
634 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
635 smaller than size, perform several stores or loads and stores
636 at dst + count - size to store or copy all of size bytes, rather
637 than just last modesize bytes.
639 2014-01-20 DJ Delorie <dj@redhat.com>
641 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
642 that CLOBBERs are REGs before propogating their values.
644 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
647 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
648 (cgraph_inline_failed_type): New function.
649 * cgraph.h (DEFCIFCODE): Add type.
650 (cgraph_inline_failed_type_t): New enum.
651 (cgraph_inline_failed_type): New prototype.
652 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
653 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
654 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
655 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
656 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
657 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
658 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
659 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
660 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
661 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
662 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
663 OPTIMIZATION_MISMATCH.
664 * tree-inline.c (expand_call_inline): Emit errors during
665 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
667 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
670 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
671 mode attribute in insn output.
673 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
675 * output.h (output_constant): Delete.
676 * varasm.c (output_constant): Make private.
678 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
680 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
682 2014-01-20 Jakub Jelinek <jakub@redhat.com>
685 * tree.h (fold_builtin_strcat): New prototype.
686 * builtins.c (fold_builtin_strcat): No longer static. Add len
687 argument, if non-NULL, don't call c_strlen. Optimize
688 directly into __builtin_memcpy instead of __builtin_strcpy.
689 (fold_builtin_2): Adjust fold_builtin_strcat caller.
690 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
692 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
694 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
695 for SImode_address_operand operands, having only a REG argument.
697 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
699 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
700 loader name using mbig-endian.
701 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
703 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
705 * doc/invoke.texi (-march): Clarify documentation for AArch64.
709 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
711 * config/aarch64/aarch64-protos.h
712 (aarch64_cannot_change_mode_class_ptr): Declare.
713 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
714 aarch64_cannot_change_mode_class_ptr): New.
715 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
716 backend hook aarch64_cannot_change_mode_class.
718 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
720 * common/config/aarch64/aarch64-common.c
721 (aarch64_handle_option): Don't handle any option order logic here.
722 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
723 selected_cpu, warn on architecture version mismatch.
724 (aarch64_override_options): Fix parsing order for option strings.
726 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
727 Iain Sandoe <iain@codesourcery.com>
730 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
731 warning. Amend comment to reflect current functionality.
733 2014-01-20 Richard Biener <rguenther@suse.de>
736 * builtins.c (fold_builtin_strcat): Remove case better handled
737 by tree-ssa-strlen.c.
739 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
741 * config/aarch64/aarch64.opt
742 (mcpu, march, mtune): Make case-insensitive.
744 2014-01-20 Jakub Jelinek <jakub@redhat.com>
747 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
748 if operands[1] is a REG or ZERO_EXTEND of a REG.
750 2014-01-19 Jan Hubicka <jh@suse.cz>
752 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
754 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
756 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
757 long non-pic millicode calls.
759 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
761 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
763 2014-01-19 Kito Cheng <kito@0xlab.org>
765 * builtins.c (expand_movstr): Check movstr expand done or fail.
767 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
768 H.J. Lu <hongjiu.lu@intel.com>
771 * config/i386/i386.md (*lea<mode>): Zero-extend return register
772 to DImode for zero-extended addresses.
774 2014-01-19 Jakub Jelinek <jakub@redhat.com>
776 PR rtl-optimization/57763
777 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
778 on the new indirect jump_insn and increment LABEL_NUSES (label).
780 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
784 * config.gcc (x86_archs): New variable.
785 (x86_64_archs): Likewise.
786 (x86_cpus): Likewise.
787 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
788 --with-arch/--with-cpu= options.
789 Support --with-arch=/--with-cpu={nehalem,westmere,
790 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
792 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
794 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
795 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
797 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
799 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
801 2014-01-18 Jakub Jelinek <jakub@redhat.com>
804 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
805 clear cpp_get_options (parse_in)->warn_unused_macros for
806 ix86_target_macros_internal with cpp_define.
808 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
810 * jump.c (delete_related_insns): Keep (use (insn))s.
811 * reorg.c (redundant_insn): Check for barriers too.
813 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
815 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
817 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
819 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
820 call to $$dyncall when TARGET_LONG_CALLS is true.
822 2014-01-17 Jeff Law <law@redhat.com>
824 * ree.c (combine_set_extension): Temporarily disable test for
825 changing number of hard registers.
827 2014-01-17 Jan Hubicka <jh@suse.cz>
830 * ipa-inline-analysis.c (inline_free_summary):
831 Do not free summary of aliases.
833 2014-01-17 Jakub Jelinek <jakub@redhat.com>
836 * gimplify.c (gimplify_expr): Use create_tmp_var
837 instead of create_tmp_var_raw. If cond doesn't have
838 integral type, don't add the IFN_ANNOTATE builtin at all.
840 2014-01-17 Martin Jambor <mjambor@suse.cz>
843 * ipa-cp.c (prev_edge_clone): New variable.
844 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
845 Also resize prev_edge_clone vector.
846 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
847 (ipcp_edge_removal_hook): New function.
848 (ipcp_driver): Register ipcp_edge_removal_hook.
850 2014-01-17 Andrew Pinski <apinski@cavium.com>
851 Steve Ellcey <sellcey@mips.com>
854 * config/mips/mips.c (mips_print_operand): Check operand mode instead
857 2014-01-17 Jeff Law <law@redhat.com>
860 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
861 so that pass_ccp runs first.
863 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
865 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
866 (ix86_adjust_cost): Use !TARGET_XXX.
867 (do_reorder_for_imul): Likewise.
868 (swap_top_of_ready_list): Likewise.
869 (ix86_sched_reorder): Likewise.
871 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
873 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
874 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
875 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
876 (intel_memset): New. Duplicate slm_memset.
877 (intel_cost): New. Duplicate slm_cost.
878 (m_INTEL): New macro.
879 (processor_target_table): Add "intel".
880 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
881 with PROCESSOR_INTEL for "intel".
882 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
883 PROCESSOR_SILVERMONT.
884 (ix86_issue_rate): Likewise.
885 (ix86_adjust_cost): Likewise.
886 (ia32_multipass_dfa_lookahead): Likewise.
887 (swap_top_of_ready_list): Likewise.
888 (ix86_sched_reorder): Likewise.
889 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
890 instead of TARGET_OPT_AGU.
891 * config/i386/i386.h (TARGET_INTEL): New.
892 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
893 (processor_type): Add PROCESSOR_INTEL.
894 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
895 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
897 2014-01-17 Marek Polacek <polacek@redhat.com>
900 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
903 2014-01-17 Richard Biener <rguenther@suse.de>
905 PR tree-optimization/46590
906 * opts.c (default_options_table): Add entries for
907 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
908 all enabled at -O1 but not for -Og.
909 * common.opt (fbranch-count-reg): Remove Init(1).
910 (fmove-loop-invariants): Likewise.
911 (ftree-pta): Likewise.
913 2014-01-17 Jakub Jelinek <jakub@redhat.com>
915 * config/i386/i386.c (ix86_data_alignment): For compatibility with
916 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
917 decls to at least the GCC 4.8 used alignments.
920 * tree-nested.c (convert_nonlocal_reference_stmt,
921 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
922 of GIMPLE_BIND stmts, adjust associated decls.
924 2014-01-17 Richard Biener <rguenther@suse.de>
926 PR tree-optimization/46590
927 * vec.h (vec<>::bseach): New member function implementing
928 binary search according to C89 bsearch.
929 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
930 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
931 bitmap pointer again. Make accesses_in_loop a flat array.
932 (mem_ref_obstack): New global.
933 (outermost_indep_loop): Adjust for mem_ref->stored changes.
934 (mark_ref_stored): Likewise.
935 (ref_indep_loop_p_2): Likewise.
936 (set_ref_stored_in_loop): New helper function.
937 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
938 (memref_free): Adjust.
939 (record_mem_ref_loc): Simplify.
940 (gather_mem_refs_stmt): Adjust.
941 (sort_locs_in_loop_postorder_cmp): New function.
942 (analyze_memory_references): Sort accesses_in_loop after
943 loop postorder number.
944 (find_ref_loc_in_loop_cmp): New function.
945 (for_all_locs_in_loop): Find relevant cluster of locs in
946 accesses_in_loop and iterate without recursion.
947 (execute_sm): Avoid uninit warning.
948 (struct ref_always_accessed): Simplify.
949 (ref_always_accessed::operator ()): Likewise.
950 (ref_always_accessed_p): Likewise.
951 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
952 loop postorder numbers here.
953 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
956 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
959 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
960 on decls for which assemble_alias has been called.
962 2014-01-17 Nick Clifton <nickc@redhat.com>
964 * config/msp430/msp430.opt: (mcpu): New option.
965 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
966 (msp430_option_override): Parse target_cpu. If the MCU name
967 matches a generic string, clear target_mcu.
968 (msp430_attr): Allow numeric interrupt values up to 63.
969 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
970 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
972 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
974 * config/msp430/msp430.md (popm): Use %J rather than %I.
975 (addsi3): Use msp430_nonimmediate_operand for operand 2.
976 (addhi_cy_i): Use immediate_operand for operand 2.
977 * doc/invoke.texi: Document -mcpu option.
979 2014-01-17 Richard Biener <rguenther@suse.de>
981 PR rtl-optimization/38518
982 * df.h (df_analyze_loop): Declare.
983 * df-core.c: Include cfgloop.h.
984 (df_analyze_1): Split out main part of df_analyze.
985 (df_analyze): Adjust.
986 (loop_inverted_post_order_compute): New function.
987 (loop_post_order_compute): Likewise.
988 (df_analyze_loop): New function avoiding whole-function
990 * loop-invariant.c (find_defs): Use df_analyze_loop.
991 (find_invariants): Adjust.
992 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
994 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
996 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
997 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
999 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1001 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1002 traversal when removing references.
1004 2014-01-16 Jan Hubicka <jh@suse.cz>
1007 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1009 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1012 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1013 pushing a reload for an autoinc when we had previously reloaded an
1014 inner part of the address.
1016 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1018 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1020 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1021 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1022 when not giving up or versioning for alias only because of
1024 (vect_analyze_data_ref_dependences): Set to true.
1025 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1027 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1028 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1032 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1035 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1036 operand 0 predicate for gathers, use a new pseudo as subtarget.
1038 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1041 * lra-constraints.c (process_alt_operands): Add printing debug info.
1042 Check absence of input/output reloads for matched operands too.
1044 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1046 PR rtl-optimization/59835
1047 * ira.c (ira_init_register_move_cost): Increase cost for
1050 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1052 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1054 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1057 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1058 non-register objects. Use gen_(high/low)part more consistently.
1061 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1064 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1065 endian support, remove tests for WORDS_BIG_ENDIAN.
1066 (p8_mfvsrd_3_<mode>): Likewise.
1067 (reload_gpr_from_vsx<mode>): Likewise.
1068 (reload_gpr_from_vsxsf): Likewise.
1069 (p8_mfvsrd_4_disf): Likewise.
1071 2014-01-16 Richard Biener <rguenther@suse.de>
1073 PR rtl-optimization/46590
1074 * lcm.c (compute_antinout_edge): Use postorder iteration.
1075 (compute_laterin): Use inverted postorder iteration.
1077 2014-01-16 Nick Clifton <nickc@redhat.com>
1080 * varasm.c (output_constant): Return the number of bytes actually
1082 (output_constructor_array_range): Update the field size with the
1083 number of bytes emitted by output_constant.
1084 (output_constructor_regular_field): Likewise. Also do not
1085 complain if the total number of bytes emitted is now greater
1086 than the expected fieldpos.
1087 * output.h (output_constant): Update prototype and descriptive comment.
1089 2014-01-16 Marek Polacek <polacek@redhat.com>
1092 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1093 it is error_mark_node.
1095 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1097 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1098 VALID_AVX256_REG_OR_OI_MODE.
1100 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1102 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1103 current procedure should be profiled.
1105 2014-01-15 Andrew Pinski <apinski@cavium.com>
1107 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1108 of moving from/to the STACK_REG register class.
1110 2014-01-15 Richard Henderson <rth@redhat.com>
1113 * reginfo.c (global_regs_decl): Globalize.
1114 * rtl.h (global_regs_decl): Declare.
1115 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1116 reserved via global_regs.
1118 2014-01-15 Teresa Johnson <tejohnson@google.com>
1120 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1122 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1124 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1125 and vmulosh rather than call gen_vec_widen_smult_*.
1126 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1127 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1128 (vec_widen_smult_even_v16qi): Likewise.
1129 (vec_widen_umult_even_v8hi): Likewise.
1130 (vec_widen_smult_even_v8hi): Likewise.
1131 (vec_widen_umult_odd_v16qi): Likewise.
1132 (vec_widen_smult_odd_v16qi): Likewise.
1133 (vec_widen_umult_odd_v8hi): Likewise.
1134 (vec_widen_smult_odd_v8hi): Likewise.
1135 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1136 vmuloub rather than call gen_vec_widen_umult_*.
1137 (vec_widen_umult_lo_v16qi): Likewise.
1138 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1139 vmulosb rather than call gen_vec_widen_smult_*.
1140 (vec_widen_smult_lo_v16qi): Likewise.
1141 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1142 rather than call gen_vec_widen_umult_*.
1143 (vec_widen_umult_lo_v8hi): Likewise.
1144 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1145 rather than call gen_vec_widen_smult_*.
1146 (vec_widen_smult_lo_v8hi): Likewise.
1148 2014-01-15 Jeff Law <law@redhat.com>
1150 PR tree-optimization/59747
1151 * ree.c (find_and_remove_re): Properly handle case where a second
1152 eliminated extension requires widening a copy created for elimination
1153 of a prior extension.
1154 (combine_set_extension): Ensure that the number of hard regs needed
1155 for a destination register does not change when we widen it.
1157 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1159 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1160 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1161 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1162 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1163 (avr-*-rtems*): Likewise.
1164 (bfin*-rtems*): Likewise.
1165 (moxie-*-rtems*): Likewise.
1166 (h8300-*-rtems*): Likewise.
1167 (i[34567]86-*-rtems*): Likewise.
1168 (lm32-*-rtems*): Likewise.
1169 (m32r-*-rtems*): Likewise.
1170 (m68k-*-rtems*): Likewise.
1171 (microblaze*-*-rtems*): Likewise.
1172 (mips*-*-rtems*): Likewise.
1173 (powerpc-*-rtems*): Likewise.
1174 (sh-*-rtems*): Likewise.
1175 (sparc-*-rtems*): Likewise.
1176 (sparc64-*-rtems*): Likewise.
1177 (v850-*-rtems*): Likewise.
1178 (m32c-*-rtems*): Likewise.
1180 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1182 PR rtl-optimization/59511
1183 * ira.c (ira_init_register_move_cost): Use memory costs for some
1184 cases of register move cost calculations.
1185 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1186 instead of BB frequency.
1187 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1188 * lra-assigns.c (find_hard_regno_for): Ditto.
1190 2014-01-15 Richard Biener <rguenther@suse.de>
1192 PR tree-optimization/59822
1193 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1194 (vectorizable_load): Use it to hoist defs of uses of invariant
1195 loads out of the loop.
1197 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1198 Kugan Vivekanandarajah <kuganv@linaro.org>
1201 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1204 2014-01-15 Richard Biener <rguenther@suse.de>
1206 PR rtl-optimization/59802
1207 * lcm.c (compute_available): Use inverted postorder to seed
1208 the initial worklist.
1210 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1213 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1214 ADDR_REGS for invalid symrefs in non-PIC code.
1216 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1219 * builtins.c (determine_block_size): Initialize *probable_max_size
1220 even if len_rtx is CONST_INT.
1222 2014-01-14 Andrew Pinski <apinski@cavium.com>
1224 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1225 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1226 (cortexa53_tunings): Likewise.
1227 (aarch64_sched_issue_rate): New function.
1228 (TARGET_SCHED_ISSUE_RATE): Define.
1230 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1232 * ira-costs.c (find_costs_and_classes): Add missed
1233 ira_init_register_move_cost_if_necessary.
1235 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1238 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1240 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1243 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1244 to indicate if type is used for function return value. Warn ABI
1245 change if the vector mode isn't available for function return value.
1246 (ix86_function_arg_advance): Pass false to type_natural_mode.
1247 (ix86_function_arg): Likewise.
1248 (ix86_gimplify_va_arg): Likewise.
1249 (function_arg_32): Don't warn ABI change.
1250 (ix86_function_value): Pass true to type_natural_mode.
1251 (ix86_return_in_memory): Likewise.
1252 (ix86_struct_value_rtx): Removed.
1253 (TARGET_STRUCT_VALUE_RTX): Likewise.
1255 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1257 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1258 converting a conditional jump into a conditional return.
1260 2014-01-14 Richard Biener <rguenther@suse.de>
1262 PR tree-optimization/58921
1263 PR tree-optimization/59006
1264 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1265 hoisting invariant stmts.
1266 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1267 invariant loads on the preheader edge if possible.
1269 2014-01-14 Joey Ye <joey.ye@arm.com>
1271 * doc/plugin.texi (Building GCC plugins): Update to C++.
1273 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1275 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1276 (_mm_rcp28_round_ss): Ditto.
1277 (_mm_rsqrt28_round_sd): Ditto.
1278 (_mm_rsqrt28_round_ss): Ditto.
1279 (_mm_rcp28_sd): Ditto.
1280 (_mm_rcp28_ss): Ditto.
1281 (_mm_rsqrt28_sd): Ditto.
1282 (_mm_rsqrt28_ss): Ditto.
1283 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1284 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1285 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1286 (IX86_BUILTIN_RCP28SD): Ditto.
1287 (IX86_BUILTIN_RCP28SS): Ditto.
1288 (IX86_BUILTIN_RSQRT28SD): Ditto.
1289 (IX86_BUILTIN_RSQRT28SS): Ditto.
1290 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1291 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1292 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1293 (ix86_expand_special_args_builtin): Expand new FTYPE.
1294 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1295 (srcp14<mode>): Make insn unary.
1296 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1297 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1298 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1299 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1300 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1301 Fix rounding: make it SAE only.
1302 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1304 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1306 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1307 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1308 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1309 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1310 (round_saeonly_mask_scalar_operand4): Ditto.
1311 (round_saeonly_mask_scalar_op3): Ditto.
1312 (round_saeonly_mask_scalar_op4): Ditto.
1314 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1316 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1317 Implement -maltivec=be for vec_insert and vec_extract.
1319 2014-01-10 DJ Delorie <dj@redhat.com>
1321 * config/msp430/msp430.md (call_internal): Don't allow memory
1322 references with SP as the base register.
1323 (call_value_internal): Likewise.
1324 * config/msp430/constraints.md (Yc): New. For memory references
1325 that don't use SP as a base register.
1327 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1328 "an integer without a # prefix"
1329 * config/msp430/msp430.md (epilogue_helper): Use it.
1331 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1334 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1335 AVX512F gather builtins.
1336 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1337 on gather decls with INTEGER_TYPE masktype.
1338 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1339 directly into the builtin rather than hoisting it before loop.
1341 PR tree-optimization/59387
1342 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1343 (scev_const_prop): If folded_casts and type has undefined overflow,
1344 use force_gimple_operand instead of force_gimple_operand_gsi and
1345 for each added stmt if it is assign with
1346 arith_code_with_undefined_signed_overflow, call
1347 rewrite_to_defined_overflow.
1348 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1349 gimple-fold.h instead.
1350 (arith_code_with_undefined_signed_overflow,
1351 rewrite_to_defined_overflow): Moved to ...
1352 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1353 rewrite_to_defined_overflow): ... here. No longer static.
1354 Include gimplify-me.h.
1355 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1356 rewrite_to_defined_overflow): New prototypes.
1358 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1360 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1362 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1364 * builtins.c (get_object_alignment_2): Minor tweak.
1365 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1367 2014-01-13 Christian Bruel <christian.bruel@st.com>
1369 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1370 optimized non constant lengths.
1372 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1375 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1376 load as __atomic_load_N if possible.
1378 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1380 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1382 (rs6000_expand_builtin): Adjust call.
1384 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1387 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1388 * config/rs6000/rs6000.c: Include target-globals.h.
1389 (rs6000_set_current_function): Instead of doing target_reinit
1390 unconditionally, use save_target_globals_default_opts and
1391 restore_target_globals.
1393 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1395 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1396 (rs6000_expand_builtin): Handle mffs and mtfsf.
1397 (rs6000_init_builtins): Define mffs and mtfsf.
1398 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1399 (rs6000_mffs): New pattern.
1400 (rs6000_mtfsf): New pattern.
1402 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1404 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1405 Start narrowing with START. Apply candidate-use pair
1406 and check overall cost in narrowing.
1407 (iv_ca_prune): Pass new argument.
1409 2014-01-10 Jeff Law <law@redhat.com>
1412 * ree.c (combine_reaching_defs): Ensure the defining statement
1413 occurs before the extension when optimizing extensions with
1414 different source and destination hard registers.
1416 2014-01-10 Jan Hubicka <jh@suse.cz>
1419 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1420 vtables into the type inheritance graph.
1422 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1424 PR rtl-optimization/59754
1425 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1426 modes in the REGNO != REGNO case.
1428 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1430 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1432 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1434 PR tree-optimization/59745
1435 * tree-predcom.c (tree_predictive_commoning_loop): Call
1436 free_affine_expand_cache if giving up because components is NULL.
1438 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1439 GC in payload of target_globals struct instead of allocating them on
1440 the heap and the larger structs separately using GC.
1441 * target-globals.h (struct target_globals): Make regs, hard_regs,
1442 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1443 of GTY((skip)) and change type to void *.
1444 (reset_target_globals): Cast loads from those fields to corresponding
1447 2014-01-10 Steve Ellcey <sellcey@mips.com>
1450 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1451 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1452 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1454 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1457 * aarch64-modes.def (CC_Zmode): New flags mode.
1458 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1459 represents an equality.
1460 (aarch64_get_condition_code): Handle CC_Zmode.
1461 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1463 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1465 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1466 extraction in good case.
1468 2014-01-10 Richard Biener <rguenther@suse.de>
1470 PR tree-optimization/59374
1471 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1472 checking after SLP discovery. Mark stmts not participating
1473 in any SLP instance properly.
1475 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1477 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1478 when handling a SET rtx.
1480 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1482 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1483 (cortex-a57): Likewise.
1484 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1486 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1488 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1489 non-iwmmxt builtins.
1491 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1495 * ipa-devirt.c record_target_from_binfo): Take as argument
1496 stack of binfos and lookup matching one for virtual inheritance.
1497 (possible_polymorphic_call_targets_1): Update.
1499 2014-01-10 Huacai Chen <chenhc@lemote.com>
1501 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1502 kernel strings for Loongson-2E/2F/3A.
1504 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1507 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1508 is_gimple_call before calling gimple_call_internal_p.
1510 2014-01-09 Steve Ellcey <sellcey@mips.com>
1512 * Makefile.in (TREE_FLOW_H): Remove.
1513 (TREE_SSA_H): Add file names from tree-flow.h.
1514 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1515 * tree.h: Remove tree-flow.h reference.
1516 * hash-table.h: Remove tree-flow.h reference.
1517 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1518 reference with tree-ssa-loop.h.
1520 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1522 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1523 default element-order behavior for -maltivec.
1524 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1525 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1526 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1527 when targeting big endian, at least for now.
1528 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1530 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1533 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1534 var satisfies use_register_for_decl, just take into account type
1535 alignment, rather than decl alignment.
1537 PR tree-optimization/59622
1538 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1539 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1540 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1541 Don't devirtualize for inplace at all. For targets.length () == 1,
1542 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1544 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1546 * config/i386/i386.md (cpu): Remove the unused btver1.
1548 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1550 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1552 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1555 * tree-core.h (struct target_globals): New forward declaration.
1556 (struct tree_target_option): Add globals field.
1557 * tree.h (TREE_TARGET_GLOBALS): Define.
1558 (prepare_target_option_nodes_for_pch): New prototype.
1559 * target-globals.h (struct target_globals): Define even if
1561 * tree.c (prepare_target_option_node_for_pch,
1562 prepare_target_option_nodes_for_pch): New functions.
1563 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1564 * config/i386/i386.c: Include target-globals.h.
1565 (ix86_set_current_function): Instead of doing target_reinit
1566 unconditionally, use save_target_globals_default_opts and
1567 restore_target_globals.
1569 2014-01-09 Richard Biener <rguenther@suse.de>
1571 PR tree-optimization/59715
1572 * tree-cfg.h (split_critical_edges): Declare.
1573 * tree-cfg.c (split_critical_edges): Export.
1574 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1576 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1578 * cfgexpand.c (expand_stack_vars): Optionally disable
1579 asan stack protection.
1580 (expand_used_vars): Likewise.
1581 (partition_stack_vars): Likewise.
1582 * asan.c (asan_emit_stack_protection): Optionally disable
1583 after return stack usage.
1584 (instrument_derefs): Optionally disable memory access instrumentation.
1585 (instrument_builtin_call): Likewise.
1586 (instrument_strlen_call): Likewise.
1587 (asan_protect_global): Optionally disable global variables protection.
1588 * doc/invoke.texi: Added doc for new options.
1589 * params.def: Added new options.
1590 * params.h: Likewise.
1592 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1594 PR rtl-optimization/59724
1595 * ifcvt.c (cond_exec_process_if_block): Don't call
1596 flow_find_head_matching_sequence with 0 longest_match.
1597 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1598 non-active insns if !stop_after.
1599 (try_head_merge_bb): Revert 2014-01-07 changes.
1601 2014-01-08 Jeff Law <law@redhat.com>
1603 * ree.c (get_sub_rtx): New function, extracted from...
1604 (merge_def_and_ext): Here.
1605 (combine_reaching_defs): Use get_sub_rtx.
1607 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1609 * cgraph.h (varpool_variable_node): Do not choke on null node.
1611 2014-01-08 Catherine Moore <clm@codesourcery.com>
1613 * config/mips/mips.md (simple_return): Attempt to use JRC
1615 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1617 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1619 PR rtl-optimization/59137
1620 * reorg.c (steal_delay_list_from_target): Call update_block for
1622 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1624 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1627 two duplicate entries.
1629 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1632 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1634 * config/mips/mips.c (mips_truncated_op_cost): New function.
1635 (mips_rtx_costs): Adjust test for BADDU.
1636 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1638 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1640 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1641 (*baddu_si): ...this new pattern.
1643 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1646 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1648 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1651 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1653 (expand_expr, expand_normal): Adjust.
1654 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1655 inner_reference_p. Use inner_reference_p to expand inner references.
1656 (store_expr): Adjust.
1657 * cfgexpand.c (expand_call_stmt): Adjust.
1659 2014-01-08 Rong Xu <xur@google.com>
1661 * gcov-io.c (gcov_var): Move from gcov-io.h.
1662 (gcov_position): Ditto.
1663 (gcov_is_error): Ditto.
1664 (gcov_rewrite): Ditto.
1665 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1666 only part to libgcc/libgcov.h.
1668 2014-01-08 Marek Polacek <polacek@redhat.com>
1671 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1673 2014-01-08 Marek Polacek <polacek@redhat.com>
1676 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1678 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1680 PR rtl-optimization/59649
1681 * stor-layout.c (get_mode_bounds): For BImode return
1682 0 and STORE_FLAG_VALUE.
1684 2014-01-08 Richard Biener <rguenther@suse.de>
1687 * gimple.h (is_gimple_builtin_call): Remove.
1688 (gimple_builtin_call_types_compatible_p): New.
1689 (gimple_call_builtin_p): New overload.
1690 * gimple.c (is_gimple_builtin_call): Remove.
1691 (validate_call): Rename to ...
1692 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1694 (validate_type): New static function.
1695 (gimple_call_builtin_p): New overload and adjust.
1696 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1697 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1698 (gimple_fold_stmt_to_constant_1): Likewise.
1699 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1701 2014-01-08 Richard Biener <rguenther@suse.de>
1704 * gimplify.c (gimplify_expr): Gimplify register-register type
1705 VIEW_CONVERT_EXPRs to separate stmts.
1707 2014-01-07 Jeff Law <law@redhat.com>
1710 * ree.c (combine_set_extension): Handle case where source
1711 and destination registers in an extension insn are different.
1712 (combine_reaching_defs): Allow source and destination registers
1713 in extension to be different under limited circumstances.
1714 (add_removable_extension): Remove restriction that the
1715 source and destination registers in the extension are the same.
1716 (find_and_remove_re): Emit a copy from the extension's
1717 destination to its source after the defining insn if
1718 the source and destination registers are different.
1721 * ifcvt.c (merge_if_block): If we are merging a block with more than
1722 one successor with a block with no successors, remove any BARRIER
1723 after the second block.
1725 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1727 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1729 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1732 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1733 for 14-bit register offsets when INT14_OK_STRICT is false.
1735 2014-01-07 Roland Stigge <stigge@antcom.de>
1736 Michael Meissner <meissner@linux.vnet.ibm.com>
1739 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1740 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1742 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1744 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1747 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1749 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1750 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1751 rtx is const0_rtx or not.
1753 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1756 * target-globals.c (save_target_globals): Remove this_fn_optab
1758 * toplev.c: Include optabs.h.
1759 (target_reinit): Temporarily restore the global options if another
1760 set of options are in force.
1762 2014-01-07 Jakub Jelinek <jakub@redhat.com>
1764 PR rtl-optimization/58668
1765 * cfgcleanup.c (flow_find_cross_jump): Don't count
1766 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
1767 to determine what is counted.
1768 (flow_find_head_matching_sequence): Use active_insn_p to determine
1770 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1772 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1773 determine what is counted.
1775 PR tree-optimization/59643
1776 * tree-predcom.c (split_data_refs_to_components): If one dr is
1777 read and one write, determine_offset fails and the write isn't
1778 in the bad component, just put the read into the bad component.
1780 2014-01-07 Mike Stump <mikestump@comcast.net>
1781 Jakub Jelinek <jakub@redhat.com>
1784 * tree-core.h (struct tree_optimization_option): Change optabs
1785 type from unsigned char * to void *.
1786 * optabs.c (init_tree_optimization_optabs): Adjust
1787 TREE_OPTIMIZATION_OPTABS initialization.
1789 2014-01-06 Jakub Jelinek <jakub@redhat.com>
1792 * config/i386/i386.h (struct machine_function): Add
1793 no_drap_save_restore field.
1794 * config/i386/i386.c (ix86_save_reg): Use
1795 !cfun->machine->no_drap_save_restore instead of
1796 crtl->stack_realign_needed.
1797 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
1798 this function clears frame_pointer_needed. Set
1799 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
1800 and DRAP reg is needed.
1802 2014-01-06 Marek Polacek <polacek@redhat.com>
1805 * doc/implement-c.texi: Mention that other integer types are
1806 permitted as bit-field types in strictly conforming mode.
1808 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
1810 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
1813 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
1815 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
1817 2014-01-06 Martin Jambor <mjambor@suse.cz>
1820 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
1822 * ipa-prop.c (ipa_print_node_params): Fix indentation.
1824 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
1828 * var-tracking.c (add_stores): Preserve the value of the source even if
1829 we don't record the store.
1831 2014-01-06 Terry Guo <terry.guo@arm.com>
1833 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
1835 2014-01-05 Iain Sandoe <iain@codesourcery.com>
1838 * config/darwin.c (darwin_function_section): Adjust return values to
1839 correspond to optimisation changes made in r206070.
1841 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
1843 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
1844 from prefetch_block tune setting.
1845 (nocona_cost): Correct size of prefetch block to 64.
1847 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
1849 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
1850 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
1851 used to save the static chain register in the computation of the offset
1852 from which the FP registers need to be restored.
1854 2014-01-04 Jakub Jelinek <jakub@redhat.com>
1856 PR tree-optimization/59519
1857 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
1858 ICE if get_current_def (current_new_name) is already non-NULL, as long
1859 as it is a phi result of some other phi in *new_exit_bb that has
1862 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
1863 or vmovdqu* for misaligned_operand.
1864 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
1865 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
1866 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
1867 aligned_mem for AVX512F masked aligned load and store builtins and for
1870 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
1872 PR tree-optimization/59651
1873 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
1874 Address range for negative step should be added by TYPE_SIZE_UNIT.
1876 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
1878 * config/m68k/m68k.c (handle_move_double): Handle pushes with
1879 overlapping registers also for registers other than the stack pointer.
1881 2014-01-03 Marek Polacek <polacek@redhat.com>
1884 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
1887 2014-01-03 Jakub Jelinek <jakub@redhat.com>
1890 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
1893 * config/i386/i386.md (MODE_SIZE): New mode attribute.
1894 (push splitter): Use <P:MODE_SIZE> instead of
1895 GET_MODE_SIZE (<P:MODE>mode).
1896 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
1897 (mov -1, reg peephole2): Likewise.
1898 * config/i386/sse.md (*mov<mode>_internal,
1899 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
1900 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
1901 *<code><mode>3, *andnot<mode>3<mask_name>,
1902 <mask_codefor><code><mode>3<mask_name>): Likewise.
1903 * config/i386/subst.md (mask_mode512bit_condition,
1904 sd_mask_mode512bit_condition): Likewise.
1906 2014-01-02 Xinliang David Li <davidxl@google.com>
1908 PR tree-optimization/59303
1909 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
1910 (dump_predicates): Better output format.
1911 (pred_equal_p): New function.
1912 (is_neq_relop_p): Ditto.
1913 (is_neq_zero_form_p): Ditto.
1914 (pred_expr_equal_p): Ditto.
1915 (pred_neg_p): Ditto.
1916 (simplify_pred): Ditto.
1917 (simplify_preds_2): Ditto.
1918 (simplify_preds_3): Ditto.
1919 (simplify_preds_4): Ditto.
1920 (simplify_preds): Ditto.
1922 (push_to_worklist): Ditto.
1923 (get_pred_info_from_cmp): Ditto.
1924 (is_degenerated_phi): Ditto.
1925 (normalize_one_pred_1): Ditto.
1926 (normalize_one_pred): Ditto.
1927 (normalize_one_pred_chain): Ditto.
1928 (normalize_preds): Ditto.
1929 (normalize_cond_1): Remove function.
1930 (normalize_cond): Ditto.
1931 (is_gcond_subset_of): Ditto.
1932 (is_subset_of_any): Ditto.
1933 (is_or_set_subset_of): Ditto.
1934 (is_and_set_subset_of): Ditto.
1935 (is_norm_cond_subset_of): Ditto.
1936 (pred_chain_length_cmp): Ditto.
1937 (convert_control_dep_chain_into_preds): Type change.
1938 (find_predicates): Ditto.
1939 (find_def_preds): Ditto.
1940 (destroy_predicates_vecs): Ditto.
1941 (find_matching_predicates_in_rest_chains): Ditto.
1942 (use_pred_not_overlap_with_undef_path_pred): Ditto.
1943 (is_pred_expr_subset): Ditto.
1944 (is_pred_chain_subset_of): Ditto.
1945 (is_included_in): Ditto.
1946 (is_superset_of): Ditto.
1948 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1950 Update copyright years.
1952 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1954 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
1955 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
1956 config/arc/arc.md, config/arc/arc.opt,
1957 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
1958 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
1959 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
1960 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
1961 config/linux-protos.h, config/linux.c, config/winnt-c.c,
1962 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
1963 vtable-verify.c, vtable-verify.h: Use the standard form for the
1966 2014-01-02 Tobias Burnus <burnus@net-b.de>
1968 * gcc.c (process_command): Update copyright notice dates.
1969 * gcov-dump.c: Ditto.
1971 * doc/cpp.texi: Bump @copying's copyright year.
1972 * doc/cppinternals.texi: Ditto.
1973 * doc/gcc.texi: Ditto.
1974 * doc/gccint.texi: Ditto.
1975 * doc/gcov.texi: Ditto.
1976 * doc/install.texi: Ditto.
1977 * doc/invoke.texi: Ditto.
1979 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1981 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
1983 2014-01-01 Jakub Jelinek <jakub@redhat.com>
1985 * config/i386/sse.md (*mov<mode>_internal): Guard
1986 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
1988 PR rtl-optimization/59647
1989 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
1990 new_rtx into UNSIGNED_FLOAT rtxes.
1992 Copyright (C) 2014 Free Software Foundation, Inc.
1994 Copying and distribution of this file, with or without modification,
1995 are permitted in any medium without royalty provided the copyright
1996 notice and this notice are preserved.