1 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd): New.
4 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5 (_mm512_prefetch_i32scatter_pd): Ditto.
6 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
7 (_mm512_prefetch_i64scatter_pd): Ditto.
8 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
9 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
10 (_mm512_mask_prefetch_i64gather_ps): Ditto.
11 (_mm512_prefetch_i32scatter_ps): Ditto.
12 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
13 (_mm512_prefetch_i64scatter_ps): Ditto.
14 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
15 * config/i386/i386-builtin-types.def: Define
16 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
17 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
18 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
19 IX86_BUILTIN_SCATTERPFQPD.
20 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
21 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
22 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
23 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
24 __builtin_ia32_scatterpfqps.
25 (ix86_expand_builtin): Expand new built-ins.
26 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
27 fix memory access data type.
28 (*avx512pf_gatherpf<mode>_mask): Ditto.
29 (*avx512pf_gatherpf<mode>): Ditto.
30 (avx512pf_scatterpf<mode>): Ditto.
31 (*avx512pf_scatterpf<mode>_mask): Ditto.
32 (*avx512pf_scatterpf<mode>): Ditto.
33 (GATHER_SCATTER_SF_MEM_MODE): New.
34 (avx512pf_gatherpf<mode>df): Ditto.
35 (*avx512pf_gatherpf<mode>df_mask): Ditto.
36 (*avx512pf_scatterpf<mode>df): Ditto.
38 2014-01-27 Jakub Jelinek <jakub@redhat.com>
41 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
42 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
45 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
47 * common/config/arm/arm-common.c
48 (arm_rewrite_mcpu): Handle multiple names.
50 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
52 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
54 * gimple-builder.h (create_gimple_tmp): Delete.
56 2014-01-27 Christian Bruel <christian.bruel@st.com>
58 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
61 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
63 * config/pa/pa.md (call): Generate indirect long calls to non-local
64 functions when outputing 32-bit code.
65 (call_value): Likewise except for special call to buggy powf function.
67 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
68 portable runtime and PIC indirect calls.
69 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
70 and PIC call sequences. Use ldo instead of blr to set return register
73 2014-01-25 Walter Lee <walt@tilera.com>
75 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
76 avoid clobbering a live register.
78 2014-01-25 Walter Lee <walt@tilera.com>
80 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
81 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
82 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
83 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
85 2014-01-25 Walter Lee <walt@tilera.com>
87 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
88 arguments on even registers.
89 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
91 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
92 (BIGGEST_ALIGNMENT): Ditto.
93 (BIGGEST_FIELD_ALIGNMENT): Ditto.
95 2014-01-25 Walter Lee <walt@tilera.com>
97 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
98 insns before bundling.
99 * config/tilegx/tilegx.md (tile_network_barrier): Update
102 2014-01-25 Walter Lee <walt@tilera.com>
104 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
105 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
106 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
108 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
110 * config/mips/constraints.md (kl): Delete.
111 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
112 define expands, using...
113 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
114 instructions for MIPS16.
115 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
116 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
118 2014-01-25 Walter Lee <walt@tilera.com>
120 * config/tilepro/tilepro.md (ctzdi2): Use register_operand
125 2014-01-25 Walter Lee <walt@tilera.com>
127 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
128 (TARGET_EXPAND_TO_RTL_HOOK): Define.
130 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
132 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
135 2014-01-25 Jakub Jelinek <jakub@redhat.com>
137 * print-rtl.c (in_call_function_usage): New var.
138 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
139 EXPR_LIST mode as mode and not as reg note name.
142 * cfgloopmanip.c (copy_loop_info): If
143 loop->warned_aggressive_loop_optimizations, make sure
144 the flag is set in target loop too.
146 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
148 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
150 * builtins.def: Likewise.
151 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
152 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
153 * ira.c (ira_setup_eliminable_regset): Likewise.
154 * omp-low.c (gate_expand_omp): Likewise.
155 (execute_lower_omp): Likewise.
156 (diagnose_sb_0): Likewise.
157 (gate_diagnose_omp_blocks): Likewise.
158 (simd_clone_clauses_extract): Likewise.
161 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
163 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
164 correction for little endian...
165 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
168 2014-01-24 Jeff Law <law@redhat.com>
170 PR tree-optimization/59919
171 * tree-vrp.c (find_assert_locations_1): Do not register asserts
172 for non-returning calls.
174 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
176 * common/config/aarch64/aarch64-common.c
177 (aarch64_rewrite_mcpu): Handle multiple names.
178 * config/aarch64/aarch64.h
179 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
181 2014-01-24 Dodji Seketeli <dodji@redhat.com>
183 * input.c (add_file_to_cache_tab): Handle the case where fopen
186 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
189 * config/i386/i386.md (pushsf splitter): Get stack adjustment
190 from push operand if code of push isn't PRE_DEC.
192 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
195 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
196 -mquad-memory-atomic. Update -mquad-memory documentation to say
197 it is only used for non-atomic loads/stores.
199 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
200 -mquad-memory or -mquad-memory-atomic switches.
202 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
203 -mquad-memory-atomic to ISA 2.07 support.
205 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
206 to separate support of normal quad word memory operations (ldq,
207 stq) from the atomic quad word memory operations.
209 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
210 support to separate non-atomic quad word operations from atomic
211 quad word operations. Disable non-atomic quad word operations in
212 little endian mode so that we don't have to swap words after the
213 load and before the store.
214 (quad_load_store_p): Add comment about atomic quad word support.
215 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
216 options printed with -mdebug=reg.
218 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
219 -mquad-memory-atomic as the test for whether we have quad word
221 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic,
222 -mquad-memory, or -mp8-vector are used, allow byte/half-word
225 * config/rs6000/sync.md (load_lockedti): Insure that the address
226 is a proper indexed or indirect address for the lqarx instruction.
227 On little endian systems, swap the hi/lo registers after the lqarx
229 (load_lockedpti): Use indexed_or_indirect_operand predicate to
230 insure the address is valid for the lqarx instruction.
231 (store_conditionalti): Insure that the address is a proper indexed
232 or indirect address for the stqcrx. instruction. On little endian
233 systems, swap the hi/lo registers before doing the stqcrx.
235 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
236 insure the address is valid for the stqcrx. instruction.
238 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
239 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
240 type of quad memory support is available.
242 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
245 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
246 there is a danger of looping.
248 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
251 force flag_ira_loop_pressure if set via command line.
253 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
255 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
256 (ashr_simd): New builtin handling DI mode.
257 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
258 (aarch64_sshr_simddi): New match pattern.
259 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
260 (vshrd_n_s64): Likewise.
261 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
263 2014-01-23 Nick Clifton <nickc@redhat.com>
265 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
266 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
267 favour of mcu specific scripts.
268 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
271 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
272 Alex Velenko <Alex.Velenko@arm.com>
274 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
275 (vaddv_s16): Likewise.
276 (vaddv_s32): Likewise.
277 (vaddv_u8): Likewise.
278 (vaddv_u16): Likewise.
279 (vaddv_u32): Likewise.
280 (vaddvq_s8): Likewise.
281 (vaddvq_s16): Likewise.
282 (vaddvq_s32): Likewise.
283 (vaddvq_s64): Likewise.
284 (vaddvq_u8): Likewise.
285 (vaddvq_u16): Likewise.
286 (vaddvq_u32): Likewise.
287 (vaddvq_u64): Likewise.
288 (vaddv_f32): Likewise.
289 (vaddvq_f32): Likewise.
290 (vaddvq_f64): Likewise.
291 (vmaxv_f32): Likewise.
292 (vmaxv_s8): Likewise.
293 (vmaxv_s16): Likewise.
294 (vmaxv_s32): Likewise.
295 (vmaxv_u8): Likewise.
296 (vmaxv_u16): Likewise.
297 (vmaxv_u32): Likewise.
298 (vmaxvq_f32): Likewise.
299 (vmaxvq_f64): Likewise.
300 (vmaxvq_s8): Likewise.
301 (vmaxvq_s16): Likewise.
302 (vmaxvq_s32): Likewise.
303 (vmaxvq_u8): Likewise.
304 (vmaxvq_u16): Likewise.
305 (vmaxvq_u32): Likewise.
306 (vmaxnmv_f32): Likewise.
307 (vmaxnmvq_f32): Likewise.
308 (vmaxnmvq_f64): Likewise.
309 (vminv_f32): Likewise.
310 (vminv_s8): Likewise.
311 (vminv_s16): Likewise.
312 (vminv_s32): Likewise.
313 (vminv_u8): Likewise.
314 (vminv_u16): Likewise.
315 (vminv_u32): Likewise.
316 (vminvq_f32): Likewise.
317 (vminvq_f64): Likewise.
318 (vminvq_s8): Likewise.
319 (vminvq_s16): Likewise.
320 (vminvq_s32): Likewise.
321 (vminvq_u8): Likewise.
322 (vminvq_u16): Likewise.
323 (vminvq_u32): Likewise.
324 (vminnmv_f32): Likewise.
325 (vminnmvq_f32): Likewise.
326 (vminnmvq_f64): Likewise.
328 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
330 * config/aarch64/aarch64-simd.md
331 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
332 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
333 (*aarch64_mul3_elt<mode>): Likewise.
334 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
335 (*aarch64_mul3_elt_to_64v2df): Likewise.
336 (*aarch64_mla_elt<mode>): Likewise.
337 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
338 (*aarch64_mls_elt<mode>): Likewise.
339 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
340 (*aarch64_fma4_elt<mode>): Likewise.
341 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
342 (*aarch64_fma4_elt_to_64v2df): Likewise.
343 (*aarch64_fnma4_elt<mode>): Likewise.
344 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
345 (*aarch64_fnma4_elt_to_64v2df): Likewise.
346 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
347 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
348 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
349 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
350 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
351 (aarch64_sqdmull_lane<mode>_internal): Likewise.
352 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
354 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
356 * config/aarch64/aarch64-simd.md
357 (aarch64_be_checked_get_lane<mode>): New define_expand.
358 * config/aarch64/aarch64-simd-builtins.def
359 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0):
360 New builtin definition.
361 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
362 Use new safe be builtin.
364 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
366 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
368 (aarch64_be_st1<mode>): Likewise.
369 (aarch_ld1<VALL:mode>): Define_expand modified.
370 (aarch_st1<VALL:mode>): Likewise.
371 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
372 (UNSPEC_ST1): Likewise.
374 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
376 * config/microblaze/microblaze.md: Add trap insn and attribute
378 2014-01-23 Dodji Seketeli <dodji@redhat.com>
380 PR preprocessor/58580
381 * input.h (location_get_source_line): Take an additional line_size
383 (void diagnostics_file_cache_fini): Declare new function.
384 * input.c (struct fcache): New type.
385 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
386 New static constants.
387 (diagnostic_file_cache_init, total_lines_num)
388 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
389 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
390 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
391 (get_next_line, read_next_line, goto_next_line, read_line_num):
392 New static function definitions.
393 (diagnostic_file_cache_fini): New function.
394 (location_get_source_line): Take an additional output line_len
395 parameter. Re-write using lookup_or_add_file_to_cache_tab and
397 * diagnostic.c (diagnostic_finish): Call
398 diagnostic_file_cache_fini.
399 (adjust_line): Take an additional input parameter for the length
400 of the line, rather than calculating it with strlen.
401 (diagnostic_show_locus): Adjust the use of
402 location_get_source_line and adjust_line with respect to their new
403 signature. While displaying a line now, do not stop at the first
404 null byte. Rather, display the zero byte as a space and keep
405 going until we reach the size of the line.
406 * Makefile.in: Add vec.o to OBJS-libcommon
408 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
409 Ilya Tocar <ilya.tocar@intel.com>
411 * config/i386/avx512fintrin.h (_mm512_kmov): New.
412 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
413 (__builtin_ia32_kmov16): Ditto.
414 * config/i386/i386.md (UNSPEC_KMOV): New.
417 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
419 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
420 (_mm512_storeu_si512): Ditto.
422 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
425 * rtl.h (get_referenced_operands): Declare.
426 * recog.c (get_referenced_operands): New function.
427 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
428 operands have been referenced when recording LO_SUM references.
430 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
432 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
434 2014-01-22 Jan Hubicka <jh@suse.cz>
436 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
437 Enable for generic and recent AMD targets.
439 2014-01-22 Jan Hubicka <jh@suse.cz>
441 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
442 ARG_SIZE note when adjustment was eliminated.
444 2014-01-22 Jeff Law <law@redhat.com>
446 PR tree-optimization/59597
447 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
448 in file. Accept new argument REGISTERING and use it to modify
449 dump output appropriately.
450 (register_jump_thread): Corresponding changes.
451 (mark_threaded_blocks): Reinstate code to cancel unprofitable
452 thread paths involving joiner blocks. Add code to dump cancelled
453 jump threading paths.
455 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
457 PR rtl-optimization/59477
458 * lra-constraints.c (inherit_in_ebb): Process call for living hard
459 regs. Update reloads_num and potential_reload_hard_regs for all
462 2014-01-22 Tom Tromey <tromey@redhat.com>
464 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
466 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
468 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
470 PR rtl-optimization/59896
471 * lra-constraints.c (process_alt_operands): Check unused note for
472 matched operands of insn with no output reloads.
474 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
476 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
477 (mips_move_from_gpr_cost): Likewise.
479 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
481 PR rtl-optimization/59858
482 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
483 ira_class_hard_regs_num.
484 (process_alt_operands): Increase reject for dying matched operand.
486 2014-01-21 Jakub Jelinek <jakub@redhat.com>
489 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
490 smaller than size, perform several stores or loads and stores
491 at dst + count - size to store or copy all of size bytes, rather
492 than just last modesize bytes.
494 2014-01-20 DJ Delorie <dj@redhat.com>
496 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
497 that CLOBBERs are REGs before propogating their values.
499 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
502 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
503 (cgraph_inline_failed_type): New function.
504 * cgraph.h (DEFCIFCODE): Add type.
505 (cgraph_inline_failed_type_t): New enum.
506 (cgraph_inline_failed_type): New prototype.
507 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
508 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
509 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
510 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
511 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
512 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
513 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
514 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
515 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
516 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
517 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
518 OPTIMIZATION_MISMATCH.
519 * tree-inline.c (expand_call_inline): Emit errors during
520 early_inlining if cgraph_inline_failed_type returns
523 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
526 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
527 mode attribute in insn output.
529 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
531 * output.h (output_constant): Delete.
532 * varasm.c (output_constant): Make private.
534 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
536 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
538 2014-01-20 Jakub Jelinek <jakub@redhat.com>
541 * tree.h (fold_builtin_strcat): New prototype.
542 * builtins.c (fold_builtin_strcat): No longer static. Add len
543 argument, if non-NULL, don't call c_strlen. Optimize
544 directly into __builtin_memcpy instead of __builtin_strcpy.
545 (fold_builtin_2): Adjust fold_builtin_strcat caller.
546 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
548 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
550 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
551 for SImode_address_operand operands, having only a REG argument.
553 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
555 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
556 loader name using mbig-endian.
557 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
559 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
561 * doc/invoke.texi (-march): Clarify documentation for AArch64.
565 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
567 * config/aarch64/aarch64-protos.h
568 (aarch64_cannot_change_mode_class_ptr): Declare.
569 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
570 aarch64_cannot_change_mode_class_ptr): New.
571 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
572 backend hook aarch64_cannot_change_mode_class.
574 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
576 * common/config/aarch64/aarch64-common.c
577 (aarch64_handle_option): Don't handle any option order logic here.
578 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
579 selected_cpu, warn on architecture version mismatch.
580 (aarch64_override_options): Fix parsing order for option strings.
582 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
583 Iain Sandoe <iain@codesourcery.com>
586 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
587 warning. Amend comment to reflect current functionality.
589 2014-01-20 Richard Biener <rguenther@suse.de>
592 * builtins.c (fold_builtin_strcat): Remove case better handled
593 by tree-ssa-strlen.c.
595 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
597 * config/aarch64/aarch64.opt
598 (mcpu, march, mtune): Make case-insensitive.
600 2014-01-20 Jakub Jelinek <jakub@redhat.com>
603 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
604 if operands[1] is a REG or ZERO_EXTEND of a REG.
606 2014-01-19 Jan Hubicka <jh@suse.cz>
608 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
610 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
612 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
613 long non-pic millicode calls.
615 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
617 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
619 2014-01-19 Kito Cheng <kito@0xlab.org>
621 * builtins.c (expand_movstr): Check movstr expand done or fail.
623 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
624 H.J. Lu <hongjiu.lu@intel.com>
627 * config/i386/i386.md (*lea<mode>): Zero-extend return register
628 to DImode for zero-extended addresses.
630 2014-01-19 Jakub Jelinek <jakub@redhat.com>
632 PR rtl-optimization/57763
633 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
634 on the new indirect jump_insn and increment LABEL_NUSES (label).
636 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
640 * config.gcc (x86_archs): New variable.
641 (x86_64_archs): Likewise.
642 (x86_cpus): Likewise.
643 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
644 --with-arch/--with-cpu= options.
645 Support --with-arch=/--with-cpu={nehalem,westmere,
646 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
648 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
650 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
651 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
653 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
655 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
657 2014-01-18 Jakub Jelinek <jakub@redhat.com>
660 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
661 clear cpp_get_options (parse_in)->warn_unused_macros for
662 ix86_target_macros_internal with cpp_define.
664 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
666 * jump.c (delete_related_insns): Keep (use (insn))s.
667 * reorg.c (redundant_insn): Check for barriers too.
669 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
671 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
673 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
675 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
676 call to $$dyncall when TARGET_LONG_CALLS is true.
678 2014-01-17 Jeff Law <law@redhat.com>
680 * ree.c (combine_set_extension): Temporarily disable test for
681 changing number of hard registers.
683 2014-01-17 Jan Hubicka <jh@suse.cz>
686 * ipa-inline-analysis.c (inline_free_summary):
687 Do not free summary of aliases.
689 2014-01-17 Jakub Jelinek <jakub@redhat.com>
692 * gimplify.c (gimplify_expr): Use create_tmp_var
693 instead of create_tmp_var_raw. If cond doesn't have
694 integral type, don't add the IFN_ANNOTATE builtin at all.
696 2014-01-17 Martin Jambor <mjambor@suse.cz>
699 * ipa-cp.c (prev_edge_clone): New variable.
700 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
701 Also resize prev_edge_clone vector.
702 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
703 (ipcp_edge_removal_hook): New function.
704 (ipcp_driver): Register ipcp_edge_removal_hook.
706 2014-01-17 Andrew Pinski <apinski@cavium.com>
707 Steve Ellcey <sellcey@mips.com>
710 * config/mips/mips.c (mips_print_operand): Check operand mode instead
713 2014-01-17 Jeff Law <law@redhat.com>
716 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
717 so that pass_ccp runs first.
719 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
721 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
722 (ix86_adjust_cost): Use !TARGET_XXX.
723 (do_reorder_for_imul): Likewise.
724 (swap_top_of_ready_list): Likewise.
725 (ix86_sched_reorder): Likewise.
727 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
729 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
730 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
731 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
732 (intel_memset): New. Duplicate slm_memset.
733 (intel_cost): New. Duplicate slm_cost.
734 (m_INTEL): New macro.
735 (processor_target_table): Add "intel".
736 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
737 with PROCESSOR_INTEL for "intel".
738 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
739 PROCESSOR_SILVERMONT.
740 (ix86_issue_rate): Likewise.
741 (ix86_adjust_cost): Likewise.
742 (ia32_multipass_dfa_lookahead): Likewise.
743 (swap_top_of_ready_list): Likewise.
744 (ix86_sched_reorder): Likewise.
745 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
746 instead of TARGET_OPT_AGU.
747 * config/i386/i386.h (TARGET_INTEL): New.
748 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
749 (processor_type): Add PROCESSOR_INTEL.
750 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
751 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
753 2014-01-17 Marek Polacek <polacek@redhat.com>
756 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
759 2014-01-17 Richard Biener <rguenther@suse.de>
761 PR tree-optimization/46590
762 * opts.c (default_options_table): Add entries for
763 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
764 all enabled at -O1 but not for -Og.
765 * common.opt (fbranch-count-reg): Remove Init(1).
766 (fmove-loop-invariants): Likewise.
767 (ftree-pta): Likewise.
769 2014-01-17 Jakub Jelinek <jakub@redhat.com>
771 * config/i386/i386.c (ix86_data_alignment): For compatibility with
772 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
773 decls to at least the GCC 4.8 used alignments.
776 * tree-nested.c (convert_nonlocal_reference_stmt,
777 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
778 of GIMPLE_BIND stmts, adjust associated decls.
780 2014-01-17 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/46590
783 * vec.h (vec<>::bseach): New member function implementing
784 binary search according to C89 bsearch.
785 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
786 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
787 bitmap pointer again. Make accesses_in_loop a flat array.
788 (mem_ref_obstack): New global.
789 (outermost_indep_loop): Adjust for mem_ref->stored changes.
790 (mark_ref_stored): Likewise.
791 (ref_indep_loop_p_2): Likewise.
792 (set_ref_stored_in_loop): New helper function.
793 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
794 (memref_free): Adjust.
795 (record_mem_ref_loc): Simplify.
796 (gather_mem_refs_stmt): Adjust.
797 (sort_locs_in_loop_postorder_cmp): New function.
798 (analyze_memory_references): Sort accesses_in_loop after
799 loop postorder number.
800 (find_ref_loc_in_loop_cmp): New function.
801 (for_all_locs_in_loop): Find relevant cluster of locs in
802 accesses_in_loop and iterate without recursion.
803 (execute_sm): Avoid uninit warning.
804 (struct ref_always_accessed): Simplify.
805 (ref_always_accessed::operator ()): Likewise.
806 (ref_always_accessed_p): Likewise.
807 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
808 loop postorder numbers here.
809 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
812 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
815 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
816 on decls for which assemble_alias has been called.
818 2014-01-17 Nick Clifton <nickc@redhat.com>
820 * config/msp430/msp430.opt: (mcpu): New option.
821 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
822 (msp430_option_override): Parse target_cpu. If the MCU name
823 matches a generic string, clear target_mcu.
824 (msp430_attr): Allow numeric interrupt values up to 63.
825 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
826 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
828 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
830 * config/msp430/msp430.md (popm): Use %J rather than %I.
831 (addsi3): Use msp430_nonimmediate_operand for operand 2.
832 (addhi_cy_i): Use immediate_operand for operand 2.
833 * doc/invoke.texi: Document -mcpu option.
835 2014-01-17 Richard Biener <rguenther@suse.de>
837 PR rtl-optimization/38518
838 * df.h (df_analyze_loop): Declare.
839 * df-core.c: Include cfgloop.h.
840 (df_analyze_1): Split out main part of df_analyze.
841 (df_analyze): Adjust.
842 (loop_inverted_post_order_compute): New function.
843 (loop_post_order_compute): Likewise.
844 (df_analyze_loop): New function avoiding whole-function
846 * loop-invariant.c (find_defs): Use df_analyze_loop.
847 (find_invariants): Adjust.
848 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
850 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
852 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
853 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
855 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
857 * ipa-ref.c (ipa_remove_stmt_references): Fix references
858 traversal when removing references.
860 2014-01-16 Jan Hubicka <jh@suse.cz>
863 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
865 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
868 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
869 pushing a reload for an autoinc when we had previously reloaded an
870 inner part of the address.
872 2014-01-16 Jakub Jelinek <jakub@redhat.com>
874 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
876 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
877 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
878 when not giving up or versioning for alias only because of
880 (vect_analyze_data_ref_dependences): Set to true.
881 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
883 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
884 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
888 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
891 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
892 operand 0 predicate for gathers, use a new pseudo as subtarget.
894 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
897 * lra-constraints.c (process_alt_operands): Add printing debug info.
898 Check absence of input/output reloads for matched operands too.
900 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
902 PR rtl-optimization/59835
903 * ira.c (ira_init_register_move_cost): Increase cost for
906 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
908 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
910 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
913 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
914 non-register objects. Use gen_(high/low)part more consistently.
917 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
920 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
921 endian support, remove tests for WORDS_BIG_ENDIAN.
922 (p8_mfvsrd_3_<mode>): Likewise.
923 (reload_gpr_from_vsx<mode>): Likewise.
924 (reload_gpr_from_vsxsf): Likewise.
925 (p8_mfvsrd_4_disf): Likewise.
927 2014-01-16 Richard Biener <rguenther@suse.de>
929 PR rtl-optimization/46590
930 * lcm.c (compute_antinout_edge): Use postorder iteration.
931 (compute_laterin): Use inverted postorder iteration.
933 2014-01-16 Nick Clifton <nickc@redhat.com>
936 * varasm.c (output_constant): Return the number of bytes actually
938 (output_constructor_array_range): Update the field size with the
939 number of bytes emitted by output_constant.
940 (output_constructor_regular_field): Likewise. Also do not
941 complain if the total number of bytes emitted is now greater
942 than the expected fieldpos.
943 * output.h (output_constant): Update prototype and descriptive comment.
945 2014-01-16 Marek Polacek <polacek@redhat.com>
948 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
949 it is error_mark_node.
951 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
953 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
954 VALID_AVX256_REG_OR_OI_MODE.
956 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
958 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
959 current procedure should be profiled.
961 2014-01-15 Andrew Pinski <apinski@cavium.com>
963 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
964 of moving from/to the STACK_REG register class.
966 2014-01-15 Richard Henderson <rth@redhat.com>
969 * reginfo.c (global_regs_decl): Globalize.
970 * rtl.h (global_regs_decl): Declare.
971 * ira.c (do_reload): Diagnose frame_pointer_needed and it
972 reserved via global_regs.
974 2014-01-15 Teresa Johnson <tejohnson@google.com>
976 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
978 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
980 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
981 and vmulosh rather than call gen_vec_widen_smult_*.
982 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
983 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
984 (vec_widen_smult_even_v16qi): Likewise.
985 (vec_widen_umult_even_v8hi): Likewise.
986 (vec_widen_smult_even_v8hi): Likewise.
987 (vec_widen_umult_odd_v16qi): Likewise.
988 (vec_widen_smult_odd_v16qi): Likewise.
989 (vec_widen_umult_odd_v8hi): Likewise.
990 (vec_widen_smult_odd_v8hi): Likewise.
991 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
992 vmuloub rather than call gen_vec_widen_umult_*.
993 (vec_widen_umult_lo_v16qi): Likewise.
994 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
995 vmulosb rather than call gen_vec_widen_smult_*.
996 (vec_widen_smult_lo_v16qi): Likewise.
997 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
998 rather than call gen_vec_widen_umult_*.
999 (vec_widen_umult_lo_v8hi): Likewise.
1000 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1001 rather than call gen_vec_widen_smult_*.
1002 (vec_widen_smult_lo_v8hi): Likewise.
1004 2014-01-15 Jeff Law <law@redhat.com>
1006 PR tree-optimization/59747
1007 * ree.c (find_and_remove_re): Properly handle case where a second
1008 eliminated extension requires widening a copy created for elimination
1009 of a prior extension.
1010 (combine_set_extension): Ensure that the number of hard regs needed
1011 for a destination register does not change when we widen it.
1013 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1015 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1016 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1017 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1018 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1019 (avr-*-rtems*): Likewise.
1020 (bfin*-rtems*): Likewise.
1021 (moxie-*-rtems*): Likewise.
1022 (h8300-*-rtems*): Likewise.
1023 (i[34567]86-*-rtems*): Likewise.
1024 (lm32-*-rtems*): Likewise.
1025 (m32r-*-rtems*): Likewise.
1026 (m68k-*-rtems*): Likewise.
1027 (microblaze*-*-rtems*): Likewise.
1028 (mips*-*-rtems*): Likewise.
1029 (powerpc-*-rtems*): Likewise.
1030 (sh-*-rtems*): Likewise.
1031 (sparc-*-rtems*): Likewise.
1032 (sparc64-*-rtems*): Likewise.
1033 (v850-*-rtems*): Likewise.
1034 (m32c-*-rtems*): Likewise.
1036 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1038 PR rtl-optimization/59511
1039 * ira.c (ira_init_register_move_cost): Use memory costs for some
1040 cases of register move cost calculations.
1041 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1042 instead of BB frequency.
1043 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1044 * lra-assigns.c (find_hard_regno_for): Ditto.
1046 2014-01-15 Richard Biener <rguenther@suse.de>
1048 PR tree-optimization/59822
1049 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1050 (vectorizable_load): Use it to hoist defs of uses of invariant
1051 loads out of the loop.
1053 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1054 Kugan Vivekanandarajah <kuganv@linaro.org>
1057 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1060 2014-01-15 Richard Biener <rguenther@suse.de>
1062 PR rtl-optimization/59802
1063 * lcm.c (compute_available): Use inverted postorder to seed
1064 the initial worklist.
1066 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1069 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1070 ADDR_REGS for invalid symrefs in non-PIC code.
1072 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1075 * builtins.c (determine_block_size): Initialize *probable_max_size
1076 even if len_rtx is CONST_INT.
1078 2014-01-14 Andrew Pinski <apinski@cavium.com>
1080 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1081 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1082 (cortexa53_tunings): Likewise.
1083 (aarch64_sched_issue_rate): New function.
1084 (TARGET_SCHED_ISSUE_RATE): Define.
1086 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1088 * ira-costs.c (find_costs_and_classes): Add missed
1089 ira_init_register_move_cost_if_necessary.
1091 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1094 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1096 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1099 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1100 to indicate if type is used for function return value. Warn ABI
1101 change if the vector mode isn't available for function return value.
1102 (ix86_function_arg_advance): Pass false to type_natural_mode.
1103 (ix86_function_arg): Likewise.
1104 (ix86_gimplify_va_arg): Likewise.
1105 (function_arg_32): Don't warn ABI change.
1106 (ix86_function_value): Pass true to type_natural_mode.
1107 (ix86_return_in_memory): Likewise.
1108 (ix86_struct_value_rtx): Removed.
1109 (TARGET_STRUCT_VALUE_RTX): Likewise.
1111 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1113 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1114 converting a conditional jump into a conditional return.
1116 2014-01-14 Richard Biener <rguenther@suse.de>
1118 PR tree-optimization/58921
1119 PR tree-optimization/59006
1120 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1121 hoisting invariant stmts.
1122 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1123 invariant loads on the preheader edge if possible.
1125 2014-01-14 Joey Ye <joey.ye@arm.com>
1127 * doc/plugin.texi (Building GCC plugins): Update to C++.
1129 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1131 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1132 (_mm_rcp28_round_ss): Ditto.
1133 (_mm_rsqrt28_round_sd): Ditto.
1134 (_mm_rsqrt28_round_ss): Ditto.
1135 (_mm_rcp28_sd): Ditto.
1136 (_mm_rcp28_ss): Ditto.
1137 (_mm_rsqrt28_sd): Ditto.
1138 (_mm_rsqrt28_ss): Ditto.
1139 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1140 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1141 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1142 (IX86_BUILTIN_RCP28SD): Ditto.
1143 (IX86_BUILTIN_RCP28SS): Ditto.
1144 (IX86_BUILTIN_RSQRT28SD): Ditto.
1145 (IX86_BUILTIN_RSQRT28SS): Ditto.
1146 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1147 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1148 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1149 (ix86_expand_special_args_builtin): Expand new FTYPE.
1150 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1151 (srcp14<mode>): Make insn unary.
1152 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1153 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1154 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1155 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1156 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1157 Fix rounding: make it SAE only.
1158 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1160 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1162 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1163 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1164 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1165 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1166 (round_saeonly_mask_scalar_operand4): Ditto.
1167 (round_saeonly_mask_scalar_op3): Ditto.
1168 (round_saeonly_mask_scalar_op4): Ditto.
1170 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1172 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1173 Implement -maltivec=be for vec_insert and vec_extract.
1175 2014-01-10 DJ Delorie <dj@redhat.com>
1177 * config/msp430/msp430.md (call_internal): Don't allow memory
1178 references with SP as the base register.
1179 (call_value_internal): Likewise.
1180 * config/msp430/constraints.md (Yc): New. For memory references
1181 that don't use SP as a base register.
1183 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1184 "an integer without a # prefix"
1185 * config/msp430/msp430.md (epilogue_helper): Use it.
1187 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1190 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1191 AVX512F gather builtins.
1192 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1193 on gather decls with INTEGER_TYPE masktype.
1194 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1195 directly into the builtin rather than hoisting it before loop.
1197 PR tree-optimization/59387
1198 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1199 (scev_const_prop): If folded_casts and type has undefined overflow,
1200 use force_gimple_operand instead of force_gimple_operand_gsi and
1201 for each added stmt if it is assign with
1202 arith_code_with_undefined_signed_overflow, call
1203 rewrite_to_defined_overflow.
1204 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1205 gimple-fold.h instead.
1206 (arith_code_with_undefined_signed_overflow,
1207 rewrite_to_defined_overflow): Moved to ...
1208 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1209 rewrite_to_defined_overflow): ... here. No longer static.
1210 Include gimplify-me.h.
1211 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1212 rewrite_to_defined_overflow): New prototypes.
1214 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1216 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1218 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1220 * builtins.c (get_object_alignment_2): Minor tweak.
1221 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1223 2014-01-13 Christian Bruel <christian.bruel@st.com>
1225 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1226 optimized non constant lengths.
1228 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1231 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1232 load as __atomic_load_N if possible.
1234 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1236 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1238 (rs6000_expand_builtin): Adjust call.
1240 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1243 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1244 * config/rs6000/rs6000.c: Include target-globals.h.
1245 (rs6000_set_current_function): Instead of doing target_reinit
1246 unconditionally, use save_target_globals_default_opts and
1247 restore_target_globals.
1249 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1251 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1252 (rs6000_expand_builtin): Handle mffs and mtfsf.
1253 (rs6000_init_builtins): Define mffs and mtfsf.
1254 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1255 (rs6000_mffs): New pattern.
1256 (rs6000_mtfsf): New pattern.
1258 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1260 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1261 Start narrowing with START. Apply candidate-use pair
1262 and check overall cost in narrowing.
1263 (iv_ca_prune): Pass new argument.
1265 2014-01-10 Jeff Law <law@redhat.com>
1268 * ree.c (combine_reaching_defs): Ensure the defining statement
1269 occurs before the extension when optimizing extensions with
1270 different source and destination hard registers.
1272 2014-01-10 Jan Hubicka <jh@suse.cz>
1275 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1276 vtables into the type inheritance graph.
1278 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1280 PR rtl-optimization/59754
1281 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1282 modes in the REGNO != REGNO case.
1284 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1286 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1288 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1290 PR tree-optimization/59745
1291 * tree-predcom.c (tree_predictive_commoning_loop): Call
1292 free_affine_expand_cache if giving up because components is NULL.
1294 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1295 GC in payload of target_globals struct instead of allocating them on
1296 the heap and the larger structs separately using GC.
1297 * target-globals.h (struct target_globals): Make regs, hard_regs,
1298 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1299 of GTY((skip)) and change type to void *.
1300 (reset_target_globals): Cast loads from those fields to corresponding
1303 2014-01-10 Steve Ellcey <sellcey@mips.com>
1306 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1307 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1308 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1310 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1313 * aarch64-modes.def (CC_Zmode): New flags mode.
1314 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1315 represents an equality.
1316 (aarch64_get_condition_code): Handle CC_Zmode.
1317 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1319 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1321 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1322 extraction in good case.
1324 2014-01-10 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/59374
1327 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1328 checking after SLP discovery. Mark stmts not participating
1329 in any SLP instance properly.
1331 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1333 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1334 when handling a SET rtx.
1336 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1338 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1339 (cortex-a57): Likewise.
1340 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1342 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1344 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1345 non-iwmmxt builtins.
1347 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1351 * ipa-devirt.c record_target_from_binfo): Take as argument
1352 stack of binfos and lookup matching one for virtual inheritance.
1353 (possible_polymorphic_call_targets_1): Update.
1355 2014-01-10 Huacai Chen <chenhc@lemote.com>
1357 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1358 kernel strings for Loongson-2E/2F/3A.
1360 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1363 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1364 is_gimple_call before calling gimple_call_internal_p.
1366 2014-01-09 Steve Ellcey <sellcey@mips.com>
1368 * Makefile.in (TREE_FLOW_H): Remove.
1369 (TREE_SSA_H): Add file names from tree-flow.h.
1370 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1371 * tree.h: Remove tree-flow.h reference.
1372 * hash-table.h: Remove tree-flow.h reference.
1373 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1374 reference with tree-ssa-loop.h.
1376 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1378 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1379 default element-order behavior for -maltivec.
1380 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1382 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1383 when targeting big endian, at least for now.
1384 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1386 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1389 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1390 var satisfies use_register_for_decl, just take into account type
1391 alignment, rather than decl alignment.
1393 PR tree-optimization/59622
1394 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1395 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1396 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1397 Don't devirtualize for inplace at all. For targets.length () == 1,
1398 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1400 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1402 * config/i386/i386.md (cpu): Remove the unused btver1.
1404 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1406 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1408 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1411 * tree-core.h (struct target_globals): New forward declaration.
1412 (struct tree_target_option): Add globals field.
1413 * tree.h (TREE_TARGET_GLOBALS): Define.
1414 (prepare_target_option_nodes_for_pch): New prototype.
1415 * target-globals.h (struct target_globals): Define even if
1417 * tree.c (prepare_target_option_node_for_pch,
1418 prepare_target_option_nodes_for_pch): New functions.
1419 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1420 * config/i386/i386.c: Include target-globals.h.
1421 (ix86_set_current_function): Instead of doing target_reinit
1422 unconditionally, use save_target_globals_default_opts and
1423 restore_target_globals.
1425 2014-01-09 Richard Biener <rguenther@suse.de>
1427 PR tree-optimization/59715
1428 * tree-cfg.h (split_critical_edges): Declare.
1429 * tree-cfg.c (split_critical_edges): Export.
1430 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1432 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1434 * cfgexpand.c (expand_stack_vars): Optionally disable
1435 asan stack protection.
1436 (expand_used_vars): Likewise.
1437 (partition_stack_vars): Likewise.
1438 * asan.c (asan_emit_stack_protection): Optionally disable
1439 after return stack usage.
1440 (instrument_derefs): Optionally disable memory access instrumentation.
1441 (instrument_builtin_call): Likewise.
1442 (instrument_strlen_call): Likewise.
1443 (asan_protect_global): Optionally disable global variables protection.
1444 * doc/invoke.texi: Added doc for new options.
1445 * params.def: Added new options.
1446 * params.h: Likewise.
1448 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1450 PR rtl-optimization/59724
1451 * ifcvt.c (cond_exec_process_if_block): Don't call
1452 flow_find_head_matching_sequence with 0 longest_match.
1453 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1454 non-active insns if !stop_after.
1455 (try_head_merge_bb): Revert 2014-01-07 changes.
1457 2014-01-08 Jeff Law <law@redhat.com>
1459 * ree.c (get_sub_rtx): New function, extracted from...
1460 (merge_def_and_ext): Here.
1461 (combine_reaching_defs): Use get_sub_rtx.
1463 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1465 * cgraph.h (varpool_variable_node): Do not choke on null node.
1467 2014-01-08 Catherine Moore <clm@codesourcery.com>
1469 * config/mips/mips.md (simple_return): Attempt to use JRC
1471 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1473 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1475 PR rtl-optimization/59137
1476 * reorg.c (steal_delay_list_from_target): Call update_block for
1478 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1480 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1482 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1483 two duplicate entries.
1485 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1488 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1490 * config/mips/mips.c (mips_truncated_op_cost): New function.
1491 (mips_rtx_costs): Adjust test for BADDU.
1492 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1494 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1496 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1497 (*baddu_si): ...this new pattern.
1499 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1502 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1504 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1507 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1509 (expand_expr, expand_normal): Adjust.
1510 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1511 inner_reference_p. Use inner_reference_p to expand inner references.
1512 (store_expr): Adjust.
1513 * cfgexpand.c (expand_call_stmt): Adjust.
1515 2014-01-08 Rong Xu <xur@google.com>
1517 * gcov-io.c (gcov_var): Move from gcov-io.h.
1518 (gcov_position): Ditto.
1519 (gcov_is_error): Ditto.
1520 (gcov_rewrite): Ditto.
1521 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1522 only part to libgcc/libgcov.h.
1524 2014-01-08 Marek Polacek <polacek@redhat.com>
1527 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1529 2014-01-08 Marek Polacek <polacek@redhat.com>
1532 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1534 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1536 PR rtl-optimization/59649
1537 * stor-layout.c (get_mode_bounds): For BImode return
1538 0 and STORE_FLAG_VALUE.
1540 2014-01-08 Richard Biener <rguenther@suse.de>
1543 * gimple.h (is_gimple_builtin_call): Remove.
1544 (gimple_builtin_call_types_compatible_p): New.
1545 (gimple_call_builtin_p): New overload.
1546 * gimple.c (is_gimple_builtin_call): Remove.
1547 (validate_call): Rename to ...
1548 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1550 (validate_type): New static function.
1551 (gimple_call_builtin_p): New overload and adjust.
1552 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1553 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1554 (gimple_fold_stmt_to_constant_1): Likewise.
1555 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1557 2014-01-08 Richard Biener <rguenther@suse.de>
1560 * gimplify.c (gimplify_expr): Gimplify register-register type
1561 VIEW_CONVERT_EXPRs to separate stmts.
1563 2014-01-07 Jeff Law <law@redhat.com>
1566 * ree.c (combine_set_extension): Handle case where source
1567 and destination registers in an extension insn are different.
1568 (combine_reaching_defs): Allow source and destination registers
1569 in extension to be different under limited circumstances.
1570 (add_removable_extension): Remove restriction that the
1571 source and destination registers in the extension are the same.
1572 (find_and_remove_re): Emit a copy from the extension's
1573 destination to its source after the defining insn if
1574 the source and destination registers are different.
1577 * ifcvt.c (merge_if_block): If we are merging a block with more than
1578 one successor with a block with no successors, remove any BARRIER
1579 after the second block.
1581 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1583 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1585 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1588 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1589 for 14-bit register offsets when INT14_OK_STRICT is false.
1591 2014-01-07 Roland Stigge <stigge@antcom.de>
1592 Michael Meissner <meissner@linux.vnet.ibm.com>
1595 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1596 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1598 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1600 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1603 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1605 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1606 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1607 rtx is const0_rtx or not.
1609 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1612 * target-globals.c (save_target_globals): Remove this_fn_optab
1614 * toplev.c: Include optabs.h.
1615 (target_reinit): Temporarily restore the global options if another
1616 set of options are in force.
1618 2014-01-07 Jakub Jelinek <jakub@redhat.com>
1620 PR rtl-optimization/58668
1621 * cfgcleanup.c (flow_find_cross_jump): Don't count
1622 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
1623 to determine what is counted.
1624 (flow_find_head_matching_sequence): Use active_insn_p to determine
1626 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1628 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1629 determine what is counted.
1631 PR tree-optimization/59643
1632 * tree-predcom.c (split_data_refs_to_components): If one dr is
1633 read and one write, determine_offset fails and the write isn't
1634 in the bad component, just put the read into the bad component.
1636 2014-01-07 Mike Stump <mikestump@comcast.net>
1637 Jakub Jelinek <jakub@redhat.com>
1640 * tree-core.h (struct tree_optimization_option): Change optabs
1641 type from unsigned char * to void *.
1642 * optabs.c (init_tree_optimization_optabs): Adjust
1643 TREE_OPTIMIZATION_OPTABS initialization.
1645 2014-01-06 Jakub Jelinek <jakub@redhat.com>
1648 * config/i386/i386.h (struct machine_function): Add
1649 no_drap_save_restore field.
1650 * config/i386/i386.c (ix86_save_reg): Use
1651 !cfun->machine->no_drap_save_restore instead of
1652 crtl->stack_realign_needed.
1653 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
1654 this function clears frame_pointer_needed. Set
1655 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
1656 and DRAP reg is needed.
1658 2014-01-06 Marek Polacek <polacek@redhat.com>
1661 * doc/implement-c.texi: Mention that other integer types are
1662 permitted as bit-field types in strictly conforming mode.
1664 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
1666 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
1669 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
1671 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
1673 2014-01-06 Martin Jambor <mjambor@suse.cz>
1676 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
1678 * ipa-prop.c (ipa_print_node_params): Fix indentation.
1680 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
1684 * var-tracking.c (add_stores): Preserve the value of the source even if
1685 we don't record the store.
1687 2014-01-06 Terry Guo <terry.guo@arm.com>
1689 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
1691 2014-01-05 Iain Sandoe <iain@codesourcery.com>
1694 * config/darwin.c (darwin_function_section): Adjust return values to
1695 correspond to optimisation changes made in r206070.
1697 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
1699 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
1700 from prefetch_block tune setting.
1701 (nocona_cost): Correct size of prefetch block to 64.
1703 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
1705 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
1706 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
1707 used to save the static chain register in the computation of the offset
1708 from which the FP registers need to be restored.
1710 2014-01-04 Jakub Jelinek <jakub@redhat.com>
1712 PR tree-optimization/59519
1713 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
1714 ICE if get_current_def (current_new_name) is already non-NULL, as long
1715 as it is a phi result of some other phi in *new_exit_bb that has
1718 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
1719 or vmovdqu* for misaligned_operand.
1720 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
1721 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
1722 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
1723 aligned_mem for AVX512F masked aligned load and store builtins and for
1726 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
1728 PR tree-optimization/59651
1729 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
1730 Address range for negative step should be added by TYPE_SIZE_UNIT.
1732 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
1734 * config/m68k/m68k.c (handle_move_double): Handle pushes with
1735 overlapping registers also for registers other than the stack pointer.
1737 2014-01-03 Marek Polacek <polacek@redhat.com>
1740 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
1743 2014-01-03 Jakub Jelinek <jakub@redhat.com>
1746 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
1749 * config/i386/i386.md (MODE_SIZE): New mode attribute.
1750 (push splitter): Use <P:MODE_SIZE> instead of
1751 GET_MODE_SIZE (<P:MODE>mode).
1752 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
1753 (mov -1, reg peephole2): Likewise.
1754 * config/i386/sse.md (*mov<mode>_internal,
1755 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
1756 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
1757 *<code><mode>3, *andnot<mode>3<mask_name>,
1758 <mask_codefor><code><mode>3<mask_name>): Likewise.
1759 * config/i386/subst.md (mask_mode512bit_condition,
1760 sd_mask_mode512bit_condition): Likewise.
1762 2014-01-02 Xinliang David Li <davidxl@google.com>
1764 PR tree-optimization/59303
1765 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
1766 (dump_predicates): Better output format.
1767 (pred_equal_p): New function.
1768 (is_neq_relop_p): Ditto.
1769 (is_neq_zero_form_p): Ditto.
1770 (pred_expr_equal_p): Ditto.
1771 (pred_neg_p): Ditto.
1772 (simplify_pred): Ditto.
1773 (simplify_preds_2): Ditto.
1774 (simplify_preds_3): Ditto.
1775 (simplify_preds_4): Ditto.
1776 (simplify_preds): Ditto.
1778 (push_to_worklist): Ditto.
1779 (get_pred_info_from_cmp): Ditto.
1780 (is_degenerated_phi): Ditto.
1781 (normalize_one_pred_1): Ditto.
1782 (normalize_one_pred): Ditto.
1783 (normalize_one_pred_chain): Ditto.
1784 (normalize_preds): Ditto.
1785 (normalize_cond_1): Remove function.
1786 (normalize_cond): Ditto.
1787 (is_gcond_subset_of): Ditto.
1788 (is_subset_of_any): Ditto.
1789 (is_or_set_subset_of): Ditto.
1790 (is_and_set_subset_of): Ditto.
1791 (is_norm_cond_subset_of): Ditto.
1792 (pred_chain_length_cmp): Ditto.
1793 (convert_control_dep_chain_into_preds): Type change.
1794 (find_predicates): Ditto.
1795 (find_def_preds): Ditto.
1796 (destroy_predicates_vecs): Ditto.
1797 (find_matching_predicates_in_rest_chains): Ditto.
1798 (use_pred_not_overlap_with_undef_path_pred): Ditto.
1799 (is_pred_expr_subset): Ditto.
1800 (is_pred_chain_subset_of): Ditto.
1801 (is_included_in): Ditto.
1802 (is_superset_of): Ditto.
1804 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1806 Update copyright years.
1808 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1810 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
1811 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
1812 config/arc/arc.md, config/arc/arc.opt,
1813 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
1814 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
1815 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
1816 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
1817 config/linux-protos.h, config/linux.c, config/winnt-c.c,
1818 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
1819 vtable-verify.c, vtable-verify.h: Use the standard form for the
1822 2014-01-02 Tobias Burnus <burnus@net-b.de>
1824 * gcc.c (process_command): Update copyright notice dates.
1825 * gcov-dump.c: Ditto.
1827 * doc/cpp.texi: Bump @copying's copyright year.
1828 * doc/cppinternals.texi: Ditto.
1829 * doc/gcc.texi: Ditto.
1830 * doc/gccint.texi: Ditto.
1831 * doc/gcov.texi: Ditto.
1832 * doc/install.texi: Ditto.
1833 * doc/invoke.texi: Ditto.
1835 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1837 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
1839 2014-01-01 Jakub Jelinek <jakub@redhat.com>
1841 * config/i386/sse.md (*mov<mode>_internal): Guard
1842 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
1844 PR rtl-optimization/59647
1845 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
1846 new_rtx into UNSIGNED_FLOAT rtxes.
1848 Copyright (C) 2014 Free Software Foundation, Inc.
1850 Copying and distribution of this file, with or without modification,
1851 are permitted in any medium without royalty provided the copyright
1852 notice and this notice are preserved.