1 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3 * common/config/arm/arm-common.c
4 (arm_rewrite_mcpu): Handle multiple names.
6 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
10 * gimple-builder.h (create_gimple_tmp): Delete.
12 2014-01-27 Christian Bruel <christian.bruel@st.com>
14 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
17 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
19 * config/pa/pa.md (call): Generate indirect long calls to non-local
20 functions when outputing 32-bit code.
21 (call_value): Likewise except for special call to buggy powf function.
23 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
24 portable runtime and PIC indirect calls.
25 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
26 and PIC call sequences. Use ldo instead of blr to set return register
29 2014-01-25 Walter Lee <walt@tilera.com>
31 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
32 avoid clobbering a live register.
34 2014-01-25 Walter Lee <walt@tilera.com>
36 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
37 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
38 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
39 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
41 2014-01-25 Walter Lee <walt@tilera.com>
43 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
44 arguments on even registers.
45 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
47 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
48 (BIGGEST_ALIGNMENT): Ditto.
49 (BIGGEST_FIELD_ALIGNMENT): Ditto.
51 2014-01-25 Walter Lee <walt@tilera.com>
53 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
54 insns before bundling.
55 * config/tilegx/tilegx.md (tile_network_barrier): Update
58 2014-01-25 Walter Lee <walt@tilera.com>
60 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
61 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
62 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
64 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
66 * config/mips/constraints.md (kl): Delete.
67 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
68 define expands, using...
69 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
70 instructions for MIPS16.
71 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
72 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
74 2014-01-25 Walter Lee <walt@tilera.com>
76 * config/tilepro/tilepro.md (ctzdi2): Use register_operand
81 2014-01-25 Walter Lee <walt@tilera.com>
83 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
84 (TARGET_EXPAND_TO_RTL_HOOK): Define.
86 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
88 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
91 2014-01-25 Jakub Jelinek <jakub@redhat.com>
93 * print-rtl.c (in_call_function_usage): New var.
94 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
95 EXPR_LIST mode as mode and not as reg note name.
98 * cfgloopmanip.c (copy_loop_info): If
99 loop->warned_aggressive_loop_optimizations, make sure
100 the flag is set in target loop too.
102 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
104 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
106 * builtins.def: Likewise.
107 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
108 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
109 * ira.c (ira_setup_eliminable_regset): Likewise.
110 * omp-low.c (gate_expand_omp): Likewise.
111 (execute_lower_omp): Likewise.
112 (diagnose_sb_0): Likewise.
113 (gate_diagnose_omp_blocks): Likewise.
114 (simd_clone_clauses_extract): Likewise.
117 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
119 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
120 correction for little endian...
121 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
124 2014-01-24 Jeff Law <law@redhat.com>
126 PR tree-optimization/59919
127 * tree-vrp.c (find_assert_locations_1): Do not register asserts
128 for non-returning calls.
130 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
132 * common/config/aarch64/aarch64-common.c
133 (aarch64_rewrite_mcpu): Handle multiple names.
134 * config/aarch64/aarch64.h
135 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
137 2014-01-24 Dodji Seketeli <dodji@redhat.com>
139 * input.c (add_file_to_cache_tab): Handle the case where fopen
142 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
145 * config/i386/i386.md (pushsf splitter): Get stack adjustment
146 from push operand if code of push isn't PRE_DEC.
148 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
151 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
152 -mquad-memory-atomic. Update -mquad-memory documentation to say
153 it is only used for non-atomic loads/stores.
155 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
156 -mquad-memory or -mquad-memory-atomic switches.
158 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
159 -mquad-memory-atomic to ISA 2.07 support.
161 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
162 to separate support of normal quad word memory operations (ldq,
163 stq) from the atomic quad word memory operations.
165 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
166 support to separate non-atomic quad word operations from atomic
167 quad word operations. Disable non-atomic quad word operations in
168 little endian mode so that we don't have to swap words after the
169 load and before the store.
170 (quad_load_store_p): Add comment about atomic quad word support.
171 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
172 options printed with -mdebug=reg.
174 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
175 -mquad-memory-atomic as the test for whether we have quad word
177 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic,
178 -mquad-memory, or -mp8-vector are used, allow byte/half-word
181 * config/rs6000/sync.md (load_lockedti): Insure that the address
182 is a proper indexed or indirect address for the lqarx instruction.
183 On little endian systems, swap the hi/lo registers after the lqarx
185 (load_lockedpti): Use indexed_or_indirect_operand predicate to
186 insure the address is valid for the lqarx instruction.
187 (store_conditionalti): Insure that the address is a proper indexed
188 or indirect address for the stqcrx. instruction. On little endian
189 systems, swap the hi/lo registers before doing the stqcrx.
191 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
192 insure the address is valid for the stqcrx. instruction.
194 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
195 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
196 type of quad memory support is available.
198 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
201 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
202 there is a danger of looping.
204 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
206 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
207 force flag_ira_loop_pressure if set via command line.
209 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
211 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
212 (ashr_simd): New builtin handling DI mode.
213 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
214 (aarch64_sshr_simddi): New match pattern.
215 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
216 (vshrd_n_s64): Likewise.
217 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
219 2014-01-23 Nick Clifton <nickc@redhat.com>
221 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
222 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
223 favour of mcu specific scripts.
224 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
227 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
228 Alex Velenko <Alex.Velenko@arm.com>
230 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
231 (vaddv_s16): Likewise.
232 (vaddv_s32): Likewise.
233 (vaddv_u8): Likewise.
234 (vaddv_u16): Likewise.
235 (vaddv_u32): Likewise.
236 (vaddvq_s8): Likewise.
237 (vaddvq_s16): Likewise.
238 (vaddvq_s32): Likewise.
239 (vaddvq_s64): Likewise.
240 (vaddvq_u8): Likewise.
241 (vaddvq_u16): Likewise.
242 (vaddvq_u32): Likewise.
243 (vaddvq_u64): Likewise.
244 (vaddv_f32): Likewise.
245 (vaddvq_f32): Likewise.
246 (vaddvq_f64): Likewise.
247 (vmaxv_f32): Likewise.
248 (vmaxv_s8): Likewise.
249 (vmaxv_s16): Likewise.
250 (vmaxv_s32): Likewise.
251 (vmaxv_u8): Likewise.
252 (vmaxv_u16): Likewise.
253 (vmaxv_u32): Likewise.
254 (vmaxvq_f32): Likewise.
255 (vmaxvq_f64): Likewise.
256 (vmaxvq_s8): Likewise.
257 (vmaxvq_s16): Likewise.
258 (vmaxvq_s32): Likewise.
259 (vmaxvq_u8): Likewise.
260 (vmaxvq_u16): Likewise.
261 (vmaxvq_u32): Likewise.
262 (vmaxnmv_f32): Likewise.
263 (vmaxnmvq_f32): Likewise.
264 (vmaxnmvq_f64): Likewise.
265 (vminv_f32): Likewise.
266 (vminv_s8): Likewise.
267 (vminv_s16): Likewise.
268 (vminv_s32): Likewise.
269 (vminv_u8): Likewise.
270 (vminv_u16): Likewise.
271 (vminv_u32): Likewise.
272 (vminvq_f32): Likewise.
273 (vminvq_f64): Likewise.
274 (vminvq_s8): Likewise.
275 (vminvq_s16): Likewise.
276 (vminvq_s32): Likewise.
277 (vminvq_u8): Likewise.
278 (vminvq_u16): Likewise.
279 (vminvq_u32): Likewise.
280 (vminnmv_f32): Likewise.
281 (vminnmvq_f32): Likewise.
282 (vminnmvq_f64): Likewise.
284 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
286 * config/aarch64/aarch64-simd.md
287 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
288 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
289 (*aarch64_mul3_elt<mode>): Likewise.
290 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
291 (*aarch64_mul3_elt_to_64v2df): Likewise.
292 (*aarch64_mla_elt<mode>): Likewise.
293 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
294 (*aarch64_mls_elt<mode>): Likewise.
295 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
296 (*aarch64_fma4_elt<mode>): Likewise.
297 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
298 (*aarch64_fma4_elt_to_64v2df): Likewise.
299 (*aarch64_fnma4_elt<mode>): Likewise.
300 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
301 (*aarch64_fnma4_elt_to_64v2df): Likewise.
302 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
303 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
304 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
305 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
306 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
307 (aarch64_sqdmull_lane<mode>_internal): Likewise.
308 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
310 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
312 * config/aarch64/aarch64-simd.md
313 (aarch64_be_checked_get_lane<mode>): New define_expand.
314 * config/aarch64/aarch64-simd-builtins.def
315 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0):
316 New builtin definition.
317 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
318 Use new safe be builtin.
320 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
322 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
324 (aarch64_be_st1<mode>): Likewise.
325 (aarch_ld1<VALL:mode>): Define_expand modified.
326 (aarch_st1<VALL:mode>): Likewise.
327 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
328 (UNSPEC_ST1): Likewise.
330 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
332 * config/microblaze/microblaze.md: Add trap insn and attribute
334 2014-01-23 Dodji Seketeli <dodji@redhat.com>
336 PR preprocessor/58580
337 * input.h (location_get_source_line): Take an additional line_size
339 (void diagnostics_file_cache_fini): Declare new function.
340 * input.c (struct fcache): New type.
341 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
342 New static constants.
343 (diagnostic_file_cache_init, total_lines_num)
344 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
345 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
346 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
347 (get_next_line, read_next_line, goto_next_line, read_line_num):
348 New static function definitions.
349 (diagnostic_file_cache_fini): New function.
350 (location_get_source_line): Take an additional output line_len
351 parameter. Re-write using lookup_or_add_file_to_cache_tab and
353 * diagnostic.c (diagnostic_finish): Call
354 diagnostic_file_cache_fini.
355 (adjust_line): Take an additional input parameter for the length
356 of the line, rather than calculating it with strlen.
357 (diagnostic_show_locus): Adjust the use of
358 location_get_source_line and adjust_line with respect to their new
359 signature. While displaying a line now, do not stop at the first
360 null byte. Rather, display the zero byte as a space and keep
361 going until we reach the size of the line.
362 * Makefile.in: Add vec.o to OBJS-libcommon
364 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
365 Ilya Tocar <ilya.tocar@intel.com>
367 * config/i386/avx512fintrin.h (_mm512_kmov): New.
368 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
369 (__builtin_ia32_kmov16): Ditto.
370 * config/i386/i386.md (UNSPEC_KMOV): New.
373 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
375 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
376 (_mm512_storeu_si512): Ditto.
378 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
381 * rtl.h (get_referenced_operands): Declare.
382 * recog.c (get_referenced_operands): New function.
383 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
384 operands have been referenced when recording LO_SUM references.
386 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
388 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
390 2014-01-22 Jan Hubicka <jh@suse.cz>
392 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
393 Enable for generic and recent AMD targets.
395 2014-01-22 Jan Hubicka <jh@suse.cz>
397 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
398 ARG_SIZE note when adjustment was eliminated.
400 2014-01-22 Jeff Law <law@redhat.com>
402 PR tree-optimization/59597
403 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
404 in file. Accept new argument REGISTERING and use it to modify
405 dump output appropriately.
406 (register_jump_thread): Corresponding changes.
407 (mark_threaded_blocks): Reinstate code to cancel unprofitable
408 thread paths involving joiner blocks. Add code to dump cancelled
409 jump threading paths.
411 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
413 PR rtl-optimization/59477
414 * lra-constraints.c (inherit_in_ebb): Process call for living hard
415 regs. Update reloads_num and potential_reload_hard_regs for all
418 2014-01-22 Tom Tromey <tromey@redhat.com>
420 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
422 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
424 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
426 PR rtl-optimization/59896
427 * lra-constraints.c (process_alt_operands): Check unused note for
428 matched operands of insn with no output reloads.
430 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
432 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
433 (mips_move_from_gpr_cost): Likewise.
435 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
437 PR rtl-optimization/59858
438 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
439 ira_class_hard_regs_num.
440 (process_alt_operands): Increase reject for dying matched operand.
442 2014-01-21 Jakub Jelinek <jakub@redhat.com>
445 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
446 smaller than size, perform several stores or loads and stores
447 at dst + count - size to store or copy all of size bytes, rather
448 than just last modesize bytes.
450 2014-01-20 DJ Delorie <dj@redhat.com>
452 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
453 that CLOBBERs are REGs before propogating their values.
455 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
458 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
459 (cgraph_inline_failed_type): New function.
460 * cgraph.h (DEFCIFCODE): Add type.
461 (cgraph_inline_failed_type_t): New enum.
462 (cgraph_inline_failed_type): New prototype.
463 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
464 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
465 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
466 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
467 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
468 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
469 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
470 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
471 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
472 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
473 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
474 OPTIMIZATION_MISMATCH.
475 * tree-inline.c (expand_call_inline): Emit errors during
476 early_inlining if cgraph_inline_failed_type returns
479 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
482 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
483 mode attribute in insn output.
485 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
487 * output.h (output_constant): Delete.
488 * varasm.c (output_constant): Make private.
490 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
492 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
494 2014-01-20 Jakub Jelinek <jakub@redhat.com>
497 * tree.h (fold_builtin_strcat): New prototype.
498 * builtins.c (fold_builtin_strcat): No longer static. Add len
499 argument, if non-NULL, don't call c_strlen. Optimize
500 directly into __builtin_memcpy instead of __builtin_strcpy.
501 (fold_builtin_2): Adjust fold_builtin_strcat caller.
502 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
504 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
506 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
507 for SImode_address_operand operands, having only a REG argument.
509 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
511 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
512 loader name using mbig-endian.
513 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
515 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
517 * doc/invoke.texi (-march): Clarify documentation for AArch64.
521 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
523 * config/aarch64/aarch64-protos.h
524 (aarch64_cannot_change_mode_class_ptr): Declare.
525 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
526 aarch64_cannot_change_mode_class_ptr): New.
527 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
528 backend hook aarch64_cannot_change_mode_class.
530 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
532 * common/config/aarch64/aarch64-common.c
533 (aarch64_handle_option): Don't handle any option order logic here.
534 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
535 selected_cpu, warn on architecture version mismatch.
536 (aarch64_override_options): Fix parsing order for option strings.
538 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
539 Iain Sandoe <iain@codesourcery.com>
542 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
543 warning. Amend comment to reflect current functionality.
545 2014-01-20 Richard Biener <rguenther@suse.de>
548 * builtins.c (fold_builtin_strcat): Remove case better handled
549 by tree-ssa-strlen.c.
551 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
553 * config/aarch64/aarch64.opt
554 (mcpu, march, mtune): Make case-insensitive.
556 2014-01-20 Jakub Jelinek <jakub@redhat.com>
559 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
560 if operands[1] is a REG or ZERO_EXTEND of a REG.
562 2014-01-19 Jan Hubicka <jh@suse.cz>
564 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
566 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
568 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
569 long non-pic millicode calls.
571 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
573 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
575 2014-01-19 Kito Cheng <kito@0xlab.org>
577 * builtins.c (expand_movstr): Check movstr expand done or fail.
579 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
580 H.J. Lu <hongjiu.lu@intel.com>
583 * config/i386/i386.md (*lea<mode>): Zero-extend return register
584 to DImode for zero-extended addresses.
586 2014-01-19 Jakub Jelinek <jakub@redhat.com>
588 PR rtl-optimization/57763
589 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
590 on the new indirect jump_insn and increment LABEL_NUSES (label).
592 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
596 * config.gcc (x86_archs): New variable.
597 (x86_64_archs): Likewise.
598 (x86_cpus): Likewise.
599 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
600 --with-arch/--with-cpu= options.
601 Support --with-arch=/--with-cpu={nehalem,westmere,
602 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
604 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
606 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
607 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
609 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
611 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
613 2014-01-18 Jakub Jelinek <jakub@redhat.com>
616 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
617 clear cpp_get_options (parse_in)->warn_unused_macros for
618 ix86_target_macros_internal with cpp_define.
620 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
622 * jump.c (delete_related_insns): Keep (use (insn))s.
623 * reorg.c (redundant_insn): Check for barriers too.
625 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
627 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
629 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
631 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
632 call to $$dyncall when TARGET_LONG_CALLS is true.
634 2014-01-17 Jeff Law <law@redhat.com>
636 * ree.c (combine_set_extension): Temporarily disable test for
637 changing number of hard registers.
639 2014-01-17 Jan Hubicka <jh@suse.cz>
642 * ipa-inline-analysis.c (inline_free_summary):
643 Do not free summary of aliases.
645 2014-01-17 Jakub Jelinek <jakub@redhat.com>
648 * gimplify.c (gimplify_expr): Use create_tmp_var
649 instead of create_tmp_var_raw. If cond doesn't have
650 integral type, don't add the IFN_ANNOTATE builtin at all.
652 2014-01-17 Martin Jambor <mjambor@suse.cz>
655 * ipa-cp.c (prev_edge_clone): New variable.
656 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
657 Also resize prev_edge_clone vector.
658 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
659 (ipcp_edge_removal_hook): New function.
660 (ipcp_driver): Register ipcp_edge_removal_hook.
662 2014-01-17 Andrew Pinski <apinski@cavium.com>
663 Steve Ellcey <sellcey@mips.com>
666 * config/mips/mips.c (mips_print_operand): Check operand mode instead
669 2014-01-17 Jeff Law <law@redhat.com>
672 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
673 so that pass_ccp runs first.
675 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
677 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
678 (ix86_adjust_cost): Use !TARGET_XXX.
679 (do_reorder_for_imul): Likewise.
680 (swap_top_of_ready_list): Likewise.
681 (ix86_sched_reorder): Likewise.
683 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
685 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
686 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
687 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
688 (intel_memset): New. Duplicate slm_memset.
689 (intel_cost): New. Duplicate slm_cost.
690 (m_INTEL): New macro.
691 (processor_target_table): Add "intel".
692 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
693 with PROCESSOR_INTEL for "intel".
694 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
695 PROCESSOR_SILVERMONT.
696 (ix86_issue_rate): Likewise.
697 (ix86_adjust_cost): Likewise.
698 (ia32_multipass_dfa_lookahead): Likewise.
699 (swap_top_of_ready_list): Likewise.
700 (ix86_sched_reorder): Likewise.
701 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
702 instead of TARGET_OPT_AGU.
703 * config/i386/i386.h (TARGET_INTEL): New.
704 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
705 (processor_type): Add PROCESSOR_INTEL.
706 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
707 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
709 2014-01-17 Marek Polacek <polacek@redhat.com>
712 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
715 2014-01-17 Richard Biener <rguenther@suse.de>
717 PR tree-optimization/46590
718 * opts.c (default_options_table): Add entries for
719 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
720 all enabled at -O1 but not for -Og.
721 * common.opt (fbranch-count-reg): Remove Init(1).
722 (fmove-loop-invariants): Likewise.
723 (ftree-pta): Likewise.
725 2014-01-17 Jakub Jelinek <jakub@redhat.com>
727 * config/i386/i386.c (ix86_data_alignment): For compatibility with
728 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
729 decls to at least the GCC 4.8 used alignments.
732 * tree-nested.c (convert_nonlocal_reference_stmt,
733 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
734 of GIMPLE_BIND stmts, adjust associated decls.
736 2014-01-17 Richard Biener <rguenther@suse.de>
738 PR tree-optimization/46590
739 * vec.h (vec<>::bseach): New member function implementing
740 binary search according to C89 bsearch.
741 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
742 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
743 bitmap pointer again. Make accesses_in_loop a flat array.
744 (mem_ref_obstack): New global.
745 (outermost_indep_loop): Adjust for mem_ref->stored changes.
746 (mark_ref_stored): Likewise.
747 (ref_indep_loop_p_2): Likewise.
748 (set_ref_stored_in_loop): New helper function.
749 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
750 (memref_free): Adjust.
751 (record_mem_ref_loc): Simplify.
752 (gather_mem_refs_stmt): Adjust.
753 (sort_locs_in_loop_postorder_cmp): New function.
754 (analyze_memory_references): Sort accesses_in_loop after
755 loop postorder number.
756 (find_ref_loc_in_loop_cmp): New function.
757 (for_all_locs_in_loop): Find relevant cluster of locs in
758 accesses_in_loop and iterate without recursion.
759 (execute_sm): Avoid uninit warning.
760 (struct ref_always_accessed): Simplify.
761 (ref_always_accessed::operator ()): Likewise.
762 (ref_always_accessed_p): Likewise.
763 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
764 loop postorder numbers here.
765 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
768 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
771 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
772 on decls for which assemble_alias has been called.
774 2014-01-17 Nick Clifton <nickc@redhat.com>
776 * config/msp430/msp430.opt: (mcpu): New option.
777 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
778 (msp430_option_override): Parse target_cpu. If the MCU name
779 matches a generic string, clear target_mcu.
780 (msp430_attr): Allow numeric interrupt values up to 63.
781 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
782 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
784 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
786 * config/msp430/msp430.md (popm): Use %J rather than %I.
787 (addsi3): Use msp430_nonimmediate_operand for operand 2.
788 (addhi_cy_i): Use immediate_operand for operand 2.
789 * doc/invoke.texi: Document -mcpu option.
791 2014-01-17 Richard Biener <rguenther@suse.de>
793 PR rtl-optimization/38518
794 * df.h (df_analyze_loop): Declare.
795 * df-core.c: Include cfgloop.h.
796 (df_analyze_1): Split out main part of df_analyze.
797 (df_analyze): Adjust.
798 (loop_inverted_post_order_compute): New function.
799 (loop_post_order_compute): Likewise.
800 (df_analyze_loop): New function avoiding whole-function
802 * loop-invariant.c (find_defs): Use df_analyze_loop.
803 (find_invariants): Adjust.
804 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
806 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
808 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
809 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
811 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
813 * ipa-ref.c (ipa_remove_stmt_references): Fix references
814 traversal when removing references.
816 2014-01-16 Jan Hubicka <jh@suse.cz>
819 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
821 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
824 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
825 pushing a reload for an autoinc when we had previously reloaded an
826 inner part of the address.
828 2014-01-16 Jakub Jelinek <jakub@redhat.com>
830 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
832 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
833 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
834 when not giving up or versioning for alias only because of
836 (vect_analyze_data_ref_dependences): Set to true.
837 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
839 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
840 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
844 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
847 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
848 operand 0 predicate for gathers, use a new pseudo as subtarget.
850 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
853 * lra-constraints.c (process_alt_operands): Add printing debug info.
854 Check absence of input/output reloads for matched operands too.
856 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
858 PR rtl-optimization/59835
859 * ira.c (ira_init_register_move_cost): Increase cost for
862 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
864 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
866 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
869 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
870 non-register objects. Use gen_(high/low)part more consistently.
873 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
876 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
877 endian support, remove tests for WORDS_BIG_ENDIAN.
878 (p8_mfvsrd_3_<mode>): Likewise.
879 (reload_gpr_from_vsx<mode>): Likewise.
880 (reload_gpr_from_vsxsf): Likewise.
881 (p8_mfvsrd_4_disf): Likewise.
883 2014-01-16 Richard Biener <rguenther@suse.de>
885 PR rtl-optimization/46590
886 * lcm.c (compute_antinout_edge): Use postorder iteration.
887 (compute_laterin): Use inverted postorder iteration.
889 2014-01-16 Nick Clifton <nickc@redhat.com>
892 * varasm.c (output_constant): Return the number of bytes actually
894 (output_constructor_array_range): Update the field size with the
895 number of bytes emitted by output_constant.
896 (output_constructor_regular_field): Likewise. Also do not
897 complain if the total number of bytes emitted is now greater
898 than the expected fieldpos.
899 * output.h (output_constant): Update prototype and descriptive comment.
901 2014-01-16 Marek Polacek <polacek@redhat.com>
904 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
905 it is error_mark_node.
907 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
909 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
910 VALID_AVX256_REG_OR_OI_MODE.
912 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
914 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
915 current procedure should be profiled.
917 2014-01-15 Andrew Pinski <apinski@cavium.com>
919 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
920 of moving from/to the STACK_REG register class.
922 2014-01-15 Richard Henderson <rth@redhat.com>
925 * reginfo.c (global_regs_decl): Globalize.
926 * rtl.h (global_regs_decl): Declare.
927 * ira.c (do_reload): Diagnose frame_pointer_needed and it
928 reserved via global_regs.
930 2014-01-15 Teresa Johnson <tejohnson@google.com>
932 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
934 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
936 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
937 and vmulosh rather than call gen_vec_widen_smult_*.
938 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
939 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
940 (vec_widen_smult_even_v16qi): Likewise.
941 (vec_widen_umult_even_v8hi): Likewise.
942 (vec_widen_smult_even_v8hi): Likewise.
943 (vec_widen_umult_odd_v16qi): Likewise.
944 (vec_widen_smult_odd_v16qi): Likewise.
945 (vec_widen_umult_odd_v8hi): Likewise.
946 (vec_widen_smult_odd_v8hi): Likewise.
947 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
948 vmuloub rather than call gen_vec_widen_umult_*.
949 (vec_widen_umult_lo_v16qi): Likewise.
950 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
951 vmulosb rather than call gen_vec_widen_smult_*.
952 (vec_widen_smult_lo_v16qi): Likewise.
953 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
954 rather than call gen_vec_widen_umult_*.
955 (vec_widen_umult_lo_v8hi): Likewise.
956 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
957 rather than call gen_vec_widen_smult_*.
958 (vec_widen_smult_lo_v8hi): Likewise.
960 2014-01-15 Jeff Law <law@redhat.com>
962 PR tree-optimization/59747
963 * ree.c (find_and_remove_re): Properly handle case where a second
964 eliminated extension requires widening a copy created for elimination
965 of a prior extension.
966 (combine_set_extension): Ensure that the number of hard regs needed
967 for a destination register does not change when we widen it.
969 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
971 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
972 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
973 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
974 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
975 (avr-*-rtems*): Likewise.
976 (bfin*-rtems*): Likewise.
977 (moxie-*-rtems*): Likewise.
978 (h8300-*-rtems*): Likewise.
979 (i[34567]86-*-rtems*): Likewise.
980 (lm32-*-rtems*): Likewise.
981 (m32r-*-rtems*): Likewise.
982 (m68k-*-rtems*): Likewise.
983 (microblaze*-*-rtems*): Likewise.
984 (mips*-*-rtems*): Likewise.
985 (powerpc-*-rtems*): Likewise.
986 (sh-*-rtems*): Likewise.
987 (sparc-*-rtems*): Likewise.
988 (sparc64-*-rtems*): Likewise.
989 (v850-*-rtems*): Likewise.
990 (m32c-*-rtems*): Likewise.
992 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
994 PR rtl-optimization/59511
995 * ira.c (ira_init_register_move_cost): Use memory costs for some
996 cases of register move cost calculations.
997 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
998 instead of BB frequency.
999 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1000 * lra-assigns.c (find_hard_regno_for): Ditto.
1002 2014-01-15 Richard Biener <rguenther@suse.de>
1004 PR tree-optimization/59822
1005 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1006 (vectorizable_load): Use it to hoist defs of uses of invariant
1007 loads out of the loop.
1009 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1010 Kugan Vivekanandarajah <kuganv@linaro.org>
1013 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1016 2014-01-15 Richard Biener <rguenther@suse.de>
1018 PR rtl-optimization/59802
1019 * lcm.c (compute_available): Use inverted postorder to seed
1020 the initial worklist.
1022 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1025 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1026 ADDR_REGS for invalid symrefs in non-PIC code.
1028 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1031 * builtins.c (determine_block_size): Initialize *probable_max_size
1032 even if len_rtx is CONST_INT.
1034 2014-01-14 Andrew Pinski <apinski@cavium.com>
1036 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1037 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1038 (cortexa53_tunings): Likewise.
1039 (aarch64_sched_issue_rate): New function.
1040 (TARGET_SCHED_ISSUE_RATE): Define.
1042 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1044 * ira-costs.c (find_costs_and_classes): Add missed
1045 ira_init_register_move_cost_if_necessary.
1047 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1050 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1052 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1055 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1056 to indicate if type is used for function return value. Warn ABI
1057 change if the vector mode isn't available for function return value.
1058 (ix86_function_arg_advance): Pass false to type_natural_mode.
1059 (ix86_function_arg): Likewise.
1060 (ix86_gimplify_va_arg): Likewise.
1061 (function_arg_32): Don't warn ABI change.
1062 (ix86_function_value): Pass true to type_natural_mode.
1063 (ix86_return_in_memory): Likewise.
1064 (ix86_struct_value_rtx): Removed.
1065 (TARGET_STRUCT_VALUE_RTX): Likewise.
1067 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1069 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1070 converting a conditional jump into a conditional return.
1072 2014-01-14 Richard Biener <rguenther@suse.de>
1074 PR tree-optimization/58921
1075 PR tree-optimization/59006
1076 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1077 hoisting invariant stmts.
1078 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1079 invariant loads on the preheader edge if possible.
1081 2014-01-14 Joey Ye <joey.ye@arm.com>
1083 * doc/plugin.texi (Building GCC plugins): Update to C++.
1085 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1087 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1088 (_mm_rcp28_round_ss): Ditto.
1089 (_mm_rsqrt28_round_sd): Ditto.
1090 (_mm_rsqrt28_round_ss): Ditto.
1091 (_mm_rcp28_sd): Ditto.
1092 (_mm_rcp28_ss): Ditto.
1093 (_mm_rsqrt28_sd): Ditto.
1094 (_mm_rsqrt28_ss): Ditto.
1095 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1096 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1097 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1098 (IX86_BUILTIN_RCP28SD): Ditto.
1099 (IX86_BUILTIN_RCP28SS): Ditto.
1100 (IX86_BUILTIN_RSQRT28SD): Ditto.
1101 (IX86_BUILTIN_RSQRT28SS): Ditto.
1102 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1103 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1104 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1105 (ix86_expand_special_args_builtin): Expand new FTYPE.
1106 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1107 (srcp14<mode>): Make insn unary.
1108 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1109 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1110 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1111 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1112 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1113 Fix rounding: make it SAE only.
1114 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1116 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1118 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1119 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1120 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1121 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1122 (round_saeonly_mask_scalar_operand4): Ditto.
1123 (round_saeonly_mask_scalar_op3): Ditto.
1124 (round_saeonly_mask_scalar_op4): Ditto.
1126 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1128 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1129 Implement -maltivec=be for vec_insert and vec_extract.
1131 2014-01-10 DJ Delorie <dj@redhat.com>
1133 * config/msp430/msp430.md (call_internal): Don't allow memory
1134 references with SP as the base register.
1135 (call_value_internal): Likewise.
1136 * config/msp430/constraints.md (Yc): New. For memory references
1137 that don't use SP as a base register.
1139 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1140 "an integer without a # prefix"
1141 * config/msp430/msp430.md (epilogue_helper): Use it.
1143 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1146 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1147 AVX512F gather builtins.
1148 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1149 on gather decls with INTEGER_TYPE masktype.
1150 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1151 directly into the builtin rather than hoisting it before loop.
1153 PR tree-optimization/59387
1154 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1155 (scev_const_prop): If folded_casts and type has undefined overflow,
1156 use force_gimple_operand instead of force_gimple_operand_gsi and
1157 for each added stmt if it is assign with
1158 arith_code_with_undefined_signed_overflow, call
1159 rewrite_to_defined_overflow.
1160 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1161 gimple-fold.h instead.
1162 (arith_code_with_undefined_signed_overflow,
1163 rewrite_to_defined_overflow): Moved to ...
1164 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1165 rewrite_to_defined_overflow): ... here. No longer static.
1166 Include gimplify-me.h.
1167 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1168 rewrite_to_defined_overflow): New prototypes.
1170 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1172 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1174 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1176 * builtins.c (get_object_alignment_2): Minor tweak.
1177 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1179 2014-01-13 Christian Bruel <christian.bruel@st.com>
1181 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1182 optimized non constant lengths.
1184 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1187 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1188 load as __atomic_load_N if possible.
1190 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1192 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1194 (rs6000_expand_builtin): Adjust call.
1196 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1199 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1200 * config/rs6000/rs6000.c: Include target-globals.h.
1201 (rs6000_set_current_function): Instead of doing target_reinit
1202 unconditionally, use save_target_globals_default_opts and
1203 restore_target_globals.
1205 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1207 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1208 (rs6000_expand_builtin): Handle mffs and mtfsf.
1209 (rs6000_init_builtins): Define mffs and mtfsf.
1210 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1211 (rs6000_mffs): New pattern.
1212 (rs6000_mtfsf): New pattern.
1214 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1216 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1217 Start narrowing with START. Apply candidate-use pair
1218 and check overall cost in narrowing.
1219 (iv_ca_prune): Pass new argument.
1221 2014-01-10 Jeff Law <law@redhat.com>
1224 * ree.c (combine_reaching_defs): Ensure the defining statement
1225 occurs before the extension when optimizing extensions with
1226 different source and destination hard registers.
1228 2014-01-10 Jan Hubicka <jh@suse.cz>
1231 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1232 vtables into the type inheritance graph.
1234 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1236 PR rtl-optimization/59754
1237 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1238 modes in the REGNO != REGNO case.
1240 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1242 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1244 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1246 PR tree-optimization/59745
1247 * tree-predcom.c (tree_predictive_commoning_loop): Call
1248 free_affine_expand_cache if giving up because components is NULL.
1250 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1251 GC in payload of target_globals struct instead of allocating them on
1252 the heap and the larger structs separately using GC.
1253 * target-globals.h (struct target_globals): Make regs, hard_regs,
1254 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1255 of GTY((skip)) and change type to void *.
1256 (reset_target_globals): Cast loads from those fields to corresponding
1259 2014-01-10 Steve Ellcey <sellcey@mips.com>
1262 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1263 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1264 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1266 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1269 * aarch64-modes.def (CC_Zmode): New flags mode.
1270 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1271 represents an equality.
1272 (aarch64_get_condition_code): Handle CC_Zmode.
1273 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1275 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1277 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1278 extraction in good case.
1280 2014-01-10 Richard Biener <rguenther@suse.de>
1282 PR tree-optimization/59374
1283 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1284 checking after SLP discovery. Mark stmts not participating
1285 in any SLP instance properly.
1287 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1289 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1290 when handling a SET rtx.
1292 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1294 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1295 (cortex-a57): Likewise.
1296 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1298 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1300 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1301 non-iwmmxt builtins.
1303 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1307 * ipa-devirt.c record_target_from_binfo): Take as argument
1308 stack of binfos and lookup matching one for virtual inheritance.
1309 (possible_polymorphic_call_targets_1): Update.
1311 2014-01-10 Huacai Chen <chenhc@lemote.com>
1313 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1314 kernel strings for Loongson-2E/2F/3A.
1316 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1319 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1320 is_gimple_call before calling gimple_call_internal_p.
1322 2014-01-09 Steve Ellcey <sellcey@mips.com>
1324 * Makefile.in (TREE_FLOW_H): Remove.
1325 (TREE_SSA_H): Add file names from tree-flow.h.
1326 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1327 * tree.h: Remove tree-flow.h reference.
1328 * hash-table.h: Remove tree-flow.h reference.
1329 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1330 reference with tree-ssa-loop.h.
1332 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1334 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1335 default element-order behavior for -maltivec.
1336 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1337 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1338 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1339 when targeting big endian, at least for now.
1340 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1342 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1345 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1346 var satisfies use_register_for_decl, just take into account type
1347 alignment, rather than decl alignment.
1349 PR tree-optimization/59622
1350 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1351 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1352 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1353 Don't devirtualize for inplace at all. For targets.length () == 1,
1354 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1356 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1358 * config/i386/i386.md (cpu): Remove the unused btver1.
1360 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1362 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1364 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1367 * tree-core.h (struct target_globals): New forward declaration.
1368 (struct tree_target_option): Add globals field.
1369 * tree.h (TREE_TARGET_GLOBALS): Define.
1370 (prepare_target_option_nodes_for_pch): New prototype.
1371 * target-globals.h (struct target_globals): Define even if
1373 * tree.c (prepare_target_option_node_for_pch,
1374 prepare_target_option_nodes_for_pch): New functions.
1375 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1376 * config/i386/i386.c: Include target-globals.h.
1377 (ix86_set_current_function): Instead of doing target_reinit
1378 unconditionally, use save_target_globals_default_opts and
1379 restore_target_globals.
1381 2014-01-09 Richard Biener <rguenther@suse.de>
1383 PR tree-optimization/59715
1384 * tree-cfg.h (split_critical_edges): Declare.
1385 * tree-cfg.c (split_critical_edges): Export.
1386 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1388 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1390 * cfgexpand.c (expand_stack_vars): Optionally disable
1391 asan stack protection.
1392 (expand_used_vars): Likewise.
1393 (partition_stack_vars): Likewise.
1394 * asan.c (asan_emit_stack_protection): Optionally disable
1395 after return stack usage.
1396 (instrument_derefs): Optionally disable memory access instrumentation.
1397 (instrument_builtin_call): Likewise.
1398 (instrument_strlen_call): Likewise.
1399 (asan_protect_global): Optionally disable global variables protection.
1400 * doc/invoke.texi: Added doc for new options.
1401 * params.def: Added new options.
1402 * params.h: Likewise.
1404 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1406 PR rtl-optimization/59724
1407 * ifcvt.c (cond_exec_process_if_block): Don't call
1408 flow_find_head_matching_sequence with 0 longest_match.
1409 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1410 non-active insns if !stop_after.
1411 (try_head_merge_bb): Revert 2014-01-07 changes.
1413 2014-01-08 Jeff Law <law@redhat.com>
1415 * ree.c (get_sub_rtx): New function, extracted from...
1416 (merge_def_and_ext): Here.
1417 (combine_reaching_defs): Use get_sub_rtx.
1419 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1421 * cgraph.h (varpool_variable_node): Do not choke on null node.
1423 2014-01-08 Catherine Moore <clm@codesourcery.com>
1425 * config/mips/mips.md (simple_return): Attempt to use JRC
1427 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1429 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1431 PR rtl-optimization/59137
1432 * reorg.c (steal_delay_list_from_target): Call update_block for
1434 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1436 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1438 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1439 two duplicate entries.
1441 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1444 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1446 * config/mips/mips.c (mips_truncated_op_cost): New function.
1447 (mips_rtx_costs): Adjust test for BADDU.
1448 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1450 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1452 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1453 (*baddu_si): ...this new pattern.
1455 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1458 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1460 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1463 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1465 (expand_expr, expand_normal): Adjust.
1466 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1467 inner_reference_p. Use inner_reference_p to expand inner references.
1468 (store_expr): Adjust.
1469 * cfgexpand.c (expand_call_stmt): Adjust.
1471 2014-01-08 Rong Xu <xur@google.com>
1473 * gcov-io.c (gcov_var): Move from gcov-io.h.
1474 (gcov_position): Ditto.
1475 (gcov_is_error): Ditto.
1476 (gcov_rewrite): Ditto.
1477 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1478 only part to libgcc/libgcov.h.
1480 2014-01-08 Marek Polacek <polacek@redhat.com>
1483 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1485 2014-01-08 Marek Polacek <polacek@redhat.com>
1488 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1490 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1492 PR rtl-optimization/59649
1493 * stor-layout.c (get_mode_bounds): For BImode return
1494 0 and STORE_FLAG_VALUE.
1496 2014-01-08 Richard Biener <rguenther@suse.de>
1499 * gimple.h (is_gimple_builtin_call): Remove.
1500 (gimple_builtin_call_types_compatible_p): New.
1501 (gimple_call_builtin_p): New overload.
1502 * gimple.c (is_gimple_builtin_call): Remove.
1503 (validate_call): Rename to ...
1504 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1506 (validate_type): New static function.
1507 (gimple_call_builtin_p): New overload and adjust.
1508 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1509 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1510 (gimple_fold_stmt_to_constant_1): Likewise.
1511 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1513 2014-01-08 Richard Biener <rguenther@suse.de>
1516 * gimplify.c (gimplify_expr): Gimplify register-register type
1517 VIEW_CONVERT_EXPRs to separate stmts.
1519 2014-01-07 Jeff Law <law@redhat.com>
1522 * ree.c (combine_set_extension): Handle case where source
1523 and destination registers in an extension insn are different.
1524 (combine_reaching_defs): Allow source and destination registers
1525 in extension to be different under limited circumstances.
1526 (add_removable_extension): Remove restriction that the
1527 source and destination registers in the extension are the same.
1528 (find_and_remove_re): Emit a copy from the extension's
1529 destination to its source after the defining insn if
1530 the source and destination registers are different.
1533 * ifcvt.c (merge_if_block): If we are merging a block with more than
1534 one successor with a block with no successors, remove any BARRIER
1535 after the second block.
1537 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1539 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1541 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1544 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1545 for 14-bit register offsets when INT14_OK_STRICT is false.
1547 2014-01-07 Roland Stigge <stigge@antcom.de>
1548 Michael Meissner <meissner@linux.vnet.ibm.com>
1551 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1552 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1554 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1556 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1559 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1561 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1562 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1563 rtx is const0_rtx or not.
1565 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1568 * target-globals.c (save_target_globals): Remove this_fn_optab
1570 * toplev.c: Include optabs.h.
1571 (target_reinit): Temporarily restore the global options if another
1572 set of options are in force.
1574 2014-01-07 Jakub Jelinek <jakub@redhat.com>
1576 PR rtl-optimization/58668
1577 * cfgcleanup.c (flow_find_cross_jump): Don't count
1578 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
1579 to determine what is counted.
1580 (flow_find_head_matching_sequence): Use active_insn_p to determine
1582 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1584 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1585 determine what is counted.
1587 PR tree-optimization/59643
1588 * tree-predcom.c (split_data_refs_to_components): If one dr is
1589 read and one write, determine_offset fails and the write isn't
1590 in the bad component, just put the read into the bad component.
1592 2014-01-07 Mike Stump <mikestump@comcast.net>
1593 Jakub Jelinek <jakub@redhat.com>
1596 * tree-core.h (struct tree_optimization_option): Change optabs
1597 type from unsigned char * to void *.
1598 * optabs.c (init_tree_optimization_optabs): Adjust
1599 TREE_OPTIMIZATION_OPTABS initialization.
1601 2014-01-06 Jakub Jelinek <jakub@redhat.com>
1604 * config/i386/i386.h (struct machine_function): Add
1605 no_drap_save_restore field.
1606 * config/i386/i386.c (ix86_save_reg): Use
1607 !cfun->machine->no_drap_save_restore instead of
1608 crtl->stack_realign_needed.
1609 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
1610 this function clears frame_pointer_needed. Set
1611 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
1612 and DRAP reg is needed.
1614 2014-01-06 Marek Polacek <polacek@redhat.com>
1617 * doc/implement-c.texi: Mention that other integer types are
1618 permitted as bit-field types in strictly conforming mode.
1620 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
1622 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
1625 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
1627 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
1629 2014-01-06 Martin Jambor <mjambor@suse.cz>
1632 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
1634 * ipa-prop.c (ipa_print_node_params): Fix indentation.
1636 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
1640 * var-tracking.c (add_stores): Preserve the value of the source even if
1641 we don't record the store.
1643 2014-01-06 Terry Guo <terry.guo@arm.com>
1645 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
1647 2014-01-05 Iain Sandoe <iain@codesourcery.com>
1650 * config/darwin.c (darwin_function_section): Adjust return values to
1651 correspond to optimisation changes made in r206070.
1653 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
1655 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
1656 from prefetch_block tune setting.
1657 (nocona_cost): Correct size of prefetch block to 64.
1659 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
1661 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
1662 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
1663 used to save the static chain register in the computation of the offset
1664 from which the FP registers need to be restored.
1666 2014-01-04 Jakub Jelinek <jakub@redhat.com>
1668 PR tree-optimization/59519
1669 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
1670 ICE if get_current_def (current_new_name) is already non-NULL, as long
1671 as it is a phi result of some other phi in *new_exit_bb that has
1674 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
1675 or vmovdqu* for misaligned_operand.
1676 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
1677 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
1678 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
1679 aligned_mem for AVX512F masked aligned load and store builtins and for
1682 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
1684 PR tree-optimization/59651
1685 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
1686 Address range for negative step should be added by TYPE_SIZE_UNIT.
1688 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
1690 * config/m68k/m68k.c (handle_move_double): Handle pushes with
1691 overlapping registers also for registers other than the stack pointer.
1693 2014-01-03 Marek Polacek <polacek@redhat.com>
1696 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
1699 2014-01-03 Jakub Jelinek <jakub@redhat.com>
1702 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
1705 * config/i386/i386.md (MODE_SIZE): New mode attribute.
1706 (push splitter): Use <P:MODE_SIZE> instead of
1707 GET_MODE_SIZE (<P:MODE>mode).
1708 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
1709 (mov -1, reg peephole2): Likewise.
1710 * config/i386/sse.md (*mov<mode>_internal,
1711 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
1712 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
1713 *<code><mode>3, *andnot<mode>3<mask_name>,
1714 <mask_codefor><code><mode>3<mask_name>): Likewise.
1715 * config/i386/subst.md (mask_mode512bit_condition,
1716 sd_mask_mode512bit_condition): Likewise.
1718 2014-01-02 Xinliang David Li <davidxl@google.com>
1720 PR tree-optimization/59303
1721 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
1722 (dump_predicates): Better output format.
1723 (pred_equal_p): New function.
1724 (is_neq_relop_p): Ditto.
1725 (is_neq_zero_form_p): Ditto.
1726 (pred_expr_equal_p): Ditto.
1727 (pred_neg_p): Ditto.
1728 (simplify_pred): Ditto.
1729 (simplify_preds_2): Ditto.
1730 (simplify_preds_3): Ditto.
1731 (simplify_preds_4): Ditto.
1732 (simplify_preds): Ditto.
1734 (push_to_worklist): Ditto.
1735 (get_pred_info_from_cmp): Ditto.
1736 (is_degenerated_phi): Ditto.
1737 (normalize_one_pred_1): Ditto.
1738 (normalize_one_pred): Ditto.
1739 (normalize_one_pred_chain): Ditto.
1740 (normalize_preds): Ditto.
1741 (normalize_cond_1): Remove function.
1742 (normalize_cond): Ditto.
1743 (is_gcond_subset_of): Ditto.
1744 (is_subset_of_any): Ditto.
1745 (is_or_set_subset_of): Ditto.
1746 (is_and_set_subset_of): Ditto.
1747 (is_norm_cond_subset_of): Ditto.
1748 (pred_chain_length_cmp): Ditto.
1749 (convert_control_dep_chain_into_preds): Type change.
1750 (find_predicates): Ditto.
1751 (find_def_preds): Ditto.
1752 (destroy_predicates_vecs): Ditto.
1753 (find_matching_predicates_in_rest_chains): Ditto.
1754 (use_pred_not_overlap_with_undef_path_pred): Ditto.
1755 (is_pred_expr_subset): Ditto.
1756 (is_pred_chain_subset_of): Ditto.
1757 (is_included_in): Ditto.
1758 (is_superset_of): Ditto.
1760 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1762 Update copyright years.
1764 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1766 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
1767 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
1768 config/arc/arc.md, config/arc/arc.opt,
1769 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
1770 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
1771 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
1772 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
1773 config/linux-protos.h, config/linux.c, config/winnt-c.c,
1774 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
1775 vtable-verify.c, vtable-verify.h: Use the standard form for the
1778 2014-01-02 Tobias Burnus <burnus@net-b.de>
1780 * gcc.c (process_command): Update copyright notice dates.
1781 * gcov-dump.c: Ditto.
1783 * doc/cpp.texi: Bump @copying's copyright year.
1784 * doc/cppinternals.texi: Ditto.
1785 * doc/gcc.texi: Ditto.
1786 * doc/gccint.texi: Ditto.
1787 * doc/gcov.texi: Ditto.
1788 * doc/install.texi: Ditto.
1789 * doc/invoke.texi: Ditto.
1791 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1793 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
1795 2014-01-01 Jakub Jelinek <jakub@redhat.com>
1797 * config/i386/sse.md (*mov<mode>_internal): Guard
1798 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
1800 PR rtl-optimization/59647
1801 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
1802 new_rtx into UNSIGNED_FLOAT rtxes.
1804 Copyright (C) 2014 Free Software Foundation, Inc.
1806 Copying and distribution of this file, with or without modification,
1807 are permitted in any medium without royalty provided the copyright
1808 notice and this notice are preserved.