1 2014-01-27 Jeff Law <law@redhat.com>
3 * ipa-inline.c (inline_small_functions): Fix typo.
5 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
7 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
8 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
9 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
10 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
11 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
12 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
13 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
14 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
15 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
16 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
17 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
18 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
19 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
20 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
21 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
22 (_mm512_storeu_epi64): Ditto.
23 (_mm512_cmpge_epi32_mask): Ditto.
24 (_mm512_cmpge_epu32_mask): Ditto.
25 (_mm512_cmpge_epi64_mask): Ditto.
26 (_mm512_cmpge_epu64_mask): Ditto.
27 (_mm512_cmple_epi32_mask): Ditto.
28 (_mm512_cmple_epu32_mask): Ditto.
29 (_mm512_cmple_epi64_mask): Ditto.
30 (_mm512_cmple_epu64_mask): Ditto.
31 (_mm512_cmplt_epi32_mask): Ditto.
32 (_mm512_cmplt_epu32_mask): Ditto.
33 (_mm512_cmplt_epi64_mask): Ditto.
34 (_mm512_cmplt_epu64_mask): Ditto.
35 (_mm512_cmpneq_epi32_mask): Ditto.
36 (_mm512_cmpneq_epu32_mask): Ditto.
37 (_mm512_cmpneq_epi64_mask): Ditto.
38 (_mm512_cmpneq_epu64_mask): Ditto.
39 (_mm512_expand_pd): Ditto.
40 (_mm512_expand_ps): Ditto.
41 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
42 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
43 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
44 * config/i386/i386.c (ix86_builtins): Add
45 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
46 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
47 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
48 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
49 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
50 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
51 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
52 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
53 IX86_BUILTIN_PMOVUSQW512_MEM.
54 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
55 __builtin_ia32_pmovsqd512mem_mask,
56 __builtin_ia32_pmovqd512mem_mask,
57 __builtin_ia32_pmovusqw512mem_mask,
58 __builtin_ia32_pmovsqw512mem_mask,
59 __builtin_ia32_pmovqw512mem_mask,
60 __builtin_ia32_pmovusdw512mem_mask,
61 __builtin_ia32_pmovsdw512mem_mask,
62 __builtin_ia32_pmovdw512mem_mask,
63 __builtin_ia32_pmovqb512mem_mask,
64 __builtin_ia32_pmovusqb512mem_mask,
65 __builtin_ia32_pmovsqb512mem_mask,
66 __builtin_ia32_pmovusdb512mem_mask,
67 __builtin_ia32_pmovsdb512mem_mask,
68 __builtin_ia32_pmovdb512mem_mask.
69 (bdesc_args): Add __builtin_ia32_expanddf512,
70 __builtin_ia32_expandsf512.
71 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
72 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
73 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
74 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
75 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
76 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
77 (avx512f_<code>v8div16qi2_mask_store): This.
78 (avx512f_expand<mode>): New.
80 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
82 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd): New.
83 (_mm512_mask_prefetch_i64gather_pd): Ditto.
84 (_mm512_prefetch_i32scatter_pd): Ditto.
85 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
86 (_mm512_prefetch_i64scatter_pd): Ditto.
87 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
88 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
89 (_mm512_mask_prefetch_i64gather_ps): Ditto.
90 (_mm512_prefetch_i32scatter_ps): Ditto.
91 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
92 (_mm512_prefetch_i64scatter_ps): Ditto.
93 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
94 * config/i386/i386-builtin-types.def: Define
95 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
96 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
97 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
98 IX86_BUILTIN_SCATTERPFQPD.
99 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
100 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
101 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
102 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
103 __builtin_ia32_scatterpfqps.
104 (ix86_expand_builtin): Expand new built-ins.
105 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
106 fix memory access data type.
107 (*avx512pf_gatherpf<mode>_mask): Ditto.
108 (*avx512pf_gatherpf<mode>): Ditto.
109 (avx512pf_scatterpf<mode>): Ditto.
110 (*avx512pf_scatterpf<mode>_mask): Ditto.
111 (*avx512pf_scatterpf<mode>): Ditto.
112 (GATHER_SCATTER_SF_MEM_MODE): New.
113 (avx512pf_gatherpf<mode>df): Ditto.
114 (*avx512pf_gatherpf<mode>df_mask): Ditto.
115 (*avx512pf_scatterpf<mode>df): Ditto.
117 2014-01-27 Jakub Jelinek <jakub@redhat.com>
120 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
121 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
124 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
126 * common/config/arm/arm-common.c
127 (arm_rewrite_mcpu): Handle multiple names.
129 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
131 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
133 * gimple-builder.h (create_gimple_tmp): Delete.
135 2014-01-27 Christian Bruel <christian.bruel@st.com>
137 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
140 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
142 * config/pa/pa.md (call): Generate indirect long calls to non-local
143 functions when outputing 32-bit code.
144 (call_value): Likewise except for special call to buggy powf function.
146 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
147 portable runtime and PIC indirect calls.
148 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
149 and PIC call sequences. Use ldo instead of blr to set return register
150 in PIC call sequence.
152 2014-01-25 Walter Lee <walt@tilera.com>
154 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
155 avoid clobbering a live register.
157 2014-01-25 Walter Lee <walt@tilera.com>
159 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
160 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
161 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
162 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
164 2014-01-25 Walter Lee <walt@tilera.com>
166 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
167 arguments on even registers.
168 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
170 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
171 (BIGGEST_ALIGNMENT): Ditto.
172 (BIGGEST_FIELD_ALIGNMENT): Ditto.
174 2014-01-25 Walter Lee <walt@tilera.com>
176 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
177 insns before bundling.
178 * config/tilegx/tilegx.md (tile_network_barrier): Update
181 2014-01-25 Walter Lee <walt@tilera.com>
183 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
184 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
185 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
187 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
189 * config/mips/constraints.md (kl): Delete.
190 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
191 define expands, using...
192 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
193 instructions for MIPS16.
194 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
195 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
197 2014-01-25 Walter Lee <walt@tilera.com>
199 * config/tilepro/tilepro.md (ctzdi2): Use register_operand
204 2014-01-25 Walter Lee <walt@tilera.com>
206 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
207 (TARGET_EXPAND_TO_RTL_HOOK): Define.
209 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
211 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
214 2014-01-25 Jakub Jelinek <jakub@redhat.com>
216 * print-rtl.c (in_call_function_usage): New var.
217 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
218 EXPR_LIST mode as mode and not as reg note name.
221 * cfgloopmanip.c (copy_loop_info): If
222 loop->warned_aggressive_loop_optimizations, make sure
223 the flag is set in target loop too.
225 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
227 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
229 * builtins.def: Likewise.
230 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
231 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
232 * ira.c (ira_setup_eliminable_regset): Likewise.
233 * omp-low.c (gate_expand_omp): Likewise.
234 (execute_lower_omp): Likewise.
235 (diagnose_sb_0): Likewise.
236 (gate_diagnose_omp_blocks): Likewise.
237 (simd_clone_clauses_extract): Likewise.
240 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
242 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
243 correction for little endian...
244 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
247 2014-01-24 Jeff Law <law@redhat.com>
249 PR tree-optimization/59919
250 * tree-vrp.c (find_assert_locations_1): Do not register asserts
251 for non-returning calls.
253 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
255 * common/config/aarch64/aarch64-common.c
256 (aarch64_rewrite_mcpu): Handle multiple names.
257 * config/aarch64/aarch64.h
258 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
260 2014-01-24 Dodji Seketeli <dodji@redhat.com>
262 * input.c (add_file_to_cache_tab): Handle the case where fopen
265 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
268 * config/i386/i386.md (pushsf splitter): Get stack adjustment
269 from push operand if code of push isn't PRE_DEC.
271 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
274 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
275 -mquad-memory-atomic. Update -mquad-memory documentation to say
276 it is only used for non-atomic loads/stores.
278 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
279 -mquad-memory or -mquad-memory-atomic switches.
281 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
282 -mquad-memory-atomic to ISA 2.07 support.
284 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
285 to separate support of normal quad word memory operations (ldq,
286 stq) from the atomic quad word memory operations.
288 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
289 support to separate non-atomic quad word operations from atomic
290 quad word operations. Disable non-atomic quad word operations in
291 little endian mode so that we don't have to swap words after the
292 load and before the store.
293 (quad_load_store_p): Add comment about atomic quad word support.
294 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
295 options printed with -mdebug=reg.
297 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
298 -mquad-memory-atomic as the test for whether we have quad word
300 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic,
301 -mquad-memory, or -mp8-vector are used, allow byte/half-word
304 * config/rs6000/sync.md (load_lockedti): Insure that the address
305 is a proper indexed or indirect address for the lqarx instruction.
306 On little endian systems, swap the hi/lo registers after the lqarx
308 (load_lockedpti): Use indexed_or_indirect_operand predicate to
309 insure the address is valid for the lqarx instruction.
310 (store_conditionalti): Insure that the address is a proper indexed
311 or indirect address for the stqcrx. instruction. On little endian
312 systems, swap the hi/lo registers before doing the stqcrx.
314 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
315 insure the address is valid for the stqcrx. instruction.
317 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
318 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
319 type of quad memory support is available.
321 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
324 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
325 there is a danger of looping.
327 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
330 force flag_ira_loop_pressure if set via command line.
332 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
334 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
335 (ashr_simd): New builtin handling DI mode.
336 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
337 (aarch64_sshr_simddi): New match pattern.
338 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
339 (vshrd_n_s64): Likewise.
340 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
342 2014-01-23 Nick Clifton <nickc@redhat.com>
344 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
345 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
346 favour of mcu specific scripts.
347 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
350 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
351 Alex Velenko <Alex.Velenko@arm.com>
353 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
354 (vaddv_s16): Likewise.
355 (vaddv_s32): Likewise.
356 (vaddv_u8): Likewise.
357 (vaddv_u16): Likewise.
358 (vaddv_u32): Likewise.
359 (vaddvq_s8): Likewise.
360 (vaddvq_s16): Likewise.
361 (vaddvq_s32): Likewise.
362 (vaddvq_s64): Likewise.
363 (vaddvq_u8): Likewise.
364 (vaddvq_u16): Likewise.
365 (vaddvq_u32): Likewise.
366 (vaddvq_u64): Likewise.
367 (vaddv_f32): Likewise.
368 (vaddvq_f32): Likewise.
369 (vaddvq_f64): Likewise.
370 (vmaxv_f32): Likewise.
371 (vmaxv_s8): Likewise.
372 (vmaxv_s16): Likewise.
373 (vmaxv_s32): Likewise.
374 (vmaxv_u8): Likewise.
375 (vmaxv_u16): Likewise.
376 (vmaxv_u32): Likewise.
377 (vmaxvq_f32): Likewise.
378 (vmaxvq_f64): Likewise.
379 (vmaxvq_s8): Likewise.
380 (vmaxvq_s16): Likewise.
381 (vmaxvq_s32): Likewise.
382 (vmaxvq_u8): Likewise.
383 (vmaxvq_u16): Likewise.
384 (vmaxvq_u32): Likewise.
385 (vmaxnmv_f32): Likewise.
386 (vmaxnmvq_f32): Likewise.
387 (vmaxnmvq_f64): Likewise.
388 (vminv_f32): Likewise.
389 (vminv_s8): Likewise.
390 (vminv_s16): Likewise.
391 (vminv_s32): Likewise.
392 (vminv_u8): Likewise.
393 (vminv_u16): Likewise.
394 (vminv_u32): Likewise.
395 (vminvq_f32): Likewise.
396 (vminvq_f64): Likewise.
397 (vminvq_s8): Likewise.
398 (vminvq_s16): Likewise.
399 (vminvq_s32): Likewise.
400 (vminvq_u8): Likewise.
401 (vminvq_u16): Likewise.
402 (vminvq_u32): Likewise.
403 (vminnmv_f32): Likewise.
404 (vminnmvq_f32): Likewise.
405 (vminnmvq_f64): Likewise.
407 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
409 * config/aarch64/aarch64-simd.md
410 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
411 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
412 (*aarch64_mul3_elt<mode>): Likewise.
413 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
414 (*aarch64_mul3_elt_to_64v2df): Likewise.
415 (*aarch64_mla_elt<mode>): Likewise.
416 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
417 (*aarch64_mls_elt<mode>): Likewise.
418 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
419 (*aarch64_fma4_elt<mode>): Likewise.
420 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
421 (*aarch64_fma4_elt_to_64v2df): Likewise.
422 (*aarch64_fnma4_elt<mode>): Likewise.
423 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
424 (*aarch64_fnma4_elt_to_64v2df): Likewise.
425 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
426 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
427 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
428 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
429 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
430 (aarch64_sqdmull_lane<mode>_internal): Likewise.
431 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
433 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
435 * config/aarch64/aarch64-simd.md
436 (aarch64_be_checked_get_lane<mode>): New define_expand.
437 * config/aarch64/aarch64-simd-builtins.def
438 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0):
439 New builtin definition.
440 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
441 Use new safe be builtin.
443 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
445 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
447 (aarch64_be_st1<mode>): Likewise.
448 (aarch_ld1<VALL:mode>): Define_expand modified.
449 (aarch_st1<VALL:mode>): Likewise.
450 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
451 (UNSPEC_ST1): Likewise.
453 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
455 * config/microblaze/microblaze.md: Add trap insn and attribute
457 2014-01-23 Dodji Seketeli <dodji@redhat.com>
459 PR preprocessor/58580
460 * input.h (location_get_source_line): Take an additional line_size
462 (void diagnostics_file_cache_fini): Declare new function.
463 * input.c (struct fcache): New type.
464 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
465 New static constants.
466 (diagnostic_file_cache_init, total_lines_num)
467 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
468 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
469 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
470 (get_next_line, read_next_line, goto_next_line, read_line_num):
471 New static function definitions.
472 (diagnostic_file_cache_fini): New function.
473 (location_get_source_line): Take an additional output line_len
474 parameter. Re-write using lookup_or_add_file_to_cache_tab and
476 * diagnostic.c (diagnostic_finish): Call
477 diagnostic_file_cache_fini.
478 (adjust_line): Take an additional input parameter for the length
479 of the line, rather than calculating it with strlen.
480 (diagnostic_show_locus): Adjust the use of
481 location_get_source_line and adjust_line with respect to their new
482 signature. While displaying a line now, do not stop at the first
483 null byte. Rather, display the zero byte as a space and keep
484 going until we reach the size of the line.
485 * Makefile.in: Add vec.o to OBJS-libcommon
487 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
488 Ilya Tocar <ilya.tocar@intel.com>
490 * config/i386/avx512fintrin.h (_mm512_kmov): New.
491 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
492 (__builtin_ia32_kmov16): Ditto.
493 * config/i386/i386.md (UNSPEC_KMOV): New.
496 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
498 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
499 (_mm512_storeu_si512): Ditto.
501 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
504 * rtl.h (get_referenced_operands): Declare.
505 * recog.c (get_referenced_operands): New function.
506 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
507 operands have been referenced when recording LO_SUM references.
509 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
511 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
513 2014-01-22 Jan Hubicka <jh@suse.cz>
515 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
516 Enable for generic and recent AMD targets.
518 2014-01-22 Jan Hubicka <jh@suse.cz>
520 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
521 ARG_SIZE note when adjustment was eliminated.
523 2014-01-22 Jeff Law <law@redhat.com>
525 PR tree-optimization/59597
526 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
527 in file. Accept new argument REGISTERING and use it to modify
528 dump output appropriately.
529 (register_jump_thread): Corresponding changes.
530 (mark_threaded_blocks): Reinstate code to cancel unprofitable
531 thread paths involving joiner blocks. Add code to dump cancelled
532 jump threading paths.
534 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
536 PR rtl-optimization/59477
537 * lra-constraints.c (inherit_in_ebb): Process call for living hard
538 regs. Update reloads_num and potential_reload_hard_regs for all
541 2014-01-22 Tom Tromey <tromey@redhat.com>
543 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
545 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
547 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
549 PR rtl-optimization/59896
550 * lra-constraints.c (process_alt_operands): Check unused note for
551 matched operands of insn with no output reloads.
553 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
555 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
556 (mips_move_from_gpr_cost): Likewise.
558 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
560 PR rtl-optimization/59858
561 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
562 ira_class_hard_regs_num.
563 (process_alt_operands): Increase reject for dying matched operand.
565 2014-01-21 Jakub Jelinek <jakub@redhat.com>
568 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
569 smaller than size, perform several stores or loads and stores
570 at dst + count - size to store or copy all of size bytes, rather
571 than just last modesize bytes.
573 2014-01-20 DJ Delorie <dj@redhat.com>
575 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
576 that CLOBBERs are REGs before propogating their values.
578 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
581 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
582 (cgraph_inline_failed_type): New function.
583 * cgraph.h (DEFCIFCODE): Add type.
584 (cgraph_inline_failed_type_t): New enum.
585 (cgraph_inline_failed_type): New prototype.
586 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
587 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
588 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
589 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
590 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
591 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
592 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
593 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
594 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
595 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
596 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
597 OPTIMIZATION_MISMATCH.
598 * tree-inline.c (expand_call_inline): Emit errors during
599 early_inlining if cgraph_inline_failed_type returns
602 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
605 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
606 mode attribute in insn output.
608 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
610 * output.h (output_constant): Delete.
611 * varasm.c (output_constant): Make private.
613 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
615 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
617 2014-01-20 Jakub Jelinek <jakub@redhat.com>
620 * tree.h (fold_builtin_strcat): New prototype.
621 * builtins.c (fold_builtin_strcat): No longer static. Add len
622 argument, if non-NULL, don't call c_strlen. Optimize
623 directly into __builtin_memcpy instead of __builtin_strcpy.
624 (fold_builtin_2): Adjust fold_builtin_strcat caller.
625 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
627 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
629 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
630 for SImode_address_operand operands, having only a REG argument.
632 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
634 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
635 loader name using mbig-endian.
636 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
638 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
640 * doc/invoke.texi (-march): Clarify documentation for AArch64.
644 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
646 * config/aarch64/aarch64-protos.h
647 (aarch64_cannot_change_mode_class_ptr): Declare.
648 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
649 aarch64_cannot_change_mode_class_ptr): New.
650 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
651 backend hook aarch64_cannot_change_mode_class.
653 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
655 * common/config/aarch64/aarch64-common.c
656 (aarch64_handle_option): Don't handle any option order logic here.
657 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
658 selected_cpu, warn on architecture version mismatch.
659 (aarch64_override_options): Fix parsing order for option strings.
661 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
662 Iain Sandoe <iain@codesourcery.com>
665 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
666 warning. Amend comment to reflect current functionality.
668 2014-01-20 Richard Biener <rguenther@suse.de>
671 * builtins.c (fold_builtin_strcat): Remove case better handled
672 by tree-ssa-strlen.c.
674 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
676 * config/aarch64/aarch64.opt
677 (mcpu, march, mtune): Make case-insensitive.
679 2014-01-20 Jakub Jelinek <jakub@redhat.com>
682 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
683 if operands[1] is a REG or ZERO_EXTEND of a REG.
685 2014-01-19 Jan Hubicka <jh@suse.cz>
687 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
689 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
691 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
692 long non-pic millicode calls.
694 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
696 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
698 2014-01-19 Kito Cheng <kito@0xlab.org>
700 * builtins.c (expand_movstr): Check movstr expand done or fail.
702 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
703 H.J. Lu <hongjiu.lu@intel.com>
706 * config/i386/i386.md (*lea<mode>): Zero-extend return register
707 to DImode for zero-extended addresses.
709 2014-01-19 Jakub Jelinek <jakub@redhat.com>
711 PR rtl-optimization/57763
712 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
713 on the new indirect jump_insn and increment LABEL_NUSES (label).
715 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
719 * config.gcc (x86_archs): New variable.
720 (x86_64_archs): Likewise.
721 (x86_cpus): Likewise.
722 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
723 --with-arch/--with-cpu= options.
724 Support --with-arch=/--with-cpu={nehalem,westmere,
725 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
727 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
729 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
730 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
732 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
734 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
736 2014-01-18 Jakub Jelinek <jakub@redhat.com>
739 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
740 clear cpp_get_options (parse_in)->warn_unused_macros for
741 ix86_target_macros_internal with cpp_define.
743 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
745 * jump.c (delete_related_insns): Keep (use (insn))s.
746 * reorg.c (redundant_insn): Check for barriers too.
748 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
750 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
752 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
754 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
755 call to $$dyncall when TARGET_LONG_CALLS is true.
757 2014-01-17 Jeff Law <law@redhat.com>
759 * ree.c (combine_set_extension): Temporarily disable test for
760 changing number of hard registers.
762 2014-01-17 Jan Hubicka <jh@suse.cz>
765 * ipa-inline-analysis.c (inline_free_summary):
766 Do not free summary of aliases.
768 2014-01-17 Jakub Jelinek <jakub@redhat.com>
771 * gimplify.c (gimplify_expr): Use create_tmp_var
772 instead of create_tmp_var_raw. If cond doesn't have
773 integral type, don't add the IFN_ANNOTATE builtin at all.
775 2014-01-17 Martin Jambor <mjambor@suse.cz>
778 * ipa-cp.c (prev_edge_clone): New variable.
779 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
780 Also resize prev_edge_clone vector.
781 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
782 (ipcp_edge_removal_hook): New function.
783 (ipcp_driver): Register ipcp_edge_removal_hook.
785 2014-01-17 Andrew Pinski <apinski@cavium.com>
786 Steve Ellcey <sellcey@mips.com>
789 * config/mips/mips.c (mips_print_operand): Check operand mode instead
792 2014-01-17 Jeff Law <law@redhat.com>
795 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
796 so that pass_ccp runs first.
798 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
800 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
801 (ix86_adjust_cost): Use !TARGET_XXX.
802 (do_reorder_for_imul): Likewise.
803 (swap_top_of_ready_list): Likewise.
804 (ix86_sched_reorder): Likewise.
806 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
808 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
809 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
810 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
811 (intel_memset): New. Duplicate slm_memset.
812 (intel_cost): New. Duplicate slm_cost.
813 (m_INTEL): New macro.
814 (processor_target_table): Add "intel".
815 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
816 with PROCESSOR_INTEL for "intel".
817 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
818 PROCESSOR_SILVERMONT.
819 (ix86_issue_rate): Likewise.
820 (ix86_adjust_cost): Likewise.
821 (ia32_multipass_dfa_lookahead): Likewise.
822 (swap_top_of_ready_list): Likewise.
823 (ix86_sched_reorder): Likewise.
824 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
825 instead of TARGET_OPT_AGU.
826 * config/i386/i386.h (TARGET_INTEL): New.
827 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
828 (processor_type): Add PROCESSOR_INTEL.
829 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
830 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
832 2014-01-17 Marek Polacek <polacek@redhat.com>
835 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
838 2014-01-17 Richard Biener <rguenther@suse.de>
840 PR tree-optimization/46590
841 * opts.c (default_options_table): Add entries for
842 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
843 all enabled at -O1 but not for -Og.
844 * common.opt (fbranch-count-reg): Remove Init(1).
845 (fmove-loop-invariants): Likewise.
846 (ftree-pta): Likewise.
848 2014-01-17 Jakub Jelinek <jakub@redhat.com>
850 * config/i386/i386.c (ix86_data_alignment): For compatibility with
851 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
852 decls to at least the GCC 4.8 used alignments.
855 * tree-nested.c (convert_nonlocal_reference_stmt,
856 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
857 of GIMPLE_BIND stmts, adjust associated decls.
859 2014-01-17 Richard Biener <rguenther@suse.de>
861 PR tree-optimization/46590
862 * vec.h (vec<>::bseach): New member function implementing
863 binary search according to C89 bsearch.
864 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
865 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
866 bitmap pointer again. Make accesses_in_loop a flat array.
867 (mem_ref_obstack): New global.
868 (outermost_indep_loop): Adjust for mem_ref->stored changes.
869 (mark_ref_stored): Likewise.
870 (ref_indep_loop_p_2): Likewise.
871 (set_ref_stored_in_loop): New helper function.
872 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
873 (memref_free): Adjust.
874 (record_mem_ref_loc): Simplify.
875 (gather_mem_refs_stmt): Adjust.
876 (sort_locs_in_loop_postorder_cmp): New function.
877 (analyze_memory_references): Sort accesses_in_loop after
878 loop postorder number.
879 (find_ref_loc_in_loop_cmp): New function.
880 (for_all_locs_in_loop): Find relevant cluster of locs in
881 accesses_in_loop and iterate without recursion.
882 (execute_sm): Avoid uninit warning.
883 (struct ref_always_accessed): Simplify.
884 (ref_always_accessed::operator ()): Likewise.
885 (ref_always_accessed_p): Likewise.
886 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
887 loop postorder numbers here.
888 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
891 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
894 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
895 on decls for which assemble_alias has been called.
897 2014-01-17 Nick Clifton <nickc@redhat.com>
899 * config/msp430/msp430.opt: (mcpu): New option.
900 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
901 (msp430_option_override): Parse target_cpu. If the MCU name
902 matches a generic string, clear target_mcu.
903 (msp430_attr): Allow numeric interrupt values up to 63.
904 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
905 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
907 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
909 * config/msp430/msp430.md (popm): Use %J rather than %I.
910 (addsi3): Use msp430_nonimmediate_operand for operand 2.
911 (addhi_cy_i): Use immediate_operand for operand 2.
912 * doc/invoke.texi: Document -mcpu option.
914 2014-01-17 Richard Biener <rguenther@suse.de>
916 PR rtl-optimization/38518
917 * df.h (df_analyze_loop): Declare.
918 * df-core.c: Include cfgloop.h.
919 (df_analyze_1): Split out main part of df_analyze.
920 (df_analyze): Adjust.
921 (loop_inverted_post_order_compute): New function.
922 (loop_post_order_compute): Likewise.
923 (df_analyze_loop): New function avoiding whole-function
925 * loop-invariant.c (find_defs): Use df_analyze_loop.
926 (find_invariants): Adjust.
927 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
929 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
931 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
932 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
934 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
936 * ipa-ref.c (ipa_remove_stmt_references): Fix references
937 traversal when removing references.
939 2014-01-16 Jan Hubicka <jh@suse.cz>
942 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
944 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
947 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
948 pushing a reload for an autoinc when we had previously reloaded an
949 inner part of the address.
951 2014-01-16 Jakub Jelinek <jakub@redhat.com>
953 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
955 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
956 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
957 when not giving up or versioning for alias only because of
959 (vect_analyze_data_ref_dependences): Set to true.
960 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
962 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
963 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
967 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
970 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
971 operand 0 predicate for gathers, use a new pseudo as subtarget.
973 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
976 * lra-constraints.c (process_alt_operands): Add printing debug info.
977 Check absence of input/output reloads for matched operands too.
979 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
981 PR rtl-optimization/59835
982 * ira.c (ira_init_register_move_cost): Increase cost for
985 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
987 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
989 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
992 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
993 non-register objects. Use gen_(high/low)part more consistently.
996 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
999 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1000 endian support, remove tests for WORDS_BIG_ENDIAN.
1001 (p8_mfvsrd_3_<mode>): Likewise.
1002 (reload_gpr_from_vsx<mode>): Likewise.
1003 (reload_gpr_from_vsxsf): Likewise.
1004 (p8_mfvsrd_4_disf): Likewise.
1006 2014-01-16 Richard Biener <rguenther@suse.de>
1008 PR rtl-optimization/46590
1009 * lcm.c (compute_antinout_edge): Use postorder iteration.
1010 (compute_laterin): Use inverted postorder iteration.
1012 2014-01-16 Nick Clifton <nickc@redhat.com>
1015 * varasm.c (output_constant): Return the number of bytes actually
1017 (output_constructor_array_range): Update the field size with the
1018 number of bytes emitted by output_constant.
1019 (output_constructor_regular_field): Likewise. Also do not
1020 complain if the total number of bytes emitted is now greater
1021 than the expected fieldpos.
1022 * output.h (output_constant): Update prototype and descriptive comment.
1024 2014-01-16 Marek Polacek <polacek@redhat.com>
1027 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1028 it is error_mark_node.
1030 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1032 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1033 VALID_AVX256_REG_OR_OI_MODE.
1035 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1037 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1038 current procedure should be profiled.
1040 2014-01-15 Andrew Pinski <apinski@cavium.com>
1042 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1043 of moving from/to the STACK_REG register class.
1045 2014-01-15 Richard Henderson <rth@redhat.com>
1048 * reginfo.c (global_regs_decl): Globalize.
1049 * rtl.h (global_regs_decl): Declare.
1050 * ira.c (do_reload): Diagnose frame_pointer_needed and it
1051 reserved via global_regs.
1053 2014-01-15 Teresa Johnson <tejohnson@google.com>
1055 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1057 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1059 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1060 and vmulosh rather than call gen_vec_widen_smult_*.
1061 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1062 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1063 (vec_widen_smult_even_v16qi): Likewise.
1064 (vec_widen_umult_even_v8hi): Likewise.
1065 (vec_widen_smult_even_v8hi): Likewise.
1066 (vec_widen_umult_odd_v16qi): Likewise.
1067 (vec_widen_smult_odd_v16qi): Likewise.
1068 (vec_widen_umult_odd_v8hi): Likewise.
1069 (vec_widen_smult_odd_v8hi): Likewise.
1070 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1071 vmuloub rather than call gen_vec_widen_umult_*.
1072 (vec_widen_umult_lo_v16qi): Likewise.
1073 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1074 vmulosb rather than call gen_vec_widen_smult_*.
1075 (vec_widen_smult_lo_v16qi): Likewise.
1076 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1077 rather than call gen_vec_widen_umult_*.
1078 (vec_widen_umult_lo_v8hi): Likewise.
1079 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1080 rather than call gen_vec_widen_smult_*.
1081 (vec_widen_smult_lo_v8hi): Likewise.
1083 2014-01-15 Jeff Law <law@redhat.com>
1085 PR tree-optimization/59747
1086 * ree.c (find_and_remove_re): Properly handle case where a second
1087 eliminated extension requires widening a copy created for elimination
1088 of a prior extension.
1089 (combine_set_extension): Ensure that the number of hard regs needed
1090 for a destination register does not change when we widen it.
1092 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1094 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1095 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1096 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1097 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1098 (avr-*-rtems*): Likewise.
1099 (bfin*-rtems*): Likewise.
1100 (moxie-*-rtems*): Likewise.
1101 (h8300-*-rtems*): Likewise.
1102 (i[34567]86-*-rtems*): Likewise.
1103 (lm32-*-rtems*): Likewise.
1104 (m32r-*-rtems*): Likewise.
1105 (m68k-*-rtems*): Likewise.
1106 (microblaze*-*-rtems*): Likewise.
1107 (mips*-*-rtems*): Likewise.
1108 (powerpc-*-rtems*): Likewise.
1109 (sh-*-rtems*): Likewise.
1110 (sparc-*-rtems*): Likewise.
1111 (sparc64-*-rtems*): Likewise.
1112 (v850-*-rtems*): Likewise.
1113 (m32c-*-rtems*): Likewise.
1115 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
1117 PR rtl-optimization/59511
1118 * ira.c (ira_init_register_move_cost): Use memory costs for some
1119 cases of register move cost calculations.
1120 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1121 instead of BB frequency.
1122 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1123 * lra-assigns.c (find_hard_regno_for): Ditto.
1125 2014-01-15 Richard Biener <rguenther@suse.de>
1127 PR tree-optimization/59822
1128 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1129 (vectorizable_load): Use it to hoist defs of uses of invariant
1130 loads out of the loop.
1132 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1133 Kugan Vivekanandarajah <kuganv@linaro.org>
1136 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1139 2014-01-15 Richard Biener <rguenther@suse.de>
1141 PR rtl-optimization/59802
1142 * lcm.c (compute_available): Use inverted postorder to seed
1143 the initial worklist.
1145 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1148 * config/s390/s390.c (s390_preferred_reload_class): Don't return
1149 ADDR_REGS for invalid symrefs in non-PIC code.
1151 2014-01-15 Jakub Jelinek <jakub@redhat.com>
1154 * builtins.c (determine_block_size): Initialize *probable_max_size
1155 even if len_rtx is CONST_INT.
1157 2014-01-14 Andrew Pinski <apinski@cavium.com>
1159 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1160 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1161 (cortexa53_tunings): Likewise.
1162 (aarch64_sched_issue_rate): New function.
1163 (TARGET_SCHED_ISSUE_RATE): Define.
1165 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1167 * ira-costs.c (find_costs_and_classes): Add missed
1168 ira_init_register_move_cost_if_necessary.
1170 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
1173 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1175 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
1178 * config/i386/i386.c (type_natural_mode): Add a bool parameter
1179 to indicate if type is used for function return value. Warn ABI
1180 change if the vector mode isn't available for function return value.
1181 (ix86_function_arg_advance): Pass false to type_natural_mode.
1182 (ix86_function_arg): Likewise.
1183 (ix86_gimplify_va_arg): Likewise.
1184 (function_arg_32): Don't warn ABI change.
1185 (ix86_function_value): Pass true to type_natural_mode.
1186 (ix86_return_in_memory): Likewise.
1187 (ix86_struct_value_rtx): Removed.
1188 (TARGET_STRUCT_VALUE_RTX): Likewise.
1190 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1192 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1193 converting a conditional jump into a conditional return.
1195 2014-01-14 Richard Biener <rguenther@suse.de>
1197 PR tree-optimization/58921
1198 PR tree-optimization/59006
1199 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1200 hoisting invariant stmts.
1201 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1202 invariant loads on the preheader edge if possible.
1204 2014-01-14 Joey Ye <joey.ye@arm.com>
1206 * doc/plugin.texi (Building GCC plugins): Update to C++.
1208 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
1210 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1211 (_mm_rcp28_round_ss): Ditto.
1212 (_mm_rsqrt28_round_sd): Ditto.
1213 (_mm_rsqrt28_round_ss): Ditto.
1214 (_mm_rcp28_sd): Ditto.
1215 (_mm_rcp28_ss): Ditto.
1216 (_mm_rsqrt28_sd): Ditto.
1217 (_mm_rsqrt28_ss): Ditto.
1218 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1219 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1220 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1221 (IX86_BUILTIN_RCP28SD): Ditto.
1222 (IX86_BUILTIN_RCP28SS): Ditto.
1223 (IX86_BUILTIN_RSQRT28SD): Ditto.
1224 (IX86_BUILTIN_RSQRT28SS): Ditto.
1225 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1226 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1227 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1228 (ix86_expand_special_args_builtin): Expand new FTYPE.
1229 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1230 (srcp14<mode>): Make insn unary.
1231 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1232 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1233 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1234 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1235 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1236 Fix rounding: make it SAE only.
1237 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1239 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1241 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1242 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1243 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1244 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1245 (round_saeonly_mask_scalar_operand4): Ditto.
1246 (round_saeonly_mask_scalar_op3): Ditto.
1247 (round_saeonly_mask_scalar_op4): Ditto.
1249 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1251 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1252 Implement -maltivec=be for vec_insert and vec_extract.
1254 2014-01-10 DJ Delorie <dj@redhat.com>
1256 * config/msp430/msp430.md (call_internal): Don't allow memory
1257 references with SP as the base register.
1258 (call_value_internal): Likewise.
1259 * config/msp430/constraints.md (Yc): New. For memory references
1260 that don't use SP as a base register.
1262 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1263 "an integer without a # prefix"
1264 * config/msp430/msp430.md (epilogue_helper): Use it.
1266 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1269 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1270 AVX512F gather builtins.
1271 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1272 on gather decls with INTEGER_TYPE masktype.
1273 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1274 directly into the builtin rather than hoisting it before loop.
1276 PR tree-optimization/59387
1277 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1278 (scev_const_prop): If folded_casts and type has undefined overflow,
1279 use force_gimple_operand instead of force_gimple_operand_gsi and
1280 for each added stmt if it is assign with
1281 arith_code_with_undefined_signed_overflow, call
1282 rewrite_to_defined_overflow.
1283 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1284 gimple-fold.h instead.
1285 (arith_code_with_undefined_signed_overflow,
1286 rewrite_to_defined_overflow): Moved to ...
1287 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1288 rewrite_to_defined_overflow): ... here. No longer static.
1289 Include gimplify-me.h.
1290 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1291 rewrite_to_defined_overflow): New prototypes.
1293 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1295 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1297 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
1299 * builtins.c (get_object_alignment_2): Minor tweak.
1300 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1302 2014-01-13 Christian Bruel <christian.bruel@st.com>
1304 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1305 optimized non constant lengths.
1307 2014-01-13 Jakub Jelinek <jakub@redhat.com>
1310 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1311 load as __atomic_load_N if possible.
1313 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1315 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1317 (rs6000_expand_builtin): Adjust call.
1319 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
1322 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1323 * config/rs6000/rs6000.c: Include target-globals.h.
1324 (rs6000_set_current_function): Instead of doing target_reinit
1325 unconditionally, use save_target_globals_default_opts and
1326 restore_target_globals.
1328 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1330 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1331 (rs6000_expand_builtin): Handle mffs and mtfsf.
1332 (rs6000_init_builtins): Define mffs and mtfsf.
1333 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1334 (rs6000_mffs): New pattern.
1335 (rs6000_mtfsf): New pattern.
1337 2014-01-11 Bin Cheng <bin.cheng@arm.com>
1339 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1340 Start narrowing with START. Apply candidate-use pair
1341 and check overall cost in narrowing.
1342 (iv_ca_prune): Pass new argument.
1344 2014-01-10 Jeff Law <law@redhat.com>
1347 * ree.c (combine_reaching_defs): Ensure the defining statement
1348 occurs before the extension when optimizing extensions with
1349 different source and destination hard registers.
1351 2014-01-10 Jan Hubicka <jh@suse.cz>
1354 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1355 vtables into the type inheritance graph.
1357 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1359 PR rtl-optimization/59754
1360 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1361 modes in the REGNO != REGNO case.
1363 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1365 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1367 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1369 PR tree-optimization/59745
1370 * tree-predcom.c (tree_predictive_commoning_loop): Call
1371 free_affine_expand_cache if giving up because components is NULL.
1373 * target-globals.c (save_target_globals): Allocate < 4KB structs using
1374 GC in payload of target_globals struct instead of allocating them on
1375 the heap and the larger structs separately using GC.
1376 * target-globals.h (struct target_globals): Make regs, hard_regs,
1377 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1378 of GTY((skip)) and change type to void *.
1379 (reset_target_globals): Cast loads from those fields to corresponding
1382 2014-01-10 Steve Ellcey <sellcey@mips.com>
1385 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1386 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1387 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1389 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
1392 * aarch64-modes.def (CC_Zmode): New flags mode.
1393 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1394 represents an equality.
1395 (aarch64_get_condition_code): Handle CC_Zmode.
1396 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1398 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1400 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1401 extraction in good case.
1403 2014-01-10 Richard Biener <rguenther@suse.de>
1405 PR tree-optimization/59374
1406 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1407 checking after SLP discovery. Mark stmts not participating
1408 in any SLP instance properly.
1410 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1412 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1413 when handling a SET rtx.
1415 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1417 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1418 (cortex-a57): Likewise.
1419 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1421 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1423 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1424 non-iwmmxt builtins.
1426 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
1430 * ipa-devirt.c record_target_from_binfo): Take as argument
1431 stack of binfos and lookup matching one for virtual inheritance.
1432 (possible_polymorphic_call_targets_1): Update.
1434 2014-01-10 Huacai Chen <chenhc@lemote.com>
1436 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1437 kernel strings for Loongson-2E/2F/3A.
1439 2014-01-10 Jakub Jelinek <jakub@redhat.com>
1442 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1443 is_gimple_call before calling gimple_call_internal_p.
1445 2014-01-09 Steve Ellcey <sellcey@mips.com>
1447 * Makefile.in (TREE_FLOW_H): Remove.
1448 (TREE_SSA_H): Add file names from tree-flow.h.
1449 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1450 * tree.h: Remove tree-flow.h reference.
1451 * hash-table.h: Remove tree-flow.h reference.
1452 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1453 reference with tree-ssa-loop.h.
1455 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1457 * doc/invoke.texi: Add -maltivec={be,le} options, and document
1458 default element-order behavior for -maltivec.
1459 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1460 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1461 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1462 when targeting big endian, at least for now.
1463 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1465 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1468 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1469 var satisfies use_register_for_decl, just take into account type
1470 alignment, rather than decl alignment.
1472 PR tree-optimization/59622
1473 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
1474 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1475 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1476 Don't devirtualize for inplace at all. For targets.length () == 1,
1477 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1479 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1481 * config/i386/i386.md (cpu): Remove the unused btver1.
1483 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
1485 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1487 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1490 * tree-core.h (struct target_globals): New forward declaration.
1491 (struct tree_target_option): Add globals field.
1492 * tree.h (TREE_TARGET_GLOBALS): Define.
1493 (prepare_target_option_nodes_for_pch): New prototype.
1494 * target-globals.h (struct target_globals): Define even if
1496 * tree.c (prepare_target_option_node_for_pch,
1497 prepare_target_option_nodes_for_pch): New functions.
1498 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1499 * config/i386/i386.c: Include target-globals.h.
1500 (ix86_set_current_function): Instead of doing target_reinit
1501 unconditionally, use save_target_globals_default_opts and
1502 restore_target_globals.
1504 2014-01-09 Richard Biener <rguenther@suse.de>
1506 PR tree-optimization/59715
1507 * tree-cfg.h (split_critical_edges): Declare.
1508 * tree-cfg.c (split_critical_edges): Export.
1509 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1511 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
1513 * cfgexpand.c (expand_stack_vars): Optionally disable
1514 asan stack protection.
1515 (expand_used_vars): Likewise.
1516 (partition_stack_vars): Likewise.
1517 * asan.c (asan_emit_stack_protection): Optionally disable
1518 after return stack usage.
1519 (instrument_derefs): Optionally disable memory access instrumentation.
1520 (instrument_builtin_call): Likewise.
1521 (instrument_strlen_call): Likewise.
1522 (asan_protect_global): Optionally disable global variables protection.
1523 * doc/invoke.texi: Added doc for new options.
1524 * params.def: Added new options.
1525 * params.h: Likewise.
1527 2014-01-09 Jakub Jelinek <jakub@redhat.com>
1529 PR rtl-optimization/59724
1530 * ifcvt.c (cond_exec_process_if_block): Don't call
1531 flow_find_head_matching_sequence with 0 longest_match.
1532 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1533 non-active insns if !stop_after.
1534 (try_head_merge_bb): Revert 2014-01-07 changes.
1536 2014-01-08 Jeff Law <law@redhat.com>
1538 * ree.c (get_sub_rtx): New function, extracted from...
1539 (merge_def_and_ext): Here.
1540 (combine_reaching_defs): Use get_sub_rtx.
1542 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
1544 * cgraph.h (varpool_variable_node): Do not choke on null node.
1546 2014-01-08 Catherine Moore <clm@codesourcery.com>
1548 * config/mips/mips.md (simple_return): Attempt to use JRC
1550 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1552 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1554 PR rtl-optimization/59137
1555 * reorg.c (steal_delay_list_from_target): Call update_block for
1557 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1559 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1561 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1562 two duplicate entries.
1564 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
1567 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
1569 * config/mips/mips.c (mips_truncated_op_cost): New function.
1570 (mips_rtx_costs): Adjust test for BADDU.
1571 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1573 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
1575 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1576 (*baddu_si): ...this new pattern.
1578 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1581 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1583 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1586 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1588 (expand_expr, expand_normal): Adjust.
1589 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1590 inner_reference_p. Use inner_reference_p to expand inner references.
1591 (store_expr): Adjust.
1592 * cfgexpand.c (expand_call_stmt): Adjust.
1594 2014-01-08 Rong Xu <xur@google.com>
1596 * gcov-io.c (gcov_var): Move from gcov-io.h.
1597 (gcov_position): Ditto.
1598 (gcov_is_error): Ditto.
1599 (gcov_rewrite): Ditto.
1600 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1601 only part to libgcc/libgcov.h.
1603 2014-01-08 Marek Polacek <polacek@redhat.com>
1606 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1608 2014-01-08 Marek Polacek <polacek@redhat.com>
1611 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1613 2014-01-08 Jakub Jelinek <jakub@redhat.com>
1615 PR rtl-optimization/59649
1616 * stor-layout.c (get_mode_bounds): For BImode return
1617 0 and STORE_FLAG_VALUE.
1619 2014-01-08 Richard Biener <rguenther@suse.de>
1622 * gimple.h (is_gimple_builtin_call): Remove.
1623 (gimple_builtin_call_types_compatible_p): New.
1624 (gimple_call_builtin_p): New overload.
1625 * gimple.c (is_gimple_builtin_call): Remove.
1626 (validate_call): Rename to ...
1627 (gimple_builtin_call_types_compatible_p): ... this and export. Also
1629 (validate_type): New static function.
1630 (gimple_call_builtin_p): New overload and adjust.
1631 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1632 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
1633 (gimple_fold_stmt_to_constant_1): Likewise.
1634 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1636 2014-01-08 Richard Biener <rguenther@suse.de>
1639 * gimplify.c (gimplify_expr): Gimplify register-register type
1640 VIEW_CONVERT_EXPRs to separate stmts.
1642 2014-01-07 Jeff Law <law@redhat.com>
1645 * ree.c (combine_set_extension): Handle case where source
1646 and destination registers in an extension insn are different.
1647 (combine_reaching_defs): Allow source and destination registers
1648 in extension to be different under limited circumstances.
1649 (add_removable_extension): Remove restriction that the
1650 source and destination registers in the extension are the same.
1651 (find_and_remove_re): Emit a copy from the extension's
1652 destination to its source after the defining insn if
1653 the source and destination registers are different.
1656 * ifcvt.c (merge_if_block): If we are merging a block with more than
1657 one successor with a block with no successors, remove any BARRIER
1658 after the second block.
1660 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
1662 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1664 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
1667 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1668 for 14-bit register offsets when INT14_OK_STRICT is false.
1670 2014-01-07 Roland Stigge <stigge@antcom.de>
1671 Michael Meissner <meissner@linux.vnet.ibm.com>
1674 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1675 Only check TFmode for SPE constants. Don't check TImode or TDmode.
1677 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
1679 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1682 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
1684 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1685 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1686 rtx is const0_rtx or not.
1688 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
1691 * target-globals.c (save_target_globals): Remove this_fn_optab
1693 * toplev.c: Include optabs.h.
1694 (target_reinit): Temporarily restore the global options if another
1695 set of options are in force.
1697 2014-01-07 Jakub Jelinek <jakub@redhat.com>
1699 PR rtl-optimization/58668
1700 * cfgcleanup.c (flow_find_cross_jump): Don't count
1701 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
1702 to determine what is counted.
1703 (flow_find_head_matching_sequence): Use active_insn_p to determine
1705 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1707 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1708 determine what is counted.
1710 PR tree-optimization/59643
1711 * tree-predcom.c (split_data_refs_to_components): If one dr is
1712 read and one write, determine_offset fails and the write isn't
1713 in the bad component, just put the read into the bad component.
1715 2014-01-07 Mike Stump <mikestump@comcast.net>
1716 Jakub Jelinek <jakub@redhat.com>
1719 * tree-core.h (struct tree_optimization_option): Change optabs
1720 type from unsigned char * to void *.
1721 * optabs.c (init_tree_optimization_optabs): Adjust
1722 TREE_OPTIMIZATION_OPTABS initialization.
1724 2014-01-06 Jakub Jelinek <jakub@redhat.com>
1727 * config/i386/i386.h (struct machine_function): Add
1728 no_drap_save_restore field.
1729 * config/i386/i386.c (ix86_save_reg): Use
1730 !cfun->machine->no_drap_save_restore instead of
1731 crtl->stack_realign_needed.
1732 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
1733 this function clears frame_pointer_needed. Set
1734 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
1735 and DRAP reg is needed.
1737 2014-01-06 Marek Polacek <polacek@redhat.com>
1740 * doc/implement-c.texi: Mention that other integer types are
1741 permitted as bit-field types in strictly conforming mode.
1743 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
1745 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
1748 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
1750 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
1752 2014-01-06 Martin Jambor <mjambor@suse.cz>
1755 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
1757 * ipa-prop.c (ipa_print_node_params): Fix indentation.
1759 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
1763 * var-tracking.c (add_stores): Preserve the value of the source even if
1764 we don't record the store.
1766 2014-01-06 Terry Guo <terry.guo@arm.com>
1768 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
1770 2014-01-05 Iain Sandoe <iain@codesourcery.com>
1773 * config/darwin.c (darwin_function_section): Adjust return values to
1774 correspond to optimisation changes made in r206070.
1776 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
1778 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
1779 from prefetch_block tune setting.
1780 (nocona_cost): Correct size of prefetch block to 64.
1782 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
1784 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
1785 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
1786 used to save the static chain register in the computation of the offset
1787 from which the FP registers need to be restored.
1789 2014-01-04 Jakub Jelinek <jakub@redhat.com>
1791 PR tree-optimization/59519
1792 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
1793 ICE if get_current_def (current_new_name) is already non-NULL, as long
1794 as it is a phi result of some other phi in *new_exit_bb that has
1797 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
1798 or vmovdqu* for misaligned_operand.
1799 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
1800 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
1801 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
1802 aligned_mem for AVX512F masked aligned load and store builtins and for
1805 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
1807 PR tree-optimization/59651
1808 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
1809 Address range for negative step should be added by TYPE_SIZE_UNIT.
1811 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
1813 * config/m68k/m68k.c (handle_move_double): Handle pushes with
1814 overlapping registers also for registers other than the stack pointer.
1816 2014-01-03 Marek Polacek <polacek@redhat.com>
1819 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
1822 2014-01-03 Jakub Jelinek <jakub@redhat.com>
1825 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
1828 * config/i386/i386.md (MODE_SIZE): New mode attribute.
1829 (push splitter): Use <P:MODE_SIZE> instead of
1830 GET_MODE_SIZE (<P:MODE>mode).
1831 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
1832 (mov -1, reg peephole2): Likewise.
1833 * config/i386/sse.md (*mov<mode>_internal,
1834 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
1835 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
1836 *<code><mode>3, *andnot<mode>3<mask_name>,
1837 <mask_codefor><code><mode>3<mask_name>): Likewise.
1838 * config/i386/subst.md (mask_mode512bit_condition,
1839 sd_mask_mode512bit_condition): Likewise.
1841 2014-01-02 Xinliang David Li <davidxl@google.com>
1843 PR tree-optimization/59303
1844 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
1845 (dump_predicates): Better output format.
1846 (pred_equal_p): New function.
1847 (is_neq_relop_p): Ditto.
1848 (is_neq_zero_form_p): Ditto.
1849 (pred_expr_equal_p): Ditto.
1850 (pred_neg_p): Ditto.
1851 (simplify_pred): Ditto.
1852 (simplify_preds_2): Ditto.
1853 (simplify_preds_3): Ditto.
1854 (simplify_preds_4): Ditto.
1855 (simplify_preds): Ditto.
1857 (push_to_worklist): Ditto.
1858 (get_pred_info_from_cmp): Ditto.
1859 (is_degenerated_phi): Ditto.
1860 (normalize_one_pred_1): Ditto.
1861 (normalize_one_pred): Ditto.
1862 (normalize_one_pred_chain): Ditto.
1863 (normalize_preds): Ditto.
1864 (normalize_cond_1): Remove function.
1865 (normalize_cond): Ditto.
1866 (is_gcond_subset_of): Ditto.
1867 (is_subset_of_any): Ditto.
1868 (is_or_set_subset_of): Ditto.
1869 (is_and_set_subset_of): Ditto.
1870 (is_norm_cond_subset_of): Ditto.
1871 (pred_chain_length_cmp): Ditto.
1872 (convert_control_dep_chain_into_preds): Type change.
1873 (find_predicates): Ditto.
1874 (find_def_preds): Ditto.
1875 (destroy_predicates_vecs): Ditto.
1876 (find_matching_predicates_in_rest_chains): Ditto.
1877 (use_pred_not_overlap_with_undef_path_pred): Ditto.
1878 (is_pred_expr_subset): Ditto.
1879 (is_pred_chain_subset_of): Ditto.
1880 (is_included_in): Ditto.
1881 (is_superset_of): Ditto.
1883 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1885 Update copyright years.
1887 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
1889 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
1890 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
1891 config/arc/arc.md, config/arc/arc.opt,
1892 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
1893 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
1894 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
1895 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
1896 config/linux-protos.h, config/linux.c, config/winnt-c.c,
1897 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
1898 vtable-verify.c, vtable-verify.h: Use the standard form for the
1901 2014-01-02 Tobias Burnus <burnus@net-b.de>
1903 * gcc.c (process_command): Update copyright notice dates.
1904 * gcov-dump.c: Ditto.
1906 * doc/cpp.texi: Bump @copying's copyright year.
1907 * doc/cppinternals.texi: Ditto.
1908 * doc/gcc.texi: Ditto.
1909 * doc/gccint.texi: Ditto.
1910 * doc/gcov.texi: Ditto.
1911 * doc/install.texi: Ditto.
1912 * doc/invoke.texi: Ditto.
1914 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1916 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
1918 2014-01-01 Jakub Jelinek <jakub@redhat.com>
1920 * config/i386/sse.md (*mov<mode>_internal): Guard
1921 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
1923 PR rtl-optimization/59647
1924 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
1925 new_rtx into UNSIGNED_FLOAT rtxes.
1927 Copyright (C) 2014 Free Software Foundation, Inc.
1929 Copying and distribution of this file, with or without modification,
1930 are permitted in any medium without royalty provided the copyright
1931 notice and this notice are preserved.