gcc/ChangeLog

   1 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
   2
   3         PR preprocessor/59935
   4         * input.c (location_get_source_line): Bail out on when line number
   5         is zero, and test the return value of
   6         lookup_or_add_file_to_cache_tab.
   7
   8 2014-01-28  Richard Biener  <rguenther@suse.de>
   9
  10         PR tree-optimization/58742
  11         * tree-ssa-forwprop.c (associate_plusminus): Handle
  12         pointer subtraction of the form (T)(P + A) - (T)P.
  13
  14 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  15
  16         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
  17         at const_int_cost.
  18
  19 2014-01-28  Richard Biener  <rguenther@suse.de>
  20
  21         Revert
  22         2014-01-28  Richard Biener  <rguenther@suse.de>
  23
  24         PR rtl-optimization/45364
  25         PR rtl-optimization/59890
  26         * var-tracking.c (local_get_addr_clear_given_value): Handle
  27         already cleared slot.
  28         (val_reset): Handle not allocated local_get_addr_cache.
  29         (vt_find_locations): Use post-order on the inverted CFG.
  30
  31 2014-01-28  Richard Biener  <rguenther@suse.de>
  32
  33         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
  34         Remove.
  35
  36 2014-01-28  Richard Biener  <rguenther@suse.de>
  37
  38         PR rtl-optimization/45364
  39         PR rtl-optimization/59890
  40         * var-tracking.c (local_get_addr_clear_given_value): Handle
  41         already cleared slot.
  42         (val_reset): Handle not allocated local_get_addr_cache.
  43         (vt_find_locations): Use post-order on the inverted CFG.
  44
  45 2014-01-28  Alan Modra  <amodra@gmail.com>
  46
  47         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
  48         * configure.ac <recursive call for build != host>: Define
  49         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
  50         and LD_FOR_BUILD too.
  51         * configure: Regenerate.
  52
  53 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
  54
  55         * config/i386/i386.c (get_builtin_code_for_version): Separate
  56         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
  57         Broadwell from Haswell.
  58
  59 2014-01-27  Steve Ellcey  <sellcey@mips.com>
  60
  61         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
  62         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
  63         * config/mips/mips.c (mips_option_override): Change setting
  64         of TARGET_DSP.
  65         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
  66         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
  67         Change from Mask to Var.
  68
  69 2014-01-27  Jeff Law  <law@redhat.com>
  70
  71         * ipa-inline.c (inline_small_functions): Fix typo.
  72
  73 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
  74
  75         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
  76         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
  77         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
  78         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
  79         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
  80         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
  81         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
  82         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
  83         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
  84         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
  85         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
  86         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
  87         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
  88         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
  89         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
  90         (_mm512_storeu_epi64): Ditto.
  91         (_mm512_cmpge_epi32_mask): Ditto.
  92         (_mm512_cmpge_epu32_mask): Ditto.
  93         (_mm512_cmpge_epi64_mask): Ditto.
  94         (_mm512_cmpge_epu64_mask): Ditto.
  95         (_mm512_cmple_epi32_mask): Ditto.
  96         (_mm512_cmple_epu32_mask): Ditto.
  97         (_mm512_cmple_epi64_mask): Ditto.
  98         (_mm512_cmple_epu64_mask): Ditto.
  99         (_mm512_cmplt_epi32_mask): Ditto.
 100         (_mm512_cmplt_epu32_mask): Ditto.
 101         (_mm512_cmplt_epi64_mask): Ditto.
 102         (_mm512_cmplt_epu64_mask): Ditto.
 103         (_mm512_cmpneq_epi32_mask): Ditto.
 104         (_mm512_cmpneq_epu32_mask): Ditto.
 105         (_mm512_cmpneq_epi64_mask): Ditto.
 106         (_mm512_cmpneq_epu64_mask): Ditto.
 107         (_mm512_expand_pd): Ditto.
 108         (_mm512_expand_ps): Ditto.
 109         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
 110         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
 111         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
 112         * config/i386/i386.c (ix86_builtins): Add
 113         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
 114         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
 115         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
 116         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
 117         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
 118         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
 119         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
 120         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
 121         IX86_BUILTIN_PMOVUSQW512_MEM.
 122         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
 123         __builtin_ia32_pmovsqd512mem_mask,
 124         __builtin_ia32_pmovqd512mem_mask,
 125         __builtin_ia32_pmovusqw512mem_mask,
 126         __builtin_ia32_pmovsqw512mem_mask,
 127         __builtin_ia32_pmovqw512mem_mask,
 128         __builtin_ia32_pmovusdw512mem_mask,
 129         __builtin_ia32_pmovsdw512mem_mask,
 130         __builtin_ia32_pmovdw512mem_mask,
 131         __builtin_ia32_pmovqb512mem_mask,
 132         __builtin_ia32_pmovusqb512mem_mask,
 133         __builtin_ia32_pmovsqb512mem_mask,
 134         __builtin_ia32_pmovusdb512mem_mask,
 135         __builtin_ia32_pmovsdb512mem_mask,
 136         __builtin_ia32_pmovdb512mem_mask.
 137         (bdesc_args): Add __builtin_ia32_expanddf512,
 138         __builtin_ia32_expandsf512.
 139         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
 140         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
 141         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
 142         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
 143         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
 144         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
 145         (avx512f_<code>v8div16qi2_mask_store): This.
 146         (avx512f_expand<mode>): New.
 147
 148 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
 149
 150         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
 151         New.
 152         (_mm512_mask_prefetch_i64gather_pd): Ditto.
 153         (_mm512_prefetch_i32scatter_pd): Ditto.
 154         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
 155         (_mm512_prefetch_i64scatter_pd): Ditto.
 156         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
 157         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
 158         (_mm512_mask_prefetch_i64gather_ps): Ditto.
 159         (_mm512_prefetch_i32scatter_ps): Ditto.
 160         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
 161         (_mm512_prefetch_i64scatter_ps): Ditto.
 162         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
 163         * config/i386/i386-builtin-types.def: Define
 164         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
 165         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
 166         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
 167         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
 168         IX86_BUILTIN_SCATTERPFQPD.
 169         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
 170         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
 171         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
 172         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
 173         __builtin_ia32_scatterpfqps.
 174         (ix86_expand_builtin): Expand new built-ins.
 175         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
 176         fix memory access data type.
 177         (*avx512pf_gatherpf<mode>_mask): Ditto.
 178         (*avx512pf_gatherpf<mode>): Ditto.
 179         (avx512pf_scatterpf<mode>): Ditto.
 180         (*avx512pf_scatterpf<mode>_mask): Ditto.
 181         (*avx512pf_scatterpf<mode>): Ditto.
 182         (GATHER_SCATTER_SF_MEM_MODE): New.
 183         (avx512pf_gatherpf<mode>df): Ditto.
 184         (*avx512pf_gatherpf<mode>df_mask): Ditto.
 185         (*avx512pf_scatterpf<mode>df): Ditto.
 186
 187 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
 188
 189         PR bootstrap/59934
 190         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
 191         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
 192         reached.
 193
 194 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
 195
 196         * common/config/arm/arm-common.c
 197         (arm_rewrite_mcpu): Handle multiple names.
 198         * config/arm/arm.h
 199         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
 200
 201 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
 202
 203         * gimple-builder.h (create_gimple_tmp): Delete.
 204
 205 2014-01-27  Christian Bruel  <christian.bruel@st.com>
 206
 207         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
 208         words comparisons.
 209
 210 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
 211
 212         * config/pa/pa.md (call): Generate indirect long calls to non-local
 213         functions when outputing 32-bit code.
 214         (call_value): Likewise except for special call to buggy powf function.
 215
 216         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
 217         portable runtime and PIC indirect calls.
 218         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
 219         and PIC call sequences.  Use ldo instead of blr to set return register
 220         in PIC call sequence.
 221
 222 2014-01-25  Walter Lee  <walt@tilera.com>
 223
 224         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
 225         avoid clobbering a live register.
 226
 227 2014-01-25  Walter Lee  <walt@tilera.com>
 228
 229         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
 230         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
 231         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
 232         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
 233
 234 2014-01-25  Walter Lee  <walt@tilera.com>
 235
 236         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
 237         arguments on even registers.
 238         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
 239         STACK_BOUNDARY.
 240         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
 241         (BIGGEST_ALIGNMENT): Ditto.
 242         (BIGGEST_FIELD_ALIGNMENT): Ditto.
 243
 244 2014-01-25  Walter Lee  <walt@tilera.com>
 245
 246         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
 247         insns before bundling.
 248         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
 249
 250 2014-01-25  Walter Lee  <walt@tilera.com>
 251
 252         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
 253         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
 254         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
 255
 256 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
 257
 258         * config/mips/constraints.md (kl): Delete.
 259         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
 260         define expands, using...
 261         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
 262         instructions for MIPS16.
 263         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
 264         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
 265
 266 2014-01-25  Walter Lee  <walt@tilera.com>
 267
 268         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
 269         (clzdi2): Ditto.
 270         (ffsdi2): Ditto.
 271
 272 2014-01-25  Walter Lee  <walt@tilera.com>
 273
 274         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
 275         (TARGET_EXPAND_TO_RTL_HOOK): Define.
 276
 277 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
 278
 279         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
 280         Handle XOR.
 281
 282 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
 283
 284         * print-rtl.c (in_call_function_usage): New var.
 285         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
 286         EXPR_LIST mode as mode and not as reg note name.
 287
 288         PR middle-end/59561
 289         * cfgloopmanip.c (copy_loop_info): If
 290         loop->warned_aggressive_loop_optimizations, make sure
 291         the flag is set in target loop too.
 292
 293 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
 294
 295         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
 296         flag_cilkplus.
 297         * builtins.def: Likewise.
 298         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
 299         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
 300         * ira.c (ira_setup_eliminable_regset): Likewise.
 301         * omp-low.c (gate_expand_omp): Likewise.
 302         (execute_lower_omp): Likewise.
 303         (diagnose_sb_0): Likewise.
 304         (gate_diagnose_omp_blocks): Likewise.
 305         (simd_clone_clauses_extract): Likewise.
 306         (gate): Likewise.
 307
 308 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 309
 310         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
 311         correction for little endian...
 312         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
 313         here.
 314
 315 2014-01-24  Jeff Law  <law@redhat.com>
 316
 317         PR tree-optimization/59919
 318         * tree-vrp.c (find_assert_locations_1): Do not register asserts
 319         for non-returning calls.
 320
 321 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
 322
 323         * common/config/aarch64/aarch64-common.c
 324         (aarch64_rewrite_mcpu): Handle multiple names.
 325         * config/aarch64/aarch64.h
 326         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
 327
 328 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
 329
 330         * input.c (add_file_to_cache_tab): Handle the case where fopen
 331         returns NULL.
 332
 333 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
 334
 335         PR target/59929
 336         * config/i386/i386.md (pushsf splitter): Get stack adjustment
 337         from push operand if code of push isn't PRE_DEC.
 338
 339 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
 340
 341         PR target/59909
 342         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
 343         -mquad-memory-atomic.  Update -mquad-memory documentation to say
 344         it is only used for non-atomic loads/stores.
 345
 346         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
 347         -mquad-memory or -mquad-memory-atomic switches.
 348
 349         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
 350         -mquad-memory-atomic to ISA 2.07 support.
 351
 352         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
 353         to separate support of normal quad word memory operations (ldq, stq)
 354         from the atomic quad word memory operations.
 355
 356         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
 357         support to separate non-atomic quad word operations from atomic
 358         quad word operations.  Disable non-atomic quad word operations in
 359         little endian mode so that we don't have to swap words after the
 360         load and before the store.
 361         (quad_load_store_p): Add comment about atomic quad word support.
 362         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
 363         options printed with -mdebug=reg.
 364
 365         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
 366         -mquad-memory-atomic as the test for whether we have quad word
 367         atomic instructions.
 368         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
 369         or -mp8-vector are used, allow byte/half-word atomic operations.
 370
 371         * config/rs6000/sync.md (load_lockedti): Insure that the address
 372         is a proper indexed or indirect address for the lqarx instruction.
 373         On little endian systems, swap the hi/lo registers after the lqarx
 374         instruction.
 375         (load_lockedpti): Use indexed_or_indirect_operand predicate to
 376         insure the address is valid for the lqarx instruction.
 377         (store_conditionalti): Insure that the address is a proper indexed
 378         or indirect address for the stqcrx. instruction.  On little endian
 379         systems, swap the hi/lo registers before doing the stqcrx.
 380         instruction.
 381         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
 382         insure the address is valid for the stqcrx. instruction.
 383
 384         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
 385         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
 386         type of quad memory support is available.
 387
 388 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
 389
 390         PR regression/59915
 391         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
 392         there is a danger of looping.
 393
 394 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
 395
 396         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
 397         force flag_ira_loop_pressure if set via command line.
 398
 399 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
 400
 401         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
 402         (ashr_simd): New builtin handling DI mode.
 403         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
 404         (aarch64_sshr_simddi): New match pattern.
 405         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
 406         (vshrd_n_s64): Likewise.
 407         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
 408
 409 2014-01-23  Nick Clifton  <nickc@redhat.com>
 410
 411         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
 412         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
 413         favour of mcu specific scripts.
 414         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
 415         430x multilibs.
 416
 417 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
 418             Alex Velenko  <Alex.Velenko@arm.com>
 419
 420         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
 421         (vaddv_s16): Likewise.
 422         (vaddv_s32): Likewise.
 423         (vaddv_u8): Likewise.
 424         (vaddv_u16): Likewise.
 425         (vaddv_u32): Likewise.
 426         (vaddvq_s8): Likewise.
 427         (vaddvq_s16): Likewise.
 428         (vaddvq_s32): Likewise.
 429         (vaddvq_s64): Likewise.
 430         (vaddvq_u8): Likewise.
 431         (vaddvq_u16): Likewise.
 432         (vaddvq_u32): Likewise.
 433         (vaddvq_u64): Likewise.
 434         (vaddv_f32): Likewise.
 435         (vaddvq_f32): Likewise.
 436         (vaddvq_f64): Likewise.
 437         (vmaxv_f32): Likewise.
 438         (vmaxv_s8): Likewise.
 439         (vmaxv_s16): Likewise.
 440         (vmaxv_s32): Likewise.
 441         (vmaxv_u8): Likewise.
 442         (vmaxv_u16): Likewise.
 443         (vmaxv_u32): Likewise.
 444         (vmaxvq_f32): Likewise.
 445         (vmaxvq_f64): Likewise.
 446         (vmaxvq_s8): Likewise.
 447         (vmaxvq_s16): Likewise.
 448         (vmaxvq_s32): Likewise.
 449         (vmaxvq_u8): Likewise.
 450         (vmaxvq_u16): Likewise.
 451         (vmaxvq_u32): Likewise.
 452         (vmaxnmv_f32): Likewise.
 453         (vmaxnmvq_f32): Likewise.
 454         (vmaxnmvq_f64): Likewise.
 455         (vminv_f32): Likewise.
 456         (vminv_s8): Likewise.
 457         (vminv_s16): Likewise.
 458         (vminv_s32): Likewise.
 459         (vminv_u8): Likewise.
 460         (vminv_u16): Likewise.
 461         (vminv_u32): Likewise.
 462         (vminvq_f32): Likewise.
 463         (vminvq_f64): Likewise.
 464         (vminvq_s8): Likewise.
 465         (vminvq_s16): Likewise.
 466         (vminvq_s32): Likewise.
 467         (vminvq_u8): Likewise.
 468         (vminvq_u16): Likewise.
 469         (vminvq_u32): Likewise.
 470         (vminnmv_f32): Likewise.
 471         (vminnmvq_f32): Likewise.
 472         (vminnmvq_f64): Likewise.
 473
 474 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
 475
 476         * config/aarch64/aarch64-simd.md
 477         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
 478         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
 479         (*aarch64_mul3_elt<mode>): Likewise.
 480         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
 481         (*aarch64_mul3_elt_to_64v2df): Likewise.
 482         (*aarch64_mla_elt<mode>): Likewise.
 483         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
 484         (*aarch64_mls_elt<mode>): Likewise.
 485         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
 486         (*aarch64_fma4_elt<mode>): Likewise.
 487         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
 488         (*aarch64_fma4_elt_to_64v2df): Likewise.
 489         (*aarch64_fnma4_elt<mode>): Likewise.
 490         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
 491         (*aarch64_fnma4_elt_to_64v2df): Likewise.
 492         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
 493         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
 494         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 495         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 496         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
 497         (aarch64_sqdmull_lane<mode>_internal): Likewise.
 498         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
 499
 500 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
 501
 502         * config/aarch64/aarch64-simd.md
 503         (aarch64_be_checked_get_lane<mode>): New define_expand.
 504         * config/aarch64/aarch64-simd-builtins.def
 505         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
 506         New builtin definition.
 507         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
 508         Use new safe be builtin.
 509
 510 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
 511
 512         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
 513         New define_insn.
 514         (aarch64_be_st1<mode>): Likewise.
 515         (aarch_ld1<VALL:mode>): Define_expand modified.
 516         (aarch_st1<VALL:mode>): Likewise.
 517         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
 518         (UNSPEC_ST1): Likewise.
 519
 520 2014-01-23  David Holsgrove <david.holsgrove@xilinx.com>
 521
 522         * config/microblaze/microblaze.md: Add trap insn and attribute
 523
 524 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
 525
 526         PR preprocessor/58580
 527         * input.h (location_get_source_line): Take an additional line_size
 528         parameter.
 529         (void diagnostics_file_cache_fini): Declare new function.
 530         * input.c (struct fcache): New type.
 531         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
 532         New static constants.
 533         (diagnostic_file_cache_init, total_lines_num)
 534         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
 535         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
 536         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
 537         (get_next_line, read_next_line, goto_next_line, read_line_num):
 538         New static function definitions.
 539         (diagnostic_file_cache_fini): New function.
 540         (location_get_source_line): Take an additional output line_len
 541         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
 542         read_line_num.
 543         * diagnostic.c (diagnostic_finish): Call
 544         diagnostic_file_cache_fini.
 545         (adjust_line): Take an additional input parameter for the length
 546         of the line, rather than calculating it with strlen.
 547         (diagnostic_show_locus): Adjust the use of
 548         location_get_source_line and adjust_line with respect to their new
 549         signature.  While displaying a line now, do not stop at the first
 550         null byte.  Rather, display the zero byte as a space and keep
 551         going until we reach the size of the line.
 552         * Makefile.in: Add vec.o to OBJS-libcommon
 553
 554 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
 555             Ilya Tocar     <ilya.tocar@intel.com>
 556
 557         * config/i386/avx512fintrin.h (_mm512_kmov): New.
 558         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
 559         (__builtin_ia32_kmov16): Ditto.
 560         * config/i386/i386.md (UNSPEC_KMOV): New.
 561         (kmovw): Ditto.
 562
 563 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
 564
 565         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
 566         (_mm512_storeu_si512): Ditto.
 567
 568 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
 569
 570         PR target/52125
 571         * rtl.h (get_referenced_operands): Declare.
 572         * recog.c (get_referenced_operands): New function.
 573         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
 574         operands have been referenced when recording LO_SUM references.
 575
 576 2014-01-22  David Holsgrove <david.holsgrove@xilinx.com>
 577
 578         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
 579
 580 2014-01-22  Jan Hubicka  <jh@suse.cz>
 581
 582         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
 583         Enable for generic and recent AMD targets.
 584
 585 2014-01-22  Jan Hubicka  <jh@suse.cz>
 586
 587         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
 588         ARG_SIZE note when adjustment was eliminated.
 589
 590 2014-01-22  Jeff Law  <law@redhat.com>
 591
 592         PR tree-optimization/59597
 593         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
 594         in file.  Accept new argument REGISTERING and use it to modify
 595         dump output appropriately.
 596         (register_jump_thread): Corresponding changes.
 597         (mark_threaded_blocks): Reinstate code to cancel unprofitable
 598         thread paths involving joiner blocks.  Add code to dump cancelled
 599         jump threading paths.
 600
 601 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
 602
 603         PR rtl-optimization/59477
 604         * lra-constraints.c (inherit_in_ebb): Process call for living hard
 605         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
 606
 607 2014-01-22  Tom Tromey  <tromey@redhat.com>
 608
 609         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
 610         PARAMS.
 611         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
 612
 613 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
 614
 615         PR rtl-optimization/59896
 616         * lra-constraints.c (process_alt_operands): Check unused note for
 617         matched operands of insn with no output reloads.
 618
 619 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
 620
 621         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
 622         (mips_move_from_gpr_cost): Likewise.
 623
 624 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
 625
 626         PR rtl-optimization/59858
 627         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
 628         ira_class_hard_regs_num.
 629         (process_alt_operands): Increase reject for dying matched operand.
 630
 631 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
 632
 633         PR target/59003
 634         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
 635         smaller than size, perform several stores or loads and stores
 636         at dst + count - size to store or copy all of size bytes, rather
 637         than just last modesize bytes.
 638
 639 2014-01-20  DJ Delorie  <dj@redhat.com>
 640
 641         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
 642         that CLOBBERs are REGs before propogating their values.
 643
 644 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
 645
 646         PR middle-end/59789
 647         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
 648         (cgraph_inline_failed_type): New function.
 649         * cgraph.h (DEFCIFCODE): Add type.
 650         (cgraph_inline_failed_type_t): New enum.
 651         (cgraph_inline_failed_type): New prototype.
 652         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
 653         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
 654         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
 655         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
 656         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
 657         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
 658         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
 659         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
 660         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
 661         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
 662         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
 663         OPTIMIZATION_MISMATCH.
 664         * tree-inline.c (expand_call_inline): Emit errors during
 665         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
 666
 667 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
 668
 669         PR target/59685
 670         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
 671         mode attribute in insn output.
 672
 673 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
 674
 675         * output.h (output_constant): Delete.
 676         * varasm.c (output_constant): Make private.
 677
 678 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
 679
 680         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
 681
 682 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
 683
 684         PR middle-end/59860
 685         * tree.h (fold_builtin_strcat): New prototype.
 686         * builtins.c (fold_builtin_strcat): No longer static.  Add len
 687         argument, if non-NULL, don't call c_strlen.  Optimize
 688         directly into __builtin_memcpy instead of __builtin_strcpy.
 689         (fold_builtin_2): Adjust fold_builtin_strcat caller.
 690         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
 691
 692 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
 693
 694         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
 695         for SImode_address_operand operands, having only a REG argument.
 696
 697 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
 698
 699         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
 700         loader name using mbig-endian.
 701         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
 702
 703 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
 704
 705         * doc/invoke.texi (-march): Clarify documentation for AArch64.
 706         (-mtune): Likewise.
 707         (-mcpu): Likewise.
 708
 709 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
 710
 711         * config/aarch64/aarch64-protos.h
 712         (aarch64_cannot_change_mode_class_ptr): Declare.
 713         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
 714         aarch64_cannot_change_mode_class_ptr): New.
 715         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
 716         backend hook aarch64_cannot_change_mode_class.
 717
 718 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
 719
 720         * common/config/aarch64/aarch64-common.c
 721         (aarch64_handle_option): Don't handle any option order logic here.
 722         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
 723         selected_cpu, warn on architecture version mismatch.
 724         (aarch64_override_options): Fix parsing order for option strings.
 725
 726 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
 727             Iain Sandoe  <iain@codesourcery.com>
 728
 729         PR bootstrap/59496
 730         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
 731         warning.  Amend comment to reflect current functionality.
 732
 733 2014-01-20  Richard Biener  <rguenther@suse.de>
 734
 735         PR middle-end/59860
 736         * builtins.c (fold_builtin_strcat): Remove case better handled
 737         by tree-ssa-strlen.c.
 738
 739 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
 740
 741         * config/aarch64/aarch64.opt
 742         (mcpu, march, mtune): Make case-insensitive.
 743
 744 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
 745
 746         PR target/59880
 747         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
 748         if operands[1] is a REG or ZERO_EXTEND of a REG.
 749
 750 2014-01-19  Jan Hubicka  <jh@suse.cz>
 751
 752         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
 753
 754 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
 755
 756         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
 757         long non-pic millicode calls.
 758
 759 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
 760
 761         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
 762
 763 2014-01-19  Kito Cheng  <kito@0xlab.org>
 764
 765         * builtins.c (expand_movstr): Check movstr expand done or fail.
 766
 767 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
 768             H.J. Lu  <hongjiu.lu@intel.com>
 769
 770         PR target/59379
 771         * config/i386/i386.md (*lea<mode>): Zero-extend return register
 772         to DImode for zero-extended addresses.
 773
 774 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
 775
 776         PR rtl-optimization/57763
 777         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
 778         on the new indirect jump_insn and increment LABEL_NUSES (label).
 779
 780 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
 781
 782         PR bootstrap/59580
 783         PR bootstrap/59583
 784         * config.gcc (x86_archs): New variable.
 785         (x86_64_archs): Likewise.
 786         (x86_cpus): Likewise.
 787         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
 788         --with-arch/--with-cpu= options.
 789         Support --with-arch=/--with-cpu={nehalem,westmere,
 790         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
 791
 792 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
 793
 794         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
 795         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
 796
 797 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
 798
 799         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
 800
 801 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
 802
 803         PR target/58944
 804         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
 805         clear cpp_get_options (parse_in)->warn_unused_macros for
 806         ix86_target_macros_internal with cpp_define.
 807
 808 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
 809
 810         * jump.c (delete_related_insns): Keep (use (insn))s.
 811         * reorg.c (redundant_insn): Check for barriers too.
 812
 813 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
 814
 815         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
 816
 817 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
 818
 819         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
 820         call to $$dyncall when TARGET_LONG_CALLS is true.
 821
 822 2014-01-17  Jeff Law  <law@redhat.com>
 823
 824         * ree.c (combine_set_extension): Temporarily disable test for
 825         changing number of hard registers.
 826
 827 2014-01-17  Jan Hubicka  <jh@suse.cz>
 828
 829         PR middle-end/58125
 830         * ipa-inline-analysis.c (inline_free_summary):
 831         Do not free summary of aliases.
 832
 833 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
 834
 835         PR middle-end/59706
 836         * gimplify.c (gimplify_expr): Use create_tmp_var
 837         instead of create_tmp_var_raw.  If cond doesn't have
 838         integral type, don't add the IFN_ANNOTATE builtin at all.
 839
 840 2014-01-17  Martin Jambor  <mjambor@suse.cz>
 841
 842         PR ipa/59736
 843         * ipa-cp.c (prev_edge_clone): New variable.
 844         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
 845         Also resize prev_edge_clone vector.
 846         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
 847         (ipcp_edge_removal_hook): New function.
 848         (ipcp_driver): Register ipcp_edge_removal_hook.
 849
 850 2014-01-17  Andrew Pinski <apinski@cavium.com>
 851             Steve Ellcey  <sellcey@mips.com>
 852
 853         PR target/59462
 854         * config/mips/mips.c (mips_print_operand): Check operand mode instead
 855         of operator mode.
 856
 857 2014-01-17  Jeff Law  <law@redhat.com>
 858
 859         PR middle-end/57904
 860         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
 861         so that pass_ccp runs first.
 862
 863 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
 864
 865         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
 866         (ix86_adjust_cost): Use !TARGET_XXX.
 867         (do_reorder_for_imul): Likewise.
 868         (swap_top_of_ready_list): Likewise.
 869         (ix86_sched_reorder): Likewise.
 870
 871 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
 872
 873         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
 874         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
 875         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
 876         (intel_memset): New.  Duplicate slm_memset.
 877         (intel_cost): New.  Duplicate slm_cost.
 878         (m_INTEL): New macro.
 879         (processor_target_table): Add "intel".
 880         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
 881         with PROCESSOR_INTEL for "intel".
 882         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
 883         PROCESSOR_SILVERMONT.
 884         (ix86_issue_rate): Likewise.
 885         (ix86_adjust_cost): Likewise.
 886         (ia32_multipass_dfa_lookahead): Likewise.
 887         (swap_top_of_ready_list): Likewise.
 888         (ix86_sched_reorder): Likewise.
 889         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
 890         instead of TARGET_OPT_AGU.
 891         * config/i386/i386.h (TARGET_INTEL): New.
 892         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
 893         (processor_type): Add PROCESSOR_INTEL.
 894         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
 895         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
 896
 897 2014-01-17  Marek Polacek  <polacek@redhat.com>
 898
 899         PR c/58346
 900         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
 901         size is zero.
 902
 903 2014-01-17  Richard Biener  <rguenther@suse.de>
 904
 905         PR tree-optimization/46590
 906         * opts.c (default_options_table): Add entries for
 907         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
 908         all enabled at -O1 but not for -Og.
 909         * common.opt (fbranch-count-reg): Remove Init(1).
 910         (fmove-loop-invariants): Likewise.
 911         (ftree-pta): Likewise.
 912
 913 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
 914
 915         * config/i386/i386.c (ix86_data_alignment): For compatibility with
 916         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
 917         decls to at least the GCC 4.8 used alignments.
 918
 919         PR fortran/59440
 920         * tree-nested.c (convert_nonlocal_reference_stmt,
 921         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
 922         of GIMPLE_BIND stmts, adjust associated decls.
 923
 924 2014-01-17  Richard Biener  <rguenther@suse.de>
 925
 926         PR tree-optimization/46590
 927         * vec.h (vec<>::bseach): New member function implementing
 928         binary search according to C89 bsearch.
 929         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
 930         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
 931         bitmap pointer again.  Make accesses_in_loop a flat array.
 932         (mem_ref_obstack): New global.
 933         (outermost_indep_loop): Adjust for mem_ref->stored changes.
 934         (mark_ref_stored): Likewise.
 935         (ref_indep_loop_p_2): Likewise.
 936         (set_ref_stored_in_loop): New helper function.
 937         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
 938         (memref_free): Adjust.
 939         (record_mem_ref_loc): Simplify.
 940         (gather_mem_refs_stmt): Adjust.
 941         (sort_locs_in_loop_postorder_cmp): New function.
 942         (analyze_memory_references): Sort accesses_in_loop after
 943         loop postorder number.
 944         (find_ref_loc_in_loop_cmp): New function.
 945         (for_all_locs_in_loop): Find relevant cluster of locs in
 946         accesses_in_loop and iterate without recursion.
 947         (execute_sm): Avoid uninit warning.
 948         (struct ref_always_accessed): Simplify.
 949         (ref_always_accessed::operator ()): Likewise.
 950         (ref_always_accessed_p): Likewise.
 951         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
 952         loop postorder numbers here.
 953         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
 954         numbers.
 955
 956 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
 957
 958         PR c++/57945
 959         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
 960         on decls for which assemble_alias has been called.
 961
 962 2014-01-17  Nick Clifton  <nickc@redhat.com>
 963
 964         * config/msp430/msp430.opt: (mcpu): New option.
 965         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
 966         (msp430_option_override): Parse target_cpu.  If the MCU name
 967         matches a generic string, clear target_mcu.
 968         (msp430_attr): Allow numeric interrupt values up to 63.
 969         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
 970         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
 971         option.
 972         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
 973         Add mcpu matches.
 974         * config/msp430/msp430.md (popm): Use %J rather than %I.
 975         (addsi3): Use msp430_nonimmediate_operand for operand 2.
 976         (addhi_cy_i): Use immediate_operand for operand 2.
 977         * doc/invoke.texi: Document -mcpu option.
 978
 979 2014-01-17  Richard Biener  <rguenther@suse.de>
 980
 981         PR rtl-optimization/38518
 982         * df.h (df_analyze_loop): Declare.
 983         * df-core.c: Include cfgloop.h.
 984         (df_analyze_1): Split out main part of df_analyze.
 985         (df_analyze): Adjust.
 986         (loop_inverted_post_order_compute): New function.
 987         (loop_post_order_compute): Likewise.
 988         (df_analyze_loop): New function avoiding whole-function
 989         postorder computes.
 990         * loop-invariant.c (find_defs): Use df_analyze_loop.
 991         (find_invariants): Adjust.
 992         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
 993
 994 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
 995
 996         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
 997         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
 998
 999 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
1000
1001         * ipa-ref.c (ipa_remove_stmt_references): Fix references
1002         traversal when removing references.
1003
1004 2014-01-16  Jan Hubicka  <jh@suse.cz>
1005
1006         PR ipa/59775
1007         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1008
1009 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
1010
1011         PR middle-end/56791
1012         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1013         pushing a reload for an autoinc when we had previously reloaded an
1014         inner part of the address.
1015
1016 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
1017
1018         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1019         field.
1020         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1021         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1022         when not giving up or versioning for alias only because of
1023         loop->safelen.
1024         (vect_analyze_data_ref_dependences): Set to true.
1025         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1026         is a GIMPLE_PHI.
1027         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1028         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1029         to the condition.
1030
1031         PR middle-end/58344
1032         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1033
1034         PR target/59839
1035         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1036         operand 0 predicate for gathers, use a new pseudo as subtarget.
1037
1038 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
1039
1040         PR middle-end/59609
1041         * lra-constraints.c (process_alt_operands): Add printing debug info.
1042         Check absence of input/output reloads for matched operands too.
1043
1044 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
1045
1046         PR rtl-optimization/59835
1047         * ira.c (ira_init_register_move_cost): Increase cost for
1048         impossible modes.
1049
1050 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
1051
1052         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1053
1054 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
1055
1056         PR target/59780
1057         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1058         non-register objects.  Use gen_(high/low)part more consistently.
1059         Fix assertions.
1060
1061 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1062
1063         PR target/59844
1064         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1065         endian support, remove tests for WORDS_BIG_ENDIAN.
1066         (p8_mfvsrd_3_<mode>): Likewise.
1067         (reload_gpr_from_vsx<mode>): Likewise.
1068         (reload_gpr_from_vsxsf): Likewise.
1069         (p8_mfvsrd_4_disf): Likewise.
1070
1071 2014-01-16  Richard Biener  <rguenther@suse.de>
1072
1073         PR rtl-optimization/46590
1074         * lcm.c (compute_antinout_edge): Use postorder iteration.
1075         (compute_laterin): Use inverted postorder iteration.
1076
1077 2014-01-16  Nick Clifton  <nickc@redhat.com>
1078
1079         PR middle-end/28865
1080         * varasm.c (output_constant): Return the number of bytes actually
1081         emitted.
1082         (output_constructor_array_range): Update the field size with the
1083         number of bytes emitted by output_constant.
1084         (output_constructor_regular_field): Likewise.  Also do not
1085         complain if the total number of bytes emitted is now greater
1086         than the expected fieldpos.
1087         * output.h (output_constant): Update prototype and descriptive comment.
1088
1089 2014-01-16  Marek Polacek  <polacek@redhat.com>
1090
1091         PR middle-end/59827
1092         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1093         it is error_mark_node.
1094
1095 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
1096
1097         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1098         VALID_AVX256_REG_OR_OI_MODE.
1099
1100 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
1101
1102         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1103         current procedure should be profiled.
1104
1105 2014-01-15  Andrew Pinski  <apinski@cavium.com>
1106
1107         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1108         of moving from/to the STACK_REG register class.
1109
1110 2014-01-15  Richard Henderson <rth@redhat.com>
1111
1112         PR debug/54694
1113         * reginfo.c (global_regs_decl): Globalize.
1114         * rtl.h (global_regs_decl): Declare.
1115         * ira.c (do_reload): Diagnose frame_pointer_needed and it
1116         reserved via global_regs.
1117
1118 2014-01-15  Teresa Johnson  <tejohnson@google.com>
1119
1120         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1121
1122 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
1123
1124         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1125         and vmulosh rather than call gen_vec_widen_smult_*.
1126         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1127         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1128         (vec_widen_smult_even_v16qi): Likewise.
1129         (vec_widen_umult_even_v8hi): Likewise.
1130         (vec_widen_smult_even_v8hi): Likewise.
1131         (vec_widen_umult_odd_v16qi): Likewise.
1132         (vec_widen_smult_odd_v16qi): Likewise.
1133         (vec_widen_umult_odd_v8hi): Likewise.
1134         (vec_widen_smult_odd_v8hi): Likewise.
1135         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1136         vmuloub rather than call gen_vec_widen_umult_*.
1137         (vec_widen_umult_lo_v16qi): Likewise.
1138         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1139         vmulosb rather than call gen_vec_widen_smult_*.
1140         (vec_widen_smult_lo_v16qi): Likewise.
1141         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1142         rather than call gen_vec_widen_umult_*.
1143         (vec_widen_umult_lo_v8hi): Likewise.
1144         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1145         rather than call gen_vec_widen_smult_*.
1146         (vec_widen_smult_lo_v8hi): Likewise.
1147
1148 2014-01-15  Jeff Law  <law@redhat.com>
1149
1150         PR tree-optimization/59747
1151         * ree.c (find_and_remove_re): Properly handle case where a second
1152         eliminated extension requires widening a copy created for elimination
1153         of a prior extension.
1154         (combine_set_extension): Ensure that the number of hard regs needed
1155         for a destination register does not change when we widen it.
1156
1157 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
1158
1159         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1160         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1161         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1162         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1163         (avr-*-rtems*): Likewise.
1164         (bfin*-rtems*): Likewise.
1165         (moxie-*-rtems*): Likewise.
1166         (h8300-*-rtems*): Likewise.
1167         (i[34567]86-*-rtems*): Likewise.
1168         (lm32-*-rtems*): Likewise.
1169         (m32r-*-rtems*): Likewise.
1170         (m68k-*-rtems*): Likewise.
1171         (microblaze*-*-rtems*): Likewise.
1172         (mips*-*-rtems*): Likewise.
1173         (powerpc-*-rtems*): Likewise.
1174         (sh-*-rtems*): Likewise.
1175         (sparc-*-rtems*): Likewise.
1176         (sparc64-*-rtems*): Likewise.
1177         (v850-*-rtems*): Likewise.
1178         (m32c-*-rtems*): Likewise.
1179
1180 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
1181
1182         PR rtl-optimization/59511
1183         * ira.c (ira_init_register_move_cost): Use memory costs for some
1184         cases of register move cost calculations.
1185         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1186         instead of BB frequency.
1187         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1188         * lra-assigns.c (find_hard_regno_for): Ditto.
1189
1190 2014-01-15  Richard Biener  <rguenther@suse.de>
1191
1192         PR tree-optimization/59822
1193         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1194         (vectorizable_load): Use it to hoist defs of uses of invariant
1195         loads out of the loop.
1196
1197 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1198             Kugan Vivekanandarajah  <kuganv@linaro.org>
1199
1200         PR target/59695
1201         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1202         truncation.
1203
1204 2014-01-15  Richard Biener  <rguenther@suse.de>
1205
1206         PR rtl-optimization/59802
1207         * lcm.c (compute_available): Use inverted postorder to seed
1208         the initial worklist.
1209
1210 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1211
1212         PR target/59803
1213         * config/s390/s390.c (s390_preferred_reload_class): Don't return
1214         ADDR_REGS for invalid symrefs in non-PIC code.
1215
1216 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
1217
1218         PR other/58712
1219         * builtins.c (determine_block_size): Initialize *probable_max_size
1220         even if len_rtx is CONST_INT.
1221
1222 2014-01-14  Andrew Pinski  <apinski@cavium.com>
1223
1224         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1225         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1226         (cortexa53_tunings): Likewise.
1227         (aarch64_sched_issue_rate): New function.
1228         (TARGET_SCHED_ISSUE_RATE): Define.
1229
1230 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
1231
1232         * ira-costs.c (find_costs_and_classes): Add missed
1233         ira_init_register_move_cost_if_necessary.
1234
1235 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
1236
1237         PR target/59787
1238         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1239
1240 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
1241
1242         PR target/59794
1243         * config/i386/i386.c (type_natural_mode): Add a bool parameter
1244         to indicate if type is used for function return value.  Warn ABI
1245         change if the vector mode isn't available for function return value.
1246         (ix86_function_arg_advance): Pass false to type_natural_mode.
1247         (ix86_function_arg): Likewise.
1248         (ix86_gimplify_va_arg): Likewise.
1249         (function_arg_32): Don't warn ABI change.
1250         (ix86_function_value): Pass true to type_natural_mode.
1251         (ix86_return_in_memory): Likewise.
1252         (ix86_struct_value_rtx): Removed.
1253         (TARGET_STRUCT_VALUE_RTX): Likewise.
1254
1255 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1256
1257         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1258         converting a conditional jump into a conditional return.
1259
1260 2014-01-14  Richard Biener  <rguenther@suse.de>
1261
1262         PR tree-optimization/58921
1263         PR tree-optimization/59006
1264         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1265         hoisting invariant stmts.
1266         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1267         invariant loads on the preheader edge if possible.
1268
1269 2014-01-14  Joey Ye  <joey.ye@arm.com>
1270
1271         * doc/plugin.texi (Building GCC plugins): Update to C++.
1272
1273 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
1274
1275         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1276         (_mm_rcp28_round_ss): Ditto.
1277         (_mm_rsqrt28_round_sd): Ditto.
1278         (_mm_rsqrt28_round_ss): Ditto.
1279         (_mm_rcp28_sd): Ditto.
1280         (_mm_rcp28_ss): Ditto.
1281         (_mm_rsqrt28_sd): Ditto.
1282         (_mm_rsqrt28_ss): Ditto.
1283         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1284         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1285         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1286         (IX86_BUILTIN_RCP28SD): Ditto.
1287         (IX86_BUILTIN_RCP28SS): Ditto.
1288         (IX86_BUILTIN_RSQRT28SD): Ditto.
1289         (IX86_BUILTIN_RSQRT28SS): Ditto.
1290         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1291         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1292         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1293         (ix86_expand_special_args_builtin): Expand new FTYPE.
1294         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1295         (srcp14<mode>): Make insn unary.
1296         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1297         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1298         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1299         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1300         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1301         Fix rounding: make it SAE only.
1302         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1303         Ditto.
1304         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1305         Ditto.
1306         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1307         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1308         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1309         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1310         (round_saeonly_mask_scalar_operand4): Ditto.
1311         (round_saeonly_mask_scalar_op3): Ditto.
1312         (round_saeonly_mask_scalar_op4): Ditto.
1313
1314 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1315
1316         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1317         Implement -maltivec=be for vec_insert and vec_extract.
1318
1319 2014-01-10  DJ Delorie  <dj@redhat.com>
1320
1321         * config/msp430/msp430.md (call_internal): Don't allow memory
1322         references with SP as the base register.
1323         (call_value_internal): Likewise.
1324         * config/msp430/constraints.md (Yc): New.  For memory references
1325         that don't use SP as a base register.
1326
1327         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1328         "an integer without a # prefix"
1329         * config/msp430/msp430.md (epilogue_helper): Use it.
1330
1331 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
1332
1333         PR target/59617
1334         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1335         AVX512F gather builtins.
1336         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1337         on gather decls with INTEGER_TYPE masktype.
1338         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1339         directly into the builtin rather than hoisting it before loop.
1340
1341         PR tree-optimization/59387
1342         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1343         (scev_const_prop): If folded_casts and type has undefined overflow,
1344         use force_gimple_operand instead of force_gimple_operand_gsi and
1345         for each added stmt if it is assign with
1346         arith_code_with_undefined_signed_overflow, call
1347         rewrite_to_defined_overflow.
1348         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1349         gimple-fold.h instead.
1350         (arith_code_with_undefined_signed_overflow,
1351         rewrite_to_defined_overflow): Moved to ...
1352         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1353         rewrite_to_defined_overflow): ... here.  No longer static.
1354         Include gimplify-me.h.
1355         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1356         rewrite_to_defined_overflow): New prototypes.
1357
1358 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1359
1360         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1361
1362 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
1363
1364         * builtins.c (get_object_alignment_2): Minor tweak.
1365         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1366
1367 2014-01-13  Christian Bruel  <christian.bruel@st.com>
1368
1369         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1370         optimized non constant lengths.
1371
1372 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
1373
1374         PR libgomp/59194
1375         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1376         load as __atomic_load_N if possible.
1377
1378 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
1379
1380         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1381         target parameter.
1382         (rs6000_expand_builtin): Adjust call.
1383
1384 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
1385
1386         PR target/58115
1387         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1388         * config/rs6000/rs6000.c: Include target-globals.h.
1389         (rs6000_set_current_function): Instead of doing target_reinit
1390         unconditionally, use save_target_globals_default_opts and
1391         restore_target_globals.
1392
1393         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1394         FPSCR.
1395         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1396         (rs6000_expand_builtin): Handle mffs and mtfsf.
1397         (rs6000_init_builtins): Define mffs and mtfsf.
1398         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1399         (rs6000_mffs): New pattern.
1400         (rs6000_mtfsf): New pattern.
1401
1402 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
1403
1404         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1405         Start narrowing with START.  Apply candidate-use pair
1406         and check overall cost in narrowing.
1407         (iv_ca_prune): Pass new argument.
1408
1409 2014-01-10  Jeff Law  <law@redhat.com>
1410
1411         PR middle-end/59743
1412         * ree.c (combine_reaching_defs): Ensure the defining statement
1413         occurs before the extension when optimizing extensions with
1414         different source and destination hard registers.
1415
1416 2014-01-10  Jan Hubicka  <jh@suse.cz>
1417
1418         PR ipa/58585
1419         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1420         vtables into the type inheritance graph.
1421
1422 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1423
1424         PR rtl-optimization/59754
1425         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1426         modes in the REGNO != REGNO case.
1427
1428 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1429
1430         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1431
1432 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1433
1434         PR tree-optimization/59745
1435         * tree-predcom.c (tree_predictive_commoning_loop): Call
1436         free_affine_expand_cache if giving up because components is NULL.
1437
1438         * target-globals.c (save_target_globals): Allocate < 4KB structs using
1439         GC in payload of target_globals struct instead of allocating them on
1440         the heap and the larger structs separately using GC.
1441         * target-globals.h (struct target_globals): Make regs, hard_regs,
1442         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1443         of GTY((skip)) and change type to void *.
1444         (reset_target_globals): Cast loads from those fields to corresponding
1445         types.
1446
1447 2014-01-10  Steve Ellcey  <sellcey@mips.com>
1448
1449         PR plugins/59335
1450         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1451         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1452         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1453
1454 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
1455
1456         PR target/59744
1457         * aarch64-modes.def (CC_Zmode): New flags mode.
1458         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1459         represents an equality.
1460         (aarch64_get_condition_code): Handle CC_Zmode.
1461         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1462
1463 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1464
1465         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1466         extraction in good case.
1467
1468 2014-01-10  Richard Biener  <rguenther@suse.de>
1469
1470         PR tree-optimization/59374
1471         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1472         checking after SLP discovery.  Mark stmts not participating
1473         in any SLP instance properly.
1474
1475 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1476
1477         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1478         when handling a SET rtx.
1479
1480 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1481
1482         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1483         (cortex-a57): Likewise.
1484         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1485
1486 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1487
1488         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1489         non-iwmmxt builtins.
1490
1491 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
1492
1493         PR ipa/58252
1494         PR ipa/59226
1495         * ipa-devirt.c record_target_from_binfo): Take as argument
1496         stack of binfos and lookup matching one for virtual inheritance.
1497         (possible_polymorphic_call_targets_1): Update.
1498
1499 2014-01-10  Huacai Chen  <chenhc@lemote.com>
1500
1501         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1502         kernel strings for Loongson-2E/2F/3A.
1503
1504 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1505
1506         PR middle-end/59670
1507         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1508         is_gimple_call before calling gimple_call_internal_p.
1509
1510 2014-01-09  Steve Ellcey  <sellcey@mips.com>
1511
1512         * Makefile.in (TREE_FLOW_H): Remove.
1513         (TREE_SSA_H): Add file names from tree-flow.h.
1514         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1515         * tree.h: Remove tree-flow.h reference.
1516         * hash-table.h: Remove tree-flow.h reference.
1517         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1518         reference with tree-ssa-loop.h.
1519
1520 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1521
1522         * doc/invoke.texi: Add -maltivec={be,le} options, and document
1523         default element-order behavior for -maltivec.
1524         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1525         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1526         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1527         when targeting big endian, at least for now.
1528         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1529
1530 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1531
1532         PR middle-end/47735
1533         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1534         var satisfies use_register_for_decl, just take into account type
1535         alignment, rather than decl alignment.
1536
1537         PR tree-optimization/59622
1538         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
1539         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1540         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1541         Don't devirtualize for inplace at all.  For targets.length () == 1,
1542         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1543
1544 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
1545
1546         * config/i386/i386.md (cpu): Remove the unused btver1.
1547
1548 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
1549
1550         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1551
1552 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1553
1554         PR target/58115
1555         * tree-core.h (struct target_globals): New forward declaration.
1556         (struct tree_target_option): Add globals field.
1557         * tree.h (TREE_TARGET_GLOBALS): Define.
1558         (prepare_target_option_nodes_for_pch): New prototype.
1559         * target-globals.h (struct target_globals): Define even if
1560         !SWITCHABLE_TARGET.
1561         * tree.c (prepare_target_option_node_for_pch,
1562         prepare_target_option_nodes_for_pch): New functions.
1563         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1564         * config/i386/i386.c: Include target-globals.h.
1565         (ix86_set_current_function): Instead of doing target_reinit
1566         unconditionally, use save_target_globals_default_opts and
1567         restore_target_globals.
1568
1569 2014-01-09  Richard Biener  <rguenther@suse.de>
1570
1571         PR tree-optimization/59715
1572         * tree-cfg.h (split_critical_edges): Declare.
1573         * tree-cfg.c (split_critical_edges): Export.
1574         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1575
1576 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
1577
1578         * cfgexpand.c (expand_stack_vars): Optionally disable
1579         asan stack protection.
1580         (expand_used_vars): Likewise.
1581         (partition_stack_vars): Likewise.
1582         * asan.c (asan_emit_stack_protection): Optionally disable
1583         after return stack usage.
1584         (instrument_derefs): Optionally disable memory access instrumentation.
1585         (instrument_builtin_call): Likewise.
1586         (instrument_strlen_call): Likewise.
1587         (asan_protect_global): Optionally disable global variables protection.
1588         * doc/invoke.texi: Added doc for new options.
1589         * params.def: Added new options.
1590         * params.h: Likewise.
1591
1592 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1593
1594         PR rtl-optimization/59724
1595         * ifcvt.c (cond_exec_process_if_block): Don't call
1596         flow_find_head_matching_sequence with 0 longest_match.
1597         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1598         non-active insns if !stop_after.
1599         (try_head_merge_bb): Revert 2014-01-07 changes.
1600
1601 2014-01-08  Jeff Law  <law@redhat.com>
1602
1603         * ree.c (get_sub_rtx): New function, extracted from...
1604         (merge_def_and_ext): Here.
1605         (combine_reaching_defs): Use get_sub_rtx.
1606
1607 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
1608
1609         * cgraph.h (varpool_variable_node): Do not choke on null node.
1610
1611 2014-01-08  Catherine Moore  <clm@codesourcery.com>
1612
1613         * config/mips/mips.md (simple_return): Attempt to use JRC
1614         for microMIPS.
1615         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1616
1617 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
1618
1619         PR rtl-optimization/59137
1620         * reorg.c (steal_delay_list_from_target): Call update_block for
1621         elided insns.
1622         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1623
1624 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1625
1626         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1627         two duplicate entries.
1628
1629 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
1630
1631         Revert:
1632         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
1633
1634         * config/mips/mips.c (mips_truncated_op_cost): New function.
1635         (mips_rtx_costs): Adjust test for BADDU.
1636         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1637
1638         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
1639
1640         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1641         (*baddu_si): ...this new pattern.
1642
1643 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
1644
1645         PR ipa/59722
1646         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1647
1648 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1649
1650         PR middle-end/57748
1651         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1652         inner_reference_p.
1653         (expand_expr, expand_normal): Adjust.
1654         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1655         inner_reference_p. Use inner_reference_p to expand inner references.
1656         (store_expr): Adjust.
1657         * cfgexpand.c (expand_call_stmt): Adjust.
1658
1659 2014-01-08  Rong Xu  <xur@google.com>
1660
1661         * gcov-io.c (gcov_var): Move from gcov-io.h.
1662         (gcov_position): Ditto.
1663         (gcov_is_error): Ditto.
1664         (gcov_rewrite): Ditto.
1665         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1666         only part to libgcc/libgcov.h.
1667
1668 2014-01-08  Marek Polacek  <polacek@redhat.com>
1669
1670         PR middle-end/59669
1671         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1672
1673 2014-01-08  Marek Polacek  <polacek@redhat.com>
1674
1675         PR sanitizer/59667
1676         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
1677
1678 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
1679
1680         PR rtl-optimization/59649
1681         * stor-layout.c (get_mode_bounds): For BImode return
1682         0 and STORE_FLAG_VALUE.
1683
1684 2014-01-08  Richard Biener  <rguenther@suse.de>
1685
1686         PR middle-end/59630
1687         * gimple.h (is_gimple_builtin_call): Remove.
1688         (gimple_builtin_call_types_compatible_p): New.
1689         (gimple_call_builtin_p): New overload.
1690         * gimple.c (is_gimple_builtin_call): Remove.
1691         (validate_call): Rename to ...
1692         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
1693         check return types.
1694         (validate_type): New static function.
1695         (gimple_call_builtin_p): New overload and adjust.
1696         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
1697         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
1698         (gimple_fold_stmt_to_constant_1): Likewise.
1699         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
1700
1701 2014-01-08  Richard Biener  <rguenther@suse.de>
1702
1703         PR middle-end/59471
1704         * gimplify.c (gimplify_expr): Gimplify register-register type
1705         VIEW_CONVERT_EXPRs to separate stmts.
1706
1707 2014-01-07  Jeff Law  <law@redhat.com>
1708
1709         PR middle-end/53623
1710         * ree.c (combine_set_extension): Handle case where source
1711         and destination registers in an extension insn are different.
1712         (combine_reaching_defs): Allow source and destination registers
1713         in extension to be different under limited circumstances.
1714         (add_removable_extension): Remove restriction that the
1715         source and destination registers in the extension are the same.
1716         (find_and_remove_re): Emit a copy from the extension's
1717         destination to its source after the defining insn if
1718         the source and destination registers are different.
1719
1720         PR middle-end/59285
1721         * ifcvt.c (merge_if_block): If we are merging a block with more than
1722         one successor with a block with no successors, remove any BARRIER
1723         after the second block.
1724
1725 2014-01-07  Dan Xio Qiang <ziyan01@163.com>
1726
1727         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
1728
1729 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
1730
1731         PR target/59652
1732         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
1733         for 14-bit register offsets when INT14_OK_STRICT is false.
1734
1735 2014-01-07  Roland Stigge  <stigge@antcom.de>
1736             Michael Meissner  <meissner@linux.vnet.ibm.com>
1737
1738         PR 57386/target
1739         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1740         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
1741
1742 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
1743
1744         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
1745         -mcpu.
1746
1747 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
1748
1749         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
1750         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
1751         rtx is const0_rtx or not.
1752
1753 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
1754
1755         PR target/58115
1756         * target-globals.c (save_target_globals): Remove this_fn_optab
1757         handling.
1758         * toplev.c: Include optabs.h.
1759         (target_reinit): Temporarily restore the global options if another
1760         set of options are in force.
1761
1762 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
1763
1764         PR rtl-optimization/58668
1765         * cfgcleanup.c (flow_find_cross_jump): Don't count
1766         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
1767         to determine what is counted.
1768         (flow_find_head_matching_sequence): Use active_insn_p to determine
1769         what is counted.
1770         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
1771         counting change.
1772         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
1773         determine what is counted.
1774
1775         PR tree-optimization/59643
1776         * tree-predcom.c (split_data_refs_to_components): If one dr is
1777         read and one write, determine_offset fails and the write isn't
1778         in the bad component, just put the read into the bad component.
1779
1780 2014-01-07  Mike Stump  <mikestump@comcast.net>
1781             Jakub Jelinek  <jakub@redhat.com>
1782
1783         PR pch/59436
1784         * tree-core.h (struct tree_optimization_option): Change optabs
1785         type from unsigned char * to void *.
1786         * optabs.c (init_tree_optimization_optabs): Adjust
1787         TREE_OPTIMIZATION_OPTABS initialization.
1788
1789 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
1790
1791         PR target/59644
1792         * config/i386/i386.h (struct machine_function): Add
1793         no_drap_save_restore field.
1794         * config/i386/i386.c (ix86_save_reg): Use
1795         !cfun->machine->no_drap_save_restore instead of
1796         crtl->stack_realign_needed.
1797         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
1798         this function clears frame_pointer_needed.  Set
1799         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
1800         and DRAP reg is needed.
1801
1802 2014-01-06  Marek Polacek  <polacek@redhat.com>
1803
1804         PR c/57773
1805         * doc/implement-c.texi: Mention that other integer types are
1806         permitted as bit-field types in strictly conforming mode.
1807
1808 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
1809
1810         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
1811         is newly allocated.
1812
1813 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
1814
1815         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
1816
1817 2014-01-06  Martin Jambor  <mjambor@suse.cz>
1818
1819         PR ipa/59008
1820         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
1821         to int.
1822         * ipa-prop.c (ipa_print_node_params): Fix indentation.
1823
1824 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
1825
1826         PR debug/59350
1827         PR debug/59510
1828         * var-tracking.c (add_stores): Preserve the value of the source even if
1829         we don't record the store.
1830
1831 2014-01-06  Terry Guo  <terry.guo@arm.com>
1832
1833         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
1834
1835 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
1836
1837         PR bootstrap/59541
1838         * config/darwin.c (darwin_function_section): Adjust return values to
1839         correspond to optimisation changes made in r206070.
1840
1841 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
1842
1843         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
1844         from prefetch_block tune setting.
1845         (nocona_cost): Correct size of prefetch block to 64.
1846
1847 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
1848
1849         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
1850         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
1851         used to save the static chain register in the computation of the offset
1852         from which the FP registers need to be restored.
1853
1854 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
1855
1856         PR tree-optimization/59519
1857         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
1858         ICE if get_current_def (current_new_name) is already non-NULL, as long
1859         as it is a phi result of some other phi in *new_exit_bb that has
1860         the same argument.
1861
1862         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
1863         or vmovdqu* for misaligned_operand.
1864         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
1865         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
1866         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
1867         aligned_mem for AVX512F masked aligned load and store builtins and for
1868         non-temporal moves.
1869
1870 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
1871
1872         PR tree-optimization/59651
1873         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
1874         Address range for negative step should be added by TYPE_SIZE_UNIT.
1875
1876 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
1877
1878         * config/m68k/m68k.c (handle_move_double): Handle pushes with
1879         overlapping registers also for registers other than the stack pointer.
1880
1881 2014-01-03  Marek Polacek  <polacek@redhat.com>
1882
1883         PR other/59661
1884         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
1885         __builtin_FILE.
1886
1887 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
1888
1889         PR target/59625
1890         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
1891         asm goto as jump.
1892
1893         * config/i386/i386.md (MODE_SIZE): New mode attribute.
1894         (push splitter): Use <P:MODE_SIZE> instead of
1895         GET_MODE_SIZE (<P:MODE>mode).
1896         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
1897         (mov -1, reg peephole2): Likewise.
1898         * config/i386/sse.md (*mov<mode>_internal,
1899         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
1900         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
1901         *<code><mode>3, *andnot<mode>3<mask_name>,
1902         <mask_codefor><code><mode>3<mask_name>): Likewise.
1903         * config/i386/subst.md (mask_mode512bit_condition,
1904         sd_mask_mode512bit_condition): Likewise.
1905
1906 2014-01-02  Xinliang David Li  <davidxl@google.com>
1907
1908         PR tree-optimization/59303
1909         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
1910         (dump_predicates): Better output format.
1911         (pred_equal_p): New function.
1912         (is_neq_relop_p): Ditto.
1913         (is_neq_zero_form_p): Ditto.
1914         (pred_expr_equal_p): Ditto.
1915         (pred_neg_p): Ditto.
1916         (simplify_pred): Ditto.
1917         (simplify_preds_2): Ditto.
1918         (simplify_preds_3): Ditto.
1919         (simplify_preds_4): Ditto.
1920         (simplify_preds): Ditto.
1921         (push_pred): Ditto.
1922         (push_to_worklist): Ditto.
1923         (get_pred_info_from_cmp): Ditto.
1924         (is_degenerated_phi): Ditto.
1925         (normalize_one_pred_1): Ditto.
1926         (normalize_one_pred): Ditto.
1927         (normalize_one_pred_chain): Ditto.
1928         (normalize_preds): Ditto.
1929         (normalize_cond_1): Remove function.
1930         (normalize_cond): Ditto.
1931         (is_gcond_subset_of): Ditto.
1932         (is_subset_of_any): Ditto.
1933         (is_or_set_subset_of): Ditto.
1934         (is_and_set_subset_of): Ditto.
1935         (is_norm_cond_subset_of): Ditto.
1936         (pred_chain_length_cmp): Ditto.
1937         (convert_control_dep_chain_into_preds): Type change.
1938         (find_predicates): Ditto.
1939         (find_def_preds): Ditto.
1940         (destroy_predicates_vecs): Ditto.
1941         (find_matching_predicates_in_rest_chains): Ditto.
1942         (use_pred_not_overlap_with_undef_path_pred): Ditto.
1943         (is_pred_expr_subset): Ditto.
1944         (is_pred_chain_subset_of): Ditto.
1945         (is_included_in): Ditto.
1946         (is_superset_of): Ditto.
1947
1948 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
1949
1950         Update copyright years.
1951
1952 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
1953
1954         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
1955         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
1956         config/arc/arc.md, config/arc/arc.opt,
1957         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
1958         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
1959         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
1960         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
1961         config/linux-protos.h, config/linux.c, config/winnt-c.c,
1962         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
1963         vtable-verify.c, vtable-verify.h: Use the standard form for the
1964         copyright notice.
1965
1966 2014-01-02  Tobias Burnus  <burnus@net-b.de>
1967
1968         * gcc.c (process_command): Update copyright notice dates.
1969         * gcov-dump.c: Ditto.
1970         * gcov.c: Ditto.
1971         * doc/cpp.texi: Bump @copying's copyright year.
1972         * doc/cppinternals.texi: Ditto.
1973         * doc/gcc.texi: Ditto.
1974         * doc/gccint.texi: Ditto.
1975         * doc/gcov.texi: Ditto.
1976         * doc/install.texi: Ditto.
1977         * doc/invoke.texi: Ditto.
1978
1979 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1980
1981         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
1982
1983 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
1984
1985         * config/i386/sse.md (*mov<mode>_internal): Guard
1986         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
1987
1988         PR rtl-optimization/59647
1989         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
1990         new_rtx into UNSIGNED_FLOAT rtxes.
1991 \f
1992 Copyright (C) 2014 Free Software Foundation, Inc.
1993
1994 Copying and distribution of this file, with or without modification,
1995 are permitted in any medium without royalty provided the copyright
1996 notice and this notice are preserved.