gcc/ChangeLog

   1 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
   2
   3         * lra-constraints.c (valid_address_p): Move earlier in file.
   4         Add a constraint argument to the address_info version.
   5         (satisfies_memory_constraint_p): New function.
   6         (satisfies_address_constraint_p): Likewise.
   7         (process_alt_operands, curr_insn_transform): Use them.
   8         (process_address): Pass the constraint to valid_address_p when
   9         checking address operands.
  10
  11 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
  12
  13         * config/mips/mips.c (mips_isa_rev): New variable.
  14         (mips_set_architecture): Set it.
  15         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
  16         from mips_isa_rev.
  17         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
  18         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
  19         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
  20         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
  21         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
  22         conditions in terms of mips_isa_rev.
  23         (mips_isa_rev): Declare.
  24
  25 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
  26
  27         * config/sh/sh-mem.cc: Use tabs instead of spaces.
  28         (prob_unlikely, prob_likely): Make variables const.
  29
  30 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
  31
  32         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
  33
  34 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
  35
  36         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
  37
  38 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
  39
  40         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
  41         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
  42         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
  43         functions.
  44         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
  45         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
  46         sh_pass_in_reg_p.
  47         Replace usage of ROUND_REG with sh_round_reg.
  48         Use CEIL instead of ROUND_ADVANCE.
  49
  50 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
  51
  52         PR target/61026
  53         * config/sh/sh.c: Include stdlib headers before everything else.
  54
  55 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
  56
  57         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
  58         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
  59         (gimplify_adjust_omp_clauses): Simd region is never
  60         directly nested in combined parallel.  Instead, for linear
  61         with copyin/copyout, if in combined for simd loop, make decl
  62         firstprivate/lastprivate on OMP_FOR.
  63         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
  64         expand_omp_for_static_chunk): When setting endvar, also set
  65         fd->loop.v to the same value.
  66
  67 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
  68
  69         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
  70
  71 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
  72
  73         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
  74         expression.
  75
  76 2014-05-02  Marek Polacek  <polacek@redhat.com>
  77
  78         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
  79
  80 2014-02-26  Kito Cheng  <kito@0xlab.org>
  81
  82         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
  83         to a C expression marco.
  84         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
  85         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
  86         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
  87         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
  88         HONOR_REG_ALLOC_ORDER.
  89         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
  90
  91 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
  92
  93         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
  94
  95 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
  96
  97         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
  98
  99 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
 100
 101         * tree-if-conv.c (is_cond_scalar_reduction): New function.
 102         (convert_scalar_cond_reduction): Likewise.
 103         (predicate_scalar_phi): Add recognition and transformation
 104         of simple conditioanl reduction to be vectorizable.
 105
 106 2014-05-01  Marek Polacek  <polacek@redhat.com>
 107
 108         PR c/43245
 109         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
 110
 111 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
 112
 113         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
 114         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
 115         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
 116         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
 117         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
 118         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
 119         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
 120         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
 121
 122 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
 123
 124         * config/arc/arc.opt (mlra): Move comment above option name
 125         to avoid mis-parsing as language options.
 126
 127 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 128
 129         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
 130         * config/sol2.h: ... here.
 131         * config/sol2-10.h: Remove.
 132
 133         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
 134         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
 135         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
 136         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
 137         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
 138         * config/sol2.h: ... here.
 139         (SECTION_NAME_FORMAT): Don't redefine.
 140         (STARTFILE_ARCH32_SPEC): Rename to ...
 141         (STARTFILE_ARCH_SPEC): ... this.
 142         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
 143         * config/sparc/sol2.h: ... here.
 144         (SECTION_NAME_FORMAT): Don't undef.
 145         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
 146         (SUBTARGET_EXTRA_SPECS): Remove.
 147         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
 148
 149         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
 150         (MD_STARTFILE_PREFIX): Remove.
 151         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
 152         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
 153         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
 154         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
 155         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
 156         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
 157         * config/i386/sol2.h: ... here.
 158         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
 159         * config/i386/sol2-bi.h: Remove.
 160         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
 161         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
 162
 163         * config/i386/t-sol2-64: Rename to ...
 164         * config/i386/t-sol2: ... this.
 165         * config/sparc/t-sol2-64: Rename to ...
 166         * config/sparc/t-sol2: ... this.
 167
 168         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
 169         sol2_tm_file_head, sol2_tm_file_tail.
 170         Include ${cpu_type}/sol2.h before sol2.h.
 171         Remove sol2-10.h.
 172         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
 173         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
 174         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
 175         Reflect i386/t-sol2-64 renaming.
 176         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
 177         Reflect sparc/t-sol2-64 renaming.
 178
 179 2014-04-30  Richard Biener  <rguenther@suse.de>
 180
 181         * passes.c (execute_function_todo): Move TODO_verify_stmts
 182         and TODO_verify_ssa under the TODO_verify_il umbrella.
 183         * tree-ssa.h (verify_ssa): Adjust prototype.
 184         * tree-ssa.c (verify_ssa): Add parameter to tell whether
 185         we should verify SSA operands.
 186         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
 187         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
 188         whether we should verify whether not throwing stmts have EH info.
 189         * graphite-scop-detection.c (create_sese_edges): Adjust.
 190         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
 191         * tree-eh.c (lower_try_finally_switch): Do not add the
 192         default case label twice.
 193
 194 2014-04-30  Marek Polacek  <polacek@redhat.com>
 195
 196         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
 197         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
 198         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
 199         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
 200
 201 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
 202
 203         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
 204         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
 205         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
 206         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
 207         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
 208         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
 209         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
 210         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
 211
 212 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
 213
 214         * tree-cfg.c (dump_function_to_file): Dump the return type of
 215         functions, in a line to itself before the function body, mimicking
 216         the layout of a C function.
 217
 218 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
 219
 220         PR tree-optimization/60971
 221         * tree-tailcall.c (process_assignment): Reject conversions which
 222         reduce precision.
 223
 224 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
 225
 226         * calls.c (initialize_argument_information): Always treat
 227         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
 228         (expand_call): Likewise.
 229         (emit_library_call_calue_1): Likewise.
 230         * expr.c (PUSH_ARGS_REVERSED): Do not define.
 231         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
 232         code accordingly.
 233
 234 2014-04-29  Nick Clifton  <nickc@redhat.com>
 235
 236         * config/msp430/msp430.md (umulsidi): Fix typo.
 237         (mulhisi3): Enable even inside interrupt handlers.
 238         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
 239         bigger return address pushed in large mode.
 240
 241 2014-04-29  Nick Clifton  <nickc@redhat.com>
 242
 243         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
 244         (arc_init_reg_tables): Use a machine_mode enum to iterate over
 245         available modes.
 246         * config/m32r/m32r.c (init_reg_tables): Likewise.
 247         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
 248         enum to hold the modes.
 249
 250 2014-04-29  Richard Biener  <rguenther@suse.de>
 251
 252         * dominance.c (free_dominance_info): Add overload with
 253         function parameter.
 254         (dom_info_state): Likewise.
 255         (dom_info_available_p): Likewise.
 256         * basic-block.h (free_dominance_info, dom_info_state,
 257         dom_info_available_p): Declare overloads.
 258         * passes.c (execute_function_todo): Verify that verifiers
 259         don't change dominator info state.  Drop dominator info
 260         for IPA pass invocations.
 261         * cgraph.c (release_function_body): Restore asserts that
 262         dominator information is released.
 263
 264 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
 265
 266         * doc/invoke.texi: Fix typo.
 267         * tree-vrp.c: Fix typos.
 268         * gimple.c (infer_nonnull_range): Reorder operands of an &&
 269         condition.
 270
 271 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 272
 273         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
 274
 275 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
 276
 277         * config/aarch64/aarch64-builtins.c
 278         (aarch64_types_storestruct_lane_qualifiers): New.
 279         (TYPES_STORESTRUCT_LANE): Likewise.
 280         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
 281         (st3_lane): Likewise.
 282         (st4_lane): Likewise.
 283         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
 284         (vec_store_lanesci_lane<mode>): Likewise.
 285         (vec_store_lanesxi_lane<mode>): Likewise.
 286         (aarch64_st2_lane<VQ:mode>): Likewise.
 287         (aarch64_st3_lane<VQ:mode>): Likewise.
 288         (aarch64_st4_lane<VQ:mode>): Likewise.
 289         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
 290         * config/aarch64/arm_neon.h
 291         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
 292         use new macro arguments.
 293         (__ST3_LANE_FUNC): Likewise.
 294         (__ST4_LANE_FUNC): Likewise.
 295         * config/aarch64/iterators.md (V_TWO_ELEM): New.
 296         (V_THREE_ELEM): Likewise.
 297         (V_FOUR_ELEM): Likewise.
 298
 299 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
 300
 301         * doc/gimple.texi: Replace the description of the now-defunct
 302         union gimple_statement_d with a diagram showing the
 303         gimple_statement_base class hierarchy and its relationships to
 304         the GSS_ and GIMPLE_ enums.
 305
 306 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
 307
 308         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
 309         * config/aarch64/aarch64.c
 310         (aarch64_cannot_change_mode_class): Weaken conditions.
 311         (aarch64_modes_tieable_p): New.
 312         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
 313
 314 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
 315
 316         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
 317         (loadsync_<mode>): Change mode.
 318         (load_quadpti, store_quadpti): New.
 319         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
 320         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
 321
 322 2014-04-28  Martin Jambor  <mjambor@suse.cz>
 323
 324         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
 325         same alias type as the original statement.
 326         (subreplacement_assignment_data): New type.
 327         (handle_unscalarized_data_in_subtree): New type of parameter,
 328         generate new memory accesses with same alias type as the original
 329         statement.
 330         (load_assign_lhs_subreplacements): Likewise.
 331         (sra_modify_constructor_assign): Generate new memory accesses with
 332         same alias type as the original statement.
 333
 334 2014-04-28  Richard Biener  <rguenther@suse.de>
 335
 336         * tree-pass.h (TODO_verify_il): Define.
 337         (TODO_verify_all): Complete properly.
 338         * passes.c (execute_function_todo): Move existing loop-closed
 339         SSA verification under TODO_verify_il.
 340         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
 341         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
 342         Fix tree sharing issue.
 343
 344 2014-04-28  Richard Biener  <rguenther@suse.de>
 345
 346         PR middle-end/60092
 347         * builtins.def (DEF_C11_BUILTIN): Add.
 348         (BUILT_IN_ALIGNED_ALLOC): Likewise.
 349         * coretypes.h (enum function_class): Add function_c11_misc.
 350         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
 351         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
 352         (call_may_clobber_ref_p_1): Likewise.
 353         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
 354         (mark_all_reaching_defs_necessary_1): Likewise.
 355         (propagate_necessity): Likewise.
 356         (eliminate_unnecessary_stmts): Likewise.
 357         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
 358
 359 2014-04-28  Richard Biener  <rguenther@suse.de>
 360
 361         * tree-vrp.c (vrp_var_may_overflow): Remove.
 362         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
 363         with overflow immediately bump to one before that value and
 364         let iteration figure out overflow status.
 365
 366 2014-04-28  Richard Biener  <rguenther@suse.de>
 367
 368         * configure.ac: Do valgrind header checks unconditionally.
 369         Add --enable-valgrind-annotations.
 370         * system.h: Guard valgrind header inclusion with
 371         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
 372         * alloc-pool.c (pool_alloc, pool_free): Use
 373         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
 374         to guard possibly dead code.
 375         * config.in: Regenerated.
 376         * configure: Likewise.
 377
 378 2014-04-28  Jeff Law  <law@redhat.com>
 379
 380         PR tree-optimization/60902
 381         * tree-ssa-threadedge.c
 382         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
 383         over real defs when invalidating outputs from statements that do not
 384         produce useful outputs for threading.
 385
 386 2014-04-28  Richard Biener  <rguenther@suse.de>
 387
 388         PR tree-optimization/60979
 389         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
 390         SCOPs that end in a block with a successor with abnormal
 391         predecessors.
 392
 393 2014-04-28  Richard Biener  <rguenther@suse.de>
 394
 395         * tree-pass.h (execute_pass_list): Adjust prototype.
 396         * passes.c (pass_manager::execute_early_local_passes):
 397         Adjust.
 398         (do_per_function): Change callback signature, push all actual
 399         work to the callbals.
 400         (do_per_function_toporder): Likewise.
 401         (execute_function_dump): Adjust.
 402         (execute_function_todo): Likewise.
 403         (clear_last_verified): Likewise.
 404         (verify_curr_properties): Likewise.
 405         (update_properties_after_pass): Likewise.
 406         (execute_pass_list_1): Split out from ...
 407         (execute_pass_list): ... here.  Adjust.
 408         (execute_ipa_pass_list): Likewise.
 409         * cgraphunit.c (cgraph_add_new_function): Adjust.
 410         (analyze_function): Likewise.
 411         (expand_function): Likewise.
 412         * cgraph.c (release_function_body): Free dominance info
 413         here instead of asserting it was magically freed elsewhere.
 414
 415 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
 416
 417         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
 418         * configure: Regenerate.
 419         * config/sparc/sparc.opt (muser-mode): New option.
 420         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
 421         for LEON3.
 422         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
 423         * doc/invoke.texi (SPARC options): Document -muser-mode.
 424
 425 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
 426
 427         * cselib.c (find_slot_memmode): Delete.
 428         (cselib_hasher): Change compare_type to a struct.
 429         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
 430         constants.
 431         (preserve_constants_and_equivs): Adjust for new compare_type.
 432         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
 433         (wrap_constant): Delete.
 434         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
 435
 436 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
 437
 438         * doc/install.texi (Building with profile feedback): Remove
 439         outdated sentence.
 440
 441 2014-04-26  Tom de Vries  <tom@codesourcery.com>
 442
 443         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
 444         array accesses.
 445
 446 2014-04-25  Cary Coutant  <ccoutant@google.com>
 447
 448         PR debug/60929
 449         * dwarf2out.c (should_move_die_to_comdat): A type definition
 450         can contain a subprogram definition, but don't move it to a
 451         comdat unit.
 452         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
 453         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
 454         from original DIE.
 455         (clone_tree_hash): Rename to...
 456         (clone_tree_partial): ...this; change callers.  Copy
 457         DW_TAG_subprogram DIEs as declarations.
 458         (copy_decls_walk): Don't copy children of a declaration into a
 459         type unit.
 460
 461 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
 462
 463         PR target/60969
 464         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
 465         alternative 12.
 466
 467 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
 468
 469         * config/arm/predicates.md (call_insn_operand): Add long_call check.
 470         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
 471         reg for long_call.
 472         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
 473         restriction.
 474
 475 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 476
 477         * config/arm/arm.c (arm_cortex_a8_tune): Initialise
 478         T16-related fields.
 479
 480 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 481
 482         PR tree-optimization/60930
 483         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
 484         creating a multiply candidate by folding two constant
 485         multiplicands when the result overflows.
 486
 487 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
 488
 489         PR tree-optimization/60960
 490         * tree-vect-generic.c (expand_vector_operation): Only call
 491         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
 492
 493 2014-04-25  Tom de Vries  <tom@codesourcery.com>
 494
 495         * expr.c (clobber_reg_mode): New function.
 496         * expr.h (clobber_reg): New function.
 497
 498 2014-04-25  Tom de Vries  <tom@codesourcery.com>
 499
 500         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
 501         clobbers.
 502
 503 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
 504             Tom de Vries  <tom@codesourcery.com>
 505
 506         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
 507         handle.
 508         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
 509         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
 510         new argument to find_all_hard_reg_sets call.
 511
 512 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 513
 514         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
 515         Use HOST_WIDE_INT_C for mask literal.
 516         (aarch_rev16_shleft_mask_imm_p): Likewise.
 517
 518 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
 519
 520         PR target/60941
 521         * config/sparc/sparc.md (ashlsi3_extend): Delete.
 522
 523 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
 524
 525         PR preprocessor/56540
 526         * config/i386/i386-c.c (ix86_target_macros): Define
 527         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
 528
 529 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 530
 531         * configure.ac (tga_func): Remove.
 532         (LIB_TLS_SPEC): Remove.
 533         * configure: Regenerate.
 534         * config.in: Regenerate.
 535         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
 536
 537 2014-04-25  Richard Biener  <rguenther@suse.de>
 538
 539         PR ipa/60912
 540         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
 541         call stmt use/clobber sets during stmt walk instead of
 542         walking the possibly incomplete set of caller edges.
 543
 544 2014-04-25  Richard Biener  <rguenther@suse.de>
 545
 546         PR ipa/60911
 547         * passes.c (apply_ipa_transforms): Inline into only caller ...
 548         (execute_one_pass): ... here.  Properly bring in function
 549         bodies for nodes we want to apply IPA transforms to.
 550
 551 2014-04-24  Cong Hou  <congh@google.com>
 552
 553         PR tree-optimization/60896
 554         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
 555         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
 556         (vect_mark_pattern_stmts): Set the def type of all statements in
 557         PATTERN_DEF_SEQ as vect_internal_def.
 558
 559 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
 560
 561         * doc/extend.texi (PowerPC Built-in Functions): Document new
 562         powerpc extended divide, bcd, pack/unpack 128-bit, builtin
 563         functions.
 564         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
 565
 566         * config/rs6000/predicates.md (const_0_to_3_operand): New
 567         predicate to match 0..3 integer constants.
 568
 569         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
 570         to support adding miscellaneous builtin functions.
 571         (BU_DFP_MISC_2): Likewise.
 572         (BU_P7_MISC_1): Likewise.
 573         (BU_P7_MISC_2): Likewise.
 574         (BU_P8V_MISC_3): Likewise.
 575         (BU_MISC_1): Likewise.
 576         (BU_MISC_2): Likewise.
 577         (DIVWE): Add extended divide builtin functions.
 578         (DIVWEO): Likewise.
 579         (DIVWEU): Likewise.
 580         (DIVWEUO): Likewise.
 581         (DIVDE): Likewise.
 582         (DIVDEO): Likewise.
 583         (DIVDEU): Likewise.
 584         (DIVDEUO): Likewise.
 585         (DXEX): Add decimal floating-point builtin functions.
 586         (DXEXQ): Likewise.
 587         (DDEDPD): Likewise.
 588         (DDEDPDQ): Likewise.
 589         (DENBCD): Likewise.
 590         (DENBCDQ): Likewise.
 591         (DIEX): Likewise.
 592         (DIEXQ): Likewise.
 593         (DSCLI): Likewise.
 594         (DSCLIQ): Likewise.
 595         (DSCRI): Likewise.
 596         (DSCRIQ): Likewise.
 597         (CDTBCD): Add new BCD builtin functions.
 598         (CBCDTD): Likewise.
 599         (ADDG6S): Likewise.
 600         (BCDADD): Likewise.
 601         (BCDADD_LT): Likewise.
 602         (BCDADD_EQ): Likewise.
 603         (BCDADD_GT): Likewise.
 604         (BCDADD_OV): Likewise.
 605         (BCDSUB): Likewise.
 606         (BCDSUB_LT): Likewise.
 607         (BCDSUB_EQ): Likewise.
 608         (BCDSUB_GT): Likewise.
 609         (BCDSUB_OV): Likewise.
 610         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
 611         (UNPACK_TD): Likewise.
 612         (PACK_TF): Likewise.
 613         (UNPACK_TF): Likewise.
 614         (UNPACK_TF_0): Likewise.
 615         (UNPACK_TF_1): Likewise.
 616         (PACK_V1TI): Likewise.
 617         (UNPACK_V1TI): Likewise.
 618
 619         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
 620         support for decimal floating point builtin functions.
 621         (rs6000_expand_ternop_builtin): Add checks for the new builtin
 622         functions that take constant arguments.
 623         (rs6000_invalid_builtin): Add decimal floating point builtin
 624         support.
 625         (rs6000_init_builtins): Setup long double, _Decimal64, and
 626         _Decimal128 types for new builtin functions.
 627         (builtin_function_type): Set the unsigned flags appropriately for
 628         the new builtin functions.
 629         (rs6000_opt_masks): Add support for decimal floating point builtin
 630         functions.
 631
 632         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
 633         floating point builtin functions.
 634         (RS6000_BTM_COMMON): Likewise.
 635         (RS6000_BTI_long_double): Likewise.
 636         (RS6000_BTI_dfloat64): Likewise.
 637         (RS6000_BTI_dfloat128): Likewise.
 638         (long_double_type_internal_node): Likewise.
 639         (dfloat64_type_internal_node): Likewise.
 640         (dfloat128_type_internal_node): Likewise.
 641
 642         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
 643         2.07 bcd arithmetic instructions.
 644         (UNSPEC_BCDSUB): Likewise.
 645         (UNSPEC_BCD_OVERFLOW): Likewise.
 646         (UNSPEC_BCD_ADD_SUB): Likewise.
 647         (bcd_add_sub): Likewise.
 648         (BCD_TEST): Likewise.
 649         (bcd<bcd_add_sub>): Likewise.
 650         (bcd<bcd_add_sub>_test): Likewise.
 651         (bcd<bcd_add_sub>_test2): Likewise.
 652         (bcd<bcd_add_sub>_<code>): Likewise.
 653         (peephole2 for combined bcd ops): Likewise.
 654
 655         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
 656         decimal floating point builtin functions.
 657         (UNSPEC_DENBCD): Likewise.
 658         (UNSPEC_DXEX): Likewise.
 659         (UNSPEC_DIEX): Likewise.
 660         (UNSPEC_DSCLI): Likewise.
 661         (UNSPEC_DSCRI): Likewise.
 662         (D64_D128): Likewise.
 663         (dfp_suffix): Likewise.
 664         (dfp_ddedpd_<mode>): Likewise.
 665         (dfp_denbcd_<mode>): Likewise.
 666         (dfp_dxex_<mode>): Likewise.
 667         (dfp_diex_<mode>): Likewise.
 668         (dfp_dscli_<mode>): Likewise.
 669         (dfp_dscri_<mode>): Likewise.
 670
 671         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
 672         builtin functions.
 673         (UNSPEC_CDTBCD): Likewise.
 674         (UNSPEC_CBCDTD): Likewise.
 675         (UNSPEC_DIVE): Add support for new extended divide builtin
 676         functions.
 677         (UNSPEC_DIVEO): Likewise.
 678         (UNSPEC_DIVEU): Likewise.
 679         (UNSPEC_DIVEUO): Likewise.
 680         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
 681         pack/unpack 128-bit types.
 682         (UNSPEC_PACK_128BIT): Likewise.
 683         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
 684         (udiv<mode>3): Use idiv_ldiv mode attribute.
 685         (div<mode>3): Likewise.
 686         (addg6s): Add new BCD builtin functions.
 687         (cdtbcd): Likewise.
 688         (cbcdtd): Likewise.
 689         (UNSPEC_DIV_EXTEND): Add support for new extended divide
 690         instructions.
 691         (div_extend): Likewise.
 692         (div<div_extend>_<mode>"): Likewise.
 693         (FP128_64): Add support for new builtin functions to pack/unpack
 694         128-bit types.
 695         (unpack<mode>): Likewise.
 696         (unpacktf_0): Likewise.
 697         (unpacktf_1): Likewise.
 698         (unpack<mode>_dm): Likewise.
 699         (unpack<mode>_nodm): Likewise.
 700         (pack<mode>): Likewise.
 701         (unpackv1ti): Likewise.
 702         (packv1ti): Likewise.
 703
 704 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
 705
 706         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
 707         is disabled.
 708
 709 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
 710
 711         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
 712         * gimplify.c (omp_is_private): Change last argument's type to int.
 713         Only diagnose lastprivate if the simd argument is 1, only diagnose
 714         linear if the simd argument is 2.
 715         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
 716         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
 717         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
 718         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
 719         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
 720         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 721         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
 722         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 723         * tree-nested.c (convert_nonlocal_omp_clauses,
 724         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
 725
 726 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
 727
 728         PR target/60822
 729         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
 730         operand 1.
 731
 732 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
 733
 734         * flag-types.h (enum ivar_visibility): Add.
 735
 736 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
 737
 738         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
 739         function * argument.
 740
 741 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
 742
 743         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
 744
 745 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 746             Tom de Vries  <tom@codesourcery.com>
 747
 748         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
 749         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
 750         reg-note.
 751         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
 752         * emit-rtl.c (try_split): Same.
 753
 754 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 755             Tom de Vries  <tom@codesourcery.com>
 756
 757         * common.opt (fuse-caller-save): New option.
 758
 759 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
 760
 761         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
 762         for big-endian.
 763
 764 2014-04-24  Richard Biener  <rguenther@suse.de>
 765
 766         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
 767         during TER and instead use the sepops interface for expanding
 768         non-GIMPLE_SINGLE_RHS.
 769
 770 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 771
 772         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
 773         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
 774
 775 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 776
 777         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
 778         assembler 64-bit option.
 779         * configure: Regenerate.
 780
 781 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 782
 783         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
 784         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
 785         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
 786         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
 787         (TARGET_CRYPTO): Take TARGET_SIMD into account.
 788
 789 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 790
 791         * config/aarch64/aarch64-builtins.c
 792         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
 793         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
 794         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
 795         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
 796         builtins.
 797         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
 798         (Vrevsuff): New mode attribute.
 799
 800 2014-04-24  Terry Guo  <terry.guo@arm.com>
 801
 802         * config/arm/arm.h (machine_function): Define variable
 803         after_arm_reorg here.
 804         * config/arm/arm.c (after_arm_reorg): Remove the definition.
 805         (arm_split_constant): Update the way to access variable
 806         after_arm_reorg.
 807         (arm_reorg): Ditto.
 808         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
 809
 810 2014-04-23  Tom de Vries  <tom@codesourcery.com>
 811
 812         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
 813
 814 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
 815
 816         * is-a.h: Update comments to reflect the following changes to the
 817         "pointerness" of the API, making the template parameter match the
 818         return type, allowing use of is-a.h with typedefs of pointers.
 819         (is_a_helper::cast): Return a T rather then a pointer to a T, so
 820         that the return type matches the parameter to the is_a_helper.
 821         (as_a): Likewise.
 822         (dyn_cast): Likewise.
 823
 824         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
 825         pointer from the is-a.h API.
 826
 827         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
 828         (is_a_helper <cgraph_node *>::test): ...this, matching change to
 829         is-a.h API.
 830         (is_a_helper <varpool_node>::test): Likewise, convert to...
 831         (is_a_helper <varpool_node *>::test): ...this.
 832
 833         (varpool_first_variable): Update for removal of implicit pointer
 834         from the is-a.h API.
 835         (varpool_next_variable): Likewise.
 836         (varpool_first_static_initializer): Likewise.
 837         (varpool_next_static_initializer): Likewise.
 838         (varpool_first_defined_variable): Likewise.
 839         (varpool_next_defined_variable): Likewise.
 840         (cgraph_first_defined_function): Likewise.
 841         (cgraph_next_defined_function): Likewise.
 842         (cgraph_first_function): Likewise.
 843         (cgraph_next_function): Likewise.
 844         (cgraph_first_function_with_gimple_body): Likewise.
 845         (cgraph_next_function_with_gimple_body): Likewise.
 846         (cgraph_alias_target): Likewise.
 847         (varpool_alias_target): Likewise.
 848         (cgraph_function_or_thunk_node): Likewise.
 849         (varpool_variable_node): Likewise.
 850         (symtab_real_symbol_p): Likewise.
 851         * cgraphunit.c (referred_to_p): Likewise.
 852         (analyze_functions): Likewise.
 853         (handle_alias_pairs): Likewise.
 854         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
 855         * gimple-ssa.h (gimple_vuse_op): Likewise.
 856         (gimple_vdef_op): Likewise.
 857         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
 858         * gimple.c (gimple_build_asm_1): Likewise.
 859         (gimple_build_try): Likewise.
 860         (gimple_build_resx): Likewise.
 861         (gimple_build_eh_dispatch): Likewise.
 862         (gimple_build_omp_for): Likewise.
 863         (gimple_omp_for_set_clauses): Likewise.
 864
 865         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
 866         (is_a_helper <gimple_statement_asm *>::test): ...this.
 867         (is_a_helper <gimple_statement_bind>::test): Convert to...
 868         (is_a_helper <gimple_statement_bind *>::test): ...this.
 869         (is_a_helper <gimple_statement_call>::test): Convert to...
 870         (is_a_helper <gimple_statement_call *>::test): ...this.
 871         (is_a_helper <gimple_statement_catch>::test): Convert to...
 872         (is_a_helper <gimple_statement_catch *>::test): ...this.
 873         (is_a_helper <gimple_statement_resx>::test): Convert to...
 874         (is_a_helper <gimple_statement_resx *>::test): ...this.
 875         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
 876         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
 877         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
 878         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
 879         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
 880         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
 881         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
 882         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
 883         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
 884         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
 885         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
 886         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
 887         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
 888         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
 889         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
 890         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
 891         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
 892         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
 893         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
 894         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
 895         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
 896         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
 897         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
 898         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
 899         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
 900         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
 901         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
 902         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
 903         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
 904         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
 905         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
 906         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
 907         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
 908         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
 909         (is_a_helper <gimple_statement_phi>::test): Convert to...
 910         (is_a_helper <gimple_statement_phi *>::test): ...this.
 911         (is_a_helper <gimple_statement_transaction>::test): Convert to...
 912         (is_a_helper <gimple_statement_transaction *>::test): ...this.
 913         (is_a_helper <gimple_statement_try>::test): Convert to...
 914         (is_a_helper <gimple_statement_try *>::test): ...this.
 915         (is_a_helper <gimple_statement_wce>::test): Convert to...
 916         (is_a_helper <gimple_statement_wce *>::test): ...this.
 917         (is_a_helper <const gimple_statement_asm>::test): Convert to...
 918         (is_a_helper <const gimple_statement_asm *>::test): ...this.
 919         (is_a_helper <const gimple_statement_bind>::test): Convert to...
 920         (is_a_helper <const gimple_statement_bind *>::test): ...this.
 921         (is_a_helper <const gimple_statement_call>::test): Convert to...
 922         (is_a_helper <const gimple_statement_call *>::test): ...this.
 923         (is_a_helper <const gimple_statement_catch>::test): Convert to...
 924         (is_a_helper <const gimple_statement_catch *>::test): ...this.
 925         (is_a_helper <const gimple_statement_resx>::test): Convert to...
 926         (is_a_helper <const gimple_statement_resx *>::test): ...this.
 927         (is_a_helper <const gimple_statement_eh_dispatch>::test):
 928         Convert to...
 929         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
 930         ...this.
 931         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
 932         to...
 933         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
 934         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
 935         Convert to...
 936         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
 937         ...this.
 938         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
 939         Convert to...
 940         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
 941         ...this.
 942         (is_a_helper <const gimple_statement_omp_return>::test): Convert
 943         to...
 944         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
 945         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
 946         to...
 947         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
 948         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
 949         to...
 950         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
 951         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
 952         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
 953         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
 954         to...
 955         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
 956         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
 957         to...
 958         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
 959         (is_a_helper <const gimple_statement_omp_target>::test): Convert
 960         to...
 961         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
 962         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
 963         to...
 964         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
 965         (is_a_helper <const gimple_statement_omp_single>::test): Convert
 966         to...
 967         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
 968         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
 969         to...
 970         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
 971         (is_a_helper <const gimple_statement_omp_task>::test): Convert
 972         to...
 973         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
 974         (is_a_helper <const gimple_statement_phi>::test): Convert to...
 975         (is_a_helper <const gimple_statement_phi *>::test): ...this.
 976         (is_a_helper <const gimple_statement_transaction>::test): Convert
 977         to...
 978         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
 979         (is_a_helper <const gimple_statement_with_ops>::test): Convert
 980         to...
 981         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
 982         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
 983         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
 984         (is_a_helper <const gimple_statement_with_memory_ops>::test):
 985         Convert to...
 986         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
 987         ...this.
 988         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
 989         to...
 990         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
 991
 992         (gimple_use_ops): Update for removal of implicit pointer from the
 993         is-a.h API.
 994         (gimple_set_use_ops): Likewise.
 995         (gimple_vuse): Likewise.
 996         (gimple_vdef): Likewise.
 997         (gimple_vuse_ptr): Likewise.
 998         (gimple_vdef_ptr): Likewise.
 999         (gimple_set_vuse): Likewise.
1000         (gimple_set_vdef): Likewise.
1001         (gimple_omp_return_set_lhs): Likewise.
1002         (gimple_omp_return_lhs): Likewise.
1003         (gimple_omp_return_lhs_ptr): Likewise.
1004         (gimple_call_fntype): Likewise.
1005         (gimple_call_set_fntype): Likewise.
1006         (gimple_call_set_internal_fn): Likewise.
1007         (gimple_call_use_set): Likewise.
1008         (gimple_call_clobber_set): Likewise.
1009         (gimple_bind_vars): Likewise.
1010         (gimple_bind_set_vars): Likewise.
1011         (gimple_bind_body_ptr): Likewise.
1012         (gimple_bind_set_body): Likewise.
1013         (gimple_bind_add_stmt): Likewise.
1014         (gimple_bind_block): Likewise.
1015         (gimple_bind_set_block): Likewise.
1016         (gimple_asm_ninputs): Likewise.
1017         (gimple_asm_noutputs): Likewise.
1018         (gimple_asm_nclobbers): Likewise.
1019         (gimple_asm_nlabels): Likewise.
1020         (gimple_asm_input_op): Likewise.
1021         (gimple_asm_input_op_ptr): Likewise.
1022         (gimple_asm_output_op): Likewise.
1023         (gimple_asm_output_op_ptr): Likewise.
1024         (gimple_asm_set_output_op): Likewise.
1025         (gimple_asm_clobber_op): Likewise.
1026         (gimple_asm_set_clobber_op): Likewise.
1027         (gimple_asm_label_op): Likewise.
1028         (gimple_asm_set_label_op): Likewise.
1029         (gimple_asm_string): Likewise.
1030         (gimple_catch_types): Likewise.
1031         (gimple_catch_types_ptr): Likewise.
1032         (gimple_catch_handler_ptr): Likewise.
1033         (gimple_catch_set_types): Likewise.
1034         (gimple_catch_set_handler): Likewise.
1035         (gimple_eh_filter_types): Likewise.
1036         (gimple_eh_filter_types_ptr): Likewise.
1037         (gimple_eh_filter_failure_ptr): Likewise.
1038         (gimple_eh_filter_set_types): Likewise.
1039         (gimple_eh_filter_set_failure): Likewise.
1040         (gimple_eh_must_not_throw_fndecl): Likewise.
1041         (gimple_eh_must_not_throw_set_fndecl): Likewise.
1042         (gimple_eh_else_n_body_ptr): Likewise.
1043         (gimple_eh_else_e_body_ptr): Likewise.
1044         (gimple_eh_else_set_n_body): Likewise.
1045         (gimple_eh_else_set_e_body): Likewise.
1046         (gimple_try_eval_ptr): Likewise.
1047         (gimple_try_cleanup_ptr): Likewise.
1048         (gimple_try_set_eval): Likewise.
1049         (gimple_try_set_cleanup): Likewise.
1050         (gimple_wce_cleanup_ptr): Likewise.
1051         (gimple_wce_set_cleanup): Likewise.
1052         (gimple_phi_capacity): Likewise.
1053         (gimple_phi_num_args): Likewise.
1054         (gimple_phi_result): Likewise.
1055         (gimple_phi_result_ptr): Likewise.
1056         (gimple_phi_set_result): Likewise.
1057         (gimple_phi_arg): Likewise.
1058         (gimple_phi_set_arg): Likewise.
1059         (gimple_resx_region): Likewise.
1060         (gimple_resx_set_region): Likewise.
1061         (gimple_eh_dispatch_region): Likewise.
1062         (gimple_eh_dispatch_set_region): Likewise.
1063         (gimple_omp_critical_name): Likewise.
1064         (gimple_omp_critical_name_ptr): Likewise.
1065         (gimple_omp_critical_set_name): Likewise.
1066         (gimple_omp_for_clauses): Likewise.
1067         (gimple_omp_for_clauses_ptr): Likewise.
1068         (gimple_omp_for_set_clauses): Likewise.
1069         (gimple_omp_for_collapse): Likewise.
1070         (gimple_omp_for_index): Likewise.
1071         (gimple_omp_for_index_ptr): Likewise.
1072         (gimple_omp_for_set_index): Likewise.
1073         (gimple_omp_for_initial): Likewise.
1074         (gimple_omp_for_initial_ptr): Likewise.
1075         (gimple_omp_for_set_initial): Likewise.
1076         (gimple_omp_for_final): Likewise.
1077         (gimple_omp_for_final_ptr): Likewise.
1078         (gimple_omp_for_set_final): Likewise.
1079         (gimple_omp_for_incr): Likewise.
1080         (gimple_omp_for_incr_ptr): Likewise.
1081         (gimple_omp_for_set_incr): Likewise.
1082         (gimple_omp_for_pre_body_ptr): Likewise.
1083         (gimple_omp_for_set_pre_body): Likewise.
1084         (gimple_omp_parallel_clauses): Likewise.
1085         (gimple_omp_parallel_clauses_ptr): Likewise.
1086         (gimple_omp_parallel_set_clauses): Likewise.
1087         (gimple_omp_parallel_child_fn): Likewise.
1088         (gimple_omp_parallel_child_fn_ptr): Likewise.
1089         (gimple_omp_parallel_set_child_fn): Likewise.
1090         (gimple_omp_parallel_data_arg): Likewise.
1091         (gimple_omp_parallel_data_arg_ptr): Likewise.
1092         (gimple_omp_parallel_set_data_arg): Likewise.
1093         (gimple_omp_task_clauses): Likewise.
1094         (gimple_omp_task_clauses_ptr): Likewise.
1095         (gimple_omp_task_set_clauses): Likewise.
1096         (gimple_omp_task_child_fn): Likewise.
1097         (gimple_omp_task_child_fn_ptr): Likewise.
1098         (gimple_omp_task_set_child_fn): Likewise.
1099         (gimple_omp_task_data_arg): Likewise.
1100         (gimple_omp_task_data_arg_ptr): Likewise.
1101         (gimple_omp_task_set_data_arg): Likewise.
1102         (gimple_omp_taskreg_clauses): Likewise.
1103         (gimple_omp_taskreg_clauses_ptr): Likewise.
1104         (gimple_omp_taskreg_set_clauses): Likewise.
1105         (gimple_omp_taskreg_child_fn): Likewise.
1106         (gimple_omp_taskreg_child_fn_ptr): Likewise.
1107         (gimple_omp_taskreg_set_child_fn): Likewise.
1108         (gimple_omp_taskreg_data_arg): Likewise.
1109         (gimple_omp_taskreg_data_arg_ptr): Likewise.
1110         (gimple_omp_taskreg_set_data_arg): Likewise.
1111         (gimple_omp_task_copy_fn): Likewise.
1112         (gimple_omp_task_copy_fn_ptr): Likewise.
1113         (gimple_omp_task_set_copy_fn): Likewise.
1114         (gimple_omp_task_arg_size): Likewise.
1115         (gimple_omp_task_arg_size_ptr): Likewise.
1116         (gimple_omp_task_set_arg_size): Likewise.
1117         (gimple_omp_task_arg_align): Likewise.
1118         (gimple_omp_task_arg_align_ptr): Likewise.
1119         (gimple_omp_task_set_arg_align): Likewise.
1120         (gimple_omp_single_clauses): Likewise.
1121         (gimple_omp_single_clauses_ptr): Likewise.
1122         (gimple_omp_single_set_clauses): Likewise.
1123         (gimple_omp_target_clauses): Likewise.
1124         (gimple_omp_target_clauses_ptr): Likewise.
1125         (gimple_omp_target_set_clauses): Likewise.
1126         (gimple_omp_target_child_fn): Likewise.
1127         (gimple_omp_target_child_fn_ptr): Likewise.
1128         (gimple_omp_target_set_child_fn): Likewise.
1129         (gimple_omp_target_data_arg): Likewise.
1130         (gimple_omp_target_data_arg_ptr): Likewise.
1131         (gimple_omp_target_set_data_arg): Likewise.
1132         (gimple_omp_teams_clauses): Likewise.
1133         (gimple_omp_teams_clauses_ptr): Likewise.
1134         (gimple_omp_teams_set_clauses): Likewise.
1135         (gimple_omp_sections_clauses): Likewise.
1136         (gimple_omp_sections_clauses_ptr): Likewise.
1137         (gimple_omp_sections_set_clauses): Likewise.
1138         (gimple_omp_sections_control): Likewise.
1139         (gimple_omp_sections_control_ptr): Likewise.
1140         (gimple_omp_sections_set_control): Likewise.
1141         (gimple_omp_for_set_cond): Likewise.
1142         (gimple_omp_for_cond): Likewise.
1143         (gimple_omp_atomic_store_set_val): Likewise.
1144         (gimple_omp_atomic_store_val): Likewise.
1145         (gimple_omp_atomic_store_val_ptr): Likewise.
1146         (gimple_omp_atomic_load_set_lhs): Likewise.
1147         (gimple_omp_atomic_load_lhs): Likewise.
1148         (gimple_omp_atomic_load_lhs_ptr): Likewise.
1149         (gimple_omp_atomic_load_set_rhs): Likewise.
1150         (gimple_omp_atomic_load_rhs): Likewise.
1151         (gimple_omp_atomic_load_rhs_ptr): Likewise.
1152         (gimple_omp_continue_control_def): Likewise.
1153         (gimple_omp_continue_control_def_ptr): Likewise.
1154         (gimple_omp_continue_set_control_def): Likewise.
1155         (gimple_omp_continue_control_use): Likewise.
1156         (gimple_omp_continue_control_use_ptr): Likewise.
1157         (gimple_omp_continue_set_control_use): Likewise.
1158         (gimple_transaction_body_ptr): Likewise.
1159         (gimple_transaction_label): Likewise.
1160         (gimple_transaction_label_ptr): Likewise.
1161         (gimple_transaction_set_body): Likewise.
1162         (gimple_transaction_set_label): Likewise.
1163
1164         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
1165         * ipa-inline-analysis.c (inline_write_summary): Likewise.
1166         * ipa-ref.c (ipa_record_reference): Likewise.
1167         * ipa-reference.c (analyze_function): Likewise.
1168         (ipa_reference_write_optimization_summary): Likewise.
1169         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
1170         (address_taken_from_non_vtable_p): Likewise.
1171         (comdat_can_be_unshared_p_1): Likewise.
1172         * lto-cgraph.c (lto_output_ref): Likewise.
1173         (add_references): Likewise.
1174         (compute_ltrans_boundary): Likewise.
1175         (output_symtab): Likewise.
1176         (input_ref): Likewise.
1177         (input_cgraph_1): Likewise.
1178         (output_cgraph_opt_summary): Likewise.
1179         * lto-streamer-out.c (lto_output): Likewise.
1180         (output_symbol_p): Likewise.
1181         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
1182         (lsei_start_function_in_partition): Likewise.
1183         (lsei_next_variable_in_partition): Likewise.
1184         (lsei_start_variable_in_partition): Likewise.
1185         * symtab.c (insert_to_assembler_name_hash): Likewise.
1186         (unlink_from_assembler_name_hash): Likewise.
1187         (symtab_unregister_node): Likewise.
1188         (symtab_remove_node): Likewise.
1189         (dump_symtab_node): Likewise.
1190         (verify_symtab_base): Likewise.
1191         (verify_symtab_node): Likewise.
1192         (symtab_make_decl_local): Likewise.
1193         (symtab_alias_ultimate_target): Likewise.
1194         (symtab_resolve_alias): Likewise.
1195         (symtab_get_symbol_partitioning_class): Likewise.
1196         * tree-phinodes.c (allocate_phi_node): Likewise.
1197         (reserve_phi_args_for_new_edge): Likewise.
1198         (remove_phi_args): Likewise.
1199         * varpool.c (varpool_node_for_asm): Likewise.
1200         (varpool_remove_unreferenced_decls): Likewise.
1201
1202 2014-04-23  Jeff Law  <law@redhat.com>
1203
1204         PR tree-optimization/60902
1205         * tree-ssa-threadedge.c
1206         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
1207         invalidate outputs from statements that do not produce useful
1208         outputs for threading.
1209
1210 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
1211
1212         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
1213         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
1214         machine descriptions for Stack Smashing Protector.
1215
1216 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
1217
1218         * aarch64.md (<optab>_rol<mode>3): New pattern.
1219         (<optab>_rolsi3_uxtw): Likewise.
1220         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
1221
1222 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
1223
1224         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
1225         (arm_cortex_a12_tune): Likewise.
1226
1227 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1228
1229         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
1230
1231 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1232
1233         * config/arm/arm.md (arm_rev16si2): New pattern.
1234         (arm_rev16si2_alt): Likewise.
1235         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
1236
1237 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1238
1239        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
1240        (rev16<mode>2_alt): Likewise.
1241        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
1242        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
1243        (aarch_rev16_shleft_mask_imm_p): Likewise.
1244        (aarch_rev16_p_1): Likewise.
1245        (aarch_rev16_p): Likewise.
1246        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
1247        (aarch_rev16_shright_mask_imm_p): Likewise.
1248        (aarch_rev16_shleft_mask_imm_p): Likewise.
1249
1250 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1251
1252        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
1253        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
1254        rev cost.
1255        (cortex_a53_extra_costs): Likewise.
1256        (cortex_a57_extra_costs): Likewise.
1257        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
1258        (cortexa7_extra_costs): Likewise.
1259        (cortexa8_extra_costs): Likewise.
1260        (cortexa12_extra_costs): Likewise.
1261        (cortexa15_extra_costs): Likewise.
1262        (v7m_extra_costs): Likewise.
1263        (arm_new_rtx_costs): Handle BSWAP.
1264
1265 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1266
1267        * config/arm/arm.c (cortexa8_extra_costs): New table.
1268        (arm_cortex_a8_tune): New tuning struct.
1269        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
1270
1271 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1272
1273        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
1274
1275 2014-04-23  Richard Biener  <rguenther@suse.de>
1276
1277         * Makefile.in (OBJS): Remove loop-unswitch.o.
1278         * loop-unswitch.c: Delete.
1279         * tree-pass.h (make_pass_rtl_unswitch): Remove.
1280         * passes.def (pass_rtl_unswitch): Likewise.
1281         * loop-init.c (gate_rtl_unswitch): Likewise.
1282         (rtl_unswitch): Likewise.
1283         (pass_data_rtl_unswitch): Likewise.
1284         (pass_rtl_unswitch): Likewise.
1285         (make_pass_rtl_unswitch): Likewise.
1286         * rtl.h (reversed_condition): Likewise.
1287         (compare_and_jump_seq): Likewise.
1288         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
1289         and make static.
1290         * loop-unroll.c (compare_and_jump_seq): Likewise.
1291
1292 2014-04-23  Richard Biener  <rguenther@suse.de>
1293
1294         PR tree-optimization/60903
1295         * tree-ssa-loop-im.c (analyze_memory_references): Remove
1296         commented code block.
1297         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
1298         loop flags to newly created BBs and edges.
1299
1300 2014-04-23  Nick Clifton  <nickc@redhat.com>
1301
1302         * config/msp430/msp430.c (msp430_handle_option): Move function
1303         to msp430-common.c
1304         (msp430_option_override): Simplify mcu and mcpu option handling.
1305         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
1306         support for -mhwmult command line option.
1307         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
1308         -mhwmult command line option.
1309         (msp430_hwmult_enabled): Delete.
1310         (msp43o_output_labelref): Add support for -mhwmult command line
1311         option.
1312         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
1313         (umulsidi3): Likewise.
1314         * config/msp430/msp430.opt (mmcu): Add Report attribute.
1315         (mcpu, mlarge, msmall): Likewise.
1316         (mhwmult): New option.
1317         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
1318         prototype.
1319         (msp430_is_f5_mcu): Remove prototype.
1320         (msp430_use_f5_series_hwmult): Add prototype.
1321         * config/msp430/msp430-opts.h: New file.
1322         * common/config/msp430: New directory.
1323         * common/config/msp430/msp430-common.c: New file.
1324         * config.gcc (msp430): Remove target_has_targetm_common.
1325         * doc/invoke.texi: Document -mhwmult command line option.
1326
1327 2014-04-23  Nick Clifton  <nickc@redhat.com>
1328
1329         * config/i386/cygwin.h (ENDFILE_SPEC): Include
1330         default-manifest.o if it can be found in the search path.
1331         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
1332
1333 2014-04-23  Terry Guo  <terry.guo@arm.com>
1334
1335         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
1336
1337 2014-04-23  Richard Biener  <rguenther@suse.de>
1338
1339         PR middle-end/60895
1340         * tree-inline.c (declare_return_variable): Use mark_addressable.
1341
1342 2014-04-23  Richard Biener  <rguenther@suse.de>
1343
1344         PR middle-end/60891
1345         * loop-init.c (loop_optimizer_init): Make sure to apply
1346         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
1347
1348 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1349
1350         PR sanitizer/60275
1351         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
1352         New options.
1353         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
1354         if flag_sanitize_undefined_trap_on_error.
1355         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
1356         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
1357         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
1358         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1359         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
1360         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
1361         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
1362         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
1363         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
1364         * ubsan.c (ubsan_instrument_unreachable): Return
1365         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
1366         (ubsan_expand_null_ifn): Emit __builtin_trap ()
1367         if flag_sanitize_undefined_trap_on_error and
1368         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
1369         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
1370         instrument_bool_enum_load): Emit __builtin_trap () if
1371         flag_sanitize_undefined_trap_on_error and
1372         __builtin_handle_*_abort () if !flag_sanitize_recover.
1373         * doc/invoke.texi (-fsanitize-recover,
1374         -fsanitize-undefined-trap-on-error): Document.
1375
1376 2014-04-22  Christian Bruel  <christian.bruel@st.com>
1377
1378         * config/sh/sh.md (mov<mode>): Replace movQIHI.
1379          Force immediates to SImode.
1380
1381 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
1382
1383         * config/nios2/nios2.md (UNSPEC_ROUND): New.
1384         (lroundsfsi2): New.
1385         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
1386         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
1387         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
1388         (nios2_fpu_insn): Add entry for round.
1389         (N2FPU_NO_ERRNO_P): Define.
1390         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
1391         flag_errno_math.
1392         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
1393
1394 2014-04-22  Richard Henderson  <rth@redhat.com>
1395
1396         * config/aarch64/aarch64 (addti3, subti3): New expanders.
1397         (add<GPI>3_compare0): Remove leading * from name.
1398         (add<GPI>3_carryin): Likewise.
1399         (sub<GPI>3_compare0): Likewise.
1400         (sub<GPI>3_carryin): Likewise.
1401         (<su_optab>mulditi3): New expander.
1402         (multi3): New expander.
1403         (madd<GPI>): Remove leading * from name.
1404
1405 2014-04-22  Martin Jambor  <mjambor@suse.cz>
1406
1407         * cgraphclones.c (cgraph_function_versioning): Copy
1408         ipa_transforms_to_apply instead of asserting it is empty.
1409
1410 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
1411
1412         PR target/60868
1413         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
1414         on count_exp to get mode.
1415
1416 2014-04-22  Andrew Pinski  <apinski@cavium.com>
1417
1418         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1419         Handle TLS for ILP32.
1420         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
1421         (tlsie_small_<mode>): this and handle PTR.
1422         (tlsie_small_sidi): New pattern.
1423         (tlsle_small): Change to an expand to handle ILP32.
1424         (tlsle_small_<mode>): New pattern.
1425         (tlsdesc_small): Rename to ...
1426         (tlsdesc_small_<mode>): this and handle PTR.
1427
1428 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1429
1430         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
1431
1432 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1433
1434         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1435         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
1436         (aarch64_types_signed_poly_qualifiers): Likewise.
1437         (aarch64_types_unsigned_signed_qualifiers): Likewise.
1438         (aarch64_types_poly_signed_qualifiers): Likewise.
1439         (TYPES_REINTERP_SS): Type macro added.
1440         (TYPES_REINTERP_SU): Likewise.
1441         (TYPES_REINTERP_SP): Likewise.
1442         (TYPES_REINTERP_US): Likewise.
1443         (TYPES_REINTERP_PS): Likewise.
1444         (aarch64_fold_builtin): New expression folding added.
1445         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
1446         Declarations removed.
1447         (REINTERP_SS): Declarations added.
1448         (REINTERP_US): Likewise.
1449         (REINTERP_PS): Likewise.
1450         (REINTERP_SU): Likewise.
1451         (REINTERP_SP): Likewise.
1452         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
1453         (vreinterpretq_p8_f64): Likewise.
1454         (vreinterpret_p16_f64): Likewise.
1455         (vreinterpretq_p16_f64): Likewise.
1456         (vreinterpret_f32_f64): Likewise.
1457         (vreinterpretq_f32_f64): Likewise.
1458         (vreinterpret_f64_f32): Likewise.
1459         (vreinterpret_f64_p8): Likewise.
1460         (vreinterpret_f64_p16): Likewise.
1461         (vreinterpret_f64_s8): Likewise.
1462         (vreinterpret_f64_s16): Likewise.
1463         (vreinterpret_f64_s32): Likewise.
1464         (vreinterpret_f64_s64): Likewise.
1465         (vreinterpret_f64_u8): Likewise.
1466         (vreinterpret_f64_u16): Likewise.
1467         (vreinterpret_f64_u32): Likewise.
1468         (vreinterpret_f64_u64): Likewise.
1469         (vreinterpretq_f64_f32): Likewise.
1470         (vreinterpretq_f64_p8): Likewise.
1471         (vreinterpretq_f64_p16): Likewise.
1472         (vreinterpretq_f64_s8): Likewise.
1473         (vreinterpretq_f64_s16): Likewise.
1474         (vreinterpretq_f64_s32): Likewise.
1475         (vreinterpretq_f64_s64): Likewise.
1476         (vreinterpretq_f64_u8): Likewise.
1477         (vreinterpretq_f64_u16): Likewise.
1478         (vreinterpretq_f64_u32): Likewise.
1479         (vreinterpretq_f64_u64): Likewise.
1480         (vreinterpret_s64_f64): Likewise.
1481         (vreinterpretq_s64_f64): Likewise.
1482         (vreinterpret_u64_f64): Likewise.
1483         (vreinterpretq_u64_f64): Likewise.
1484         (vreinterpret_s8_f64): Likewise.
1485         (vreinterpretq_s8_f64): Likewise.
1486         (vreinterpret_s16_f64): Likewise.
1487         (vreinterpretq_s16_f64): Likewise.
1488         (vreinterpret_s32_f64): Likewise.
1489         (vreinterpretq_s32_f64): Likewise.
1490         (vreinterpret_u8_f64): Likewise.
1491         (vreinterpretq_u8_f64): Likewise.
1492         (vreinterpret_u16_f64): Likewise.
1493         (vreinterpretq_u16_f64): Likewise.
1494         (vreinterpret_u32_f64): Likewise.
1495         (vreinterpretq_u32_f64): Likewise.
1496
1497 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1498
1499         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1500         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
1501         (vreinterpret_p8_s8): Likewise.
1502         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
1503         (vreinterpret_p8_s16): Likewise.
1504         (vreinterpret_p8_s32): Likewise.
1505         (vreinterpret_p8_s64): Likewise.
1506         (vreinterpret_p8_f32): Likewise.
1507         (vreinterpret_p8_u8): Likewise.
1508         (vreinterpret_p8_u16): Likewise.
1509         (vreinterpret_p8_u32): Likewise.
1510         (vreinterpret_p8_u64): Likewise.
1511         (vreinterpret_p8_p16): Likewise.
1512         (vreinterpretq_p8_s8): Likewise.
1513         (vreinterpretq_p8_s16): Likewise.
1514         (vreinterpretq_p8_s32): Likewise.
1515         (vreinterpretq_p8_s64): Likewise.
1516         (vreinterpretq_p8_f32): Likewise.
1517         (vreinterpretq_p8_u8): Likewise.
1518         (vreinterpretq_p8_u16): Likewise.
1519         (vreinterpretq_p8_u32): Likewise.
1520         (vreinterpretq_p8_u64): Likewise.
1521         (vreinterpretq_p8_p16): Likewise.
1522         (vreinterpret_p16_s8): Likewise.
1523         (vreinterpret_p16_s16): Likewise.
1524         (vreinterpret_p16_s32): Likewise.
1525         (vreinterpret_p16_s64): Likewise.
1526         (vreinterpret_p16_f32): Likewise.
1527         (vreinterpret_p16_u8): Likewise.
1528         (vreinterpret_p16_u16): Likewise.
1529         (vreinterpret_p16_u32): Likewise.
1530         (vreinterpret_p16_u64): Likewise.
1531         (vreinterpret_p16_p8): Likewise.
1532         (vreinterpretq_p16_s8): Likewise.
1533         (vreinterpretq_p16_s16): Likewise.
1534         (vreinterpretq_p16_s32): Likewise.
1535         (vreinterpretq_p16_s64): Likewise.
1536         (vreinterpretq_p16_f32): Likewise.
1537         (vreinterpretq_p16_u8): Likewise.
1538         (vreinterpretq_p16_u16): Likewise.
1539         (vreinterpretq_p16_u32): Likewise.
1540         (vreinterpretq_p16_u64): Likewise.
1541         (vreinterpretq_p16_p8): Likewise.
1542         (vreinterpret_f32_s8): Likewise.
1543         (vreinterpret_f32_s16): Likewise.
1544         (vreinterpret_f32_s32): Likewise.
1545         (vreinterpret_f32_s64): Likewise.
1546         (vreinterpret_f32_u8): Likewise.
1547         (vreinterpret_f32_u16): Likewise.
1548         (vreinterpret_f32_u32): Likewise.
1549         (vreinterpret_f32_u64): Likewise.
1550         (vreinterpret_f32_p8): Likewise.
1551         (vreinterpret_f32_p16): Likewise.
1552         (vreinterpretq_f32_s8): Likewise.
1553         (vreinterpretq_f32_s16): Likewise.
1554         (vreinterpretq_f32_s32): Likewise.
1555         (vreinterpretq_f32_s64): Likewise.
1556         (vreinterpretq_f32_u8): Likewise.
1557         (vreinterpretq_f32_u16): Likewise.
1558         (vreinterpretq_f32_u32): Likewise.
1559         (vreinterpretq_f32_u64): Likewise.
1560         (vreinterpretq_f32_p8): Likewise.
1561         (vreinterpretq_f32_p16): Likewise.
1562         (vreinterpret_s64_s8): Likewise.
1563         (vreinterpret_s64_s16): Likewise.
1564         (vreinterpret_s64_s32): Likewise.
1565         (vreinterpret_s64_f32): Likewise.
1566         (vreinterpret_s64_u8): Likewise.
1567         (vreinterpret_s64_u16): Likewise.
1568         (vreinterpret_s64_u32): Likewise.
1569         (vreinterpret_s64_u64): Likewise.
1570         (vreinterpret_s64_p8): Likewise.
1571         (vreinterpret_s64_p16): Likewise.
1572         (vreinterpretq_s64_s8): Likewise.
1573         (vreinterpretq_s64_s16): Likewise.
1574         (vreinterpretq_s64_s32): Likewise.
1575         (vreinterpretq_s64_f32): Likewise.
1576         (vreinterpretq_s64_u8): Likewise.
1577         (vreinterpretq_s64_u16): Likewise.
1578         (vreinterpretq_s64_u32): Likewise.
1579         (vreinterpretq_s64_u64): Likewise.
1580         (vreinterpretq_s64_p8): Likewise.
1581         (vreinterpretq_s64_p16): Likewise.
1582         (vreinterpret_u64_s8): Likewise.
1583         (vreinterpret_u64_s16): Likewise.
1584         (vreinterpret_u64_s32): Likewise.
1585         (vreinterpret_u64_s64): Likewise.
1586         (vreinterpret_u64_f32): Likewise.
1587         (vreinterpret_u64_u8): Likewise.
1588         (vreinterpret_u64_u16): Likewise.
1589         (vreinterpret_u64_u32): Likewise.
1590         (vreinterpret_u64_p8): Likewise.
1591         (vreinterpret_u64_p16): Likewise.
1592         (vreinterpretq_u64_s8): Likewise.
1593         (vreinterpretq_u64_s16): Likewise.
1594         (vreinterpretq_u64_s32): Likewise.
1595         (vreinterpretq_u64_s64): Likewise.
1596         (vreinterpretq_u64_f32): Likewise.
1597         (vreinterpretq_u64_u8): Likewise.
1598         (vreinterpretq_u64_u16): Likewise.
1599         (vreinterpretq_u64_u32): Likewise.
1600         (vreinterpretq_u64_p8): Likewise.
1601         (vreinterpretq_u64_p16): Likewise.
1602         (vreinterpret_s8_s16): Likewise.
1603         (vreinterpret_s8_s32): Likewise.
1604         (vreinterpret_s8_s64): Likewise.
1605         (vreinterpret_s8_f32): Likewise.
1606         (vreinterpret_s8_u8): Likewise.
1607         (vreinterpret_s8_u16): Likewise.
1608         (vreinterpret_s8_u32): Likewise.
1609         (vreinterpret_s8_u64): Likewise.
1610         (vreinterpret_s8_p8): Likewise.
1611         (vreinterpret_s8_p16): Likewise.
1612         (vreinterpretq_s8_s16): Likewise.
1613         (vreinterpretq_s8_s32): Likewise.
1614         (vreinterpretq_s8_s64): Likewise.
1615         (vreinterpretq_s8_f32): Likewise.
1616         (vreinterpretq_s8_u8): Likewise.
1617         (vreinterpretq_s8_u16): Likewise.
1618         (vreinterpretq_s8_u32): Likewise.
1619         (vreinterpretq_s8_u64): Likewise.
1620         (vreinterpretq_s8_p8): Likewise.
1621         (vreinterpretq_s8_p16): Likewise.
1622         (vreinterpret_s16_s8): Likewise.
1623         (vreinterpret_s16_s32): Likewise.
1624         (vreinterpret_s16_s64): Likewise.
1625         (vreinterpret_s16_f32): Likewise.
1626         (vreinterpret_s16_u8): Likewise.
1627         (vreinterpret_s16_u16): Likewise.
1628         (vreinterpret_s16_u32): Likewise.
1629         (vreinterpret_s16_u64): Likewise.
1630         (vreinterpret_s16_p8): Likewise.
1631         (vreinterpret_s16_p16): Likewise.
1632         (vreinterpretq_s16_s8): Likewise.
1633         (vreinterpretq_s16_s32): Likewise.
1634         (vreinterpretq_s16_s64): Likewise.
1635         (vreinterpretq_s16_f32): Likewise.
1636         (vreinterpretq_s16_u8): Likewise.
1637         (vreinterpretq_s16_u16): Likewise.
1638         (vreinterpretq_s16_u32): Likewise.
1639         (vreinterpretq_s16_u64): Likewise.
1640         (vreinterpretq_s16_p8): Likewise.
1641         (vreinterpretq_s16_p16): Likewise.
1642         (vreinterpret_s32_s8): Likewise.
1643         (vreinterpret_s32_s16): Likewise.
1644         (vreinterpret_s32_s64): Likewise.
1645         (vreinterpret_s32_f32): Likewise.
1646         (vreinterpret_s32_u8): Likewise.
1647         (vreinterpret_s32_u16): Likewise.
1648         (vreinterpret_s32_u32): Likewise.
1649         (vreinterpret_s32_u64): Likewise.
1650         (vreinterpret_s32_p8): Likewise.
1651         (vreinterpret_s32_p16): Likewise.
1652         (vreinterpretq_s32_s8): Likewise.
1653         (vreinterpretq_s32_s16): Likewise.
1654         (vreinterpretq_s32_s64): Likewise.
1655         (vreinterpretq_s32_f32): Likewise.
1656         (vreinterpretq_s32_u8): Likewise.
1657         (vreinterpretq_s32_u16): Likewise.
1658         (vreinterpretq_s32_u32): Likewise.
1659         (vreinterpretq_s32_u64): Likewise.
1660         (vreinterpretq_s32_p8): Likewise.
1661         (vreinterpretq_s32_p16): Likewise.
1662         (vreinterpret_u8_s8): Likewise.
1663         (vreinterpret_u8_s16): Likewise.
1664         (vreinterpret_u8_s32): Likewise.
1665         (vreinterpret_u8_s64): Likewise.
1666         (vreinterpret_u8_f32): Likewise.
1667         (vreinterpret_u8_u16): Likewise.
1668         (vreinterpret_u8_u32): Likewise.
1669         (vreinterpret_u8_u64): Likewise.
1670         (vreinterpret_u8_p8): Likewise.
1671         (vreinterpret_u8_p16): Likewise.
1672         (vreinterpretq_u8_s8): Likewise.
1673         (vreinterpretq_u8_s16): Likewise.
1674         (vreinterpretq_u8_s32): Likewise.
1675         (vreinterpretq_u8_s64): Likewise.
1676         (vreinterpretq_u8_f32): Likewise.
1677         (vreinterpretq_u8_u16): Likewise.
1678         (vreinterpretq_u8_u32): Likewise.
1679         (vreinterpretq_u8_u64): Likewise.
1680         (vreinterpretq_u8_p8): Likewise.
1681         (vreinterpretq_u8_p16): Likewise.
1682         (vreinterpret_u16_s8): Likewise.
1683         (vreinterpret_u16_s16): Likewise.
1684         (vreinterpret_u16_s32): Likewise.
1685         (vreinterpret_u16_s64): Likewise.
1686         (vreinterpret_u16_f32): Likewise.
1687         (vreinterpret_u16_u8): Likewise.
1688         (vreinterpret_u16_u32): Likewise.
1689         (vreinterpret_u16_u64): Likewise.
1690         (vreinterpret_u16_p8): Likewise.
1691         (vreinterpret_u16_p16): Likewise.
1692         (vreinterpretq_u16_s8): Likewise.
1693         (vreinterpretq_u16_s16): Likewise.
1694         (vreinterpretq_u16_s32): Likewise.
1695         (vreinterpretq_u16_s64): Likewise.
1696         (vreinterpretq_u16_f32): Likewise.
1697         (vreinterpretq_u16_u8): Likewise.
1698         (vreinterpretq_u16_u32): Likewise.
1699         (vreinterpretq_u16_u64): Likewise.
1700         (vreinterpretq_u16_p8): Likewise.
1701         (vreinterpretq_u16_p16): Likewise.
1702         (vreinterpret_u32_s8): Likewise.
1703         (vreinterpret_u32_s16): Likewise.
1704         (vreinterpret_u32_s32): Likewise.
1705         (vreinterpret_u32_s64): Likewise.
1706         (vreinterpret_u32_f32): Likewise.
1707         (vreinterpret_u32_u8): Likewise.
1708         (vreinterpret_u32_u16): Likewise.
1709         (vreinterpret_u32_u64): Likewise.
1710         (vreinterpret_u32_p8): Likewise.
1711         (vreinterpret_u32_p16): Likewise.
1712         (vreinterpretq_u32_s8): Likewise.
1713         (vreinterpretq_u32_s16): Likewise.
1714         (vreinterpretq_u32_s32): Likewise.
1715         (vreinterpretq_u32_s64): Likewise.
1716         (vreinterpretq_u32_f32): Likewise.
1717         (vreinterpretq_u32_u8): Likewise.
1718         (vreinterpretq_u32_u16): Likewise.
1719         (vreinterpretq_u32_u64): Likewise.
1720         (vreinterpretq_u32_p8): Likewise.
1721         (vreinterpretq_u32_p16): Likewise.
1722
1723 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1724
1725         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
1726         Pattern extended.
1727         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
1728         extended.
1729         (sqabs): Likewise.
1730         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
1731         (vqnegd_s64): Likewise.
1732         (vqabs_s64): Likewise.
1733         (vqabsd_s64): Likewise.
1734
1735 2014-04-22  Richard Henderson  <rth@redhat.com>
1736
1737         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
1738         computation to the top of the loop.
1739
1740 2014-04-22  Renlin  <renlin.li@arm.com>
1741             Jiong Wang  <jiong.wang@arm.com>
1742
1743         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
1744         * config/aarch64/aarch64.c (aarch64_layout_frame)
1745         (aarch64_initial_elimination_offset): Likewise.
1746
1747 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1748
1749         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
1750         Fix indentation.
1751
1752 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
1753
1754         * machmode.h (bitwise_mode_for_mode): Declare.
1755         * stor-layout.h (bitwise_type_for_mode): Likewise.
1756         * stor-layout.c (bitwise_mode_for_mode): New function.
1757         (bitwise_type_for_mode): Likewise.
1758         * builtins.c (fold_builtin_memory_op): Use it instead of
1759         int_mode_for_mode and build_nonstandard_integer_type.
1760
1761 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1762
1763         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
1764         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
1765         (*-*-solaris2*): Simplify.
1766         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
1767         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
1768         *-*-solaris2.9* handling.
1769
1770         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
1771         as bug.
1772         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
1773         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
1774         handling, simplify.
1775         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
1776         * configure: Regenerate.
1777
1778         * config/i386/sol2-9.h: Remove.
1779
1780         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
1781         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
1782         Remove Solaris 9 references.
1783
1784 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
1785
1786         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
1787         (floatuns<GPI:mode><GPF:mode>2): Remove.
1788         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
1789         and floatuns conversions.
1790         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
1791         and floatuns conversions.
1792         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
1793         (w1,w2): New mode attributes for inequal width conversions.
1794
1795 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
1796
1797         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
1798         the output asm format.
1799
1800 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
1801
1802         * config/aarch64/aarch64-simd.md
1803         (aarch64_cm<optab>di): Always split.
1804         (*aarch64_cm<optab>di): New.
1805         (aarch64_cmtstdi): Always split.
1806         (*aarch64_cmtstdi): New.
1807
1808 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1809
1810         PR tree-optimization/60823
1811         * omp-low.c (ipa_simd_modify_function_body): Go through
1812         all SSA_NAMEs and for those refering to vector arguments
1813         which are going to be replaced adjust SSA_NAME_VAR and,
1814         if it is a default definition, change it into a non-default
1815         definition assigned at the beginning of function from new_decl.
1816         (ipa_simd_modify_stmt_ops): Rewritten.
1817         * tree-dfa.c (set_ssa_default_def): When removing default def,
1818         check for NULL loc instead of NULL *loc.
1819
1820 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1821
1822         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
1823         restrictions on core registers for DImode values in Thumb2.
1824
1825 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1826
1827         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
1828         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
1829
1830 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1831
1832         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
1833         (*iordi_notzesidi_di): Likewise.
1834         (*iordi_notsesidi_di): Likewise.
1835
1836 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1837
1838         * config/arm/arm-protos.h (tune_params): New struct members.
1839         * config/arm/arm.c: Initialise tune_params per processor.
1840         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
1841         for speed, based on new tune_params.
1842
1843 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1844
1845         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
1846         added.
1847         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
1848         macro.
1849         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
1850         corrected.
1851         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
1852         * config/aarch64/arm_neon.h (vrnd_f64): Added.
1853         (vrnda_f64): Likewise.
1854         (vrndi_f64): Likewise.
1855         (vrndm_f64): Likewise.
1856         (vrndn_f64): Likewise.
1857         (vrndp_f64): Likewise.
1858         (vrndx_f64): Likewise.
1859
1860 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1861
1862         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
1863         GET_MODE_SIZE argument is enum machine_mode.
1864
1865 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1866
1867         PR target/60910
1868         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
1869         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
1870
1871 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
1872
1873         PR middle-end/60281
1874         * asan.c (asan_emit_stack_protection): Force the base to align to
1875         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
1876         appropriate bits if STRICT_ALIGNMENT.
1877         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
1878         when asan is on.
1879         (expand_used_vars): Leave a space in the stack frame for alignment
1880         if STRICT_ALIGNMENT.
1881
1882 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
1883
1884         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
1885         than a gimple.
1886         (gimple_store_p): Likewise.
1887         (gimple_assign_load_p): Likewise.
1888         (gimple_assign_cast_p): Likewise.
1889         (gimple_clobber_p): Likewise.
1890
1891         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
1892         rather than a gimple.
1893         (gimple_assign_cast_p): Likewise.
1894
1895 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
1896
1897         PR target/60735
1898         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
1899         If mode is DDmode and TARGET_E500_DOUBLE allow move.
1900
1901         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
1902         more debug information for E500 if -mdebug=reg.
1903
1904 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
1905
1906         PR target/60909
1907         * config/i386/i386.c (ix86_expand_builtin)
1908         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
1909         register for target RTX.
1910         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
1911
1912 2014-04-18  Cong Hou  <congh@google.com>
1913
1914         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
1915         the widen-mult pattern by handling two operands with different sizes,
1916         and operands whose size is smaller than half of the result type.
1917
1918 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1919
1920         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
1921         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
1922         (do_estimate_edge_time): Compute it.
1923         * ipa-inline.c (want_inline_small_function_p): Bypass
1924         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
1925
1926 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1927
1928         * ipa-inline.c (spec_rem): New static variable.
1929         (dump_overall_stats): New function.
1930         (dump_inline_stats): New function.
1931
1932 2014-04-18  Richard Henderson  <rth@redhat.com>
1933
1934         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
1935         to GET_MODE_SIZE, not a reg_class_t.
1936
1937 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1938
1939         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
1940         (vsx_xxmrglw_<mode>): Likewise.
1941
1942 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
1943
1944         PR target/60876
1945         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
1946         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
1947         (rs6000_init_hard_regno_mode_ok): Likewise.
1948
1949 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
1950
1951         * ipa-inline.c (inline_small_functions): Account only non-cold
1952         functions.
1953         * doc/invoke.texi (inline-unit-growth): Update documentation.
1954
1955 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
1956
1957         * config/rs6000/rs6000.md (addti3, subti3): New.
1958
1959 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
1960
1961         PR target/60863
1962         * config/i386/i386.c (ix86_expand_clear): Remove outdated
1963         comment.  Check optimize_insn_for_size_p instead of
1964         optimize_insn_for_speed_p.
1965
1966 2014-04-17  Martin Jambor  <mjambor@suse.cz>
1967
1968         * gimple-iterator.c (gsi_start_edge): New function.
1969         * gimple-iterator.h (gsi_start_edge): Declare.
1970         * tree-sra.c (single_non_eh_succ): New function.
1971         (disqualify_ops_if_throwing_stmt): Renamed to
1972         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
1973         having one non-EH successor BB.
1974         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
1975         generate loads into replacements.
1976         (sra_modify_assign): Likewise and and also use the simple path for
1977         such statements.
1978         (sra_modify_function_body): Commit statements on edges.
1979
1980 2014-04-17  Richard Biener  <rguenther@suse.de>
1981
1982         PR middle-end/60849
1983         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
1984         comparison results and add clarifying comment.
1985
1986 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1987
1988         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
1989         (blank_mode): Initialize it.
1990         (emit_mode_size_inline, emit_mode_nunits_inline,
1991         emit_mode_inner_inline): New functions.
1992         (emit_insn_modes_h): Call them and surround their output with
1993         #if GCC_VERSION >= 4001 ... #endif.
1994         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
1995         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
1996         mode_* arrays if the argument is __builtin_constant_p.
1997         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
1998         is enum machine_mode.
1999
2000 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2001
2002         * passes.c (opt_pass::execute): Adjust.
2003         (pass_manager::execute_pass_mode_switching): Likewise.
2004         (early_local_passes::execute): Likewise.
2005         (execute_one_pass): Pass cfun to the pass's execute method.
2006         * tree-pass.h (opt_pass::execute): Add function * argument.
2007         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
2008         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
2009         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
2010         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
2011         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
2012         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
2013         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
2014         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
2015         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
2016         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
2017         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
2018         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
2019         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
2020         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
2021         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
2022         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
2023         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
2024         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
2025         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
2026         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
2027         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
2028         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
2029         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
2030         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
2031         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
2032         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
2033         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
2034         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
2035         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
2036         Adjust.
2037
2038 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2039
2040         * passes.c (opt_pass::gate): Take function * argument.
2041         (gate_all_early_local_passes): Merge into
2042         (early_local_passes::gate): this.
2043         (gate_all_early_optimizations): Merge into
2044         (all_early_optimizations::gate): this.
2045         (gate_all_optimizations): Mege into
2046         (all_optimizations::gate): this.
2047         (gate_all_optimizations_g): Merge into
2048         (all_optimizations_g::gate): this.
2049         (gate_rest_of_compilation): Mege into
2050         (rest_of_compilation::gate): this.
2051         (gate_postreload): Merge into
2052         (postreload::gate): this.
2053         (dump_one_pass): Pass cfun to the pass's gate method.
2054         (execute_ipa_summary_passes): Likewise.
2055         (execute_one_pass): Likewise.
2056         (ipa_write_summaries_2): Likewise.
2057         (ipa_write_optimization_summaries_1): Likewise.
2058         (ipa_read_summaries_1): Likewise.
2059         (ipa_read_optimization_summaries_1): Likewise.
2060         (execute_ipa_stmt_fixups): Likewise.
2061         * tree-pass.h (opt_pass::gate): Add function * argument.
2062         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
2063         combine-stack-adj.c, combine.c, compare-elim.c,
2064         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
2065         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
2066         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
2067         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
2068         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
2069         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
2070         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
2071         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
2072         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
2073         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
2074         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
2075         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
2076         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
2077         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
2078         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
2079         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
2080         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
2081         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
2082         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
2083         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
2084         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
2085         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
2086         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
2087         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
2088         var-tracking.c, vtable-verify.c, web.c: Adjust.
2089
2090 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2091
2092         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
2093         * configure: Regenerate.
2094
2095 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2096
2097         * passes.c (dump_one_pass): don't check pass->has_gate.
2098         (execute_ipa_summary_passes): Likewise.
2099         (execute_one_pass): Likewise.
2100         (ipa_write_summaries_2): Likewise.
2101         (ipa_write_optimization_summaries_1): Likewise.
2102         (ipa_read_optimization_summaries_1): Likewise.
2103         (execute_ipa_stmt_fixups): Likewise.
2104         * tree-pass.h (pass_data::has_gate): Remove.
2105         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
2106         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
2107         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
2108         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
2109         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
2110         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
2111         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
2112         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
2113         gimple-low.c, gimple-ssa-isolate-paths.c,
2114         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
2115         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
2116         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
2117         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
2118         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
2119         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
2120         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
2121         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
2122         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
2123         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
2124         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
2125         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
2126         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
2127         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
2128         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
2129         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
2130         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
2131         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
2132         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
2133         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
2134         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
2135         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
2136         Adjust.
2137
2138 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2139
2140         * pass_manager.h (pass_manager::register_dump_files_1): Remove
2141         declaration.
2142         * passes.c (pass_manager::register_dump_files_1): Merge into
2143         (pass_manager::register_dump_files): this, and remove its handling of
2144         properties since the pass always has the properties anyway.
2145         (pass_manager::pass_manager): Adjust.
2146
2147 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
2148
2149         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
2150         * passes.c (pass_manager::register_dump_files_1): Remove dead code
2151         dealing with properties.
2152         (pass_manager::register_dump_files): Adjust.
2153
2154 2014-03-20  Mark Wielaard  <mjw@redhat.com>
2155
2156         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
2157         then represent the bound as normal constant value.
2158
2159 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
2160
2161         PR target/60847
2162         Forward port from 4.8 branch
2163         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
2164
2165         * config/i386/bmiintrin.h (_blsi_u32): New.
2166         (_blsi_u64): Ditto.
2167         (_blsr_u32): Ditto.
2168         (_blsr_u64): Ditto.
2169         (_blsmsk_u32): Ditto.
2170         (_blsmsk_u64): Ditto.
2171         (_tzcnt_u32): Ditto.
2172         (_tzcnt_u64): Ditto.
2173
2174 2014-04-17  Kito Cheng  <kito@0xlab.org>
2175
2176         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
2177
2178 2014-04-17  Richard Biener  <rguenther@suse.de>
2179
2180         PR middle-end/60849
2181         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
2182         boolean results for comparisons.
2183
2184 2014-04-17  Richard Biener  <rguenther@suse.de>
2185
2186         PR tree-optimization/60836
2187         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
2188         initial PHI args to be gimple values.
2189
2190 2014-04-17   Richard Biener  <rguenther@suse.de>
2191
2192         PR tree-optimization/60841
2193         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
2194         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
2195         of stmts to SLP build.
2196         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
2197         (vect_analyze_slp): Likewise.
2198         (vect_analyze_slp_instance): Likewise.
2199         (vect_build_slp_tree): Limit overall SLP tree growth.
2200         * tree-vectorizer.h (vect_analyze_data_refs,
2201         vect_analyze_slp): Adjust prototypes.
2202
2203 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2204
2205         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
2206         Silvermont.
2207
2208 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2209
2210         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
2211         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
2212         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
2213         for TARGET_SLOW_PSHUFB
2214
2215 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2216
2217         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
2218         * config/i386/i386.c (intel_cost): Ditto.
2219
2220 2014-04-17  Joey Ye  <joey.ye@arm.com>
2221
2222         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
2223
2224 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2225
2226         * opts.c (common_handle_option): Disable -fipa-reference coorectly
2227         with -fuse-profile.
2228
2229 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2230
2231         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
2232         (type_all_derivations_known_p): New predicate.
2233         (type_all_ctors_visible_p): New predicate.
2234         (type_possibly_instantiated_p): New predicate.
2235         (get_odr_type): Compute all_derivations_known.
2236         (dump_odr_type): Dump the flag.
2237         (maybe_record_type): Cleanup.
2238         (record_target_from_binfo): Add bases_to_consider array;
2239         record bases for types w/o instances and skip CXX destructor.
2240         (possible_polymorphic_call_targets_1): Add bases_to_consider
2241         and consider_construction parameters; check if type may have instance.
2242         (get_polymorphic_call_info): Set maybe_in_construction to true
2243         when we know nothing.
2244         (record_targets_from_bases): Skip CXX destructors; they are
2245         never called for types in construction.
2246         (possible_polymorphic_call_targets): Do not record target when
2247         type may not have instance.
2248
2249 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2250
2251         PR ipa/60854
2252         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
2253         external aliases alive, too.
2254
2255 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
2256
2257         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
2258         definition.
2259
2260 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
2261
2262         * final.c (compute_alignments): Do not apply loop alignment to a block
2263         falling through to the exit.
2264
2265 2014-04-16  Catherine Moore  <clm@codesourcery.com>
2266
2267         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
2268         Adjust constraints for microMIPS store patterns.
2269
2270 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2271
2272         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
2273
2274 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
2275
2276         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
2277         (append_use): Run at -O0.
2278         (append_vdef): Likewise.
2279         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
2280         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
2281
2282 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
2283
2284         PR tree-optimization/60844
2285         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
2286         (propagate_op_to_single_use, remove_visited_stmt_chain,
2287         linearize_expr, repropagate_negates, reassociate_bb): Use it
2288         instead of gsi_remove.
2289
2290 2014-04-16  Martin Jambor  <mjambor@suse.cz>
2291
2292         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
2293         ipa_transforms_to_apply.
2294         (cgraph_function_versioning): Assert that old_node has empty
2295         ipa_transforms_to_apply.
2296         * trans-mem.c (ipa_tm_create_version): Likewise.
2297         * tree-inline.c (tree_function_versioning): Do not duplicate
2298         ipa_transforms_to_apply.
2299
2300 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2301
2302         PR target/60817
2303         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
2304         x86_64-*-* cases.
2305         Pass necessary as flags on 64-bit Solaris/x86.
2306         Use lowercase relocs for x86_64-*-*.
2307         * configure: Regenerate.
2308
2309 2014-04-15  Jan Hubicka  <jh@suse.cz>
2310
2311         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
2312         (maybe_record_node, likely_target_p): Use it.
2313
2314 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2315
2316         PR target/60839
2317         Revert following patch
2318
2319         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2320
2321         PR target/60735
2322         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2323         software floating point or no floating point registers, do not
2324         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2325         in GPRs that occurs after we tested for GPRs that would never be
2326         true.
2327
2328         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2329         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2330         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2331         specifically allow DDmode, since that does not use the SPE SIMD
2332         instructions.
2333
2334 2014-03-21  Mark Wielaard  <mjw@redhat.com>
2335
2336         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
2337         as unsigned or int depending on type and value used.
2338
2339 2014-04-15  Richard Biener  <rguenther@suse.de>
2340
2341         PR rtl-optimization/56965
2342         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
2343         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
2344         ... here.
2345         * alias.c (true_dependence_1): Do not call
2346         nonoverlapping_component_refs_p.
2347         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
2348         nonoverlapping_component_refs_p.
2349         (indirect_refs_may_alias_p): Likewise.
2350
2351 2014-04-15  Teresa Johnson  <tejohnson@google.com>
2352
2353         * cfg.c (dump_bb_info): Fix flags check.
2354         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
2355
2356 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2357
2358         PR rtl-optimization/60663
2359         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
2360         avoid 0 cost.
2361
2362 2014-04-15  Richard Biener  <rguenther@suse.de>
2363
2364         * lto-streamer.h (LTO_major_version): Bump to 4.
2365
2366 2014-04-15  Richard Biener  <rguenther@suse.de>
2367
2368         * common.opt (lto_partition_model): New enum.
2369         (flto-partition=): Merge separate options with a single with argument,
2370         add -flto-partition=one support.
2371         * flag-types.h (enum lto_partition_model): Declare.
2372         * opts.c (finish_options): Remove duplicate -flto-partition=
2373         option check.
2374         * lto-wrapper.c (run_gcc): Adjust.
2375
2376 2014-04-15  Richard Biener  <rguenther@suse.de>
2377
2378         * alias.c (ncr_compar): New function.
2379         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
2380
2381 2014-04-15  Richard Biener  <rguenther@suse.de>
2382
2383         * alias.c (record_component_aliases): Do not walk BINFOs.
2384
2385 2014-04-15  Richard Biener  <rguenther@suse.de>
2386
2387         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2388         Add struct function argument and adjust.
2389         (find_func_aliases_for_call): Likewise.
2390         (find_func_aliases): Likewise.
2391         (find_func_clobbers): Likewise.
2392         (intra_create_variable_infos): Likewise.
2393         (compute_points_to_sets): Likewise.
2394         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
2395
2396 2014-04-15  Richard Biener  <rguenther@suse.de>
2397
2398         * tree.c (iterative_hash_expr): Use enum tree_code_class
2399         to store TREE_CODE_CLASS.
2400         (tree_block): Likewise.
2401         (tree_set_block): Likewise.
2402         * tree.h (fold_build_pointer_plus_loc): Use
2403         convert_to_ptrofftype_loc.
2404
2405 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
2406
2407         PR plugins/59335
2408         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
2409         added in 4.9.
2410
2411 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
2412
2413         * cfgloop.h (struct loop): Move force_vectorize down.
2414         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
2415         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
2416         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
2417         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
2418         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
2419         * tree-core.h (enum annot_expr_kind): Add new kind values.
2420         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
2421         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
2422         kinds.
2423         * tree.def (ANNOTATE_EXPR): Tweak comment.
2424
2425 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2426
2427         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
2428         cxa_pure_virtual).
2429
2430 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
2431
2432         * tree.h (TYPE_IDENTIFIER): Declare.
2433         * tree.c (subrange_type_for_debug_p): Use it.
2434         * godump.c (go_format_type): Likewise.
2435         * dwarf2out.c (is_cxx_auto, modified_type_die,
2436         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
2437         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
2438
2439 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
2440
2441         PR lto/60820
2442         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
2443
2444 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
2445
2446         * config/i386/i386.c (examine_argument): Return bool.  Return true if
2447         parameter should be passed in memory.
2448         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
2449         (construct_container): Update calls to examine_argument.
2450         (function_arg_advance_64): Ditto.
2451         (return_in_memory_32): Merge with ix86_return_in_memory.
2452         (return_in_memory_64): Ditto.
2453         (return_in_memory_ms_64): Ditto.
2454
2455 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2456
2457         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
2458         * coverage.c (coverage_compute_profile_id): Handle externally visible
2459         symbols.
2460
2461 2014-04-14  Martin Jambor  <mjambor@suse.cz>
2462
2463         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
2464         DECL_DISREGARD_INLINE_LIMITS functions.
2465
2466 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2467
2468         PR target/60827
2469         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
2470
2471 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2472
2473         PR target/60827
2474         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
2475         optimize_insn_for_speed_p instead of
2476         optimize_function_for_speed_p.
2477
2478 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
2479
2480         * doc/invoke.texi (free): Document AArch64.
2481
2482 2014-04-14  Richard Biener  <rguenther@suse.de>
2483
2484         PR tree-optimization/60042
2485         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
2486         (insert_into_preds_of_block): Do not prevent PHI insertion
2487         for REFERENCE exprs here ...
2488         (eliminate_dom_walker::before_dom_children): ... but prevent
2489         their use here under similar conditions when applied to the
2490         IL after PRE optimizations.
2491
2492 2014-04-14  Richard Biener  <rguenther@suse.de>
2493
2494         * passes.def: Move early points-to after early SRA.
2495
2496 2014-04-14  Richard Biener  <rguenther@suse.de>
2497
2498         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
2499         check for which sign-changes we allow when forwarding
2500         a converted value into a switch.
2501
2502 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2503
2504         * stor-layout.c (place_field): Finalize non-constant offset for the
2505         field, if any.
2506
2507 2014-04-14  Richard Biener  <rguenther@suse.de>
2508
2509         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
2510         as argument.
2511         (expand_switch_using_bit_tests_p): Likewise.
2512         (process_switch): Compute and pass on speed_p based on the
2513         switch stmt.
2514         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
2515         optimize_bb_for_speed_p.
2516
2517 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2518
2519         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
2520         * function.h (struct function): Rename has_force_vect_loops into
2521         has_force_vectorize_loops.
2522         * lto-streamer-in.c (input_cfg): Adjust for renaming.
2523         (input_struct_function_base): Likewise.
2524         * lto-streamer-out.c (output_cfg): Likewise.
2525         (output_struct_function_base): Likewise.
2526         * omp-low.c (expand_omp_simd): Likewise.
2527         * tree-cfg.c (move_sese_region_to_fn): Likewise.
2528         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2529         (version_loop_for_if_conversion): Likewise.
2530         (tree_if_conversion): Likewise.
2531         (main_tree_if_conversion): Likewise.
2532         (gate_tree_if_conversion): Likewise.
2533         * tree-inline.c (copy_loops): Likewise.
2534         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
2535         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
2536         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
2537         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
2538         * tree-vectorizer.c (vectorize_loops): Likewise.
2539         * tree-vectorizer.h (unlimited_cost_model): Likewise.
2540
2541 2014-04-14  Richard Biener  <rguenther@suse.de>
2542
2543         PR lto/60720
2544         * lto-streamer-out.c (wrap_refs): New function.
2545         (lto_output): Wrap symbol references in global initializes in
2546         type-preserving MEM_REFs.
2547
2548 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2549
2550         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
2551
2552 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2553
2554         * config/sh/sh.md (setmemqi): New expand pattern.
2555         * config/sh/sh.h (CLEAR_RATIO): Define.
2556         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
2557         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
2558
2559 2014-04-14  Richard Biener  <rguenther@suse.de>
2560
2561         PR middle-end/55022
2562         * fold-const.c (negate_expr_p): Don't negate directional rounding
2563         division.
2564         (fold_negate_expr): Likewise.
2565
2566 2014-04-14  Richard Biener  <rguenther@suse.de>
2567
2568         PR tree-optimization/59817
2569         PR tree-optimization/60453
2570         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
2571         recursion to catch all CHRECs in the scalar evolution and restrict
2572         the predicate for the remains appropriately.
2573
2574 2014-04-12  Catherine Moore  <clm@codesourcery.com>
2575
2576         * config/mips/constraints.md: Add new register constraint "kb".
2577         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
2578         (*movhi_internal): Likewise.
2579         (*movqi_internal): Likewise.
2580         * config/mips/mips.h (M16_STORE_REGS): New register class.
2581         (REG_CLASS_NAMES): Add M16_STORE_REGS.
2582         (REG_CLASS_CONTENTS): Likewise.
2583         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
2584
2585 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2586
2587         PR c/60194
2588         * doc/invoke.texi (-Wformat-signedness): Document it.
2589         (Wformat=2): Mention that this enables -Wformat-signedness.
2590
2591 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2592
2593         * common/config/epiphany/epiphany-common.c
2594         (epiphany_option_optimization_table): Enable section anchors by
2595         default at -O1 or higher.
2596         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
2597         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
2598         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
2599         carries no extra cost.
2600         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
2601         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
2602         * config/epiphany/predicates.md (memclob_operand): New predicate.
2603         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
2604         Use memclob_operand predicate and X constraint for operand 3.
2605
2606 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2607
2608         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
2609         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
2610         its operands.
2611
2612 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2613
2614         PR rtl-optimization/60651
2615         * mode-switching.c (optimize_mode_switching): Make sure to emit
2616         sets of a lower numbered entity before sets of a higher numbered
2617         entity to a mode of the same or lower priority.
2618         When creating a seginfo for a basic block that starts with a code
2619         label, move the insertion point past the code label.
2620         (new_seginfo): Document and enforce requirement that
2621         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
2622         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
2623         * doc/tm.texi: Regenerate.
2624
2625 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2626
2627         PR target/60811
2628         * config/arc/arc.c (arc_save_restore): Fix assert typo.
2629
2630 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
2631
2632         * BASE-VER: Set to 4.10.0.
2633
2634 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2635
2636         PR other/59055
2637         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
2638         * doc/gcc.texi (Service): Update description in the @menu
2639         * doc/invoke.texi (Option Summary): Remove misplaced and
2640         duplicated @menu.
2641
2642 2014-04-11  Steve Ellcey  <sellcey@mips.com>
2643             Jakub Jelinek  <jakub@redhat.com>
2644
2645         PR middle-end/60556
2646         * expr.c (convert_move): Use emit_store_flag_force instead of
2647         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
2648         argument to it.
2649
2650 2014-04-11  Richard Biener  <rguenther@suse.de>
2651
2652         PR middle-end/60797
2653         * varasm.c (assemble_alias): Avoid endless error reporting
2654         recursion by setting TREE_ASM_WRITTEN.
2655
2656 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2657
2658         * config/s390/s390.md: Add a splitter for NOT rtx.
2659
2660 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
2661
2662         PR rtl-optimization/60663
2663         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
2664
2665 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
2666             Jakub Jelinek  <jakub@redhat.com>
2667
2668         PR lto/60567
2669         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
2670         flag from decl_node to node.
2671
2672 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2673
2674         PR debug/60655
2675         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
2676         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
2677         ameliorating the cases where it can be.
2678
2679 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
2680
2681         Revert
2682         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2683
2684         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2685         (loadsync_<mode>): Change mode.
2686         (load_quadpti, store_quadpti): New.
2687         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2688         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2689         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
2690
2691 2014-04-09  Cong Hou  <congh@google.com>
2692
2693         PR testsuite/60773
2694         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
2695         documentation.
2696
2697 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2698
2699         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
2700         instead of vnor to exploit possible fusion opportunity in the
2701         future.
2702         (altivec_expand_vec_perm_const_le): Likewise.
2703
2704 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2705
2706         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2707         (loadsync_<mode>): Change mode.
2708         (load_quadpti, store_quadpti): New.
2709         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2710         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2711
2712 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
2713
2714         PR target/60763
2715         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
2716         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
2717         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
2718
2719 2014-04-08  Richard Biener  <rguenther@suse.de>
2720
2721         PR middle-end/60706
2722         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
2723         a 64bit widest int print double-int similar to on HWI64 hosts.
2724
2725 2014-04-08  Richard Biener  <rguenther@suse.de>
2726
2727         PR tree-optimization/60785
2728         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
2729         default defs properly.
2730
2731 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
2732
2733         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
2734         (Weffc++): Likewise.
2735
2736 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
2737
2738         * ipa-devirt.c (maybe_record_node): When node is not recorded,
2739         set completep to false rather than true.
2740
2741 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
2742
2743         PR target/60504
2744         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
2745         ARM_TARGET2_DWARF_FORMAT.
2746
2747 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
2748
2749         PR target/60609
2750         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
2751         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
2752         ADDR_DIFF_VEC.
2753
2754 2014-04-07  Richard Biener  <rguenther@suse.de>
2755
2756         PR tree-optimization/60766
2757         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
2758         (may_eliminate_iv): Convert cand_value_at result to desired type.
2759
2760 2014-04-07  Jason Merrill  <jason@redhat.com>
2761
2762         PR c++/60731
2763         * common.opt (-fno-gnu-unique): Add.
2764         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
2765
2766 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2767
2768         * haifa-sched.c: Fix outdated function reference and minor
2769         grammar errors in introductory comment.
2770
2771 2014-04-07  Richard Biener  <rguenther@suse.de>
2772
2773         PR middle-end/60750
2774         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
2775         for noreturn calls.
2776         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
2777
2778 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
2779
2780         PR debug/55794
2781         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
2782         size accounting for thunks.
2783         (pa_asm_output_mi_thunk): Use final_start_function() and
2784         final_end_function() to output function start and end directives.
2785
2786 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2787
2788         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
2789         device specific ISA/ feature information. Remove short_sp and
2790         errata_skip ds.  Add avr_device_specific_features enum to have device
2791         specific info.
2792         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
2793         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
2794         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
2795         updated device specific info.
2796         * config/avr/avr-mcus.def: Merge device specific details to
2797         dev_attribute field.
2798         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
2799         errata_skip.
2800         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
2801         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
2802         assembler if RMW isa supported by current device.
2803         * config/avr/genmultilib.awk: Update as device info structure changed.
2804         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
2805
2806 2014-04-04  Cong Hou  <congh@google.com>
2807
2808         PR tree-optimization/60656
2809         * tree-vect-stmts.c (supportable_widening_operation):
2810         Fix a bug that elements in a vector with vect_used_by_reduction
2811         property are incorrectly reordered when the operation on it is not
2812         consistant with the one in reduction operation.
2813
2814 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
2815
2816         PR rtl-optimization/60155
2817         * gcse.c (record_set_data): New function.
2818         (single_set_gcse): New function.
2819         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
2820         (hoist_code): Likewise.
2821         (get_pressure_class_and_nregs): Likewise.
2822
2823 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
2824
2825         * explow.c (probe_stack_range): Emit a final optimization blockage.
2826
2827 2014-04-04  Anthony Green  <green@moxielogic.com>
2828
2829         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
2830         typos.
2831
2832 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
2833
2834         PR ipa/59626
2835         * lto-cgraph.c (input_overwrite_node): Check that partitioning
2836         flags are set only during streaming.
2837         * ipa.c (process_references, walk_polymorphic_call_targets,
2838         symtab_remove_unreachable_nodes): Drop bodies of always inline
2839         after early inlining.
2840         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
2841
2842 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
2843         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2844
2845         PR debug/60655
2846         * dwarf2out.c (const_ok_for_output_1): Reject expressions
2847         containing a NOT.
2848
2849 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2850
2851         PR bootstrap/60743
2852         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
2853         duration.
2854         (cortex_a53_fdivd): Likewise.
2855
2856 2014-04-04  Martin Jambor  <mjambor@suse.cz>
2857
2858         PR ipa/60640
2859         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
2860         Adjust all callers.
2861         * cgraph.c (clone_of_p): Also return true if thunks match.
2862         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
2863         cgraph_function_or_thunk_node and an obsolete comment.
2864         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
2865         file.
2866         (build_function_decl_skip_args): Likewise.
2867         (set_new_clone_decl_and_node_flags): New function.
2868         (duplicate_thunk_for_node): Likewise.
2869         (redirect_edge_duplicating_thunks): Likewise.
2870         (cgraph_clone_node): New parameter args_to_skip, pass it to
2871         redirect_edge_duplicating_thunks which is called instead of
2872         cgraph_redirect_edge_callee.
2873         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
2874         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
2875
2876 2014-04-04  Jeff Law  <law@redhat.com>
2877
2878         PR target/60657
2879         * config/arm/predicates.md (const_int_I_operand): New predicate.
2880         (const_int_M_operand): Similarly.
2881         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
2882         const_int_operand.
2883         (insv_t2, extv_reg, extzv_t2): Likewise.
2884         (load_multiple_with_writeback): Similarly for const_int_I_operand.
2885         (pop_multiple_with_writeback_and_return): Likewise.
2886         (vfp_pop_multiple_with_writeback): Likewise
2887
2888 2014-04-04  Richard Biener  <rguenther@suse.de>
2889
2890         PR ipa/60746
2891         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
2892         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
2893         non-GIMPLE_LABELs.
2894         * gimplify.h (gimple_add_tmp_var_fn): Declare.
2895         * gimplify.c (gimple_add_tmp_var_fn): New function.
2896         * gimple-expr.h (create_tmp_reg_fn): Declare.
2897         * gimple-expr.c (create_tmp_reg_fn): New function.
2898         * gimple-low.c (record_vars_into): Don't change cfun.
2899         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
2900         code generation without cfun.
2901
2902 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
2903
2904         PR bootstrap/60719
2905         * Makefile.in (install-driver): Fix shell scripting.
2906
2907 2014-04-03  Cong Hou  <congh@google.com>
2908
2909         PR tree-optimization/60505
2910         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
2911         threshold of number of iterations below which no vectorization
2912         will be done.
2913         * tree-vect-loop.c (new_loop_vec_info):
2914         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
2915         * tree-vect-loop.c (vect_analyze_loop_operations):
2916         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
2917         * tree-vect-loop.c (vect_transform_loop):
2918         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
2919         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
2920         of iterations of the loop and see if we should build the epilogue.
2921
2922 2014-04-03  Richard Biener  <rguenther@suse.de>
2923
2924         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
2925         (streamer_tree_cache_create): Adjust.
2926         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
2927         to allow optional nodes array.
2928         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
2929         (streamer_tree_cache_append): Likewise.
2930         (streamer_tree_cache_create): Create nodes array optionally
2931         as specified by parameter.
2932         * lto-streamer-out.c (create_output_block): Avoid maintaining
2933         the node array in the writer cache.
2934         (DFS_write_tree): Remove assertion.
2935         (produce_asm_for_decls): Free the out decl state hash table early.
2936         * lto-streamer-in.c (lto_data_in_create): Adjust for
2937         streamer_tree_cache_create prototype change.
2938
2939 2014-04-03  Richard Biener  <rguenther@suse.de>
2940
2941         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
2942         set TREE_CHAIN to NULL_TREE.
2943
2944 2014-04-03  Richard Biener  <rguenther@suse.de>
2945
2946         PR tree-optimization/60740
2947         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
2948         over all GIMPLE_COND operands.
2949
2950 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
2951
2952         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
2953         (Weffc++): Remove Scott's numbering, merge lists and reference
2954         Wnon-virtual-dtor.
2955
2956 2014-04-03  Nick Clifton  <nickc@redhat.com>
2957
2958         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
2959         properly.
2960
2961 2014-04-03  Martin Jambor  <mjambor@suse.cz>
2962
2963         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
2964         mention gcc_unreachable before failing.
2965         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
2966         removed symbols.
2967
2968 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
2969
2970         PR ipa/60659
2971         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
2972         inconsistent code and instead mark the context inconsistent.
2973         (possible_polymorphic_call_targets): For inconsistent contexts
2974         return empty complete list.
2975
2976 2014-04-02  Anthony Green  <green@moxielogic.com>
2977
2978         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
2979         (extendqisi2, extendhisi2): Define.
2980         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
2981         (WCHAR_TYPE): Change to unsigned int.
2982
2983 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2984
2985         PR tree-optimization/60733
2986         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
2987         insertion point for PHI candidates to be the end of the feeding
2988         block for the PHI argument.
2989
2990 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
2991
2992         PR rtl-optimization/60650
2993         * lra-constraints.c (process_alt_operands): Decrease reject for
2994         earlyclobber matching.
2995
2996 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2997
2998         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
2999
3000 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3001
3002         * config/spu/spu.c (pad_bb): Do not crash when the last
3003         insn is CODE_FOR_blockage.
3004
3005 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3006
3007         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
3008         lies outside the target mode.
3009
3010 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3011
3012         PR target/60735
3013         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
3014         software floating point or no floating point registers, do not
3015         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
3016         in GPRs that occurs after we tested for GPRs that would never be
3017         true.
3018
3019         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
3020         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
3021         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
3022         specifically allow DDmode, since that does not use the SPE SIMD
3023         instructions.
3024
3025 2014-04-02  Richard Biener  <rguenther@suse.de>
3026
3027         PR middle-end/60729
3028         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
3029         MODE_INTs.  Properly use negv_optab.
3030         (expand_abs): Likewise.
3031
3032 2014-04-02  Richard Biener  <rguenther@suse.de>
3033
3034         PR bootstrap/60719
3035         * Makefile.in (install-driver): Guard extra installs with special
3036         names properly.
3037
3038 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
3039
3040         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3041         Document vec_vgbbd.
3042
3043 2014-04-01  Richard Henderson  <rth@redhat.com>
3044
3045         PR target/60704
3046         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
3047         alternative enabled before register allocation.
3048
3049 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
3050
3051         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
3052         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
3053         typo.
3054         (nios2_large_got_address): Remove unneeded 'sym' parameter.
3055         (nios2_got_address): Update nios2_large_got_address call site.
3056         (nios2_delegitimize_address): New function.
3057         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
3058         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
3059         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
3060
3061 2014-04-01  Martin Husemann  <martin@duskware.de>
3062
3063         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
3064         for -mabi=32.
3065
3066 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
3067
3068         PR rtl-optimization/60604
3069         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
3070         check from register_operand.
3071         (register_operand): Redefine in terms of general_operand.
3072         (nonmemory_operand): Use register_operand for the non-constant cases.
3073
3074 2014-04-01  Richard Biener  <rguenther@suse.de>
3075
3076         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
3077
3078 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3079
3080         * doc/invoke.texi (mapp-regs): Clarify.
3081
3082 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
3083
3084         * config/i386/avx512fintrin.h (__v32hi): Define type.
3085         (__v64qi): Likewise.
3086         (_mm512_set1_epi8): Define.
3087         (_mm512_set1_epi16): Define.
3088         (_mm512_set4_epi32): Define.
3089         (_mm512_set4_epi64): Define.
3090         (_mm512_set4_pd): Define.
3091         (_mm512_set4_ps): Define.
3092         (_mm512_setr4_epi64): Define.
3093         (_mm512_setr4_epi32): Define.
3094         (_mm512_setr4_pd): Define.
3095         (_mm512_setr4_ps): Define.
3096         (_mm512_setzero_epi32): Define.
3097
3098 2014-03-31  Martin Jambor  <mjambor@suse.cz>
3099
3100         PR middle-end/60647
3101         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
3102         callsite_arguments_match_p.  Updated all callers.  Also check types of
3103         corresponding formal parameters and actual arguments.
3104         (not_all_callers_have_enough_arguments_p) Renamed to
3105         some_callers_have_mismatched_arguments_p.
3106
3107 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
3108
3109         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
3110
3111 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
3112
3113         PR target/60034
3114         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
3115         section anchor.
3116
3117 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
3118
3119         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
3120         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
3121         Split out
3122         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
3123         Use FMAMODE_NOVF512 mode iterator.
3124         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
3125         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
3126         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
3127         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
3128         Split out
3129         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
3130         Use VF_128_256 mode iterator.
3131         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
3132         Ditto.
3133
3134 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
3135
3136         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
3137         static chain if needed.
3138
3139 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
3140
3141         PR target/60697
3142         * lra-constraints.c (index_part_to_reg): New.
3143         (process_address): Use it.
3144
3145 2014-03-27  Jeff Law  <law@redhat.com>
3146             Jakub Jelinek  <jakub@redhat.com>
3147
3148         PR target/60648
3149         * expr.c (do_tablejump): Use simplify_gen_binary rather than
3150         gen_rtx_{PLUS,MULT} to build up the address expression.
3151
3152         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
3153         creating non-canonical RTL.
3154
3155 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
3156
3157         PR ipa/60243
3158         * ipa-inline.c (want_inline_small_function_p): Short circuit large
3159         functions; reorganize to make cheap checks first.
3160         (inline_small_functions): Do not estimate growth when dumping;
3161         it is expensive.
3162         * ipa-inline.h (inline_summary): Add min_size.
3163         (growth_likely_positive): New function.
3164         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
3165         (set_cond_stmt_execution_predicate): Cleanup.
3166         (estimate_edge_size_and_time): Compute min_size.
3167         (estimate_calls_size_and_time): Likewise.
3168         (estimate_node_size_and_time): Likewise.
3169         (inline_update_overall_summary): Update min_size.
3170         (do_estimate_edge_time): Likewise.
3171         (do_estimate_edge_size): Update.
3172         (do_estimate_edge_hints): Update.
3173         (growth_likely_positive): New function.
3174
3175 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
3176
3177         PR target/60693
3178         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
3179         also if addr has VOIDmode.
3180
3181 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3182
3183         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
3184         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
3185         Declare extern.
3186         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
3187         instructions as well as AdvancedSIMD loads.
3188
3189 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3190
3191         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
3192         Use crypto_aese type.
3193         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
3194         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
3195         crypto_aese, crypto_aesmc.  Move to types.md.
3196         * config/arm/types.md (crypto_aes): Split into crypto_aese,
3197         crypto_aesmc.
3198         * config/arm/iterators.md (crypto_type): Likewise.
3199
3200 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
3201
3202         * cgraph.c: Include expr.h and tree-dfa.h.
3203         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
3204         remove LHS.
3205
3206 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
3207
3208         PR target/60675
3209         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
3210         regs from checking multi-reg pseudos.
3211
3212 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3213
3214         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
3215
3216 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3217
3218         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
3219         if it would clobber the stack pointer, even temporarily.
3220
3221 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
3222
3223         * mode-switching.c: Make small adjustments to the top comment.
3224
3225 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3226
3227         * config/rs6000/constraints.md (wD constraint): New constraint to
3228         match the constant integer to get the top DImode/DFmode out of a
3229         vector in a VSX register.
3230
3231         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
3232         match the constant integer to get the top DImode/DFmode out of a
3233         vector in a VSX register.
3234
3235         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
3236         for ISA 2.07.
3237
3238         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3239         vbpermq builtins.
3240
3241         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
3242         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
3243
3244         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
3245         Optimize vec_extract of 64-bit values, where the value being
3246         extracted is in the top word, where we can use scalar
3247         instructions.  Add direct move and store support.  Combine the big
3248         endian/little endian vector select load support into a single insn.
3249         (vsx_extract_<mode>_internal1): Likewise.
3250         (vsx_extract_<mode>_internal2): Likewise.
3251         (vsx_extract_<mode>_load): Likewise.
3252         (vsx_extract_<mode>_store): Likewise.
3253         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
3254         combined into vsx_extract_<mode>_load.
3255         (vsx_extract_<mode>_one_le): Likewise.
3256
3257         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
3258         define the top 64-bit vector element.
3259
3260         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
3261         constraint.
3262
3263         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3264         Document vec_vbpermq builtin.
3265
3266         PR target/60672
3267         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
3268         enable use of xxsldwi and xxpermdi builtin functions.
3269         (vec_xxpermdi): Likewise.
3270
3271         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3272         Document use of vec_xxsldwi and vec_xxpermdi builtins.
3273
3274 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
3275
3276         PR rtl-optimization/60650
3277         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
3278         first_p.  Use it.
3279         (find_spills_for): New.
3280         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
3281         Spill all pseudos on the second iteration.
3282
3283 2014-03-27  Marek Polacek  <polacek@redhat.com>
3284
3285         PR c/50347
3286         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
3287         types.
3288
3289 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3290
3291         * config/s390/s390.c (s390_can_use_return_insn): Check for
3292         call-saved FPRs on 31 bit.
3293
3294 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
3295
3296         PR middle-end/60682
3297         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
3298         if they need regimplification, just drop them instead of
3299         calling gimple_regimplify_operands on them.
3300
3301 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3302
3303         PR target/60580
3304         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
3305         (aarch64_frame_pointer_required): Adjust logic.
3306         (aarch64_can_eliminate): Adjust logic.
3307         (aarch64_override_options_after_change): Adjust logic.
3308
3309 2014-03-27  Dehao Chen  <dehao@google.com>
3310
3311         * ipa-inline.c (early_inliner): Update node's inline info.
3312
3313 2014-03-26  Dehao Chen  <dehao@google.com>
3314
3315         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
3316         compiler inserted conditional jumps for NAN float check.
3317
3318 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
3319
3320         * ubsan.h (ubsan_create_data): Change second argument's type
3321         to const location_t *.
3322         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
3323         _("<unknown>").
3324         (ubsan_create_data): Change second argument to const location_t *PLOC.
3325         Create Loc field whenever PLOC is non-NULL.
3326         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
3327         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
3328         callers.
3329
3330         PR other/59545
3331         * real.c (real_to_integer2): Change type of low to UHWI.
3332
3333 2014-03-26  Tobias Burnus  <burnus@net-b.de>
3334
3335         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
3336         (CILK_SELF_SPECS): New define.
3337         (driver_self_specs): Use it.
3338
3339 2014-03-26  Richard Biener  <rguenther@suse.de>
3340
3341         * tree-pretty-print.c (percent_K_format): Implement special
3342         case for LTO and its stripped down BLOCK tree.
3343
3344 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
3345
3346         PR sanitizer/60636
3347         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
3348
3349         * tree-vrp.c (simplify_internal_call_using_ranges): If only
3350         one range is range_int_cst_p, but not both, at least optimize
3351         addition/subtraction of 0 and multiplication by 0 or 1.
3352         * gimple-fold.c (gimple_fold_call): Fold
3353         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
3354         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
3355         INTEGER_CSTs, try to fold at least x * 0 and y - y.
3356
3357 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
3358
3359         PR rtl-optimization/60452
3360         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
3361         <case REG>: Return 1 for invalid offsets from the frame pointer.
3362
3363 2014-03-26  Marek Polacek  <polacek@redhat.com>
3364
3365         PR c/37428
3366         * doc/extend.texi (C Extensions): Mention variable-length arrays in
3367         a structure/union.
3368
3369 2014-03-26  Marek Polacek  <polacek@redhat.com>
3370
3371         PR c/39525
3372         * doc/extend.texi (Designated Inits): Describe what happens to omitted
3373         field members.
3374
3375 2014-03-26  Marek Polacek  <polacek@redhat.com>
3376
3377         PR other/59545
3378         * ira-color.c (update_conflict_hard_regno_costs): Perform the
3379         multiplication in unsigned type.
3380
3381 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
3382
3383         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
3384
3385 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
3386
3387         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
3388
3389 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
3390
3391         PR ipa/60315
3392         * cif-code.def (UNREACHABLE) New code.
3393         * ipa-inline.c (inline_small_functions): Skip edges to
3394         __builtlin_unreachable.
3395         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
3396         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
3397         predicate to __bulitin_unreachable.
3398         (set_cond_stmt_execution_predicate): Fix issue when
3399         invert_tree_comparison returns ERROR_MARK.
3400         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
3401         propagate to inline clones.
3402         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
3403         to unreachable.
3404         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
3405         * cgraphclones.c (cgraph_clone_node): If call destination is already
3406         ureachable, do not redirect it back.
3407         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
3408         unreachable.
3409
3410 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
3411
3412         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
3413         Do not modify inline clones.
3414
3415 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3416
3417         * config/i386/i386.md (general_sext_operand): New mode attr.
3418         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
3419         don't generate (sign_extend (const_int)).
3420         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
3421         operands[2].  Use We constraint instead of <i> and
3422         <general_sext_operand> predicate instead of <general_operand>.
3423         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
3424         * config/i386/constraints.md (We): New constraint.
3425         * config/i386/predicates.md (x86_64_sext_operand,
3426         sext_operand): New predicates.
3427
3428 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3429
3430         PR ipa/60600
3431         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
3432         inconsistent devirtualizations to __builtin_unreachable.
3433
3434 2014-03-25  Marek Polacek  <polacek@redhat.com>
3435
3436         PR c/35449
3437         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
3438
3439 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
3440
3441         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
3442         order of elements for big-endian.
3443
3444 2014-03-25  Richard Biener  <rguenther@suse.de>
3445
3446         PR middle-end/60635
3447         * gimplify-me.c (gimple_regimplify_operands): Update the
3448         re-gimplifed stmt.
3449
3450 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3451
3452         PR ipa/59176
3453         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
3454         (lto_output_varpool_node): Likewise.
3455         (input_overwrite_node): Likewise.
3456         (input_varpool_node): Likewise.
3457
3458 2014-03-25  Richard Biener  <rguenther@suse.de>
3459
3460         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
3461         (run_gcc): Likewise.
3462
3463 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3464
3465         * combine.c (simplify_compare_const): Add MODE argument.
3466         Handle mode_width 0 as very large mode_width.
3467         (try_combine, simplify_comparison): Adjust callers.
3468
3469         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
3470         type to avoid signed integer overflow.
3471         * explow.c (plus_constant): Likewise.
3472
3473 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3474
3475         * doc/generic.texi: Correct typos.
3476
3477 2014-03-24  Tobias Burnus  <burnus@net-b.de>
3478
3479         * doc/invoke.texi (-flto): Expand section about
3480         using static libraries with LTO.
3481
3482 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3483
3484         PR rtl-optimization/60501
3485         * optabs.def (addptr3_optab): New optab.
3486         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
3487         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
3488         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
3489
3490         * lra.c (emit_add3_insn): Use the addptr pattern if available.
3491
3492         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
3493
3494 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
3495
3496         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
3497         _mm512_set1_pd.
3498
3499         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
3500         (_mm256_undefined_ps): Define.
3501         (_mm256_undefined_pd): Define.
3502         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
3503         (_mm_undefined_pd): Define.
3504         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
3505         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
3506         (_mm512_undefined_ps): Define.
3507         (_mm512_undefined_pd): Define.
3508         Use _mm*_undefined_*.
3509         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
3510
3511 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
3512
3513         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
3514         (lshr_simd): DI mode added.
3515         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
3516         (aarch64_ushr_simddi): Likewise.
3517         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
3518         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
3519         (vshrd_n_u64): Likewise.
3520
3521 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3522
3523         * Makefile.in (s-macro_list): Depend on cc1.
3524
3525 2014-03-23  Teresa Johnson  <tejohnson@google.com>
3526
3527         * ipa-utils.c (ipa_print_order): Use specified dump file.
3528
3529 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
3530
3531         PR rtl-optimization/60601
3532         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
3533
3534         * gcc.c (eval_spec_function): Initialize save_growing_value.
3535
3536 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
3537
3538         PR sanitizer/60613
3539         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
3540         code == MINUS_EXPR, never swap op0 with op1.
3541
3542         * toplev.c (init_local_tick): Avoid signed integer multiplication
3543         overflow.
3544         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
3545         shift by first operand's bitsize.
3546
3547 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
3548
3549         PR target/60610
3550         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
3551         redefine to 1 or 0.
3552         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
3553         TARGET_ISA_64BIT_P(x).
3554
3555 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3556
3557         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
3558         pattern for vector nor instead of subtract from splat(-1).
3559         (altivec_expand_vec_perm_const_le): Likewise.
3560
3561 2014-03-21  Richard Henderson  <rth@twiddle.net>
3562
3563         PR target/60598
3564         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
3565         related insns after epilogue_completed.
3566
3567 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3568
3569         PR ipa/59176
3570         * cgraph.h (symtab_node): New flag body_removed.
3571         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
3572         when removing bodies.
3573         * symtab.c (dump_symtab_base): Dump body_removed flag.
3574         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
3575         had their bodies removed.
3576
3577 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3578
3579         PR ipa/60419
3580         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
3581         in the border.
3582
3583 2014-03-21  Richard Biener  <rguenther@suse.de>
3584
3585         PR tree-optimization/60577
3586         * tree-core.h (struct tree_base): Document nothrow_flag use
3587         in VAR_DECL_NONALIASED.
3588         * tree.h (VAR_DECL_NONALIASED): New.
3589         (may_be_aliased): Adjust.
3590         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
3591
3592 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3593
3594         * expr.c (expand_expr_real_1): Remove outdated comment.
3595
3596 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
3597
3598         PR middle-end/60597
3599         * ira.c (adjust_cleared_regs): Call copy_rtx on
3600         *reg_equiv[REGNO (loc)].src_p before passing it to
3601         simplify_replace_fn_rtx.
3602
3603         PR target/60568
3604         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
3605         into CONST, put pic register as first operand of PLUS.  Use
3606         gen_const_mem for both 32-bit and 64-bit PIC got loads.
3607
3608 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3609
3610         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
3611
3612 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3613
3614         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
3615         around for store forwarding issue in the FPU on the UT699.
3616         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
3617         loads and operations if -mfix-ut699 is specified.
3618         (divtf3_hq): Tweak attribute.
3619         (sqrttf2_hq): Likewise.
3620
3621 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3622
3623         * calls.c (store_one_arg): Remove incorrect const qualification on the
3624         type of the temporary.
3625         * cfgexpand.c (expand_return): Likewise.
3626         * expr.c (expand_constructor): Likewise.
3627         (expand_expr_real_1): Likewise.
3628
3629 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3630
3631         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
3632         of parts.
3633
3634 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
3635
3636         PR target/60039
3637         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
3638
3639 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
3640
3641         * config/arm/aarch-common-protos.h
3642         (alu_cost_table): Fix spelling of "extend".
3643         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
3644
3645 2014-03-19  Richard Biener  <rguenther@suse.de>
3646
3647         PR middle-end/60553
3648         * tree-core.h (tree_type_common): Re-order pointer members
3649         to reduce recursion depth during GC walks.
3650
3651 2014-03-19  Marek Polacek  <polacek@redhat.com>
3652
3653         PR sanitizer/60569
3654         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
3655         before accessing it.
3656
3657 2014-03-19  Richard Biener  <rguenther@suse.de>
3658
3659         PR lto/59543
3660         * lto-streamer-in.c (input_function): In WPA stage do not drop
3661         debug stmts.
3662
3663 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
3664
3665         PR tree-optimization/60559
3666         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
3667         with build_zero_cst assignment.
3668
3669 2014-03-18  Kai Tietz  <ktietz@redhat.com>
3670
3671         PR rtl-optimization/56356
3672         * sdbout.c (sdbout_parms): Verify that parms'
3673         incoming argument is valid.
3674         (sdbout_reg_parms): Likewise.
3675
3676 2014-03-18  Richard Henderson  <rth@redhat.com>
3677
3678         PR target/60562
3679         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
3680         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
3681         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
3682
3683 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
3684
3685         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
3686         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
3687         Italicize plugin event names in description.  Explain that
3688         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
3689         Remind that no GCC functions should be called after PLUGIN_FINISH.
3690         Explain what pragmas with expansion are.
3691
3692 2014-03-18  Martin Liska  <mliska@suse.cz>
3693
3694         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
3695         gimple call statement is update.
3696         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
3697         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
3698
3699 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3700
3701         PR sanitizer/60557
3702         * ubsan.c (ubsan_instrument_unreachable): Call
3703         initialize_sanitizer_builtins.
3704         (ubsan_pass): Likewise.
3705
3706         PR sanitizer/60535
3707         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
3708         varpool_finalize_decl instead of rest_of_decl_compilation.
3709
3710 2014-03-18  Richard Biener  <rguenther@suse.de>
3711
3712         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
3713         by using bitmap_and_compl instead of bitmap_and_compl_into.
3714         (df_rd_transfer_function): Likewise.
3715
3716 2014-03-18  Richard Biener  <rguenther@suse.de>
3717
3718         * doc/lto.texi (fresolution): Fix typo.
3719
3720 2014-03-18  Richard Biener  <rguenther@suse.de>
3721
3722         * doc/invoke.texi (flto): Update for changes in 4.9.
3723
3724 2014-03-18  Richard Biener  <rguenther@suse.de>
3725
3726         * doc/loop.texi: Remove section on the removed lambda framework.
3727         Update loop docs with recent changes in preserving loop structure.
3728
3729 2014-03-18  Richard Biener  <rguenther@suse.de>
3730
3731         * doc/lto.texi (-fresolution): Document.
3732
3733 2014-03-18  Richard Biener  <rguenther@suse.de>
3734
3735         * doc/contrib.texi: Adjust my name.
3736
3737 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3738
3739         PR ipa/58721
3740         * internal-fn.c: Include diagnostic-core.h.
3741         (expand_BUILTIN_EXPECT): New function.
3742         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
3743         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
3744         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
3745         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
3746         IFN_BUILTIN_EXPECT.
3747         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
3748         Revert 3 argument __builtin_expect code.
3749         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
3750         * gimple-fold.c (gimple_fold_call): Likewise.
3751         * tree.h (fold_builtin_expect): New prototype.
3752         * builtins.c (build_builtin_expect_predicate): Add predictor
3753         argument, if non-NULL, create 3 argument __builtin_expect.
3754         (fold_builtin_expect): No longer static.  Add ARG2 argument,
3755         pass it through to build_builtin_expect_predicate.
3756         (fold_builtin_2): Adjust caller.
3757         (fold_builtin_3): Handle BUILT_IN_EXPECT.
3758         * internal-fn.def (BUILTIN_EXPECT): New.
3759
3760 2014-03-18  Tobias Burnus  <burnus@net-b.de>
3761
3762         PR ipa/58721
3763         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
3764         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
3765         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
3766
3767 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
3768
3769         PR ipa/58721
3770         * predict.c (combine_predictions_for_bb): Fix up formatting.
3771         (expr_expected_value_1, expr_expected_value): Add predictor argument,
3772         fill what it points to if non-NULL.
3773         (tree_predict_by_opcode): Adjust caller, use the predictor.
3774         * predict.def (PRED_COMPARE_AND_SWAP): Add.
3775
3776 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
3777
3778         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
3779         proper constant for the store mode.
3780
3781 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
3782
3783         * symtab.c (change_decl_assembler_name): Fix transparent alias
3784         chain construction.
3785
3786 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
3787
3788         * config/aarch64/aarch64.c: Correct the comments about the
3789         aarch64 stack layout.
3790
3791 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
3792
3793         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
3794         check for GF_OMP_FOR_KIND_FOR.
3795
3796 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
3797
3798         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
3799         ymm and zmm register names.
3800
3801 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
3802
3803         PR target/60516
3804         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
3805         note creation for the 2010-08-31 changes.
3806
3807 2014-03-17  Marek Polacek  <polacek@redhat.com>
3808
3809         PR middle-end/60534
3810         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
3811         as -fno-tree-loop-vectorize.
3812         (expand_omp_simd): Likewise.
3813
3814 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
3815
3816         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
3817         (eligible_for_call_delay): New prototype.
3818         * config/sparc/sparc.c (tls_call_delay): Rename into...
3819         (eligible_for_call_delay): ...this.  Return false if the instruction
3820         cannot be put in the delay slot of a branch.
3821         (eligible_for_restore_insn): Simplify.
3822         (eligible_for_return_delay): Return false if the instruction cannot be
3823         put in the delay slot of a branch and simplify.
3824         (eligible_for_sibcall_delay): Return false if the instruction cannot be
3825         put in the delay slot of a branch.
3826         * config/sparc/sparc.md (fix_ut699): New attribute.
3827         (tls_call_delay): Delete.
3828         (in_call_delay): Reimplement.
3829         (eligible_for_sibcall_delay): Rename into...
3830         (in_sibcall_delay): ...this.
3831         (eligible_for_return_delay): Rename into...
3832         (in_return_delay): ...this.
3833         (in_branch_delay): Reimplement.
3834         (in_uncond_branch_delay): Delete.
3835         (in_annul_branch_delay): Delete.
3836
3837 2014-03-14  Richard Henderson  <rth@redhat.com>
3838
3839         PR target/60525
3840         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
3841         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
3842         (*floathi<X87MODEF>2_i387_with_temp): Remove.
3843         (floathi splitters): Remove.
3844         (float<SWI48x>xf2): New pattern.
3845         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
3846         code that tried to handle DImode for 32-bit, but which was excluded
3847         by the pattern's condition.  Drop allocation of stack temporary.
3848         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
3849         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
3850         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
3851         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
3852         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
3853         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
3854         (*float<SWI48><MODEF>2_sse_interunit): Remove.
3855         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
3856         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
3857         (*float<SWI48x><X87MODEF>2_i387): Remove.
3858         (all float _with_temp splitters): Remove.
3859         (*float<SWI48x><MODEF>2_i387): New pattern.
3860         (*float<SWI48><MODEF>2_sse): New pattern.
3861         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
3862         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
3863
3864 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
3865             Marek Polacek  <polacek@redhat.com>
3866
3867         PR middle-end/60484
3868         * common.opt (dump_base_name_prefixed): New Variable.
3869         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
3870         if x_dump_base_name_prefixed is already set, set it at the end.
3871
3872 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
3873
3874         PR rtl-optimization/60508
3875         * lra-constraints.c (get_reload_reg): Add new parameter
3876         in_subreg_p.
3877         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
3878         Pass the new parameter values.
3879
3880 2014-03-14  Richard Biener  <rguenther@suse.de>
3881
3882         * common.opt: Revert unintented changes from r205065.
3883         * opts.c: Likewise.
3884
3885 2014-03-14  Richard Biener  <rguenther@suse.de>
3886
3887         PR middle-end/60518
3888         * cfghooks.c (split_block): Properly adjust all loops the
3889         block was a latch of.
3890
3891 2014-03-14  Martin Jambor  <mjambor@suse.cz>
3892
3893         PR lto/60461
3894         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
3895         and simplify it.
3896
3897 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
3898
3899         PR target/59396
3900         * config/avr/avr.c (avr_set_current_function): Pass function name
3901         through default_strip_name_encoding before sanity checking instead
3902         of skipping the first char of the assembler name.
3903
3904 2014-03-13  Richard Henderson  <rth@redhat.com>
3905
3906         PR debug/60438
3907         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
3908         (ix86_force_to_memory, ix86_free_from_memory): Remove.
3909         * config/i386/i386-protos.h: Likewise.
3910         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
3911         in the expander instead of a splitter.
3912         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
3913         any possibility of requiring a memory.
3914         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
3915         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
3916         (fp branch splitters): Update for ix86_split_fp_branch.
3917         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
3918         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
3919         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
3920         (*fop_<MODEF>_2_i387): Remove f/r alternative.
3921         (*fop_<MODEF>_3_i387): Likewise.
3922         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
3923         (splitters for the fop_* register patterns): Remove.
3924         (fscalexf4_i387): Rename from *fscalexf4_i387.
3925         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
3926
3927 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3928
3929         PR tree-optimization/59779
3930         * tree-dfa.c (get_ref_base_and_extent): Use double_int
3931         type for bitsize and maxsize instead of HOST_WIDE_INT.
3932
3933 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
3934
3935         PR rtl-optimization/57320
3936         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
3937         the CFG after thread_prologue_and_epilogue_insns.
3938
3939 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
3940
3941         PR rtl-optimization/57189
3942         * lra-constraints.c (process_alt_operands): Disfavor spilling
3943         vector pseudos.
3944
3945 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
3946
3947         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
3948
3949 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3950
3951         PR tree-optimization/59025
3952         PR middle-end/60418
3953         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
3954         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
3955
3956 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
3957
3958         PR target/60486
3959         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
3960         calls of avr_out_plus_1.
3961
3962 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
3963
3964         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
3965         BB's single pred and update the father loop's latch info later.
3966
3967 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
3968
3969         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
3970         (VEC_M): Likewise.
3971         (VEC_N): Likewise.
3972         (VEC_R): Likewise.
3973         (VEC_base): Likewise.
3974         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
3975         registers, we need to swap double words in little endian mode.
3976
3977         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
3978         to be a container mode for 128-bit integer operations added in ISA
3979         2.07.  Unlike TImode and PTImode, the preferred register set is
3980         the Altivec/VMX registers for the 128-bit operations.
3981
3982         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
3983         declarations.
3984         (rs6000_split_128bit_ok_p): Likewise.
3985
3986         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
3987         macros for creating ISA 2.07 normal and overloaded builtin
3988         functions with 3 arguments.
3989         (BU_P8V_OVERLOAD_3): Likewise.
3990         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
3991         for use as overloaded functions.
3992         (VPERM_1TI_UNS): Likewise.
3993         (VSEL_1TI): Likewise.
3994         (VSEL_1TI_UNS): Likewise.
3995         (ST_INTERNAL_1ti): Likewise.
3996         (LD_INTERNAL_1ti): Likewise.
3997         (XXSEL_1TI): Likewise.
3998         (XXSEL_1TI_UNS): Likewise.
3999         (VPERM_1TI): Likewise.
4000         (VPERM_1TI_UNS): Likewise.
4001         (XXPERMDI_1TI): Likewise.
4002         (SET_1TI): Likewise.
4003         (LXVD2X_V1TI): Likewise.
4004         (STXVD2X_V1TI): Likewise.
4005         (VEC_INIT_V1TI): Likewise.
4006         (VEC_SET_V1TI): Likewise.
4007         (VEC_EXT_V1TI): Likewise.
4008         (EQV_V1TI): Likewise.
4009         (NAND_V1TI): Likewise.
4010         (ORC_V1TI): Likewise.
4011         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
4012         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
4013         overloaded builtin.
4014         (VADDUQM): Likewise.
4015         (VSUBCUQ): Likewise.
4016         (VADDEUQM): Likewise.
4017         (VADDECUQ): Likewise.
4018         (VSUBEUQM): Likewise.
4019         (VSUBECUQ): Likewise.
4020
4021         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
4022         __int128_t and __uint128_t types.
4023         (__uint128_type): Likewise.
4024         (altivec_categorize_keyword): Add support for vector __int128_t,
4025         vector __uint128_t, vector __int128, and vector unsigned __int128
4026         as a container type for TImode operations that need to be done in
4027         VSX/Altivec registers.
4028         (rs6000_macro_to_expand): Likewise.
4029         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
4030         to support 128-bit integer instructions vaddcuq, vadduqm,
4031         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
4032         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
4033
4034         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
4035         for V1TImode, and set up preferences to use VSX/Altivec registers.
4036         Setup VSX reload handlers.
4037         (rs6000_debug_reg_global): Likewise.
4038         (rs6000_init_hard_regno_mode_ok): Likewise.
4039         (rs6000_preferred_simd_mode): Likewise.
4040         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
4041         (easy_altivec_constant): Likewise.
4042         (output_vec_const_move): Likewise.
4043         (rs6000_expand_vector_set): Convert V1TImode set and extract to
4044         simple move.
4045         (rs6000_expand_vector_extract): Likewise.
4046         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
4047         addressing.
4048         (rs6000_const_vec): Add support for V1TImode.
4049         (rs6000_emit_le_vsx_load): Swap double words when loading or
4050         storing TImode/V1TImode.
4051         (rs6000_emit_le_vsx_store): Likewise.
4052         (rs6000_emit_le_vsx_move): Likewise.
4053         (rs6000_emit_move): Add support for V1TImode.
4054         (altivec_expand_ld_builtin): Likewise.
4055         (altivec_expand_st_builtin): Likewise.
4056         (altivec_expand_vec_init_builtin): Likewise.
4057         (altivec_expand_builtin): Likewise.
4058         (rs6000_init_builtins): Add support for V1TImode type.  Add
4059         support for ISA 2.07 128-bit integer builtins.  Define type names
4060         for the VSX/Altivec vector types.
4061         (altivec_init_builtins): Add support for overloaded vector
4062         functions with V1TImode type.
4063         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
4064         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
4065         external function.
4066         (rs6000_split_128bit_ok_p): Likewise.
4067         (rs6000_handle_altivec_attribute): Create V1TImode from vector
4068         __int128_t and vector __uint128_t.
4069
4070         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
4071         and mode attributes.
4072         (VSX_M): Likewise.
4073         (VSX_M2): Likewise.
4074         (VSm): Likewise.
4075         (VSs): Likewise.
4076         (VSr): Likewise.
4077         (VSv): Likewise.
4078         (VS_scalar): Likewise.
4079         (VS_double): Likewise.
4080         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
4081
4082         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
4083         we support the ISA 2.07 128-bit integer arithmetic instructions.
4084         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
4085         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
4086         and TImode types for use with the builtin functions.
4087         (V1TI_type_node): Likewise.
4088         (unsigned_V1TI_type_node): Likewise.
4089         (intTI_type_internal_node): Likewise.
4090         (uintTI_type_internal_node): Likewise.
4091
4092         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
4093         128-bit builtin functions.
4094         (UNSPEC_VADDEUQM): Likewise.
4095         (UNSPEC_VADDECUQ): Likewise.
4096         (UNSPEC_VSUBCUQ): Likewise.
4097         (UNSPEC_VSUBEUQM): Likewise.
4098         (UNSPEC_VSUBECUQ): Likewise.
4099         (VM): Add V1TImode to vector mode iterators.
4100         (VM2): Likewise.
4101         (VI_unit): Likewise.
4102         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
4103         (altivec_vaddcuq): Likewise.
4104         (altivec_vsubuqm): Likewise.
4105         (altivec_vsubcuq): Likewise.
4106         (altivec_vaddeuqm): Likewise.
4107         (altivec_vaddecuq): Likewise.
4108         (altivec_vsubeuqm): Likewise.
4109         (altivec_vsubecuq): Likewise.
4110
4111         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
4112         mode iterators.
4113         (BOOL_128): Likewise.
4114         (BOOL_REGS_OUTPUT): Likewise.
4115         (BOOL_REGS_OP1): Likewise.
4116         (BOOL_REGS_OP2): Likewise.
4117         (BOOL_REGS_UNARY): Likewise.
4118         (BOOL_REGS_AND_CR0): Likewise.
4119
4120         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
4121         128-bit integer builtin support.
4122         (vec_vadduqm): Likewise.
4123         (vec_vaddecuq): Likewise.
4124         (vec_vaddeuqm): Likewise.
4125         (vec_vsubecuq): Likewise.
4126         (vec_vsubeuqm): Likewise.
4127         (vec_vsubcuq): Likewise.
4128         (vec_vsubuqm): Likewise.
4129
4130         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4131         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
4132         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
4133         128-bit integer add/subtract to ISA 2.07.
4134
4135 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
4136
4137         * config/arc/arc.c (arc_predicate_delay_insns):
4138         Fix third argument passed to conditionalize_nonjump.
4139
4140 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
4141
4142         * config/aarch64/aarch64-builtins.c
4143         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
4144         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
4145         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
4146         instead of __builtin_lfloor.
4147         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
4148
4149 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
4150
4151         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
4152         (tree_ssa_ifcombine_bb_1): New function.
4153         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
4154         is an empty forwarder block to then_bb or vice versa and then_bb
4155         and else_bb are effectively swapped.
4156
4157 2014-03-12  Christian Bruel  <christian.bruel@st.com>
4158
4159         PR target/60264
4160         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
4161         REG_CFA_DEF_CFA note.
4162         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
4163         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
4164
4165 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
4166
4167         PR tree-optimization/60454
4168         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
4169
4170 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4171
4172         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
4173         Do not define target_cpu_default2 to generic.
4174         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
4175         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
4176         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
4177
4178 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
4179             Marc Glisse  <marc.glisse@inria.fr>
4180
4181         PR tree-optimization/60502
4182         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
4183         instead of build_low_bits_mask.
4184
4185 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
4186
4187         PR middle-end/60482
4188         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
4189         if there are multiple uses, but op doesn't live on E edge.
4190         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
4191         clobber stmts before __builtin_unreachable.
4192
4193 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
4194
4195         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
4196         hard_frame_pointer_rtx.
4197         * cse.c (cse_insn): Remove volatile check.
4198         * cselib.c (cselib_process_insn): Likewise.
4199         * dse.c (scan_insn): Likewise.
4200
4201 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4202
4203         * config/arc/arc.c (conditionalize_nonjump): New function,
4204         broken out of ...
4205         (arc_ifcvt): ... this.
4206         (arc_predicate_delay_insns): Use it.
4207
4208 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4209
4210         * config/arc/predicates.md (extend_operand): During/after reload,
4211         allow const_int_operand.
4212         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
4213         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
4214         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
4215         to "i".
4216         (umulsi3_highpart_i): Likewise.
4217
4218 2014-03-11  Richard Biener  <rguenther@suse.de>
4219
4220         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
4221         Add asserts to guard possible wrong-code bugs.
4222
4223 2014-03-11  Richard Biener  <rguenther@suse.de>
4224
4225         PR tree-optimization/60429
4226         PR tree-optimization/60485
4227         * tree-ssa-structalias.c (set_union_with_increment): Properly
4228         take into account all fields that overlap the shifted vars.
4229         (do_sd_constraint): Likewise.
4230         (do_ds_constraint): Likewise.
4231         (get_constraint_for_ptr_offset): Likewise.
4232
4233 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
4234
4235         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
4236         (nios2_compute_frame_layout):
4237         Add calculation of cfun->machine->fp_save_offset.
4238         (nios2_expand_prologue): Correct setting of frame pointer register
4239         in prologue.
4240         (nios2_expand_epilogue): Update recovery of stack pointer from
4241         frame pointer accordingly.
4242         (nios2_initial_elimination_offset): Update calculation of offset
4243         for eliminating to HARD_FRAME_POINTER_REGNUM.
4244
4245 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
4246
4247         PR ipa/60457
4248         * ipa.c (symtab_remove_unreachable_nodes): Don't call
4249         cgraph_get_create_node on VAR_DECLs.
4250
4251 2014-03-10  Richard Biener  <rguenther@suse.de>
4252
4253         PR middle-end/60474
4254         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
4255
4256 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
4257
4258         * config/vms/vms.opt (vms_float_format): New variable.
4259
4260 2014-03-08  Tobias Burnus  <burnus@net-b.de>
4261
4262         * doc/invoke.texi (-fcilkplus): Update implementation status.
4263
4264 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
4265             Richard Biener  <rguenther@suse.de>
4266
4267         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
4268         consistently accross all TUs.
4269         (run_gcc): Enable -fshort-double automatically at link at link-time
4270         and disallow override.
4271
4272 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
4273
4274         PR target/58271
4275         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
4276         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
4277         if they can't be used.
4278
4279 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4280
4281         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
4282         for Solaris 11/x86 ld.
4283         * configure: Regenerate.
4284
4285 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4286
4287         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
4288         (LIB_TLS_SPEC): Save as ld_tls_libs.
4289         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
4290         (HAVE_AS_IX86_TLSLDM): New test.
4291         * configure, config.in: Regenerate.
4292         * config/i386/i386.c (legitimize_tls_address): Fall back to
4293         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
4294         cannot support TLS_MODEL_LOCAL_DYNAMIC.
4295         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
4296         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
4297
4298 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
4299
4300         * common.opt (fira-loop-pressure): Mark as optimization.
4301
4302 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
4303
4304         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
4305         an OpenMP mappable type.
4306
4307 2014-03-06  Matthias Klose  <doko@ubuntu.com>
4308
4309         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
4310         MULTILIB_OSDIRNAMES is not defined.
4311
4312 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
4313             Meador Inge  <meadori@codesourcery.com>
4314
4315         PR target/58595
4316         * config/arm/arm.c (arm_tls_symbol_p): Remove.
4317         (arm_legitimize_address): Call legitimize_tls_address for any
4318         arm_tls_referenced_p expression, handle constant addend.  Call it
4319         before testing for !TARGET_ARM.
4320         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
4321
4322 2014-03-06  Richard Biener  <rguenther@suse.de>
4323
4324         PR middle-end/60445
4325         PR lto/60424
4326         PR lto/60427
4327         Revert
4328         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
4329
4330         * tree-streamer.c (record_common_node): Assert we don't record
4331         nodes with type double.
4332         (preload_common_node): Skip type double, complex double and double
4333         pointer since it is now frontend dependent due to fshort-double option.
4334
4335 2014-03-06  Richard Biener  <rguenther@suse.de>
4336
4337         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
4338         or -fno-lto is specified and the linker has full plugin support.
4339         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
4340         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
4341         * lto-wrapper.c (merge_and_complain): Merge compile-time
4342         optimization levels.
4343         (run_gcc): And pass it through to the link options.
4344
4345 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
4346
4347         PR debug/60381
4348         Revert:
4349         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4350         PR debug/59992
4351         * cselib.c (remove_useless_values): Skip to avoid quadratic
4352         behavior if the condition moved from...
4353         (cselib_process_insn): ... here holds.
4354
4355 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
4356
4357         PR plugins/59335
4358         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
4359         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
4360
4361         PR plugins/59335
4362         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
4363         (TM_H): Add x86-tune.def.
4364
4365 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4366
4367         * config/aarch64/aarch64.c (generic_tunings):
4368         Use cortexa57_extra_costs.
4369
4370 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
4371
4372         PR lto/60404
4373         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
4374         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
4375         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
4376         cost for in_lto_p.
4377
4378 2014-03-04  Heiher  <r@hev.cc>
4379
4380         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
4381         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
4382
4383 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
4384
4385         * config/i386/predicates.md (const2356_operand): Change to ...
4386         (const2367_operand): ... this.
4387         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
4388         const2367_operand.
4389         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4390         (*avx512pf_scatterpf<mode>sf): Ditto.
4391         (avx512pf_scatterpf<mode>df): Ditto.
4392         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4393         (*avx512pf_scatterpf<mode>df): Ditto.
4394         * config/i386/i386.c (ix86_expand_builtin): Update
4395         incorrect hint operand error message.
4396
4397 2014-03-04  Richard Biener  <rguenther@suse.de>
4398
4399         * lto-section-in.c (lto_get_section_data): Fix const cast.
4400
4401 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
4402
4403         * tree-streamer.c (record_common_node): Assert we don't record
4404         nodes with type double.
4405         (preload_common_node): Skip type double, complex double and double
4406         pointer since it is now frontend dependent due to fshort-double option.
4407
4408 2014-03-04  Richard Biener  <rguenther@suse.de>
4409
4410         PR lto/60405
4411         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
4412         (lto_input_toplevel_asms): Likewise.
4413         * lto-section-in.c (lto_get_section_data): Instead do it here
4414         for every section.
4415
4416 2014-03-04  Richard Biener  <rguenther@suse.de>
4417
4418         PR tree-optimization/60382
4419         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
4420         dead PHIs a reduction.
4421
4422 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
4423
4424         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
4425         hint value.
4426         (_mm_prefetch): Move out of GCC target("sse") pragma.
4427         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
4428         GCC target("prfchw") pragma.
4429         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
4430         for locality <= 2.
4431         * config/i386/i386.c (ix86_option_override_internal): Enable
4432         -mprfchw with -mprefetchwt1.
4433
4434 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4435
4436         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
4437         Mark as varying.
4438
4439 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4440
4441         * opts.h (CL_PCH_IGNORE): Define.
4442         * targhooks.c (option_affects_pch_p):
4443         Return false for options that have CL_PCH_IGNORE set.
4444         * opt-functions.awk: Process PchIgnore.
4445         * doc/options.texi: Document PchIgnore.
4446
4447         * config/arc/arc.opt (misize): Add PchIgnore property.
4448
4449 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4450
4451         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
4452         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
4453         constraint on constants to permit them being loaded into
4454         GENERAL_REGS or BASE_REGS.
4455
4456 2014-03-03  Nick Clifton  <nickc@redhat.com>
4457
4458         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
4459         anti-cacnonical alternatives.
4460         (negandhi3_real): New pattern.
4461         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
4462
4463 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
4464
4465         * config/avr/avr-mcus.def: Remove atxmega16x1.
4466         * config/avr/avr-tables.opt: Regenerate.
4467         * config/avr/t-multilib: Regenerate.
4468         * doc/avr-mmcu.texi: Regenerate.
4469
4470 2014-03-03  Tobias Grosser  <tobias@grosser.es>
4471             Mircea Namolaru  <mircea.namolaru@inria.fr>
4472
4473         PR tree-optimization/58028
4474         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
4475         scalar dimensions.
4476
4477 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4478
4479         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
4480         not handled by recognizers.
4481
4482 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
4483
4484         PR middle-end/60175
4485         * function.c (expand_function_end): Don't emit
4486         clobber_return_register sequence if clobber_after is a BARRIER.
4487         * cfgexpand.c (construct_exit_block): Append instructions before
4488         return_label to prev_bb.
4489
4490 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4491
4492         * config/rs6000/constraints.md: Document reserved use of "wc".
4493
4494 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4495
4496         PR ipa/60150
4497         * ipa.c (function_and_variable_visibility): When dissolving comdat
4498         group, also set all symbols to local.
4499
4500 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4501
4502         PR ipa/60306
4503
4504         Revert:
4505         2013-12-14   Jan Hubicka  <jh@suse.cz>
4506         PR middle-end/58477
4507         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
4508
4509 2014-03-02  Jon Beniston  <jon@beniston.com>
4510
4511         PR bootstrap/48230
4512         PR bootstrap/50927
4513         PR bootstrap/52466
4514         PR target/46898
4515         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
4516           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
4517         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
4518         (simple_return, *simple_return): New patterns
4519         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
4520         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
4521
4522 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
4523
4524         * dwarf2out.c (gen_subprogram_die): Tidy.
4525
4526 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
4527
4528         PR target/60071
4529         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
4530         (*mov_t_msb_neg_negc): ... this new insn.
4531
4532 2014-02-28  Jason Merrill  <jason@redhat.com>
4533
4534         PR c++/58678
4535         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
4536         function.
4537
4538 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
4539
4540         PR c++/60314
4541         * dwarf2out.c (decltype_auto_die): New static.
4542         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
4543         (gen_type_die_with_usage): Handle 'decltype(auto)'.
4544         (is_cxx_auto): Likewise.
4545
4546 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
4547
4548         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
4549         we are not using general regs only.
4550
4551 2014-02-28  Richard Biener  <rguenther@suse.de>
4552
4553         PR target/60280
4554         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
4555         previous fix and only allow to remove trivial pre-headers
4556         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
4557         (remove_forwarder_block): Properly update the latch of a loop.
4558
4559 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4560
4561         PR debug/59992
4562         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
4563         (cselib_preserved_hash_table): New.
4564         (preserve_constants_and_equivs): Move preserved vals to it.
4565         (cselib_find_slot): Look it up first.
4566         (cselib_init): Initialize it.
4567         (cselib_finish): Release it.
4568         (dump_cselib_table): Dump it.
4569
4570 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4571
4572         PR debug/59992
4573         * cselib.c (remove_useless_values): Skip to avoid quadratic
4574         behavior if the condition moved from...
4575         (cselib_process_insn): ... here holds.
4576
4577 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4578
4579         PR debug/57232
4580         * var-tracking.c (vt_initialize): Apply the same condition to
4581         preserve the CFA base value.
4582
4583 2014-02-28  Joey Ye  <joey.ye@arm.com>
4584
4585         PR target/PR60169
4586         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
4587         if reload in progress or completed.
4588
4589 2014-02-28  Tobias Burnus  <burnus@net-b.de>
4590
4591         PR middle-end/60147
4592         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
4593         NAMELIST_DECL.
4594
4595 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
4596
4597         * doc/tm.texi.in (Condition Code Status): Update documention for
4598         relative locations of cc0-setter and cc0-user.
4599
4600 2014-02-27  Jeff Law  <law@redhat.com>
4601
4602         PR rtl-optimization/52714
4603         * combine.c (try_combine): When splitting an unrecognized PARALLEL
4604         into two independent simple sets, if I3 is a jump, ensure the
4605         pattern we place into I3 is a (set (pc) ...).
4606
4607 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
4608             Jeff Law  <law@redhat.com>
4609
4610         PR rtl-optimization/49847
4611         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
4612         are in different blocks.
4613         * doc/tm.texi (Condition Code Status): Update documention for
4614         relative locations of cc0-setter and cc0-user.
4615
4616 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
4617
4618         PR target/59222
4619         * lra.c (lra_emit_add): Check SUBREG too.
4620
4621 2014-02-27  Andreas Schwab  <schwab@suse.de>
4622
4623         * config/m68k/m68k.c (m68k_option_override): Disable
4624         -flive-range-shrinkage for classic m68k.
4625         (m68k_override_options_after_change): Likewise.
4626
4627 2014-02-27  Marek Polacek  <polacek@redhat.com>
4628
4629         PR middle-end/59223
4630         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
4631         -Wmaybe-uninitialized.
4632
4633 2014-02-27  Alan Modra  <amodra@gmail.com>
4634
4635         PR target/57936
4636         * reload1.c (emit_input_reload_insns): When reload_override_in,
4637         set old to rl->in_reg when rl->in_reg is a subreg.
4638
4639 2014-02-26  Richard Biener  <rguenther@suse.de>
4640
4641         PR bootstrap/60343
4642         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
4643
4644 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4645
4646         * common/config/i386/predicates.md (const1256_operand): Remove.
4647         (const2356_operand): New.
4648         (const_1_to_2_operand): Remove.
4649         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4650         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4651         (*avx512pf_gatherpf<mode>sf): Ditto.
4652         (avx512pf_gatherpf<mode>df): Ditto.
4653         (*avx512pf_gatherpf<mode>df_mask): Ditto.
4654         (*avx512pf_gatherpf<mode>df): Ditto.
4655         (avx512pf_scatterpf<mode>sf): Ditto.
4656         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4657         (*avx512pf_scatterpf<mode>sf): Ditto.
4658         (avx512pf_scatterpf<mode>df): Ditto.
4659         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4660         (*avx512pf_scatterpf<mode>df): Ditto.
4661         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
4662
4663 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
4664
4665         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
4666         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
4667         (_mm512_mask_testn_epi64_mask): Move to ...
4668         * config/i386/avx512cdintrin.h: Here.
4669         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
4670         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
4671         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
4672         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
4673         TARGET_AVX512F from TARGET_AVX512CD.
4674
4675 2014-02-26  Richard Biener  <rguenther@suse.de>
4676
4677         PR ipa/60327
4678         * ipa.c (walk_polymorphic_call_targets): Properly guard
4679         call to inline_update_overall_summary.
4680
4681 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
4682
4683         PR target/60280
4684         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
4685         and latches only if requested.  Fix latch if it is removed.
4686         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
4687         LOOPS_HAVE_PREHEADERS.
4688
4689 2014-02-25  Andrew Pinski  <apinski@cavium.com>
4690
4691         * builtins.c (expand_builtin_thread_pointer): Create a new target
4692         when the target is NULL.
4693
4694 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
4695
4696         PR rtl-optimization/60317
4697         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4698         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4699         * lra-assigns.c: Include params.h.
4700         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
4701         other reload pseudos considerations.
4702
4703 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4704
4705         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
4706         to use canonical form for nor<mode>3.
4707
4708 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4709
4710         PR target/55426
4711         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
4712         conversions.
4713
4714 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4715
4716         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
4717         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
4718         (ix86_handle_option): Handle OPT_mprefetchwt1.
4719         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
4720         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4721         PREFETCHWT1 CPUID.
4722         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4723         OPTION_MASK_ISA_PREFETCHWT1.
4724         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
4725         (PTA_PREFETCHWT1): New.
4726         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
4727         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
4728         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
4729         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
4730         (*prefetch_avx512pf_<mode>_: Change into ...
4731         (*prefetch_prefetchwt1_<mode>: This.
4732         * config/i386/i386.opt (mprefetchwt1): New.
4733         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
4734         (_mm_prefetch): Handle intent to write.
4735         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
4736
4737 2014-02-25  Richard Biener  <rguenther@suse.de>
4738
4739         PR middle-end/60291
4740         * emit-rtl.c (mem_attrs_htab): Remove.
4741         (mem_attrs_htab_hash): Likewise.
4742         (mem_attrs_htab_eq): Likewise.
4743         (set_mem_attrs): Always allocate new mem-attrs when something changed.
4744         (init_emit_once): Do not allocate mem_attrs_htab.
4745
4746 2014-02-25  Richard Biener  <rguenther@suse.de>
4747
4748         PR lto/60319
4749         * lto-opts.c (lto_write_options): Output non-explicit conservative
4750         -fwrapv, -fno-trapv and -fno-strict-overflow.
4751         * lto-wrapper.c (merge_and_complain): Handle merging those options.
4752         (run_gcc): And pass them through.
4753
4754 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4755
4756         * sel-sched.c (calculate_new_fences): New parameter ptime.
4757         Calculate it as a maximum over all fence cycles.
4758         (sel_sched_region_2): Adjust the call to calculate_new_fences.
4759         Print the final schedule timing when sched_verbose.
4760
4761 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4762
4763         PR rtl-optimization/60292
4764         * sel-sched.c (fill_vec_av_set): Do not reset target availability
4765         bit fot the fence instruction.
4766
4767 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
4768
4769         * calls.h: Fix typo in comment.
4770
4771 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
4772
4773         * config/pa/pa.c (pa_output_move_double): Don't valididate when
4774         adjusting offsetable addresses.
4775
4776 2014-02-24  Guozhi Wei  <carrot@google.com>
4777
4778         * sparseset.h (sparseset_pop): Fix the wrong index.
4779
4780 2014-02-24  Walter Lee  <walt@tilera.com>
4781
4782         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
4783         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
4784         triplet.
4785         * common/config/tilegx/tilegx-common.c
4786         (TARGET_DEFAULT_TARGET_FLAGS): Define.
4787         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
4788         (LINK_SPEC): Ditto.
4789         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
4790         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
4791         (tilegx_gimplify_va_arg_expr): Handle big endian.
4792         (tilegx_expand_unaligned_load): Ditto.
4793         (tilegx_expand_unaligned_store): Ditto.
4794         (TARGET_RETURN_IN_MSB): New.
4795         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
4796         (TARGET_ENDIAN_DEFAULT): New.
4797         (TARGET_BIG_ENDIAN): Handle big endian.
4798         (BYTES_BIG_ENDIAN): Ditto.
4799         (WORDS_BIG_ENDIAN): Ditto.
4800         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
4801         (ENDIAN_SPEC): New.
4802         (EXTRA_SPECS): New.
4803         * config/tilegx/tilegx.md (extv): Handle big endian.
4804         (extzv): Ditto.
4805         (insn_st<n>): Ditto.
4806         (insn_st<n>_add<bitsuffix>): Ditto.
4807         (insn_stnt<n>): Ditto.
4808         (insn_stnt<n>_add<bitsuffix>):Ditto.
4809         (vec_interleave_highv8qi): Handle big endian.
4810         (vec_interleave_highv8qi_be): New.
4811         (vec_interleave_highv8qi_le): New.
4812         (insn_v1int_h): Handle big endian.
4813         (vec_interleave_lowv8qi): Handle big endian.
4814         (vec_interleave_lowv8qi_be): New.
4815         (vec_interleave_lowv8qi_le): New.
4816         (insn_v1int_l): Handle big endian.
4817         (vec_interleave_highv4hi): Handle big endian.
4818         (vec_interleave_highv4hi_be): New.
4819         (vec_interleave_highv4hi_le): New.
4820         (insn_v2int_h): Handle big endian.
4821         (vec_interleave_lowv4hi): Handle big endian.
4822         (vec_interleave_lowv4hi_be): New.
4823         (vec_interleave_lowv4hi_le): New.
4824         (insn_v2int_l): Handle big endian.
4825         (vec_interleave_highv2si): Handle big endian.
4826         (vec_interleave_highv2si_be): New.
4827         (vec_interleave_highv2si_le): New.
4828         (insn_v4int_h): Handle big endian.
4829         (vec_interleave_lowv2si): Handle big endian.
4830         (vec_interleave_lowv2si_be): New.
4831         (vec_interleave_lowv2si_le): New.
4832         (insn_v4int_l): Handle big endian.
4833         * config/tilegx/tilegx.opt (mbig-endian): New option.
4834         (mlittle-endian): New option.
4835         * doc/install.texi: Document tilegxbe-linux.
4836         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
4837
4838 2014-02-24  Martin Jambor  <mjambor@suse.cz>
4839
4840         PR ipa/60266
4841         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
4842         there are no parameter descriptors.
4843
4844 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
4845
4846         PR rtl-optimization/60268
4847         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
4848         initialization to ...
4849         (sched_rgn_init): ... here.
4850         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
4851
4852 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4853
4854         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
4855         names.
4856
4857 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
4858
4859         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
4860         definition.
4861
4862 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4863
4864         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
4865         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
4866
4867 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4868
4869         * config/microblaze/predicates.md: Add cmp_op predicate.
4870         * config/microblaze/microblaze.md: Add branch_compare instruction
4871         which uses cmp_op predicate and emits cmp insn before branch.
4872         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
4873         to microblaze_expand_conditional_branch and consolidate logic.
4874         (microblaze_expand_conditional_branch): emit branch_compare
4875         insn instead of handling cmp op separate from branch insn.
4876
4877 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4878
4879         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
4880         to permit subregs.
4881
4882 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4883
4884         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
4885         define_insn with define_expand and new define_insn
4886         *altivec_lve<VI_char>x_internal.
4887         (altivec_stve<VI_char>x): Replace define_insn with define_expand
4888         and new define_insn *altivec_stve<VI_char>x_internal.
4889         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
4890         prototype.
4891         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
4892         lve*x built-ins.
4893         (altivec_expand_stvex_be): New function.
4894
4895 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
4896
4897         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
4898         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
4899         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
4900         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
4901
4902 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
4903
4904         PR target/60298
4905         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
4906         instead of emit_move_insn.
4907
4908 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4909
4910         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
4911         vspltw with vsldoi.
4912         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
4913         gen_altivec_vsumsws.
4914
4915 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4916
4917         * config/rs6000/altivec.md (altivec_lvxl): Rename as
4918         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
4919         (altivec_lvxl_<mode>): New define_expand incorporating
4920         -maltivec=be semantics where needed.
4921         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
4922         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
4923         semantics where needed.
4924         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
4925         (altivec_stvx_<mode>): New define_expand incorporating
4926         -maltivec=be semantics where needed.
4927         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
4928         VM2 iterator instead of V4SI.
4929         (altivec_stvxl_<mode>): New define_expand incorporating
4930         -maltivec=be semantics where needed.
4931         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
4932         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
4933         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
4934         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
4935         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
4936         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
4937         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
4938         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
4939         ALTIVEC_BUILTIN_STVXL.
4940         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
4941         (altivec_expand_stvx_be): Likewise.
4942         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
4943         (altivec_expand_lvx_be): Likewise.
4944         (altivec_expand_stvx_be): Likewise.
4945         (altivec_expand_builtin): Add cases for
4946         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
4947         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
4948         (altivec_init_builtins): Add definitions for
4949         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
4950         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
4951
4952 2014-02-21  Catherine Moore  <clm@codesourcery.com>
4953
4954         * doc/invoke.texi (mvirt, mno-virt): Document.
4955         * config/mips/mips.opt (mvirt): New option.
4956         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
4957
4958 2014-02-21  Richard Biener  <rguenther@suse.de>
4959
4960         PR tree-optimization/60276
4961         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
4962         (STMT_VINFO_MIN_NEG_DIST): New macro.
4963         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
4964         STMT_VINFO_MIN_NEG_DIST.
4965         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
4966         made for negative dependence distances still hold.
4967
4968 2014-02-21  Richard Biener  <rguenther@suse.de>
4969
4970         PR middle-end/60291
4971         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
4972         DECL_INITIAL for globals not in the current function context.
4973
4974 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4975
4976         PR tree-optimization/56490
4977         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
4978         * tree-ssa-uninit.c: Include params.h.
4979         (compute_control_dep_chain): Add num_calls argument, return false
4980         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
4981         num_calls to recursive call.
4982         (find_predicates): Change dep_chain into normal array,
4983         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
4984         variable and adjust compute_control_dep_chain caller.
4985         (find_def_preds): Likewise.
4986
4987 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
4988
4989         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4990         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
4991
4992 2014-02-21  Nick Clifton  <nickc@redhat.com>
4993
4994         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
4995         (pushhi1): Likewise.
4996         (popqi1): Add mode to pre_dec.
4997         (pophi1): Likewise.
4998
4999 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
5000
5001         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
5002         mode for mask of V8SFmode permutation.
5003
5004 2014-02-20  Richard Henderson  <rth@redhat.com>
5005
5006         PR c++/60272
5007         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
5008         a new pseudo for OLDVAL.
5009
5010 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
5011
5012         PR target/57896
5013         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
5014         gen_reg_rtx if d->testing_p.
5015         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
5016         if d->testing_p and we will certainly return true.
5017         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
5018         if d->testing_p.
5019
5020 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
5021
5022         * emit-rtl.c (gen_reg_rtx): Assert that
5023         crtl->emit.regno_pointer_align_length is non-zero.
5024
5025 2014-02-20  Richard Henderson  <rth@redhat.com>
5026
5027         PR c++/60272
5028         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
5029         on failure the store back into EXPECT.
5030
5031 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
5032             Sandra Loosemore  <sandra@codesourcery.com>
5033
5034         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
5035         * config/nios2/nios2.c (nios2_function_profiler): Add
5036         -fPIC (flag_pic == 2) support.
5037         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
5038         (nios2_large_offset_p): New function.
5039         (nios2_unspec_reloc_p): Move up position, update to use
5040         nios2_large_offset_p.
5041         (nios2_unspec_address): Remove function.
5042         (nios2_unspec_offset): New function.
5043         (nios2_large_got_address): New function.
5044         (nios2_got_address): Add large offset support.
5045         (nios2_legitimize_tls_address): Update usage of removed and new
5046         functions.
5047         (nios2_symbol_binds_local_p): New function.
5048         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
5049         (nios2_legitimize_address): Update to use nios2_large_offset_p.
5050         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
5051         (nios2_print_operand): Merge H/L processing, add hiadj/lo
5052         processing for (const (unspec ...)).
5053         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
5054
5055 2014-02-20  Richard Biener  <rguenther@suse.de>
5056
5057         * tree-cfg.c (replace_uses_by): Mark altered BBs before
5058         doing the substitution.
5059         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
5060
5061 2014-02-20  Martin Jambor  <mjambor@suse.cz>
5062
5063         PR ipa/55260
5064         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
5065         info when checking whether lattices are bottom.
5066
5067 2014-02-20  Richard Biener  <rguenther@suse.de>
5068
5069         PR middle-end/60221
5070         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
5071         regions at -O0.
5072
5073 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
5074
5075         PR ipa/58555
5076         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
5077         parameter specifying the scaling.
5078         (inline_call): Update.
5079         (want_inline_recursively): Guard division by zero.
5080         (recursive_inlining): Update.
5081         * ipa-inline.h (clone_inlined_nodes): Update.
5082
5083 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
5084
5085         PR target/60204
5086         * config/i386/i386.c (classify_argument): Pass structures of size
5087         64 bytes or less in register.
5088
5089 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
5090             Kirill Yukhin  <kirill.yukhin@intel.com>
5091
5092         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
5093         (_mm_rcp28_round_ss): Ditto.
5094         (_mm_rsqrt28_round_sd): Ditto.
5095         (_mm_rsqrt28_round_ss): Ditto.
5096         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
5097         (_mm_rcp14_round_ss): Ditto.
5098         (_mm_rsqrt14_round_sd): Ditto.
5099         (_mm_rsqrt14_round_ss): Ditto.
5100         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
5101         the first input operand, get rid of match_dup.
5102         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
5103         attribute to sse.
5104         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5105         Ditto.
5106         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
5107         operand as the first input operand, set type attribute.
5108         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5109         Set type attribute.
5110         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
5111         operand as the first input operand, set type attribute.
5112
5113 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5114
5115         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
5116         bit of zero.
5117
5118 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
5119
5120         PR target/60207
5121         * config/i386/i386.c (construct_container): Remove TFmode check
5122         for X86_64_INTEGER_CLASS.
5123
5124 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
5125
5126         PR target/59794
5127         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
5128         only when -Wpsabi is enabled.
5129
5130 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
5131
5132          PR target/59799
5133         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
5134         passing arrays in registers are the same as for structs, so remove the
5135         special case for them.
5136
5137 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
5138
5139         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
5140         destination type, extract only the valid bits if the source type is not
5141         integral and has a different mode.
5142
5143 2014-02-19  Richard Biener  <rguenther@suse.de>
5144
5145         PR ipa/60243
5146         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
5147         for all calls.
5148
5149 2014-02-19  Richard Biener  <rguenther@suse.de>
5150
5151         PR ipa/60243
5152         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
5153         (ipa_modify_call_arguments): Emit an argument load explicitely and
5154         preserve virtual SSA form there and for the replacement call.
5155         Do not update SSA form nor free dominance info.
5156
5157 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
5158
5159         * ipa.c (function_and_variable_visibility): Also clear WEAK
5160         flag when disolving COMDAT_GROUP.
5161
5162 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
5163
5164         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
5165         * ipa-prop.c (ipa_set_jf_known_type): Return early when
5166         not devirtualizing.
5167         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
5168         do more sanity checks.
5169         (detect_type_change): Return true when giving up early.
5170         (compute_complex_assign_jump_func): Fix type parameter of
5171         ipa_set_ancestor_jf.
5172         (compute_complex_ancestor_jump_func): Likewise.
5173         (update_jump_functions_after_inlining): Fix updating of
5174         ancestor function.
5175         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
5176
5177 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
5178
5179         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
5180         inline clones when edge disappears.
5181
5182 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
5183
5184         PR target/60203
5185         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
5186         Split 64-bit moves into 2 patterns.  Do not allow the use of
5187         direct move for TDmode in little endian, since the decimal value
5188         has little endian bytes within a word, but the 64-bit pieces are
5189         ordered in a big endian fashion, and normal subreg's of TDmode are
5190         not allowed.
5191         (mov<mode>_64bit_dm): Likewise.
5192         (movtd_64bit_nodm): Likewise.
5193
5194 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
5195
5196         PR tree-optimization/60174
5197         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
5198         statement of an SSA_NAME that occurs in an abnormal PHI node.
5199
5200 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
5201
5202         PR sanitizer/60142
5203         * final.c (SEEN_BB): Remove.
5204         (SEEN_NOTE, SEEN_EMITTED): Renumber.
5205         (final_scan_insn): Don't force_source_line on second
5206         NOTE_INSN_BASIC_BLOCK.
5207
5208 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
5209
5210         PR target/60205
5211         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
5212         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
5213         (type_natural_mode): Warn ABI change when %zmm register is not
5214         available for AVX512F vector value passing.
5215
5216 2014-02-18  Kai Tietz  <ktietz@redhat.com>
5217
5218         PR target/60193
5219         * config/i386/i386.c (ix86_expand_prologue): Use value in
5220         rax register as displacement when restoring %r10 or %rax.
5221         Fix wrong offset when restoring both registers.
5222
5223 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
5224
5225         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
5226         assertion with conditional return.
5227
5228 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
5229             Uros Bizjak  <ubizjak@gmail.com>
5230
5231         PR driver/60233
5232         * config/i386/driver-i386.c (host_detect_local_cpu): If
5233         YMM state is not saved by the OS, also clear has_f16c.  Move
5234         CPUID 0x80000001 handling before YMM state saving checking.
5235
5236 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
5237
5238         PR rtl-optimization/58960
5239         * haifa-sched.c (alloc_global_sched_pressure_data): New,
5240         factored out from ...
5241         (sched_init): ... here.
5242         (free_global_sched_pressure_data): New, factored out from ...
5243         (sched_finish): ... here.
5244         * sched-int.h (free_global_sched_pressure_data): Declare.
5245         * sched-rgn.c (nr_regions_initial): New static global.
5246         (haifa_find_rgns): Initialize it.
5247         (schedule_region): Disable sched-pressure for the newly
5248         generated regions.
5249
5250 2014-02-17  Richard Biener  <rguenther@suse.de>
5251
5252         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
5253         release SSA defs of pattern stmts.
5254
5255 2014-02-17  Richard Biener  <rguenther@suse.de>
5256
5257         * tree-inline.c (expand_call_inline): Release the virtual
5258         operand defined by the call we are about to inline.
5259
5260 2014-02-17  Richard Biener  <rguenther@suse.de>
5261
5262         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
5263
5264 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
5265             Ilya Tocar  <ilya.tocar@intel.com>
5266
5267         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
5268         arguments order in builtin.
5269         (_mm512_permutexvar_epi64): Ditto.
5270         (_mm512_mask_permutexvar_epi64): Ditto
5271         (_mm512_maskz_permutexvar_epi32): Ditto
5272         (_mm512_permutexvar_epi32): Ditto
5273         (_mm512_mask_permutexvar_epi32): Ditto
5274
5275 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5276
5277         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
5278         (p8_vmrgow): Likewise.
5279
5280 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5281
5282         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
5283         endian targets.
5284
5285 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
5286
5287         PR target/60203
5288         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
5289         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
5290         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
5291         using direct move instructions on ISA 2.07.  Also adjust
5292         instruction length for 64-bit.
5293         (mov<mode>_64bit, TFmode/TDmode): Likewise.
5294         (mov<mode>_32bit, TFmode/TDmode): Likewise.
5295
5296 2014-02-15  Alan Modra  <amodra@gmail.com>
5297
5298         PR target/58675
5299         PR target/57935
5300         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
5301         find_replacement on parts of insn rtl that might be reloaded.
5302
5303 2014-02-15  Richard Biener  <rguenther@suse.de>
5304
5305         PR tree-optimization/60183
5306         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
5307         (tree_ssa_phiprop): Calculate and free post-dominators.
5308
5309 2014-02-14  Jeff Law  <law@redhat.com>
5310
5311         PR rtl-optimization/60131
5312         * ree.c (get_extended_src_reg): New function.
5313         (combine_reaching_defs): Use it rather than assuming location of REG.
5314         (find_and_remove_re): Verify first operand of extension is
5315         a REG before adding the insns to the copy list.
5316
5317 2014-02-14  Roland McGrath  <mcgrathr@google.com>
5318
5319         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
5320         * configure: Regenerated.
5321         * config.in: Regenerated.
5322         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
5323         instead of ASM_SHORT.
5324
5325 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
5326             Richard Earnshaw  <rearnsha@arm.com>
5327
5328         PR rtl-optimization/59535
5329         * lra-constraints.c (process_alt_operands): Encourage alternative
5330         when unassigned pseudo class is superset of the alternative class.
5331         (inherit_reload_reg): Don't inherit when optimizing for code size.
5332         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
5333         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
5334         modes not less than 4 for Thumb1.
5335
5336 2014-02-14  Kyle McMartin  <kyle@redhat.com>
5337
5338         PR pch/60010
5339         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
5340
5341 2014-02-14  Richard Biener  <rguenther@suse.de>
5342
5343         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
5344         (get_frame_arg): Drop the assert with langhook types_compatible_p.
5345         Do not strip INDIRECT_REFs.
5346
5347 2014-02-14  Richard Biener  <rguenther@suse.de>
5348
5349         PR lto/60179
5350         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
5351         DECL_FUNCTION_SPECIFIC_TARGET.
5352         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
5353         * tree-streamer-out.c (pack_ts_target_option): Remove.
5354         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
5355         (write_ts_function_decl_tree_pointers): Do not stream
5356         DECL_FUNCTION_SPECIFIC_TARGET.
5357         * tree-streamer-in.c (unpack_ts_target_option): Remove.
5358         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
5359         (lto_input_ts_function_decl_tree_pointers): Do not stream
5360         DECL_FUNCTION_SPECIFIC_TARGET.
5361
5362 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
5363
5364         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
5365         (get_initial_def_for_induction, vectorizable_induction): Ignore
5366         debug stmts when looking for exit_phi.
5367         (vectorizable_live_operation): Fix up condition.
5368
5369 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5370
5371         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
5372         nreverse() because it changes the content of original tree list.
5373
5374 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5375
5376         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
5377         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
5378
5379 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5380
5381         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
5382         GNU coding standards.
5383
5384 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
5385
5386         PR debug/60152
5387         * dwarf2out.c (gen_subprogram_die): Don't call
5388         add_calling_convention_attribute if subr_die is old_die.
5389
5390 2014-02-13  Sharad Singhai  <singhai@google.com>
5391
5392         * doc/optinfo.texi: Fix order of nodes.
5393
5394 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
5395
5396         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
5397         operands[2], not operands[3].
5398
5399 2014-02-13  Richard Biener  <rguenther@suse.de>
5400
5401         PR bootstrap/59878
5402         * doc/install.texi (ISL): Update recommended version to 0.12.2,
5403         mention the possibility of an in-tree build.
5404         (CLooG): Update recommended version to 0.18.1, mention the
5405         possibility of an in-tree build and clarify that the ISL
5406         bundled with CLooG does not work.
5407
5408 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
5409
5410         PR target/43546
5411         * expr.c (compress_float_constant): If x is a hard register,
5412         extend into a pseudo and then move to x.
5413
5414 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5415
5416         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
5417         caused by bad second argument to warning_at() with -mhotpatch and
5418         nested functions (e.g. with gfortran).
5419
5420 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
5421
5422         * opts.c (option_name): Remove "enabled by default" rider.
5423
5424 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
5425
5426         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
5427
5428 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
5429             Uros Bizjak  <ubizjak@gmail.com>
5430
5431         PR target/60151
5432         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
5433         * configure: Regenerated.
5434
5435 2014-02-12  Richard Biener  <rguenther@suse.de>
5436
5437         * vec.c (vec_prefix::calculate_allocation): Move as
5438         inline variant to vec.h.
5439         (vec_prefix::calculate_allocation_1): New out-of-line version.
5440         * vec.h (vec_prefix::calculate_allocation_1): Declare.
5441         (vec_prefix::m_has_auto_buf): Rename to ...
5442         (vec_prefix::m_using_auto_storage): ... this.
5443         (vec_prefix::calculate_allocation): Inline the easy cases
5444         and dispatch to calculate_allocation_1 which doesn't need the
5445         prefix address.
5446         (va_heap::reserve): Use gcc_checking_assert.
5447         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
5448         m_using_auto_storage.
5449         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
5450         member and adjust.
5451         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
5452         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
5453         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
5454
5455 2014-02-12  Richard Biener  <rguenther@suse.de>
5456
5457         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
5458         when we found a dependence.
5459
5460 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
5461
5462         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
5463         common code...
5464         (maybe_fold_stmt): ... into this new function.
5465         * omp-low.c (lower_omp): Update comment.
5466
5467         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
5468         last use.
5469
5470         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
5471         dereference.
5472
5473 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
5474
5475         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
5476         identifiers in comments.
5477         (cortexa53_extra_costs): Likewise.
5478         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
5479         (cortexa7_extra_costs): Likewise.
5480         (cortexa12_extra_costs): Likewise.
5481         (cortexa15_extra_costs): Likewise.
5482         (v7m_extra_costs): Likewise.
5483
5484 2014-02-12  Richard Biener  <rguenther@suse.de>
5485
5486         PR middle-end/60092
5487         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
5488         of posix_memalign being successful.
5489         (lower_stmt): Restrict lowering of posix_memalign to when
5490         -ftree-bit-ccp is enabled.
5491
5492 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5493
5494         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
5495         arg_loc.
5496         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
5497
5498 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
5499
5500         PR rtl-optimization/60116
5501         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
5502         other_insn once the combination has been validated.
5503
5504 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
5505
5506         PR lto/59468
5507         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
5508         and wrapper.
5509         * ipa-devirt.c: Include demangle.h
5510         (odr_violation_reported): New static variable.
5511         (add_type_duplicate): Update odr_violations.
5512         (maybe_record_node): Add completep parameter; update it.
5513         (record_target_from_binfo): Add COMPLETEP parameter;
5514         update it as needed.
5515         (possible_polymorphic_call_targets_1): Likewise.
5516         (struct polymorphic_call_target_d): Add nonconstruction_targets;
5517         rename FINAL to COMPLETE.
5518         (record_targets_from_bases): Sanity check we found the binfo;
5519         fix COMPLETEP updating.
5520         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
5521         parameter, fix computing of COMPLETEP.
5522         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
5523         at LTO time do demangling.
5524         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
5525         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
5526         parameter.
5527         (gimple_get_virt_method_for_binfo): Likewise.
5528         * gimple-fold.h (gimple_get_virt_method_for_binfo,
5529         gimple_get_virt_method_for_vtable): Update prototypes.
5530
5531 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
5532
5533         PR target/49008
5534         * genautomata.c (add_presence_absence): Fix typo with
5535         {final_}presence_list.
5536
5537 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5538
5539         PR target/60137
5540         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
5541         for VSX/Altivec vectors that land in GPR registers.
5542
5543 2014-02-11  Richard Henderson  <rth@redhat.com>
5544             Jakub Jelinek  <jakub@redhat.com>
5545
5546         PR debug/59776
5547         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
5548         around drhs if type conversion to lacc->type is not useless.
5549
5550 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5551
5552         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
5553         tuning struct.
5554         (cortex-a57.cortex-a53): Likewise.
5555         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
5556
5557 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5558
5559         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
5560         arm_restrict_it.
5561
5562 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
5563
5564         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
5565         add_options_for_arm_vfp3.
5566
5567 2014-02-11  Jeff Law  <law@redhat.com>
5568
5569         PR middle-end/54041
5570         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
5571         object with an undesirable mode.
5572
5573 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5574
5575         PR libgomp/60107
5576         * config/i386/sol2-9.h: New file.
5577         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
5578         *-*-solaris2.9*): Use it.
5579
5580 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5581
5582         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
5583         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
5584
5585 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5586
5587         * config/microblaze/microblaze.c: Extend mcpu version format
5588
5589 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
5590
5591         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
5592
5593 2014-02-10  Richard Henderson  <rth@redhat.com>
5594
5595         PR target/59927
5596         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
5597         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
5598         ms-abi vs -mno-accumulate-outgoing-args.
5599         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
5600         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
5601         respect to ms-abi.
5602
5603 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5604
5605         PR middle-end/60080
5606         * cfgexpand.c (expand_asm_operands): Attach source location to
5607         ASM_INPUT rtx objects.
5608         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
5609
5610 2014-02-10  Nick Clifton  <nickc@redhat.com>
5611
5612         * config/mn10300/mn10300.c (popcount): New function.
5613         (mn10300_expand_prologue): Include saved registers in stack usage
5614         count.
5615
5616 2014-02-10  Jeff Law  <law@redhat.com>
5617
5618         PR middle-end/52306
5619         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
5620         when changing the SET_DEST of a prior insn to avoid an input reload.
5621
5622 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5623
5624         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
5625         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
5626         -mcall-openbsd, or -mcall-linux.
5627         (CC1_ENDIAN_BIG_SPEC): Remove.
5628         (CC1_ENDIAN_LITTLE_SPEC): Remove.
5629         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5630         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
5631         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
5632         and %cc1_endian_default.
5633         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5634
5635 2014-02-10  Richard Biener  <rguenther@suse.de>
5636
5637         PR tree-optimization/60115
5638         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
5639         MEM_REF handling.  Properly verify that the accesses are not
5640         out of the objects bound.
5641
5642 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5643
5644         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
5645         coretex to cortex.
5646
5647 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
5648
5649         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
5650         proper constants and fix formatting.
5651         (possible_polymorphic_call_targets): Fix formatting.
5652
5653 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
5654             Ilya Tocar  <ilya.tocar@intel.com>
5655
5656         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
5657         (_mm512_loadu_epi32): Renamed into...
5658         (_mm512_loadu_si512): This.
5659         (_mm512_storeu_epi32): Renamed into...
5660         (_mm512_storeu_si512): This.
5661         (_mm512_maskz_ceil_ps): Removed.
5662         (_mm512_maskz_ceil_pd): Ditto.
5663         (_mm512_maskz_floor_ps): Ditto.
5664         (_mm512_maskz_floor_pd): Ditto.
5665         (_mm512_floor_round_ps): Ditto.
5666         (_mm512_floor_round_pd): Ditto.
5667         (_mm512_ceil_round_ps): Ditto.
5668         (_mm512_ceil_round_pd): Ditto.
5669         (_mm512_mask_floor_round_ps): Ditto.
5670         (_mm512_mask_floor_round_pd): Ditto.
5671         (_mm512_mask_ceil_round_ps): Ditto.
5672         (_mm512_mask_ceil_round_pd): Ditto.
5673         (_mm512_maskz_floor_round_ps): Ditto.
5674         (_mm512_maskz_floor_round_pd): Ditto.
5675         (_mm512_maskz_ceil_round_ps): Ditto.
5676         (_mm512_maskz_ceil_round_pd): Ditto.
5677         (_mm512_expand_pd): Ditto.
5678         (_mm512_expand_ps): Ditto.
5679         * config/i386/i386.c (ix86_builtins): Remove
5680         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
5681         (bdesc_args): Ditto.
5682         * config/i386/predicates.md (const1256_operand): New.
5683         (const_1_to_2_operand): Ditto.
5684         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
5685         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5686         (*avx512pf_gatherpf<mode>sf): Ditto.
5687         (avx512pf_gatherpf<mode>df): Ditto.
5688         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5689         (*avx512pf_gatherpf<mode>df): Ditto.
5690         (avx512pf_scatterpf<mode>sf): Ditto.
5691         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5692         (*avx512pf_scatterpf<mode>sf): Ditto.
5693         (avx512pf_scatterpf<mode>df): Ditto.
5694         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5695         (*avx512pf_scatterpf<mode>df): Ditto.
5696         (avx512f_expand<mode>): Removed.
5697         (<shift_insn><mode>3<mask_name>): Change predicate type.
5698
5699 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5700
5701         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
5702         not at the end of datarefs vector use ordered_remove to avoid
5703         reordering datarefs vector.
5704
5705         PR c/59984
5706         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
5707         mark local addressable non-static vars as GOVD_PRIVATE
5708         instead of GOVD_LOCAL.
5709         * omp-low.c (lower_omp_for): Move gimple_bind_vars
5710         and BLOCK_VARS of gimple_bind_block to new_stmt rather
5711         than copying them.
5712
5713         PR middle-end/60092
5714         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
5715         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
5716         assume_aligned or alloc_align attributes.
5717         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
5718         arguments.  Handle also assume_aligned and alloc_align attributes.
5719         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
5720         calls to functions with assume_aligned or alloc_align attributes.
5721         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
5722
5723 2014-02-08  Terry Guo  <terry.guo@arm.com>
5724
5725         * doc/invoke.texi: Document ARM -march=armv7e-m.
5726
5727 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5728
5729         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
5730         flag on __cilkrts_rethrow builtin.
5731
5732         PR ipa/60026
5733         * ipa-cp.c (determine_versionability): Fail at -O0
5734         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
5735         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
5736
5737         Revert:
5738         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5739
5740         PR ipa/60026
5741         * tree-inline.c (copy_forbidden): Fail for
5742         __attribute__((optimize (0))) functions.
5743
5744 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5745
5746         * varpool.c: Include pointer-set.h.
5747         (varpool_remove_unreferenced_decls): Variables in other partitions
5748         will not be output; be however careful to not lose information
5749         about partitioning.
5750
5751 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5752
5753         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
5754         lookup in the vtable constructor.
5755
5756 2014-02-07  Jeff Law  <law@redhat.com>
5757
5758         PR target/40977
5759         * config/m68k/m68k.md (ashldi_extsi): Turn into a
5760         define_insn_and_split.
5761
5762         * ipa-inline.c (inline_small_functions): Fix typos.
5763
5764 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5765
5766         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
5767         (s390_can_use_return_insn): Declare.
5768         * config/s390/s390.h (EPILOGUE_USES): Define.
5769         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
5770         instructions.
5771         (s390_chunkify_start): Handle return JUMP_LABELs.
5772         (s390_early_mach): Emit a main_pool instruction on the entry edge.
5773         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
5774         (s390_can_use_return_insn): New functions.
5775         (s390_fix_long_loop_prediction): Handle conditional returns.
5776         (TARGET_SET_UP_BY_PROLOGUE): Define.
5777         * config/s390/s390.md (ANY_RETURN): New code iterator.
5778         (*creturn, *csimple_return, return, simple_return): New patterns.
5779
5780 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5781
5782         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
5783         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
5784         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
5785         REG_CFA_RESTORE list when deciding not to restore a register.
5786
5787 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5788
5789         * config/s390/s390.c: Include tree-pass.h and context.h.
5790         (s390_early_mach): New function, split out from...
5791         (s390_emit_prologue): ...here.
5792         (pass_data_s390_early_mach): New pass structure.
5793         (pass_s390_early_mach): New class.
5794         (s390_option_override): Create and register early_mach pass.
5795         Move to end of file.
5796
5797 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5798
5799         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
5800         to match for the exit block.
5801
5802 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5803
5804         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
5805         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
5806         Reject misaligned operands.
5807
5808 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5809
5810         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
5811
5812 2014-02-07  Richard Biener  <rguenther@suse.de>
5813
5814         PR middle-end/60092
5815         * gimple-low.c (lower_builtin_posix_memalign): New function.
5816         (lower_stmt): Call it to lower posix_memalign in a way
5817         to make alignment info accessible.
5818
5819 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
5820
5821         PR c++/60082
5822         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
5823         __builtin_setjmp_receiver.
5824
5825 2014-02-07  Richard Biener  <rguenther@suse.de>
5826
5827         PR middle-end/60092
5828         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
5829         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
5830         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5831         Handle BUILT_IN_POSIX_MEMALIGN.
5832         (find_func_clobbers): Likewise.
5833         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5834         (call_may_clobber_ref_p_1): Likewise.
5835
5836 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5837
5838         PR ipa/59918
5839         * ipa-devirt.c (record_target_from_binfo): Remove overactive
5840         sanity check.
5841
5842 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5843
5844         PR ipa/59469
5845         * lto-cgraph.c (lto_output_node): Use
5846         symtab_get_symbol_partitioning_class.
5847         (lto_output_varpool_node): likewise.
5848         (symtab_get_symbol_partitioning_class): Move here from
5849         lto/lto-partition.c
5850         * cgraph.h (symbol_partitioning_class): Likewise.
5851         (symtab_get_symbol_partitioning_class): Declare.
5852
5853 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5854
5855         * ggc.h (ggc_internal_cleared_alloc): New macro.
5856         * vec.h (vec_safe_copy): Handle memory stats.
5857         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
5858         * target-globals.c (save_target_globals): Likewise.
5859
5860 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5861
5862         PR target/60077
5863         * expr.c (emit_move_resolve_push): Export; be bit more selective
5864         on when to clear alias set.
5865         * expr.h (emit_move_resolve_push): Declare.
5866         * function.h (struct function): Add tail_call_marked.
5867         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
5868         * config/i386/i386-protos.h (ix86_expand_push): Remove.
5869         * config/i386/i386.md (TImode move expander): De not call
5870         ix86_expand_push.
5871         (FP push expanders): Preserve memory attributes.
5872         * config/i386/sse.md (push<mode>1): Remove.
5873         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
5874         (ix86_expand_push): Remove.
5875         * config/i386/mmx.md (push<mode>1): Remove.
5876
5877 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5878
5879         PR rtl-optimization/60030
5880         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
5881         lopart with paradoxical subreg before shifting it up by hprec.
5882
5883 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5884
5885         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
5886         Remove extra newline at end of file.
5887         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
5888         (arm_issue_rate): Handle cortexa57.
5889         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
5890         (cortex-a57.cortex-a53): Likewise.
5891
5892 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5893
5894         PR target/59575
5895         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
5896         don't record in REG_FRAME_RELATED_EXPR registers not set in that
5897         bitmask.
5898         (arm_expand_prologue): Adjust all callers.
5899         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
5900         info, registers also at the lowest numbered registers side.  Use
5901         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
5902         XEXP.
5903
5904         PR debug/59992
5905         * var-tracking.c (adjust_mems): Before adding a SET to
5906         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
5907
5908 2014-02-06  Alan Modra  <amodra@gmail.com>
5909
5910         PR target/60032
5911         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
5912         change SDmode to DDmode when lra_in_progress.
5913
5914 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5915
5916         PR middle-end/59150
5917         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
5918         free_data_ref on the dr first, and before goto again also set dr
5919         to the next dr.  For simd_lane_access, free old datarefs[i] before
5920         overwriting it.  For get_vectype_for_scalar_type failure, don't
5921         free_data_ref if simd_lane_access.
5922
5923         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
5924
5925         PR target/60062
5926         * tree.h (opts_for_fn): New inline function.
5927         (opt_for_fn): Define.
5928         * config/i386/i386.c (ix86_function_regparm): Use
5929         opt_for_fn (decl, optimize) instead of optimize.
5930
5931 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5932
5933         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
5934         for SYMBOL_REF in large memory model.
5935
5936 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5937
5938         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
5939         and crypto support.
5940         (cortex-a57): Likewise.
5941         (cortex-a57.cortex-a53): Likewise.
5942
5943 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
5944             Kugan Vivekanandarajah  <kuganv@linaro.org>
5945
5946         * config/arm/arm.c (arm_vector_alignment_reachable): Check
5947         unaligned_access.
5948         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
5949
5950 2014-02-06  Richard Biener  <rguenther@suse.de>
5951
5952         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
5953         set_loop_copy and initialize_original_copy_tables.
5954
5955 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
5956
5957         * config/aarch64/aarch64-simd.md
5958         (aarch64_ashr_simddi): Change QI to SI.
5959
5960 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5961             Jakub Jelinek  <jakub@redhat.com>
5962
5963         PR middle-end/60013
5964         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
5965         of the dataflow.
5966
5967 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5968
5969         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
5970         CODE_FOR_altivec_vpku[hw]um to
5971         CODE_FOR_altivec_vpku[hw]um_direct.
5972         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
5973         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
5974         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
5975         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
5976
5977 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5978
5979         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
5980         generation for -maltivec=be.
5981         (altivec_vsumsws): Simplify redundant test.
5982
5983 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5984
5985         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
5986         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
5987         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
5988         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
5989         gen_altivec_vpkuwum.
5990         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
5991         BYTES_BIG_ENDIAN.
5992         (altivec_vpks<VI_char>ss): Likewise.
5993         (altivec_vpks<VI_char>us): Likewise.
5994         (altivec_vpku<VI_char>us): Likewise.
5995         (altivec_vpku<VI_char>um): Likewise.
5996         (altivec_vpku<VI_char>um_direct): New (copy of
5997         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
5998         internal use).
5999         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
6000         target is little endian and -maltivec=be is not specified.
6001         (*altivec_vupkhs<VU_char>_direct): New (copy of
6002         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
6003         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
6004         target is little endian and -maltivec=be is not specified.
6005         (*altivec_vupkls<VU_char>_direct): New (copy of
6006         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
6007         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
6008         little endian and -maltivec=be is not specified.
6009         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
6010         little endian and -maltivec=be is not specified.
6011
6012 2014-02-05  Richard Henderson  <rth@redhat.com>
6013
6014         PR debug/52727
6015         * combine-stack-adj.c: Revert r206943.
6016         * sched-int.h (struct deps_desc): Add last_args_size.
6017         * sched-deps.c (init_deps): Initialize it.
6018         (sched_analyze_insn): Add OUTPUT dependencies between insns that
6019         contain REG_ARGS_SIZE notes.
6020
6021 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
6022
6023         * lto-cgraph.c (asm_nodes_output): Make global.
6024         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
6025         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
6026         (driver_handle_option): Handle OPT_fwpa.
6027
6028 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
6029
6030         PR ipa/59947
6031         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
6032         a comment typo and formatting issue.  If odr_hash hasn't been
6033         created, return vNULL and set *completep to false.
6034
6035         PR middle-end/57499
6036         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
6037         bb with no successors.
6038
6039 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
6040
6041         PR target/59718
6042         * doc/invoke.texi (-march): Clarify documentation for ARM.
6043         (-mtune): Likewise.
6044         (-mcpu): Likewise.
6045
6046 2014-02-05  Richard Biener  <rguenther@suse.de>
6047
6048         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
6049         when not vectorizing because of too many alias checks.
6050         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
6051         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
6052
6053 2014-02-05  Nick Clifton  <nickc@redhat.com>
6054
6055         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
6056         accept extended registers in any mode when compiling for the MN10300.
6057
6058 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
6059
6060         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
6061         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
6062         sanitization attributes.
6063         (can_inline_edge_p): Likewise.
6064         (sanitize_attrs_match_for_inline_p): New function.
6065
6066 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
6067
6068         * ipa-prop.c (detect_type_change): Shor circuit testing of
6069         type changes on THIS pointer.
6070
6071 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
6072
6073         PR target/59777
6074         * config/pa/pa.c (legitimize_tls_address): Return original address
6075         if not passed a SYMBOL_REF rtx.
6076         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
6077         addresses.
6078         (pa_emit_move_sequence): Simplify TLS source operands.
6079         (pa_legitimate_constant_p): Reject all TLS constants.
6080         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
6081         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
6082
6083 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
6084
6085         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
6086         groups when we know they are controlled by LTO.
6087         * varasm.c (default_binds_local_p_1): If object is in other partition,
6088         it will be resolved locally.
6089
6090 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
6091
6092         * config/host-linux.c (linux_gt_pch_use_address): Don't
6093         use SSIZE_MAX because it is not always defined.
6094
6095 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
6096
6097         PR bootstrap/59913
6098         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
6099         threshold for pseudo splitting.
6100         (update_ebb_live_info): Process call argument hard registers and
6101         hard registers from insn definition too.
6102         (max_small_class_regs_num): New constant.
6103         (inherit_in_ebb): Update live hard regs through EBBs.  Update
6104         reloads_num only for small register classes.  Don't split for
6105         outputs of jumps.
6106
6107 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
6108
6109         PR ipa/60058
6110         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
6111         is non-null.
6112
6113 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
6114
6115         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
6116         visibility is safe.
6117
6118 2014-02-04  Marek Polacek  <polacek@redhat.com>
6119
6120         * gdbinit.in (pel): Define.
6121
6122 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
6123
6124         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
6125         behavior.
6126
6127 2014-02-04  Richard Biener  <rguenther@suse.de>
6128
6129         PR lto/59723
6130         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
6131         in function context local.
6132         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
6133         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
6134         similar to LTO_imported_decl_ref.
6135
6136 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
6137
6138         PR tree-optimization/60002
6139         * cgraphclones.c (build_function_decl_skip_args): Clear
6140         DECL_LANG_SPECIFIC.
6141
6142         PR tree-optimization/60023
6143         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
6144         false to gsi_replace.
6145         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
6146         has been in some EH region and vec_stmt could throw, add
6147         vec_stmt into the same EH region.
6148         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
6149         has no lhs, ignore it.
6150         * internal-fn.c (expand_MASK_LOAD): Likewise.
6151
6152         PR ipa/60026
6153         * tree-inline.c (copy_forbidden): Fail for
6154         __attribute__((optimize (0))) functions.
6155
6156         PR other/58712
6157         * omp-low.c (simd_clone_struct_copy): If from->inbranch
6158         is set, copy one less argument.
6159         (expand_simd_clones): Don't subtract clone_info->inbranch
6160         from simd_clone_struct_alloc argument.
6161
6162         PR rtl-optimization/57915
6163         * recog.c (simplify_while_replacing): If all unary/binary/relational
6164         operation arguments are constant, attempt to simplify those.
6165
6166         PR middle-end/59261
6167         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
6168         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
6169
6170 2014-02-04  Richard Biener  <rguenther@suse.de>
6171
6172         PR tree-optimization/60012
6173         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
6174         TBAA disambiguation to all DDRs.
6175
6176 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6177
6178         PR target/59788
6179         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
6180         (LINK_SPEC): Use it for -shared, -shared-libgcc.
6181
6182 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6183
6184         PR ipa/59882
6185         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
6186
6187 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6188
6189         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
6190         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
6191
6192 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6193
6194         PR ipa/59831
6195         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
6196         to figure out targets of polymorphic calls with known decl.
6197         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
6198         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
6199         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
6200         (get_polymorphic_call_info): ... here.
6201         (get_polymorphic_call_info_from_invariant): New function.
6202
6203 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6204
6205         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
6206         lookup via vtable pointer; check for type consistency
6207         and turn inconsitent facts into UNREACHABLE.
6208         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
6209         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
6210         type inconsistent querries; return UNREACHABLE instead.
6211
6212 2014-02-03  Richard Henderson  <rth@twiddle.net>
6213
6214         PR tree-opt/59924
6215         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
6216         already processed this node.
6217         (normalize_one_pred_1): Pass along mark_set.
6218         (normalize_one_pred): Create and destroy a pointer_set_t.
6219         (normalize_one_pred_chain): Likewise.
6220
6221 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
6222
6223         PR gcov-profile/58602
6224         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
6225
6226 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6227
6228         PR ipa/59831
6229         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
6230         -fno-devirtualize; try to devirtualize by the knowledge of
6231         virtual table pointer given by aggregate propagation.
6232         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
6233         (ipa_print_node_jump_functions): Dump also offset that
6234         is relevant for polymorphic calls.
6235         (determine_known_aggregate_parts): Add arg_type parameter; use it
6236         instead of determining the type from pointer type.
6237         (ipa_compute_jump_functions_for_edge): Update call of
6238         determine_known_aggregate_parts.
6239         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
6240         (gimple_get_virt_method_for_binfo): ... here; simplify using
6241         vtable_pointer_value_to_vtable.
6242         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
6243         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
6244         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
6245         (vtable_pointer_value_to_vtable): Break out from ...; handle also
6246         POINTER_PLUS_EXPR.
6247         (vtable_pointer_value_to_binfo): ... here.
6248         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
6249
6250 2014-02-03  Teresa Johnson  <tejohnson@google.com>
6251
6252         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
6253         redef of outer loop index variable.
6254
6255 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
6256
6257         PR c++/53017
6258         PR c++/59211
6259         * doc/extend.texi (Function Attributes): Typo.
6260
6261 2014-02-03  Cong Hou  <congh@google.com>
6262
6263         PR tree-optimization/60000
6264         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
6265         if the vectorized statement is a store.  A store statement can only
6266         appear at the end of pattern statements.
6267
6268 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
6269
6270         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
6271         (ix86_option_override_internal): Default long double to 64-bit for
6272         32-bit Bionic and to 128-bit for 64-bit Bionic.
6273
6274         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
6275         TARGET_LONG_DOUBLE_128 is true.
6276         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
6277
6278         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
6279         (mlong-double-64): Negate -mlong-double-128.
6280         (mlong-double-128): New option.
6281
6282         * config/i386/i386-c.c (ix86_target_macros): Define
6283         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
6284
6285         * doc/invoke.texi: Document -mlong-double-128.
6286
6287 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
6288
6289         PR rtl-optimization/60024
6290         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
6291
6292 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
6293
6294         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
6295
6296 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
6297
6298         PR rtl-optimization/57662
6299         * sel-sched.c (code_motion_path_driver): Do not mark already not
6300         existing blocks in the visiting bitmap.
6301
6302 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
6303
6304         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
6305         on the insn being emitted.
6306
6307 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
6308             Will Deacon  <will.deacon@arm.com>
6309
6310         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
6311
6312 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6313
6314         * config/arm/arm-tables.opt: Regenerate.
6315
6316 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6317
6318         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
6319         for vector types other than V16QImode.
6320         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
6321         define_expand, and call altivec_expand_vec_perm_le when producing
6322         code with little endian element order.
6323         (*altivec_vperm_<mode>_internal): New insn having previous
6324         behavior of altivec_vperm_<mode>.
6325         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
6326         altivec_expand_vec_perm_le when producing code with little endian
6327         element order.
6328         (*altivec_vperm_<mode>_uns_internal): New insn having previous
6329         behavior of altivec_vperm_<mode>_uns.
6330
6331 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6332
6333         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
6334         (altivec_vsumsws): Add handling for -maltivec=be with a little
6335         endian target.
6336         (altivec_vsumsws_direct): New.
6337         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
6338         gen_altivec_vsumsws.
6339
6340 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
6341
6342         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
6343         vtable_pointer_value_to_binfo): New functions.
6344         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
6345         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
6346
6347 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
6348
6349         * config/nios2/nios2.md (load_got_register): Initialize GOT
6350         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
6351         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
6352
6353 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
6354
6355         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
6356         preserverd by passthrough, do not propagate the type.
6357
6358 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6359
6360         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
6361         (mips_atomic_assign_expand_fenv): New function.
6362         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
6363
6364 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6365
6366         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
6367         (__builtin_mips_set_fcsr): Likewise.
6368         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
6369         MIPS_USI_FTYPE_VOID.
6370         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
6371         (mips16_expand_set_fcsr): Likewise.
6372         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
6373         (mips16_set_fcsr_stub): Likewise.
6374         (mips16_get_fcsr_one_only_stub): New class.
6375         (mips16_set_fcsr_one_only_stub): Likewise.
6376         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
6377         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
6378         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
6379         (hard_float): New availability predicate.
6380         (mips_builtins): Add get_fcsr and set_fcsr.
6381         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
6382         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
6383         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
6384         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
6385         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
6386         patterns.
6387
6388 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6389
6390         * config/mips/mips.c (mips_one_only_stub): New class.
6391         (mips_need_mips16_rdhwr_p): Replace with...
6392         (mips16_rdhwr_stub): ...this new variable.
6393         (mips16_stub_call_address): New function.
6394         (mips16_rdhwr_one_only_stub): New class.
6395         (mips_expand_thread_pointer): Use mips16_stub_call_address.
6396         (mips_output_mips16_rdhwr): Delete.
6397         (mips_finish_stub): New function.
6398         (mips_code_end): Use it to handle rdhwr stubs.
6399
6400 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
6401
6402         PR target/60017
6403         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
6404         when calculating size of integer atomic types.
6405
6406 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
6407
6408         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
6409
6410 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
6411
6412         PR tree-optimization/60003
6413         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
6414         * profile.c (branch_prob): Use gimple_call_builtin_p
6415         to check for BUILT_IN_SETJMP_RECEIVER.
6416         * tree-inline.c (copy_bb): Call notice_special_calls.
6417
6418 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
6419
6420         PR bootstrap/59985
6421         * lra-constraints.c (process_alt_operands): Update reload_sum only
6422         on the first pass.
6423
6424 2014-01-31  Richard Henderson  <rth@redhat.com>
6425
6426         PR middle-end/60004
6427         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
6428         until after else_eh is processed.
6429
6430 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6431
6432         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
6433         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
6434         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
6435         in smmintrin.h, remove them.
6436         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
6437         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
6438         * config/i386/i386.md (ROUND_SAE): Fix value.
6439         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
6440         (const48_operand): New.
6441         * config/i386/subst.md (round), (round_expand): Use
6442         const_4_or_8_to_11_operand.
6443         (round_saeonly), (round_saeonly_expand): Use const48_operand.
6444
6445 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6446
6447         * config/i386/constraints.md (Yk): Swap meaning with k.
6448         * config/i386/i386.md (movhi_internal): Change Yk to k.
6449         (movqi_internal): Ditto.
6450         (*k<logic><mode>): Ditto.
6451         (*andhi_1): Ditto.
6452         (*andqi_1): Ditto.
6453         (kandn<mode>): Ditto.
6454         (*<code>hi_1): Ditto.
6455         (*<code>qi_1): Ditto.
6456         (kxnor<mode>): Ditto.
6457         (kortestzhi): Ditto.
6458         (kortestchi): Ditto.
6459         (kunpckhi): Ditto.
6460         (*one_cmplhi2_1): Ditto.
6461         (*one_cmplqi2_1): Ditto.
6462         * config/i386/sse.md (): Change k to Yk.
6463         (avx512f_load<mode>_mask): Ditto.
6464         (avx512f_blendm<mode>): Ditto.
6465         (avx512f_store<mode>_mask): Ditto.
6466         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
6467         (avx512f_storedqu<mode>_mask): Ditto.
6468         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
6469         Ditto.
6470         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
6471         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
6472         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
6473         (avx512f_maskcmp<mode>3): Ditto.
6474         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
6475         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
6476         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
6477         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
6478         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
6479         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
6480         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
6481         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
6482         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
6483         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
6484         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
6485         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
6486         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
6487         (vec_extract_lo_<mode>_maskm): Ditto.
6488         (vec_extract_hi_<mode>_maskm): Ditto.
6489         (avx512f_vternlog<mode>_mask): Ditto.
6490         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
6491         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
6492         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
6493         (avx512f_<code>v8div16qi2_mask): Ditto.
6494         (avx512f_<code>v8div16qi2_mask_store): Ditto.
6495         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
6496         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
6497         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
6498         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
6499         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6500         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6501         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6502         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6503         (avx512cd_maskb_vec_dupv8di): Ditto.
6504         (avx512cd_maskw_vec_dupv16si): Ditto.
6505         (avx512f_vpermi2var<mode>3_maskz): Ditto.
6506         (avx512f_vpermi2var<mode>3_mask): Ditto.
6507         (avx512f_vpermi2var<mode>3_mask): Ditto.
6508         (avx512f_vpermt2var<mode>3_maskz): Ditto.
6509         (*avx512f_gathersi<mode>): Ditto.
6510         (*avx512f_gathersi<mode>_2): Ditto.
6511         (*avx512f_gatherdi<mode>): Ditto.
6512         (*avx512f_gatherdi<mode>_2): Ditto.
6513         (*avx512f_scattersi<mode>): Ditto.
6514         (*avx512f_scatterdi<mode>): Ditto.
6515         (avx512f_compress<mode>_mask): Ditto.
6516         (avx512f_compressstore<mode>_mask): Ditto.
6517         (avx512f_expand<mode>_mask): Ditto.
6518         * config/i386/subst.md (mask): Change k to Yk.
6519         (mask_scalar_merge): Ditto.
6520         (sd): Ditto.
6521
6522 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
6523
6524         * doc/extend.texi (Vector Extensions): Document ?: in C++.
6525
6526 2014-01-31  Richard Biener  <rguenther@suse.de>
6527
6528         PR middle-end/59990
6529         * builtins.c (fold_builtin_memory_op): Make sure to not
6530         use a floating-point mode or a boolean or enumeral type for
6531         the copy operation.
6532
6533 2014-01-30  DJ Delorie  <dj@redhat.com>
6534
6535         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
6536         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
6537         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6538         whenever main() has an epilogue.
6539
6540 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6541
6542         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
6543         unused variable "field".
6544         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
6545         (vsx_mergeh_<mode>): Likewise.
6546         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
6547         (altivec_vmrghh): Likewise.
6548         (altivec_vmrghw): Likewise.
6549         (altivec_vmrglb): Likewise.
6550         (altivec_vmrglh): Likewise.
6551         (altivec_vmrglw): Likewise.
6552         (altivec_vspltb): Add missing uses.
6553         (altivec_vsplth): Likewise.
6554         (altivec_vspltw): Likewise.
6555         (altivec_vspltsf): Likewise.
6556
6557 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6558
6559         PR target/59923
6560         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
6561         frame related instructions.
6562
6563 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
6564
6565         PR rtl-optimization/59959
6566         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
6567         any reload of register whose subreg is invalid.
6568
6569 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6570
6571         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
6572         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
6573         Add missing return type - void.
6574
6575 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6576
6577         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
6578         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
6579         remove element index adjustment for endian (now handled in vsx.md
6580         and altivec.md).
6581         (altivec_expand_vec_perm_const): Use
6582         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
6583         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
6584         (vsx_xxspltw_<mode>): Adjust element index for little endian.
6585         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
6586         define_expand and a new define_insn *altivec_vspltb_internal;
6587         adjust for -maltivec=be on a little endian target.
6588         (altivec_vspltb_direct): New.
6589         (altivec_vsplth): Divide into a define_expand and a new
6590         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
6591         little endian target.
6592         (altivec_vsplth_direct): New.
6593         (altivec_vspltw): Divide into a define_expand and a new
6594         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
6595         little endian target.
6596         (altivec_vspltw_direct): New.
6597         (altivec_vspltsf): Divide into a define_expand and a new
6598         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
6599         a little endian target.
6600
6601 2014-01-30  Richard Biener  <rguenther@suse.de>
6602
6603         PR tree-optimization/59993
6604         * tree-ssa-forwprop.c (associate_pointerplus): Check we
6605         can propagate form the earlier stmt and avoid the transform
6606         when the intermediate result is needed.
6607
6608 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
6609
6610         * README.Portability: Fix typo.
6611
6612 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
6613
6614         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
6615         comparison_operator with ordered_comparison_operator.
6616
6617 2014-01-30  Nick Clifton  <nickc@redhat.com>
6618
6619         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
6620         Rename to mn10300_store_multiple_regs.
6621         * config/mn10300/mn10300.c: Likewise.
6622         * config/mn10300/mn10300.md (store_movm): Fix typo: call
6623         store_multiple_regs.
6624         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
6625         Call mn10300_store_multiple_regs.
6626
6627 2014-01-30  Nick Clifton  <nickc@redhat.com>
6628             DJ Delorie  <dj@redhat.com>
6629
6630         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
6631         %fp 2 to keep registers after it properly word-aligned.
6632         (rl78_alloc_physical_registers_umul): Handle the case where both
6633         input operands are the same.
6634
6635 2014-01-30  Richard Biener  <rguenther@suse.de>
6636
6637         PR tree-optimization/59903
6638         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
6639         check properly.
6640
6641 2014-01-30  Jason Merrill  <jason@redhat.com>
6642
6643         PR c++/59633
6644         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
6645
6646         PR c++/59645
6647         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
6648
6649 2014-01-30  Richard Biener  <rguenther@suse.de>
6650
6651         PR tree-optimization/59951
6652         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
6653
6654 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
6655
6656         PR target/59784
6657         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
6658         SFmode to DFmode case.
6659
6660 2014-01-29  DJ Delorie  <dj@redhat.com>
6661
6662         * config/msp430/msp430.opt (-minrt): New.
6663         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
6664         if -minrt given.
6665         (ENDFILE_SPEC): Likewise.
6666
6667 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
6668
6669         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
6670         (estimate_function_body_sizes): Use it.
6671
6672 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
6673
6674         PR c++/58561
6675         * dwarf2out.c (is_cxx_auto): New.
6676         (is_base_type): Use it.
6677         (gen_type_die_with_usage): Likewise.
6678
6679 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6680
6681         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
6682         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
6683         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
6684         -maltivec=be with LE targets.
6685         (vsx_mergeh_<mode>): Likewise.
6686         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
6687         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
6688         (altivec_vmrghb): Replace with define_expand and new
6689         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
6690         (altivec_vmrghb_direct): New define_insn.
6691         (altivec_vmrghh): Replace with define_expand and new
6692         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
6693         (altivec_vmrghh_direct): New define_insn.
6694         (altivec_vmrghw): Replace with define_expand and new
6695         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
6696         (altivec_vmrghw_direct): New define_insn.
6697         (*altivec_vmrghsf): Adjust for endianness.
6698         (altivec_vmrglb): Replace with define_expand and new
6699         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
6700         (altivec_vmrglb_direct): New define_insn.
6701         (altivec_vmrglh): Replace with define_expand and new
6702         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
6703         (altivec_vmrglh_direct): New define_insn.
6704         (altivec_vmrglw): Replace with define_expand and new
6705         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
6706         (altivec_vmrglw_direct): New define_insn.
6707         (*altivec_vmrglsf): Adjust for endianness.
6708         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6709         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6710         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6711         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6712         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6713         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6714         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6715         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6716
6717 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6718
6719         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
6720         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
6721         whitespace.
6722
6723 2014-01-29  Richard Biener  <rguenther@suse.de>
6724
6725         PR tree-optimization/58742
6726         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
6727         associate_pointerplus_align.
6728         (associate_pointerplus_diff): New function.
6729         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
6730         and associate_pointerplus_diff.
6731
6732 2014-01-29  Richard Biener  <rguenther@suse.de>
6733
6734         * lto-streamer.h (LTO_major_version): Bump to 3.
6735         (LTO_minor_version): Reset to 0.
6736
6737 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
6738
6739         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
6740         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
6741         (arm_file_start): Generate correct asm header for armv7ve.
6742         * config/arm/bpabi.h: Add multilib support for armv7ve.
6743         * config/arm/driver-arm.c: Change the architectures of cortex-a7
6744         and cortex-a15 to armv7ve.
6745         * config/arm/t-aprofile: Add multilib support for armv7ve.
6746         * doc/invoke.texi: Document -march=armv7ve.
6747
6748 2014-01-29  Richard Biener  <rguenther@suse.de>
6749
6750         PR tree-optimization/58742
6751         * tree-ssa-forwprop.c (associate_plusminus): Return true
6752         if we changed sth, defer EH cleanup to ...
6753         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
6754         (simplify_mult): New function.
6755
6756 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6757
6758         PR middle-end/59917
6759         PR tree-optimization/59920
6760         * tree.c (build_common_builtin_nodes): Remove
6761         __builtin_setjmp_dispatcher initialization.
6762         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
6763         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
6764         instead of gsi_after_labels + manually skipping debug stmts.
6765         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
6766         ignore bbs with IFN_ABNORMAL_DISPATCHER.
6767         * tree-inline.c (copy_edges_for_bb): Remove
6768         can_make_abnormal_goto argument, instead add abnormal_goto_dest
6769         argument.  Ignore computed_goto_p stmts.  Don't call
6770         make_abnormal_goto_edges.  If a call might need abnormal edges
6771         for non-local gotos, see if it already has an edge to
6772         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
6773         with true argument, don't do anything then, otherwise add
6774         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
6775         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
6776         caller.
6777         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
6778         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
6779         (lower_stmt): Don't set data->calls_builtin_setjmp.
6780         (lower_builtin_setjmp): Adjust comment.
6781         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
6782         * tree-cfg.c (found_computed_goto): Remove.
6783         (factor_computed_gotos): Remove.
6784         (make_goto_expr_edges): Return bool, true for computed gotos.
6785         Don't call make_abnormal_goto_edges.
6786         (build_gimple_cfg): Don't set found_computed_goto, don't call
6787         factor_computed_gotos.
6788         (computed_goto_p): No longer static.
6789         (make_blocks): Don't set found_computed_goto.
6790         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
6791         (make_edges): If make_goto_expr_edges returns true, push bb
6792         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
6793         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
6794         vector.  Record mapping between bbs and OpenMP regions if there
6795         are any, adjust make_gimple_omp_edges caller.  Call
6796         handle_abnormal_edges.
6797         (make_abnormal_goto_edges): Remove.
6798         * tree-cfg.h (make_abnormal_goto_edges): Remove.
6799         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
6800         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
6801         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
6802         * internal-fn.def (ABNORMAL_DISPATCHER): New.
6803         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
6804         filling *region also set *region_idx to (*region)->entry->index.
6805
6806         PR other/58712
6807         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
6808         For REGs set ORIGINAL_REGNO.
6809
6810 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
6811
6812         * doc/md.texi: Mention that a target shouldn't implement
6813         vec_widen_(s|u)mul_even/odd pair if it is less efficient
6814         than hi/lo pair.
6815
6816 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6817
6818         PR tree-optimization/59594
6819         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
6820         a copy of the datarefs vector rather than the vector itself.
6821
6822 2014-01-28  Jason Merrill  <jason@redhat.com>
6823
6824         PR c++/53756
6825         * dwarf2out.c (auto_die): New static.
6826         (gen_type_die_with_usage): Handle C++1y 'auto'.
6827         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
6828         on definition.
6829
6830 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
6831
6832         PR target/59672
6833         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
6834         (SPEC_X32): Likewise.
6835         (SPEC_64): Likewise.
6836         * config/i386/i386.c (ix86_option_override_internal): Turn off
6837         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
6838         for TARGET_16BIT.
6839         (x86_file_start): Output .code16gcc for TARGET_16BIT.
6840         * config/i386/i386.h (TARGET_16BIT): New macro.
6841         (TARGET_16BIT_P): Likewise.
6842         * config/i386/i386.opt: Add m16.
6843         * doc/invoke.texi: Document -m16.
6844
6845 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
6846
6847         PR preprocessor/59935
6848         * input.c (location_get_source_line): Bail out on when line number
6849         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
6850
6851 2014-01-28  Richard Biener  <rguenther@suse.de>
6852
6853         PR tree-optimization/58742
6854         * tree-ssa-forwprop.c (associate_plusminus): Handle
6855         pointer subtraction of the form (T)(P + A) - (T)P.
6856
6857 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6858
6859         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
6860         at const_int_cost.
6861
6862 2014-01-28  Richard Biener  <rguenther@suse.de>
6863
6864         Revert
6865         2014-01-28  Richard Biener  <rguenther@suse.de>
6866
6867         PR rtl-optimization/45364
6868         PR rtl-optimization/59890
6869         * var-tracking.c (local_get_addr_clear_given_value): Handle
6870         already cleared slot.
6871         (val_reset): Handle not allocated local_get_addr_cache.
6872         (vt_find_locations): Use post-order on the inverted CFG.
6873
6874 2014-01-28  Richard Biener  <rguenther@suse.de>
6875
6876         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
6877
6878 2014-01-28  Richard Biener  <rguenther@suse.de>
6879
6880         PR rtl-optimization/45364
6881         PR rtl-optimization/59890
6882         * var-tracking.c (local_get_addr_clear_given_value): Handle
6883         already cleared slot.
6884         (val_reset): Handle not allocated local_get_addr_cache.
6885         (vt_find_locations): Use post-order on the inverted CFG.
6886
6887 2014-01-28  Alan Modra  <amodra@gmail.com>
6888
6889         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
6890         * configure.ac <recursive call for build != host>: Define
6891         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
6892         and LD_FOR_BUILD too.
6893         * configure: Regenerate.
6894
6895 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
6896
6897         * config/i386/i386.c (get_builtin_code_for_version): Separate
6898         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
6899         Broadwell from Haswell.
6900
6901 2014-01-27  Steve Ellcey  <sellcey@mips.com>
6902
6903         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6904         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
6905         * config/mips/mips.c (mips_option_override): Change setting
6906         of TARGET_DSP.
6907         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
6908         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
6909         Change from Mask to Var.
6910
6911 2014-01-27  Jeff Law  <law@redhat.com>
6912
6913         * ipa-inline.c (inline_small_functions): Fix typo.
6914
6915 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
6916
6917         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
6918         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
6919         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
6920         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
6921         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
6922         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
6923         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
6924         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
6925         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
6926         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
6927         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
6928         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
6929         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
6930         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
6931         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
6932         (_mm512_storeu_epi64): Ditto.
6933         (_mm512_cmpge_epi32_mask): Ditto.
6934         (_mm512_cmpge_epu32_mask): Ditto.
6935         (_mm512_cmpge_epi64_mask): Ditto.
6936         (_mm512_cmpge_epu64_mask): Ditto.
6937         (_mm512_cmple_epi32_mask): Ditto.
6938         (_mm512_cmple_epu32_mask): Ditto.
6939         (_mm512_cmple_epi64_mask): Ditto.
6940         (_mm512_cmple_epu64_mask): Ditto.
6941         (_mm512_cmplt_epi32_mask): Ditto.
6942         (_mm512_cmplt_epu32_mask): Ditto.
6943         (_mm512_cmplt_epi64_mask): Ditto.
6944         (_mm512_cmplt_epu64_mask): Ditto.
6945         (_mm512_cmpneq_epi32_mask): Ditto.
6946         (_mm512_cmpneq_epu32_mask): Ditto.
6947         (_mm512_cmpneq_epi64_mask): Ditto.
6948         (_mm512_cmpneq_epu64_mask): Ditto.
6949         (_mm512_expand_pd): Ditto.
6950         (_mm512_expand_ps): Ditto.
6951         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
6952         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
6953         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
6954         * config/i386/i386.c (ix86_builtins): Add
6955         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
6956         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
6957         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
6958         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
6959         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
6960         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
6961         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
6962         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
6963         IX86_BUILTIN_PMOVUSQW512_MEM.
6964         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
6965         __builtin_ia32_pmovsqd512mem_mask,
6966         __builtin_ia32_pmovqd512mem_mask,
6967         __builtin_ia32_pmovusqw512mem_mask,
6968         __builtin_ia32_pmovsqw512mem_mask,
6969         __builtin_ia32_pmovqw512mem_mask,
6970         __builtin_ia32_pmovusdw512mem_mask,
6971         __builtin_ia32_pmovsdw512mem_mask,
6972         __builtin_ia32_pmovdw512mem_mask,
6973         __builtin_ia32_pmovqb512mem_mask,
6974         __builtin_ia32_pmovusqb512mem_mask,
6975         __builtin_ia32_pmovsqb512mem_mask,
6976         __builtin_ia32_pmovusdb512mem_mask,
6977         __builtin_ia32_pmovsdb512mem_mask,
6978         __builtin_ia32_pmovdb512mem_mask.
6979         (bdesc_args): Add __builtin_ia32_expanddf512,
6980         __builtin_ia32_expandsf512.
6981         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
6982         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
6983         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
6984         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
6985         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
6986         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
6987         (avx512f_<code>v8div16qi2_mask_store): This.
6988         (avx512f_expand<mode>): New.
6989
6990 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
6991
6992         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
6993         New.
6994         (_mm512_mask_prefetch_i64gather_pd): Ditto.
6995         (_mm512_prefetch_i32scatter_pd): Ditto.
6996         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6997         (_mm512_prefetch_i64scatter_pd): Ditto.
6998         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6999         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
7000         (_mm512_mask_prefetch_i64gather_ps): Ditto.
7001         (_mm512_prefetch_i32scatter_ps): Ditto.
7002         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
7003         (_mm512_prefetch_i64scatter_ps): Ditto.
7004         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
7005         * config/i386/i386-builtin-types.def: Define
7006         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
7007         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
7008         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
7009         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
7010         IX86_BUILTIN_SCATTERPFQPD.
7011         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
7012         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
7013         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
7014         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
7015         __builtin_ia32_scatterpfqps.
7016         (ix86_expand_builtin): Expand new built-ins.
7017         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
7018         fix memory access data type.
7019         (*avx512pf_gatherpf<mode>_mask): Ditto.
7020         (*avx512pf_gatherpf<mode>): Ditto.
7021         (avx512pf_scatterpf<mode>): Ditto.
7022         (*avx512pf_scatterpf<mode>_mask): Ditto.
7023         (*avx512pf_scatterpf<mode>): Ditto.
7024         (GATHER_SCATTER_SF_MEM_MODE): New.
7025         (avx512pf_gatherpf<mode>df): Ditto.
7026         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7027         (*avx512pf_scatterpf<mode>df): Ditto.
7028
7029 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
7030
7031         PR bootstrap/59934
7032         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
7033         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
7034         reached.
7035
7036 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
7037
7038         * common/config/arm/arm-common.c
7039         (arm_rewrite_mcpu): Handle multiple names.
7040         * config/arm/arm.h
7041         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
7042
7043 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
7044
7045         * gimple-builder.h (create_gimple_tmp): Delete.
7046
7047 2014-01-27  Christian Bruel  <christian.bruel@st.com>
7048
7049         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
7050         words comparisons.
7051
7052 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
7053
7054         * config/pa/pa.md (call): Generate indirect long calls to non-local
7055         functions when outputing 32-bit code.
7056         (call_value): Likewise except for special call to buggy powf function.
7057
7058         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
7059         portable runtime and PIC indirect calls.
7060         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
7061         and PIC call sequences.  Use ldo instead of blr to set return register
7062         in PIC call sequence.
7063
7064 2014-01-25  Walter Lee  <walt@tilera.com>
7065
7066         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
7067         avoid clobbering a live register.
7068
7069 2014-01-25  Walter Lee  <walt@tilera.com>
7070
7071         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
7072         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
7073         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
7074         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
7075
7076 2014-01-25  Walter Lee  <walt@tilera.com>
7077
7078         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
7079         arguments on even registers.
7080         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
7081         STACK_BOUNDARY.
7082         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
7083         (BIGGEST_ALIGNMENT): Ditto.
7084         (BIGGEST_FIELD_ALIGNMENT): Ditto.
7085
7086 2014-01-25  Walter Lee  <walt@tilera.com>
7087
7088         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
7089         insns before bundling.
7090         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
7091
7092 2014-01-25  Walter Lee  <walt@tilera.com>
7093
7094         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
7095         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
7096         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
7097
7098 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
7099
7100         * config/mips/constraints.md (kl): Delete.
7101         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
7102         define expands, using...
7103         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
7104         instructions for MIPS16.
7105         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
7106         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
7107
7108 2014-01-25  Walter Lee  <walt@tilera.com>
7109
7110         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
7111         (clzdi2): Ditto.
7112         (ffsdi2): Ditto.
7113
7114 2014-01-25  Walter Lee  <walt@tilera.com>
7115
7116         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
7117         (TARGET_EXPAND_TO_RTL_HOOK): Define.
7118
7119 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
7120
7121         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
7122         Handle XOR.
7123
7124 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
7125
7126         * print-rtl.c (in_call_function_usage): New var.
7127         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
7128         EXPR_LIST mode as mode and not as reg note name.
7129
7130         PR middle-end/59561
7131         * cfgloopmanip.c (copy_loop_info): If
7132         loop->warned_aggressive_loop_optimizations, make sure
7133         the flag is set in target loop too.
7134
7135 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7136
7137         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
7138         flag_cilkplus.
7139         * builtins.def: Likewise.
7140         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
7141         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
7142         * ira.c (ira_setup_eliminable_regset): Likewise.
7143         * omp-low.c (gate_expand_omp): Likewise.
7144         (execute_lower_omp): Likewise.
7145         (diagnose_sb_0): Likewise.
7146         (gate_diagnose_omp_blocks): Likewise.
7147         (simd_clone_clauses_extract): Likewise.
7148         (gate): Likewise.
7149
7150 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7151
7152         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
7153         correction for little endian...
7154         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
7155         here.
7156
7157 2014-01-24  Jeff Law  <law@redhat.com>
7158
7159         PR tree-optimization/59919
7160         * tree-vrp.c (find_assert_locations_1): Do not register asserts
7161         for non-returning calls.
7162
7163 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
7164
7165         * common/config/aarch64/aarch64-common.c
7166         (aarch64_rewrite_mcpu): Handle multiple names.
7167         * config/aarch64/aarch64.h
7168         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
7169
7170 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
7171
7172         * input.c (add_file_to_cache_tab): Handle the case where fopen
7173         returns NULL.
7174
7175 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
7176
7177         PR target/59929
7178         * config/i386/i386.md (pushsf splitter): Get stack adjustment
7179         from push operand if code of push isn't PRE_DEC.
7180
7181 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7182
7183         PR target/59909
7184         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
7185         -mquad-memory-atomic.  Update -mquad-memory documentation to say
7186         it is only used for non-atomic loads/stores.
7187
7188         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
7189         -mquad-memory or -mquad-memory-atomic switches.
7190
7191         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
7192         -mquad-memory-atomic to ISA 2.07 support.
7193
7194         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
7195         to separate support of normal quad word memory operations (ldq, stq)
7196         from the atomic quad word memory operations.
7197
7198         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
7199         support to separate non-atomic quad word operations from atomic
7200         quad word operations.  Disable non-atomic quad word operations in
7201         little endian mode so that we don't have to swap words after the
7202         load and before the store.
7203         (quad_load_store_p): Add comment about atomic quad word support.
7204         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
7205         options printed with -mdebug=reg.
7206
7207         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
7208         -mquad-memory-atomic as the test for whether we have quad word
7209         atomic instructions.
7210         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
7211         or -mp8-vector are used, allow byte/half-word atomic operations.
7212
7213         * config/rs6000/sync.md (load_lockedti): Insure that the address
7214         is a proper indexed or indirect address for the lqarx instruction.
7215         On little endian systems, swap the hi/lo registers after the lqarx
7216         instruction.
7217         (load_lockedpti): Use indexed_or_indirect_operand predicate to
7218         insure the address is valid for the lqarx instruction.
7219         (store_conditionalti): Insure that the address is a proper indexed
7220         or indirect address for the stqcrx. instruction.  On little endian
7221         systems, swap the hi/lo registers before doing the stqcrx.
7222         instruction.
7223         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
7224         insure the address is valid for the stqcrx. instruction.
7225
7226         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7227         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
7228         type of quad memory support is available.
7229
7230 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
7231
7232         PR regression/59915
7233         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
7234         there is a danger of looping.
7235
7236 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
7237
7238         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7239         force flag_ira_loop_pressure if set via command line.
7240
7241 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7242
7243         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
7244         (ashr_simd): New builtin handling DI mode.
7245         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
7246         (aarch64_sshr_simddi): New match pattern.
7247         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
7248         (vshrd_n_s64): Likewise.
7249         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
7250
7251 2014-01-23  Nick Clifton  <nickc@redhat.com>
7252
7253         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
7254         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
7255         favour of mcu specific scripts.
7256         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
7257         430x multilibs.
7258
7259 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
7260             Alex Velenko  <Alex.Velenko@arm.com>
7261
7262         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
7263         (vaddv_s16): Likewise.
7264         (vaddv_s32): Likewise.
7265         (vaddv_u8): Likewise.
7266         (vaddv_u16): Likewise.
7267         (vaddv_u32): Likewise.
7268         (vaddvq_s8): Likewise.
7269         (vaddvq_s16): Likewise.
7270         (vaddvq_s32): Likewise.
7271         (vaddvq_s64): Likewise.
7272         (vaddvq_u8): Likewise.
7273         (vaddvq_u16): Likewise.
7274         (vaddvq_u32): Likewise.
7275         (vaddvq_u64): Likewise.
7276         (vaddv_f32): Likewise.
7277         (vaddvq_f32): Likewise.
7278         (vaddvq_f64): Likewise.
7279         (vmaxv_f32): Likewise.
7280         (vmaxv_s8): Likewise.
7281         (vmaxv_s16): Likewise.
7282         (vmaxv_s32): Likewise.
7283         (vmaxv_u8): Likewise.
7284         (vmaxv_u16): Likewise.
7285         (vmaxv_u32): Likewise.
7286         (vmaxvq_f32): Likewise.
7287         (vmaxvq_f64): Likewise.
7288         (vmaxvq_s8): Likewise.
7289         (vmaxvq_s16): Likewise.
7290         (vmaxvq_s32): Likewise.
7291         (vmaxvq_u8): Likewise.
7292         (vmaxvq_u16): Likewise.
7293         (vmaxvq_u32): Likewise.
7294         (vmaxnmv_f32): Likewise.
7295         (vmaxnmvq_f32): Likewise.
7296         (vmaxnmvq_f64): Likewise.
7297         (vminv_f32): Likewise.
7298         (vminv_s8): Likewise.
7299         (vminv_s16): Likewise.
7300         (vminv_s32): Likewise.
7301         (vminv_u8): Likewise.
7302         (vminv_u16): Likewise.
7303         (vminv_u32): Likewise.
7304         (vminvq_f32): Likewise.
7305         (vminvq_f64): Likewise.
7306         (vminvq_s8): Likewise.
7307         (vminvq_s16): Likewise.
7308         (vminvq_s32): Likewise.
7309         (vminvq_u8): Likewise.
7310         (vminvq_u16): Likewise.
7311         (vminvq_u32): Likewise.
7312         (vminnmv_f32): Likewise.
7313         (vminnmvq_f32): Likewise.
7314         (vminnmvq_f64): Likewise.
7315
7316 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
7317
7318         * config/aarch64/aarch64-simd.md
7319         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
7320         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
7321         (*aarch64_mul3_elt<mode>): Likewise.
7322         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
7323         (*aarch64_mul3_elt_to_64v2df): Likewise.
7324         (*aarch64_mla_elt<mode>): Likewise.
7325         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
7326         (*aarch64_mls_elt<mode>): Likewise.
7327         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
7328         (*aarch64_fma4_elt<mode>): Likewise.
7329         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
7330         (*aarch64_fma4_elt_to_64v2df): Likewise.
7331         (*aarch64_fnma4_elt<mode>): Likewise.
7332         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7333         (*aarch64_fnma4_elt_to_64v2df): Likewise.
7334         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
7335         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
7336         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
7337         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
7338         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7339         (aarch64_sqdmull_lane<mode>_internal): Likewise.
7340         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7341
7342 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7343
7344         * config/aarch64/aarch64-simd.md
7345         (aarch64_be_checked_get_lane<mode>): New define_expand.
7346         * config/aarch64/aarch64-simd-builtins.def
7347         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
7348         New builtin definition.
7349         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
7350         Use new safe be builtin.
7351
7352 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7353
7354         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
7355         New define_insn.
7356         (aarch64_be_st1<mode>): Likewise.
7357         (aarch_ld1<VALL:mode>): Define_expand modified.
7358         (aarch_st1<VALL:mode>): Likewise.
7359         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
7360         (UNSPEC_ST1): Likewise.
7361
7362 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
7363
7364         * config/microblaze/microblaze.md: Add trap insn and attribute
7365
7366 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
7367
7368         PR preprocessor/58580
7369         * input.h (location_get_source_line): Take an additional line_size
7370         parameter.
7371         (void diagnostics_file_cache_fini): Declare new function.
7372         * input.c (struct fcache): New type.
7373         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
7374         New static constants.
7375         (diagnostic_file_cache_init, total_lines_num)
7376         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
7377         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
7378         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
7379         (get_next_line, read_next_line, goto_next_line, read_line_num):
7380         New static function definitions.
7381         (diagnostic_file_cache_fini): New function.
7382         (location_get_source_line): Take an additional output line_len
7383         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
7384         read_line_num.
7385         * diagnostic.c (diagnostic_finish): Call
7386         diagnostic_file_cache_fini.
7387         (adjust_line): Take an additional input parameter for the length
7388         of the line, rather than calculating it with strlen.
7389         (diagnostic_show_locus): Adjust the use of
7390         location_get_source_line and adjust_line with respect to their new
7391         signature.  While displaying a line now, do not stop at the first
7392         null byte.  Rather, display the zero byte as a space and keep
7393         going until we reach the size of the line.
7394         * Makefile.in: Add vec.o to OBJS-libcommon
7395
7396 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
7397             Ilya Tocar     <ilya.tocar@intel.com>
7398
7399         * config/i386/avx512fintrin.h (_mm512_kmov): New.
7400         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
7401         (__builtin_ia32_kmov16): Ditto.
7402         * config/i386/i386.md (UNSPEC_KMOV): New.
7403         (kmovw): Ditto.
7404
7405 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
7406
7407         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
7408         (_mm512_storeu_si512): Ditto.
7409
7410 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
7411
7412         PR target/52125
7413         * rtl.h (get_referenced_operands): Declare.
7414         * recog.c (get_referenced_operands): New function.
7415         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
7416         operands have been referenced when recording LO_SUM references.
7417
7418 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
7419
7420         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
7421
7422 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7423
7424         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
7425         Enable for generic and recent AMD targets.
7426
7427 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7428
7429         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
7430         ARG_SIZE note when adjustment was eliminated.
7431
7432 2014-01-22  Jeff Law  <law@redhat.com>
7433
7434         PR tree-optimization/59597
7435         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
7436         in file.  Accept new argument REGISTERING and use it to modify
7437         dump output appropriately.
7438         (register_jump_thread): Corresponding changes.
7439         (mark_threaded_blocks): Reinstate code to cancel unprofitable
7440         thread paths involving joiner blocks.  Add code to dump cancelled
7441         jump threading paths.
7442
7443 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
7444
7445         PR rtl-optimization/59477
7446         * lra-constraints.c (inherit_in_ebb): Process call for living hard
7447         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
7448
7449 2014-01-22  Tom Tromey  <tromey@redhat.com>
7450
7451         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
7452         PARAMS.
7453         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
7454
7455 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7456
7457         PR rtl-optimization/59896
7458         * lra-constraints.c (process_alt_operands): Check unused note for
7459         matched operands of insn with no output reloads.
7460
7461 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
7462
7463         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
7464         (mips_move_from_gpr_cost): Likewise.
7465
7466 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7467
7468         PR rtl-optimization/59858
7469         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
7470         ira_class_hard_regs_num.
7471         (process_alt_operands): Increase reject for dying matched operand.
7472
7473 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
7474
7475         PR target/59003
7476         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
7477         smaller than size, perform several stores or loads and stores
7478         at dst + count - size to store or copy all of size bytes, rather
7479         than just last modesize bytes.
7480
7481 2014-01-20  DJ Delorie  <dj@redhat.com>
7482
7483         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
7484         that CLOBBERs are REGs before propogating their values.
7485
7486 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
7487
7488         PR middle-end/59789
7489         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
7490         (cgraph_inline_failed_type): New function.
7491         * cgraph.h (DEFCIFCODE): Add type.
7492         (cgraph_inline_failed_type_t): New enum.
7493         (cgraph_inline_failed_type): New prototype.
7494         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
7495         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
7496         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
7497         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
7498         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
7499         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
7500         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
7501         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
7502         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
7503         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
7504         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
7505         OPTIMIZATION_MISMATCH.
7506         * tree-inline.c (expand_call_inline): Emit errors during
7507         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
7508
7509 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7510
7511         PR target/59685
7512         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
7513         mode attribute in insn output.
7514
7515 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
7516
7517         * output.h (output_constant): Delete.
7518         * varasm.c (output_constant): Make private.
7519
7520 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
7521
7522         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
7523
7524 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7525
7526         PR middle-end/59860
7527         * tree.h (fold_builtin_strcat): New prototype.
7528         * builtins.c (fold_builtin_strcat): No longer static.  Add len
7529         argument, if non-NULL, don't call c_strlen.  Optimize
7530         directly into __builtin_memcpy instead of __builtin_strcpy.
7531         (fold_builtin_2): Adjust fold_builtin_strcat caller.
7532         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
7533
7534 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7535
7536         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7537         for SImode_address_operand operands, having only a REG argument.
7538
7539 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7540
7541         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
7542         loader name using mbig-endian.
7543         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
7544
7545 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7546
7547         * doc/invoke.texi (-march): Clarify documentation for AArch64.
7548         (-mtune): Likewise.
7549         (-mcpu): Likewise.
7550
7551 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
7552
7553         * config/aarch64/aarch64-protos.h
7554         (aarch64_cannot_change_mode_class_ptr): Declare.
7555         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
7556         aarch64_cannot_change_mode_class_ptr): New.
7557         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
7558         backend hook aarch64_cannot_change_mode_class.
7559
7560 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7561
7562         * common/config/aarch64/aarch64-common.c
7563         (aarch64_handle_option): Don't handle any option order logic here.
7564         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
7565         selected_cpu, warn on architecture version mismatch.
7566         (aarch64_override_options): Fix parsing order for option strings.
7567
7568 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7569             Iain Sandoe  <iain@codesourcery.com>
7570
7571         PR bootstrap/59496
7572         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
7573         warning.  Amend comment to reflect current functionality.
7574
7575 2014-01-20  Richard Biener  <rguenther@suse.de>
7576
7577         PR middle-end/59860
7578         * builtins.c (fold_builtin_strcat): Remove case better handled
7579         by tree-ssa-strlen.c.
7580
7581 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
7582
7583         * config/aarch64/aarch64.opt
7584         (mcpu, march, mtune): Make case-insensitive.
7585
7586 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7587
7588         PR target/59880
7589         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7590         if operands[1] is a REG or ZERO_EXTEND of a REG.
7591
7592 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
7593
7594         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
7595
7596 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
7597
7598         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
7599         long non-pic millicode calls.
7600
7601 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7602
7603         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
7604
7605 2014-01-19  Kito Cheng  <kito@0xlab.org>
7606
7607         * builtins.c (expand_movstr): Check movstr expand done or fail.
7608
7609 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7610             H.J. Lu  <hongjiu.lu@intel.com>
7611
7612         PR target/59379
7613         * config/i386/i386.md (*lea<mode>): Zero-extend return register
7614         to DImode for zero-extended addresses.
7615
7616 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
7617
7618         PR rtl-optimization/57763
7619         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
7620         on the new indirect jump_insn and increment LABEL_NUSES (label).
7621
7622 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
7623
7624         PR bootstrap/59580
7625         PR bootstrap/59583
7626         * config.gcc (x86_archs): New variable.
7627         (x86_64_archs): Likewise.
7628         (x86_cpus): Likewise.
7629         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
7630         --with-arch/--with-cpu= options.
7631         Support --with-arch=/--with-cpu={nehalem,westmere,
7632         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
7633
7634 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7635
7636         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
7637         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
7638
7639 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7640
7641         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
7642
7643 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
7644
7645         PR target/58944
7646         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
7647         clear cpp_get_options (parse_in)->warn_unused_macros for
7648         ix86_target_macros_internal with cpp_define.
7649
7650 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
7651
7652         * jump.c (delete_related_insns): Keep (use (insn))s.
7653         * reorg.c (redundant_insn): Check for barriers too.
7654
7655 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7656
7657         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
7658
7659 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
7660
7661         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
7662         call to $$dyncall when TARGET_LONG_CALLS is true.
7663
7664 2014-01-17  Jeff Law  <law@redhat.com>
7665
7666         * ree.c (combine_set_extension): Temporarily disable test for
7667         changing number of hard registers.
7668
7669 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7670
7671         PR middle-end/58125
7672         * ipa-inline-analysis.c (inline_free_summary):
7673         Do not free summary of aliases.
7674
7675 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7676
7677         PR middle-end/59706
7678         * gimplify.c (gimplify_expr): Use create_tmp_var
7679         instead of create_tmp_var_raw.  If cond doesn't have
7680         integral type, don't add the IFN_ANNOTATE builtin at all.
7681
7682 2014-01-17  Martin Jambor  <mjambor@suse.cz>
7683
7684         PR ipa/59736
7685         * ipa-cp.c (prev_edge_clone): New variable.
7686         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
7687         Also resize prev_edge_clone vector.
7688         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
7689         (ipcp_edge_removal_hook): New function.
7690         (ipcp_driver): Register ipcp_edge_removal_hook.
7691
7692 2014-01-17  Andrew Pinski  <apinski@cavium.com>
7693             Steve Ellcey  <sellcey@mips.com>
7694
7695         PR target/59462
7696         * config/mips/mips.c (mips_print_operand): Check operand mode instead
7697         of operator mode.
7698
7699 2014-01-17  Jeff Law  <law@redhat.com>
7700
7701         PR middle-end/57904
7702         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
7703         so that pass_ccp runs first.
7704
7705 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7706
7707         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
7708         (ix86_adjust_cost): Use !TARGET_XXX.
7709         (do_reorder_for_imul): Likewise.
7710         (swap_top_of_ready_list): Likewise.
7711         (ix86_sched_reorder): Likewise.
7712
7713 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7714
7715         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7716         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
7717         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
7718         (intel_memset): New.  Duplicate slm_memset.
7719         (intel_cost): New.  Duplicate slm_cost.
7720         (m_INTEL): New macro.
7721         (processor_target_table): Add "intel".
7722         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
7723         with PROCESSOR_INTEL for "intel".
7724         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
7725         PROCESSOR_SILVERMONT.
7726         (ix86_issue_rate): Likewise.
7727         (ix86_adjust_cost): Likewise.
7728         (ia32_multipass_dfa_lookahead): Likewise.
7729         (swap_top_of_ready_list): Likewise.
7730         (ix86_sched_reorder): Likewise.
7731         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
7732         instead of TARGET_OPT_AGU.
7733         * config/i386/i386.h (TARGET_INTEL): New.
7734         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
7735         (processor_type): Add PROCESSOR_INTEL.
7736         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
7737         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
7738
7739 2014-01-17  Marek Polacek  <polacek@redhat.com>
7740
7741         PR c/58346
7742         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
7743         size is zero.
7744
7745 2014-01-17  Richard Biener  <rguenther@suse.de>
7746
7747         PR tree-optimization/46590
7748         * opts.c (default_options_table): Add entries for
7749         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
7750         all enabled at -O1 but not for -Og.
7751         * common.opt (fbranch-count-reg): Remove Init(1).
7752         (fmove-loop-invariants): Likewise.
7753         (ftree-pta): Likewise.
7754
7755 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7756
7757         * config/i386/i386.c (ix86_data_alignment): For compatibility with
7758         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
7759         decls to at least the GCC 4.8 used alignments.
7760
7761         PR fortran/59440
7762         * tree-nested.c (convert_nonlocal_reference_stmt,
7763         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
7764         of GIMPLE_BIND stmts, adjust associated decls.
7765
7766 2014-01-17  Richard Biener  <rguenther@suse.de>
7767
7768         PR tree-optimization/46590
7769         * vec.h (vec<>::bseach): New member function implementing
7770         binary search according to C89 bsearch.
7771         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
7772         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
7773         bitmap pointer again.  Make accesses_in_loop a flat array.
7774         (mem_ref_obstack): New global.
7775         (outermost_indep_loop): Adjust for mem_ref->stored changes.
7776         (mark_ref_stored): Likewise.
7777         (ref_indep_loop_p_2): Likewise.
7778         (set_ref_stored_in_loop): New helper function.
7779         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
7780         (memref_free): Adjust.
7781         (record_mem_ref_loc): Simplify.
7782         (gather_mem_refs_stmt): Adjust.
7783         (sort_locs_in_loop_postorder_cmp): New function.
7784         (analyze_memory_references): Sort accesses_in_loop after
7785         loop postorder number.
7786         (find_ref_loc_in_loop_cmp): New function.
7787         (for_all_locs_in_loop): Find relevant cluster of locs in
7788         accesses_in_loop and iterate without recursion.
7789         (execute_sm): Avoid uninit warning.
7790         (struct ref_always_accessed): Simplify.
7791         (ref_always_accessed::operator ()): Likewise.
7792         (ref_always_accessed_p): Likewise.
7793         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
7794         loop postorder numbers here.
7795         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
7796         numbers.
7797
7798 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7799
7800         PR c++/57945
7801         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
7802         on decls for which assemble_alias has been called.
7803
7804 2014-01-17  Nick Clifton  <nickc@redhat.com>
7805
7806         * config/msp430/msp430.opt: (mcpu): New option.
7807         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
7808         (msp430_option_override): Parse target_cpu.  If the MCU name
7809         matches a generic string, clear target_mcu.
7810         (msp430_attr): Allow numeric interrupt values up to 63.
7811         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
7812         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
7813         option.
7814         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
7815         Add mcpu matches.
7816         * config/msp430/msp430.md (popm): Use %J rather than %I.
7817         (addsi3): Use msp430_nonimmediate_operand for operand 2.
7818         (addhi_cy_i): Use immediate_operand for operand 2.
7819         * doc/invoke.texi: Document -mcpu option.
7820
7821 2014-01-17  Richard Biener  <rguenther@suse.de>
7822
7823         PR rtl-optimization/38518
7824         * df.h (df_analyze_loop): Declare.
7825         * df-core.c: Include cfgloop.h.
7826         (df_analyze_1): Split out main part of df_analyze.
7827         (df_analyze): Adjust.
7828         (loop_inverted_post_order_compute): New function.
7829         (loop_post_order_compute): Likewise.
7830         (df_analyze_loop): New function avoiding whole-function
7831         postorder computes.
7832         * loop-invariant.c (find_defs): Use df_analyze_loop.
7833         (find_invariants): Adjust.
7834         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
7835
7836 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
7837
7838         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
7839         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
7840
7841 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
7842
7843         * ipa-ref.c (ipa_remove_stmt_references): Fix references
7844         traversal when removing references.
7845
7846 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
7847
7848         PR ipa/59775
7849         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
7850
7851 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
7852
7853         PR middle-end/56791
7854         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
7855         pushing a reload for an autoinc when we had previously reloaded an
7856         inner part of the address.
7857
7858 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
7859
7860         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
7861         field.
7862         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
7863         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
7864         when not giving up or versioning for alias only because of
7865         loop->safelen.
7866         (vect_analyze_data_ref_dependences): Set to true.
7867         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
7868         is a GIMPLE_PHI.
7869         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
7870         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
7871         to the condition.
7872
7873         PR middle-end/58344
7874         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
7875
7876         PR target/59839
7877         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
7878         operand 0 predicate for gathers, use a new pseudo as subtarget.
7879
7880 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7881
7882         PR middle-end/59609
7883         * lra-constraints.c (process_alt_operands): Add printing debug info.
7884         Check absence of input/output reloads for matched operands too.
7885
7886 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7887
7888         PR rtl-optimization/59835
7889         * ira.c (ira_init_register_move_cost): Increase cost for
7890         impossible modes.
7891
7892 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
7893
7894         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
7895
7896 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
7897
7898         PR target/59780
7899         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
7900         non-register objects.  Use gen_(high/low)part more consistently.
7901         Fix assertions.
7902
7903 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
7904
7905         PR target/59844
7906         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
7907         endian support, remove tests for WORDS_BIG_ENDIAN.
7908         (p8_mfvsrd_3_<mode>): Likewise.
7909         (reload_gpr_from_vsx<mode>): Likewise.
7910         (reload_gpr_from_vsxsf): Likewise.
7911         (p8_mfvsrd_4_disf): Likewise.
7912
7913 2014-01-16  Richard Biener  <rguenther@suse.de>
7914
7915         PR rtl-optimization/46590
7916         * lcm.c (compute_antinout_edge): Use postorder iteration.
7917         (compute_laterin): Use inverted postorder iteration.
7918
7919 2014-01-16  Nick Clifton  <nickc@redhat.com>
7920
7921         PR middle-end/28865
7922         * varasm.c (output_constant): Return the number of bytes actually
7923         emitted.
7924         (output_constructor_array_range): Update the field size with the
7925         number of bytes emitted by output_constant.
7926         (output_constructor_regular_field): Likewise.  Also do not
7927         complain if the total number of bytes emitted is now greater
7928         than the expected fieldpos.
7929         * output.h (output_constant): Update prototype and descriptive comment.
7930
7931 2014-01-16  Marek Polacek  <polacek@redhat.com>
7932
7933         PR middle-end/59827
7934         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
7935         it is error_mark_node.
7936
7937 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
7938
7939         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
7940         VALID_AVX256_REG_OR_OI_MODE.
7941
7942 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
7943
7944         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
7945         current procedure should be profiled.
7946
7947 2014-01-15  Andrew Pinski  <apinski@cavium.com>
7948
7949         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
7950         of moving from/to the STACK_REG register class.
7951
7952 2014-01-15  Richard Henderson  <rth@redhat.com>
7953
7954         PR debug/54694
7955         * reginfo.c (global_regs_decl): Globalize.
7956         * rtl.h (global_regs_decl): Declare.
7957         * ira.c (do_reload): Diagnose frame_pointer_needed and it
7958         reserved via global_regs.
7959
7960 2014-01-15  Teresa Johnson  <tejohnson@google.com>
7961
7962         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
7963
7964 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7965
7966         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
7967         and vmulosh rather than call gen_vec_widen_smult_*.
7968         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
7969         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
7970         (vec_widen_smult_even_v16qi): Likewise.
7971         (vec_widen_umult_even_v8hi): Likewise.
7972         (vec_widen_smult_even_v8hi): Likewise.
7973         (vec_widen_umult_odd_v16qi): Likewise.
7974         (vec_widen_smult_odd_v16qi): Likewise.
7975         (vec_widen_umult_odd_v8hi): Likewise.
7976         (vec_widen_smult_odd_v8hi): Likewise.
7977         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
7978         vmuloub rather than call gen_vec_widen_umult_*.
7979         (vec_widen_umult_lo_v16qi): Likewise.
7980         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
7981         vmulosb rather than call gen_vec_widen_smult_*.
7982         (vec_widen_smult_lo_v16qi): Likewise.
7983         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
7984         rather than call gen_vec_widen_umult_*.
7985         (vec_widen_umult_lo_v8hi): Likewise.
7986         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
7987         rather than call gen_vec_widen_smult_*.
7988         (vec_widen_smult_lo_v8hi): Likewise.
7989
7990 2014-01-15  Jeff Law  <law@redhat.com>
7991
7992         PR tree-optimization/59747
7993         * ree.c (find_and_remove_re): Properly handle case where a second
7994         eliminated extension requires widening a copy created for elimination
7995         of a prior extension.
7996         (combine_set_extension): Ensure that the number of hard regs needed
7997         for a destination register does not change when we widen it.
7998
7999 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8000
8001         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
8002         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
8003         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
8004         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
8005         (avr-*-rtems*): Likewise.
8006         (bfin*-rtems*): Likewise.
8007         (moxie-*-rtems*): Likewise.
8008         (h8300-*-rtems*): Likewise.
8009         (i[34567]86-*-rtems*): Likewise.
8010         (lm32-*-rtems*): Likewise.
8011         (m32r-*-rtems*): Likewise.
8012         (m68k-*-rtems*): Likewise.
8013         (microblaze*-*-rtems*): Likewise.
8014         (mips*-*-rtems*): Likewise.
8015         (powerpc-*-rtems*): Likewise.
8016         (sh-*-rtems*): Likewise.
8017         (sparc-*-rtems*): Likewise.
8018         (sparc64-*-rtems*): Likewise.
8019         (v850-*-rtems*): Likewise.
8020         (m32c-*-rtems*): Likewise.
8021
8022 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
8023
8024         PR rtl-optimization/59511
8025         * ira.c (ira_init_register_move_cost): Use memory costs for some
8026         cases of register move cost calculations.
8027         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
8028         instead of BB frequency.
8029         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
8030         * lra-assigns.c (find_hard_regno_for): Ditto.
8031
8032 2014-01-15  Richard Biener  <rguenther@suse.de>
8033
8034         PR tree-optimization/59822
8035         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
8036         (vectorizable_load): Use it to hoist defs of uses of invariant
8037         loads out of the loop.
8038
8039 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8040             Kugan Vivekanandarajah  <kuganv@linaro.org>
8041
8042         PR target/59695
8043         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
8044         truncation.
8045
8046 2014-01-15  Richard Biener  <rguenther@suse.de>
8047
8048         PR rtl-optimization/59802
8049         * lcm.c (compute_available): Use inverted postorder to seed
8050         the initial worklist.
8051
8052 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8053
8054         PR target/59803
8055         * config/s390/s390.c (s390_preferred_reload_class): Don't return
8056         ADDR_REGS for invalid symrefs in non-PIC code.
8057
8058 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
8059
8060         PR other/58712
8061         * builtins.c (determine_block_size): Initialize *probable_max_size
8062         even if len_rtx is CONST_INT.
8063
8064 2014-01-14  Andrew Pinski  <apinski@cavium.com>
8065
8066         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
8067         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
8068         (cortexa53_tunings): Likewise.
8069         (aarch64_sched_issue_rate): New function.
8070         (TARGET_SCHED_ISSUE_RATE): Define.
8071
8072 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
8073
8074         * ira-costs.c (find_costs_and_classes): Add missed
8075         ira_init_register_move_cost_if_necessary.
8076
8077 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
8078
8079         PR target/59787
8080         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
8081
8082 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
8083
8084         PR target/59794
8085         * config/i386/i386.c (type_natural_mode): Add a bool parameter
8086         to indicate if type is used for function return value.  Warn ABI
8087         change if the vector mode isn't available for function return value.
8088         (ix86_function_arg_advance): Pass false to type_natural_mode.
8089         (ix86_function_arg): Likewise.
8090         (ix86_gimplify_va_arg): Likewise.
8091         (function_arg_32): Don't warn ABI change.
8092         (ix86_function_value): Pass true to type_natural_mode.
8093         (ix86_return_in_memory): Likewise.
8094         (ix86_struct_value_rtx): Removed.
8095         (TARGET_STRUCT_VALUE_RTX): Likewise.
8096
8097 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8098
8099         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
8100         converting a conditional jump into a conditional return.
8101
8102 2014-01-14  Richard Biener  <rguenther@suse.de>
8103
8104         PR tree-optimization/58921
8105         PR tree-optimization/59006
8106         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
8107         hoisting invariant stmts.
8108         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
8109         invariant loads on the preheader edge if possible.
8110
8111 2014-01-14  Joey Ye  <joey.ye@arm.com>
8112
8113         * doc/plugin.texi (Building GCC plugins): Update to C++.
8114
8115 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
8116
8117         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
8118         (_mm_rcp28_round_ss): Ditto.
8119         (_mm_rsqrt28_round_sd): Ditto.
8120         (_mm_rsqrt28_round_ss): Ditto.
8121         (_mm_rcp28_sd): Ditto.
8122         (_mm_rcp28_ss): Ditto.
8123         (_mm_rsqrt28_sd): Ditto.
8124         (_mm_rsqrt28_ss): Ditto.
8125         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
8126         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
8127         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
8128         (IX86_BUILTIN_RCP28SD): Ditto.
8129         (IX86_BUILTIN_RCP28SS): Ditto.
8130         (IX86_BUILTIN_RSQRT28SD): Ditto.
8131         (IX86_BUILTIN_RSQRT28SS): Ditto.
8132         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
8133         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
8134         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
8135         (ix86_expand_special_args_builtin): Expand new FTYPE.
8136         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
8137         (srcp14<mode>): Make insn unary.
8138         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
8139         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
8140         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
8141         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
8142         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
8143         Fix rounding: make it SAE only.
8144         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
8145         Ditto.
8146         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
8147         Ditto.
8148         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
8149         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
8150         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
8151         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
8152         (round_saeonly_mask_scalar_operand4): Ditto.
8153         (round_saeonly_mask_scalar_op3): Ditto.
8154         (round_saeonly_mask_scalar_op4): Ditto.
8155
8156 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8157
8158         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8159         Implement -maltivec=be for vec_insert and vec_extract.
8160
8161 2014-01-10  DJ Delorie  <dj@redhat.com>
8162
8163         * config/msp430/msp430.md (call_internal): Don't allow memory
8164         references with SP as the base register.
8165         (call_value_internal): Likewise.
8166         * config/msp430/constraints.md (Yc): New.  For memory references
8167         that don't use SP as a base register.
8168
8169         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
8170         "an integer without a # prefix"
8171         * config/msp430/msp430.md (epilogue_helper): Use it.
8172
8173 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
8174
8175         PR target/59617
8176         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
8177         AVX512F gather builtins.
8178         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
8179         on gather decls with INTEGER_TYPE masktype.
8180         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
8181         directly into the builtin rather than hoisting it before loop.
8182
8183         PR tree-optimization/59387
8184         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
8185         (scev_const_prop): If folded_casts and type has undefined overflow,
8186         use force_gimple_operand instead of force_gimple_operand_gsi and
8187         for each added stmt if it is assign with
8188         arith_code_with_undefined_signed_overflow, call
8189         rewrite_to_defined_overflow.
8190         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
8191         gimple-fold.h instead.
8192         (arith_code_with_undefined_signed_overflow,
8193         rewrite_to_defined_overflow): Moved to ...
8194         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
8195         rewrite_to_defined_overflow): ... here.  No longer static.
8196         Include gimplify-me.h.
8197         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
8198         rewrite_to_defined_overflow): New prototypes.
8199
8200 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8201
8202         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
8203
8204 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
8205
8206         * builtins.c (get_object_alignment_2): Minor tweak.
8207         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
8208
8209 2014-01-13  Christian Bruel  <christian.bruel@st.com>
8210
8211         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
8212         optimized non constant lengths.
8213
8214 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
8215
8216         PR libgomp/59194
8217         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
8218         load as __atomic_load_N if possible.
8219
8220 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
8221
8222         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
8223         target parameter.
8224         (rs6000_expand_builtin): Adjust call.
8225
8226 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
8227
8228         PR target/58115
8229         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
8230         * config/rs6000/rs6000.c: Include target-globals.h.
8231         (rs6000_set_current_function): Instead of doing target_reinit
8232         unconditionally, use save_target_globals_default_opts and
8233         restore_target_globals.
8234
8235         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
8236         FPSCR.
8237         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
8238         (rs6000_expand_builtin): Handle mffs and mtfsf.
8239         (rs6000_init_builtins): Define mffs and mtfsf.
8240         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
8241         (rs6000_mffs): New pattern.
8242         (rs6000_mtfsf): New pattern.
8243
8244 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
8245
8246         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
8247         Start narrowing with START.  Apply candidate-use pair
8248         and check overall cost in narrowing.
8249         (iv_ca_prune): Pass new argument.
8250
8251 2014-01-10  Jeff Law  <law@redhat.com>
8252
8253         PR middle-end/59743
8254         * ree.c (combine_reaching_defs): Ensure the defining statement
8255         occurs before the extension when optimizing extensions with
8256         different source and destination hard registers.
8257
8258 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
8259
8260         PR ipa/58585
8261         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
8262         vtables into the type inheritance graph.
8263
8264 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8265
8266         PR rtl-optimization/59754
8267         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
8268         modes in the REGNO != REGNO case.
8269
8270 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8271
8272         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
8273
8274 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8275
8276         PR tree-optimization/59745
8277         * tree-predcom.c (tree_predictive_commoning_loop): Call
8278         free_affine_expand_cache if giving up because components is NULL.
8279
8280         * target-globals.c (save_target_globals): Allocate < 4KB structs using
8281         GC in payload of target_globals struct instead of allocating them on
8282         the heap and the larger structs separately using GC.
8283         * target-globals.h (struct target_globals): Make regs, hard_regs,
8284         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
8285         of GTY((skip)) and change type to void *.
8286         (reset_target_globals): Cast loads from those fields to corresponding
8287         types.
8288
8289 2014-01-10  Steve Ellcey  <sellcey@mips.com>
8290
8291         PR plugins/59335
8292         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
8293         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
8294         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
8295
8296 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
8297
8298         PR target/59744
8299         * aarch64-modes.def (CC_Zmode): New flags mode.
8300         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
8301         represents an equality.
8302         (aarch64_get_condition_code): Handle CC_Zmode.
8303         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
8304
8305 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8306
8307         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
8308         extraction in good case.
8309
8310 2014-01-10  Richard Biener  <rguenther@suse.de>
8311
8312         PR tree-optimization/59374
8313         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
8314         checking after SLP discovery.  Mark stmts not participating
8315         in any SLP instance properly.
8316
8317 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8318
8319         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
8320         when handling a SET rtx.
8321
8322 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8323
8324         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
8325         (cortex-a57): Likewise.
8326         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
8327
8328 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8329
8330         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
8331         non-iwmmxt builtins.
8332
8333 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
8334
8335         PR ipa/58252
8336         PR ipa/59226
8337         * ipa-devirt.c record_target_from_binfo): Take as argument
8338         stack of binfos and lookup matching one for virtual inheritance.
8339         (possible_polymorphic_call_targets_1): Update.
8340
8341 2014-01-10  Huacai Chen  <chenhc@lemote.com>
8342
8343         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
8344         kernel strings for Loongson-2E/2F/3A.
8345
8346 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8347
8348         PR middle-end/59670
8349         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
8350         is_gimple_call before calling gimple_call_internal_p.
8351
8352 2014-01-09  Steve Ellcey  <sellcey@mips.com>
8353
8354         * Makefile.in (TREE_FLOW_H): Remove.
8355         (TREE_SSA_H): Add file names from tree-flow.h.
8356         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
8357         * tree.h: Remove tree-flow.h reference.
8358         * hash-table.h: Remove tree-flow.h reference.
8359         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
8360         reference with tree-ssa-loop.h.
8361
8362 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8363
8364         * doc/invoke.texi: Add -maltivec={be,le} options, and document
8365         default element-order behavior for -maltivec.
8366         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
8367         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
8368         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
8369         when targeting big endian, at least for now.
8370         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
8371
8372 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8373
8374         PR middle-end/47735
8375         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
8376         var satisfies use_register_for_decl, just take into account type
8377         alignment, rather than decl alignment.
8378
8379         PR tree-optimization/59622
8380         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
8381         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
8382         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
8383         Don't devirtualize for inplace at all.  For targets.length () == 1,
8384         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
8385
8386 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
8387
8388         * config/i386/i386.md (cpu): Remove the unused btver1.
8389
8390 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
8391
8392         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
8393
8394 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8395
8396         PR target/58115
8397         * tree-core.h (struct target_globals): New forward declaration.
8398         (struct tree_target_option): Add globals field.
8399         * tree.h (TREE_TARGET_GLOBALS): Define.
8400         (prepare_target_option_nodes_for_pch): New prototype.
8401         * target-globals.h (struct target_globals): Define even if
8402         !SWITCHABLE_TARGET.
8403         * tree.c (prepare_target_option_node_for_pch,
8404         prepare_target_option_nodes_for_pch): New functions.
8405         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
8406         * config/i386/i386.c: Include target-globals.h.
8407         (ix86_set_current_function): Instead of doing target_reinit
8408         unconditionally, use save_target_globals_default_opts and
8409         restore_target_globals.
8410
8411 2014-01-09  Richard Biener  <rguenther@suse.de>
8412
8413         PR tree-optimization/59715
8414         * tree-cfg.h (split_critical_edges): Declare.
8415         * tree-cfg.c (split_critical_edges): Export.
8416         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
8417
8418 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
8419
8420         * cfgexpand.c (expand_stack_vars): Optionally disable
8421         asan stack protection.
8422         (expand_used_vars): Likewise.
8423         (partition_stack_vars): Likewise.
8424         * asan.c (asan_emit_stack_protection): Optionally disable
8425         after return stack usage.
8426         (instrument_derefs): Optionally disable memory access instrumentation.
8427         (instrument_builtin_call): Likewise.
8428         (instrument_strlen_call): Likewise.
8429         (asan_protect_global): Optionally disable global variables protection.
8430         * doc/invoke.texi: Added doc for new options.
8431         * params.def: Added new options.
8432         * params.h: Likewise.
8433
8434 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8435
8436         PR rtl-optimization/59724
8437         * ifcvt.c (cond_exec_process_if_block): Don't call
8438         flow_find_head_matching_sequence with 0 longest_match.
8439         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
8440         non-active insns if !stop_after.
8441         (try_head_merge_bb): Revert 2014-01-07 changes.
8442
8443 2014-01-08  Jeff Law  <law@redhat.com>
8444
8445         * ree.c (get_sub_rtx): New function, extracted from...
8446         (merge_def_and_ext): Here.
8447         (combine_reaching_defs): Use get_sub_rtx.
8448
8449 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
8450
8451         * cgraph.h (varpool_variable_node): Do not choke on null node.
8452
8453 2014-01-08  Catherine Moore  <clm@codesourcery.com>
8454
8455         * config/mips/mips.md (simple_return): Attempt to use JRC
8456         for microMIPS.
8457         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
8458
8459 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8460
8461         PR rtl-optimization/59137
8462         * reorg.c (steal_delay_list_from_target): Call update_block for
8463         elided insns.
8464         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
8465
8466 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8467
8468         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8469         two duplicate entries.
8470
8471 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8472
8473         Revert:
8474         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
8475
8476         * config/mips/mips.c (mips_truncated_op_cost): New function.
8477         (mips_rtx_costs): Adjust test for BADDU.
8478         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
8479
8480         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
8481
8482         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
8483         (*baddu_si): ...this new pattern.
8484
8485 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8486
8487         PR ipa/59722
8488         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
8489
8490 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8491
8492         PR middle-end/57748
8493         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
8494         inner_reference_p.
8495         (expand_expr, expand_normal): Adjust.
8496         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
8497         inner_reference_p. Use inner_reference_p to expand inner references.
8498         (store_expr): Adjust.
8499         * cfgexpand.c (expand_call_stmt): Adjust.
8500
8501 2014-01-08  Rong Xu  <xur@google.com>
8502
8503         * gcov-io.c (gcov_var): Move from gcov-io.h.
8504         (gcov_position): Ditto.
8505         (gcov_is_error): Ditto.
8506         (gcov_rewrite): Ditto.
8507         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
8508         only part to libgcc/libgcov.h.
8509
8510 2014-01-08  Marek Polacek  <polacek@redhat.com>
8511
8512         PR middle-end/59669
8513         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
8514
8515 2014-01-08  Marek Polacek  <polacek@redhat.com>
8516
8517         PR sanitizer/59667
8518         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
8519
8520 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8521
8522         PR rtl-optimization/59649
8523         * stor-layout.c (get_mode_bounds): For BImode return
8524         0 and STORE_FLAG_VALUE.
8525
8526 2014-01-08  Richard Biener  <rguenther@suse.de>
8527
8528         PR middle-end/59630
8529         * gimple.h (is_gimple_builtin_call): Remove.
8530         (gimple_builtin_call_types_compatible_p): New.
8531         (gimple_call_builtin_p): New overload.
8532         * gimple.c (is_gimple_builtin_call): Remove.
8533         (validate_call): Rename to ...
8534         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
8535         check return types.
8536         (validate_type): New static function.
8537         (gimple_call_builtin_p): New overload and adjust.
8538         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
8539         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
8540         (gimple_fold_stmt_to_constant_1): Likewise.
8541         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
8542
8543 2014-01-08  Richard Biener  <rguenther@suse.de>
8544
8545         PR middle-end/59471
8546         * gimplify.c (gimplify_expr): Gimplify register-register type
8547         VIEW_CONVERT_EXPRs to separate stmts.
8548
8549 2014-01-07  Jeff Law  <law@redhat.com>
8550
8551         PR middle-end/53623
8552         * ree.c (combine_set_extension): Handle case where source
8553         and destination registers in an extension insn are different.
8554         (combine_reaching_defs): Allow source and destination registers
8555         in extension to be different under limited circumstances.
8556         (add_removable_extension): Remove restriction that the
8557         source and destination registers in the extension are the same.
8558         (find_and_remove_re): Emit a copy from the extension's
8559         destination to its source after the defining insn if
8560         the source and destination registers are different.
8561
8562         PR middle-end/59285
8563         * ifcvt.c (merge_if_block): If we are merging a block with more than
8564         one successor with a block with no successors, remove any BARRIER
8565         after the second block.
8566
8567 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
8568
8569         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
8570
8571 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
8572
8573         PR target/59652
8574         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
8575         for 14-bit register offsets when INT14_OK_STRICT is false.
8576
8577 2014-01-07  Roland Stigge  <stigge@antcom.de>
8578             Michael Meissner  <meissner@linux.vnet.ibm.com>
8579
8580         PR 57386/target
8581         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
8582         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
8583
8584 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
8585
8586         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
8587         -mcpu.
8588
8589 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
8590
8591         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
8592         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
8593         rtx is const0_rtx or not.
8594
8595 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
8596
8597         PR target/58115
8598         * target-globals.c (save_target_globals): Remove this_fn_optab
8599         handling.
8600         * toplev.c: Include optabs.h.
8601         (target_reinit): Temporarily restore the global options if another
8602         set of options are in force.
8603
8604 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
8605
8606         PR rtl-optimization/58668
8607         * cfgcleanup.c (flow_find_cross_jump): Don't count
8608         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
8609         to determine what is counted.
8610         (flow_find_head_matching_sequence): Use active_insn_p to determine
8611         what is counted.
8612         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
8613         counting change.
8614         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
8615         determine what is counted.
8616
8617         PR tree-optimization/59643
8618         * tree-predcom.c (split_data_refs_to_components): If one dr is
8619         read and one write, determine_offset fails and the write isn't
8620         in the bad component, just put the read into the bad component.
8621
8622 2014-01-07  Mike Stump  <mikestump@comcast.net>
8623             Jakub Jelinek  <jakub@redhat.com>
8624
8625         PR pch/59436
8626         * tree-core.h (struct tree_optimization_option): Change optabs
8627         type from unsigned char * to void *.
8628         * optabs.c (init_tree_optimization_optabs): Adjust
8629         TREE_OPTIMIZATION_OPTABS initialization.
8630
8631 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
8632
8633         PR target/59644
8634         * config/i386/i386.h (struct machine_function): Add
8635         no_drap_save_restore field.
8636         * config/i386/i386.c (ix86_save_reg): Use
8637         !cfun->machine->no_drap_save_restore instead of
8638         crtl->stack_realign_needed.
8639         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
8640         this function clears frame_pointer_needed.  Set
8641         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
8642         and DRAP reg is needed.
8643
8644 2014-01-06  Marek Polacek  <polacek@redhat.com>
8645
8646         PR c/57773
8647         * doc/implement-c.texi: Mention that other integer types are
8648         permitted as bit-field types in strictly conforming mode.
8649
8650 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
8651
8652         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
8653         is newly allocated.
8654
8655 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
8656
8657         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
8658
8659 2014-01-06  Martin Jambor  <mjambor@suse.cz>
8660
8661         PR ipa/59008
8662         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
8663         to int.
8664         * ipa-prop.c (ipa_print_node_params): Fix indentation.
8665
8666 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
8667
8668         PR debug/59350
8669         PR debug/59510
8670         * var-tracking.c (add_stores): Preserve the value of the source even if
8671         we don't record the store.
8672
8673 2014-01-06  Terry Guo  <terry.guo@arm.com>
8674
8675         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
8676
8677 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
8678
8679         PR bootstrap/59541
8680         * config/darwin.c (darwin_function_section): Adjust return values to
8681         correspond to optimisation changes made in r206070.
8682
8683 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
8684
8685         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
8686         from prefetch_block tune setting.
8687         (nocona_cost): Correct size of prefetch block to 64.
8688
8689 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
8690
8691         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
8692         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
8693         used to save the static chain register in the computation of the offset
8694         from which the FP registers need to be restored.
8695
8696 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
8697
8698         PR tree-optimization/59519
8699         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
8700         ICE if get_current_def (current_new_name) is already non-NULL, as long
8701         as it is a phi result of some other phi in *new_exit_bb that has
8702         the same argument.
8703
8704         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
8705         or vmovdqu* for misaligned_operand.
8706         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
8707         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
8708         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
8709         aligned_mem for AVX512F masked aligned load and store builtins and for
8710         non-temporal moves.
8711
8712 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
8713
8714         PR tree-optimization/59651
8715         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
8716         Address range for negative step should be added by TYPE_SIZE_UNIT.
8717
8718 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
8719
8720         * config/m68k/m68k.c (handle_move_double): Handle pushes with
8721         overlapping registers also for registers other than the stack pointer.
8722
8723 2014-01-03  Marek Polacek  <polacek@redhat.com>
8724
8725         PR other/59661
8726         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
8727         __builtin_FILE.
8728
8729 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
8730
8731         PR target/59625
8732         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
8733         asm goto as jump.
8734
8735         * config/i386/i386.md (MODE_SIZE): New mode attribute.
8736         (push splitter): Use <P:MODE_SIZE> instead of
8737         GET_MODE_SIZE (<P:MODE>mode).
8738         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
8739         (mov -1, reg peephole2): Likewise.
8740         * config/i386/sse.md (*mov<mode>_internal,
8741         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
8742         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
8743         *<code><mode>3, *andnot<mode>3<mask_name>,
8744         <mask_codefor><code><mode>3<mask_name>): Likewise.
8745         * config/i386/subst.md (mask_mode512bit_condition,
8746         sd_mask_mode512bit_condition): Likewise.
8747
8748 2014-01-02  Xinliang David Li  <davidxl@google.com>
8749
8750         PR tree-optimization/59303
8751         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
8752         (dump_predicates): Better output format.
8753         (pred_equal_p): New function.
8754         (is_neq_relop_p): Ditto.
8755         (is_neq_zero_form_p): Ditto.
8756         (pred_expr_equal_p): Ditto.
8757         (pred_neg_p): Ditto.
8758         (simplify_pred): Ditto.
8759         (simplify_preds_2): Ditto.
8760         (simplify_preds_3): Ditto.
8761         (simplify_preds_4): Ditto.
8762         (simplify_preds): Ditto.
8763         (push_pred): Ditto.
8764         (push_to_worklist): Ditto.
8765         (get_pred_info_from_cmp): Ditto.
8766         (is_degenerated_phi): Ditto.
8767         (normalize_one_pred_1): Ditto.
8768         (normalize_one_pred): Ditto.
8769         (normalize_one_pred_chain): Ditto.
8770         (normalize_preds): Ditto.
8771         (normalize_cond_1): Remove function.
8772         (normalize_cond): Ditto.
8773         (is_gcond_subset_of): Ditto.
8774         (is_subset_of_any): Ditto.
8775         (is_or_set_subset_of): Ditto.
8776         (is_and_set_subset_of): Ditto.
8777         (is_norm_cond_subset_of): Ditto.
8778         (pred_chain_length_cmp): Ditto.
8779         (convert_control_dep_chain_into_preds): Type change.
8780         (find_predicates): Ditto.
8781         (find_def_preds): Ditto.
8782         (destroy_predicates_vecs): Ditto.
8783         (find_matching_predicates_in_rest_chains): Ditto.
8784         (use_pred_not_overlap_with_undef_path_pred): Ditto.
8785         (is_pred_expr_subset): Ditto.
8786         (is_pred_chain_subset_of): Ditto.
8787         (is_included_in): Ditto.
8788         (is_superset_of): Ditto.
8789
8790 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8791
8792         Update copyright years.
8793
8794 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8795
8796         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
8797         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
8798         config/arc/arc.md, config/arc/arc.opt,
8799         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
8800         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
8801         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
8802         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
8803         config/linux-protos.h, config/linux.c, config/winnt-c.c,
8804         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
8805         vtable-verify.c, vtable-verify.h: Use the standard form for the
8806         copyright notice.
8807
8808 2014-01-02  Tobias Burnus  <burnus@net-b.de>
8809
8810         * gcc.c (process_command): Update copyright notice dates.
8811         * gcov-dump.c: Ditto.
8812         * gcov.c: Ditto.
8813         * doc/cpp.texi: Bump @copying's copyright year.
8814         * doc/cppinternals.texi: Ditto.
8815         * doc/gcc.texi: Ditto.
8816         * doc/gccint.texi: Ditto.
8817         * doc/gcov.texi: Ditto.
8818         * doc/install.texi: Ditto.
8819         * doc/invoke.texi: Ditto.
8820
8821 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8822
8823         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
8824
8825 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
8826
8827         * config/i386/sse.md (*mov<mode>_internal): Guard
8828         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
8829
8830         PR rtl-optimization/59647
8831         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
8832         new_rtx into UNSIGNED_FLOAT rtxes.
8833 \f
8834 Copyright (C) 2014 Free Software Foundation, Inc.
8835
8836 Copying and distribution of this file, with or without modification,
8837 are permitted in any medium without royalty provided the copyright
8838 notice and this notice are preserved.