gcc/ChangeLog

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