gcc/ChangeLog

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