gcc/ChangeLog

   1 2014-05-16  David Wohlferd <dw@LimeGreenSocks.com>
   2
   3          * doc/extend.texi: (Visibility Pragmas) Fix misplaced @xref
   4
   5 2014-05-16  Jan Hubicka  <hubicka@ucw.cz>
   6
   7         * varpool.c (dump_varpool_node): Dump write-only flag.
   8         * lto-cgraph.c (lto_output_varpool_node, input_varpool_node): Stream
   9         write-only flag.
  10         * tree-cfg.c (execute_fixup_cfg): Remove statements setting write-only variables.
  11
  12 2014-05-16  Vladimir Makarov  <vmakarov@redhat.com>
  13
  14         PR rtl-optimization/60969
  15         * ira-costs.c (record_reg_classes): Allow only memory for pseudo.
  16         Calculate costs for this case.
  17
  18 2014-05-16  Eric Botcazou  <ebotcazou@adacore.com>
  19
  20         * fold-const (fold_unary_loc) <NON_LVALUE_EXPR>: New case.
  21         <CASE_CONVERT>: Pass arg0 instead of op0 to fold_convert_const.
  22
  23 2014-05-16  Richard Biener  <rguenther@suse.de>
  24
  25         PR tree-optimization/61194
  26         * tree-vect-patterns.c (adjust_bool_pattern): Also handle
  27         bool patterns ending in a COND_EXPR.
  28
  29 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  30
  31         * config/aarch64/aarch64.c (aarch64_rtx_mult_cost): Fix FNMUL case.
  32
  33 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  34
  35         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle the case
  36         where we were unable to cost an RTX.
  37
  38 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  39
  40         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost SYMBOL_REF,
  41         HIGH, LO_SUM.
  42
  43 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  44             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  45
  46         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost TRUNCATE.
  47
  48 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  49             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  50
  51         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost FMA,
  52         FLOAT_EXTEND, FLOAT_TRUNCATE, ABS, SMAX, and SMIN.
  53
  54 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  55             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  56
  57         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost comparison
  58         operators.
  59
  60 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  61             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  62
  63         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
  64         DIV/MOD.
  65
  66 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  67             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  68
  69         * config/aarch64/aarch64.c (aarch64_rtx_arith_op_extract_p): New.
  70         (aarch64_rtx_costs): Improve costs for SIGN/ZERO_EXTRACT.
  71
  72 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  73             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  74
  75         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
  76         rotates and shifts.
  77
  78 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  79             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  80
  81         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost
  82         ZERO_EXTEND and SIGN_EXTEND better.
  83
  84 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  85             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  86
  87         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve cost for
  88         logical operations.
  89
  90 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  91             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
  92
  93         * config/aarch64/aarch64.c (aarch64_rtx_costs): Use address
  94         costs when costing loads and stores to memory.
  95
  96 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
  97             Philip Tomsich  <philipp.tomsich@theobroma-systems.com>
  98
  99         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costing
 100         for SET RTX.
 101
 102 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 103
 104         * config/aarch64/aarch64.c (aarch64_rtx_costs): Set default costs.
 105
 106 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 107             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 108
 109         * config/aarch64/aarch64.c (aarch64_strip_shift_or_extend): Rename
 110         to...
 111         (aarch64_strip_extend): ...this, don't strip shifts, check RTX is
 112         well formed.
 113         (aarch64_rtx_mult_cost): New.
 114         (aarch64_rtx_costs): Use it, refactor as appropriate.
 115
 116 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 117             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 118
 119         * config/aarch64/aarch64.c (aarch64_build_constant): Conditionally
 120         emit instructions, return number of instructions which would
 121         be emitted.
 122         (aarch64_add_constant): Update call to aarch64_build_constant.
 123         (aarch64_output_mi_thunk): Likewise.
 124         (aarch64_rtx_costs): Estimate cost of a CONST_INT, cost
 125         a CONST_DOUBLE.
 126
 127 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 128
 129         * config/aarch64/aarch64.c (aarch64_rtx_costs_wrapper): New.
 130         (TARGET_RTX_COSTS): Call it.
 131
 132 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 133
 134         * config/aarch64/aarch64.c (cortexa57_addrcost_table): New.
 135         (cortexa57_vector_cost): Likewise.
 136         (cortexa57_tunings): Use them.
 137
 138 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 139
 140         * config/aarch64/aarch64-protos.h (scale_addr_mode_cost): New.
 141         (cpu_addrcost_table): Use it.
 142         * config/aarch64/aarch64.c (generic_addrcost_table): Initialize it.
 143         (aarch64_address_cost): Rewrite using aarch64_classify_address,
 144         move it.
 145
 146 2014-05-16  Richard Biener  <rguenther@suse.de>
 147
 148         * tree-ssa-sccvn.c: Include tree-cfg.h and domwalk.h.
 149         (set_ssa_val_to): Handle unexpected sets to VN_TOP.
 150         (visit_phi): Ignore edges marked as not executable.
 151         (class cond_dom_walker): New.
 152         (cond_dom_walker::before_dom_children): Value-number
 153         control statements and mark successor edges as not
 154         executable if possible.
 155         (run_scc_vn): First walk all control statements in
 156         dominator order, marking edges as not executable.
 157         * tree-inline.c (copy_edges_for_bb): Be not confused
 158         about random edge flags.
 159
 160 2014-05-16  Richard Biener  <rguenther@suse.de>
 161
 162         * tree-ssa-sccvn.c (visit_use): Also constant-fold calls.
 163
 164 2014-05-15  Peter Bergner  <bergner@vnet.ibm.com>
 165
 166         PR target/61193
 167         * config/rs6000/htmxlintrin.h (_HTM_TBEGIN_STARTED): New define.
 168         (__TM_simple_begin): Use it.
 169         (__TM_begin): Likewise.
 170
 171 2014-05-15  Martin Jambor  <mjambor@suse.cz>
 172
 173         PR ipa/61085
 174         * ipa-prop.c (update_indirect_edges_after_inlining): Check
 175         type_preserved flag when the indirect edge is polymorphic.
 176
 177 2014-05-15  Martin Jambor  <mjambor@suse.cz>
 178
 179         PR tree-optimization/61090
 180         * tree-sra.c (sra_modify_expr): Pass the current gsi to
 181         build_ref_for_model.
 182
 183 2014-05-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 184
 185         * config/arm/arm.c (arm_option_override): Use the SCHED_PRESSURE_MODEL
 186         enum name for PARAM_SCHED_PRESSURE_ALGORITHM.
 187
 188 2014-05-15  Jakub Jelinek  <jakub@redhat.com>
 189
 190         PR tree-optimization/61158
 191         * fold-const.c (fold_binary_loc): If X is zero-extended and
 192         shiftc >= prec, make sure zerobits is all ones instead of
 193         invoking undefined behavior.
 194
 195 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 196
 197         * regcprop.h: New file.
 198         * regcprop.c (skip_debug_insn_p): New decl.
 199         (replace_oldest_value_reg): Check skip_debug_insn_p.
 200         (copyprop_hardreg_forward_bb_without_debug_insn): New function.
 201         * shrink-wrap.c: Include regcprop.h.
 202         (prepare_shrink_wrap): Call
 203         copyprop_hardreg_forward_bb_without_debug_insn.
 204
 205 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 206
 207         * shrink-wrap.h: Update comment.
 208         * shrink-wrap.c: Update comment.
 209         (next_block_for_reg): Rename to live_edge_for_reg.
 210         (live_edge_for_reg): Allow live_edge->dest has two predecessors.
 211         (move_insn_for_shrink_wrap): Split live_edge.
 212         (prepre_shrink_wrap): One more parameter for move_insn_for_shrink_wrap.
 213
 214 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
 215
 216         * config/sparc/sparc-protos.h (sparc_absnegfloat_split_legitimate):
 217         Delete.
 218         * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): Likewise.
 219         * config/sparc/sparc.md (fptype_ut699): New attribute.
 220         (in_branch_delay): Return false if -mfix-ut699 is specified and
 221         fptype_ut699 is set to single.
 222         (truncdfsf2): Add fptype_ut699 attribute.
 223         (fix_truncdfsi2): Likewise.
 224         (floatsisf2): Change fptype attribute.
 225         (fix_truncsfsi2): Likewise.
 226         (negtf2_notv9): Delete.
 227         (negtf2_v9): Likewise.
 228         (negtf2_hq): New instruction.
 229         (negtf2): New instruction and splitter.
 230         (negdf2_notv9): Rewrite.
 231         (abstf2_notv9): Delete.
 232         (abstf2_hq_v9): Likewise.
 233         (abstf2_v9): Likewise.
 234         (abstf2_hq): New instruction.
 235         (abstf2): New instruction and splitter.
 236         (absdf2_notv9): Rewrite.
 237
 238 2014-05-14  Cary Coutant  <ccoutant@google.com>
 239
 240         PR debug/61013
 241         * opts.c (common_handle_option): Don't special-case "-g".
 242         (set_debug_level): Default to at least level 2 with "-g".
 243
 244 2014-05-14  DJ Delorie  <dj@redhat.com>
 245
 246         * config/msp430/msp430.c (msp430_builtin): Add
 247         MSP430_BUILTIN_DELAY_CYCLES.
 248         (msp430_init_builtins): Register void __delay_cycles(long long).
 249         (msp430_builtin_decl): Add it.
 250         (cg_magic_constant): New.
 251         (msp430_expand_delay_cycles): New.
 252         (msp430_expand_builtin): Call it.
 253         (msp430_print_operand_raw): Change integer printing from "int" to
 254         HOST_WIDE_INT.
 255         * config/msp430/msp430.md (define_constants): Add delay_cycles tags.
 256         (delay_cycles_start): New.
 257         (delay_cycles_end): New.
 258         (delay_cycles_32): New.
 259         (delay_cycles_32x): New.
 260         (delay_cycles_16): New.
 261         (delay_cycles_16x): New.
 262         (delay_cycles_2): New.
 263         (delay_cycles_1): New.
 264         * doc/extend.texi: Document __delay_cycles().
 265
 266 2014-05-14  Sandra Loosemore  <sandra@codesourcery.com>
 267
 268         * config/nios2/nios2.md (nios2_cbranch): Fix paste-o in
 269         length attribute computation.
 270
 271 2014-05-14  Richard Sandiford  <rdsandiford@googlemail.com>
 272
 273         PR debug/61188
 274         * print-rtl.c (print_rtx): Suppress uids if flag_dump_unnumbered.
 275
 276 2014-05-14  Richard Sandiford  <r.sandiford@uk.ibm.com>
 277
 278         PR target/61084
 279         * config/sparc/sparc.md: Fix types of low and high in DI constant
 280         splitter.  Use gen_int_mode in some other splitters.
 281
 282 2014-05-14  Martin Jambor  <mjambor@suse.cz>
 283
 284         PR ipa/60897
 285         * ipa-prop.c (ipa_modify_formal_parameters): Reset DECL_LANG_SPECIFIC.
 286
 287 2014-05-14  James Norris  <jnorris@codesourcery.com>
 288
 289         * omp-low.c (expand_parallel_call): Remove shadow variable.
 290         (expand_omp_taskreg): Likewise.
 291
 292 2014-05-14  Ilya Tocar  <ilya.tocar@intel.com>
 293
 294         * common/config/i386/i386-common.c
 295         (OPTION_MASK_ISA_CLFLUSHOPT_SET): Define.
 296         (OPTION_MASK_ISA_XSAVES_SET): Ditto.
 297         (OPTION_MASK_ISA_XSAVEC_SET): Ditto.
 298         (OPTION_MASK_ISA_CLFLUSHOPT_UNSET): Ditto.
 299         (OPTION_MASK_ISA_XSAVES_UNSET): Ditto.
 300         (OPTION_MASK_ISA_XSAVEC_UNSET): Ditto.
 301         (ix86_handle_option): Handle OPT_mxsavec, OPT_mxsaves,
 302         OPT_mclflushopt.
 303         * config.gcc (i[34567]86-*-*): Add clflushoptintrin.h,
 304         xsavecintrin.h, xsavesintrin.h.
 305         (x86_64-*-*): Ditto.
 306         * config/i386/clflushoptintrin.h: New.
 307         * config/i386/xsavecintrin.h: Ditto.
 308         * config/i386/xsavesintrin.h: Ditto.
 309         * config/i386/cpuid.h (bit_CLFLUSHOPT): Define.
 310         (bit_XSAVES): Ditto.
 311         (bit_XSAVES): Ditto.
 312         * config/i386/driver-i386.c (host_detect_local_cpu): Handle
 313         -mclflushopt, -mxsavec, -mxsaves, -mno-xsaves, -mno-xsavec,
 314         -mno-clflushopt.
 315         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
 316         OPTION_MASK_ISA_CLFLUSHOPT, OPTION_MASK_ISA_XSAVEC,
 317         OPTION_MASK_ISA_XSAVES.
 318         * config/i386/i386.c (ix86_target_string): Handle -mclflushopt,
 319         -mxsavec, -mxsaves.
 320         (PTA_CLFLUSHOPT) Define.
 321         (PTA_XSAVEC): Ditto.
 322         (PTA_XSAVES): Ditto.
 323         (ix86_option_override_internal): Handle new options.
 324         (ix86_valid_target_attribute_inner_p): Ditto.
 325         (ix86_builtins): Add IX86_BUILTIN_XSAVEC, IX86_BUILTIN_XSAVEC64,
 326         IX86_BUILTIN_XSAVES, IX86_BUILTIN_XRSTORS, IX86_BUILTIN_XSAVES64,
 327         IX86_BUILTIN_XRSTORS64, IX86_BUILTIN_CLFLUSHOPT.
 328         (bdesc_special_args): Add __builtin_ia32_xsaves, __builtin_ia32_xrstors,
 329         __builtin_ia32_xsavec, __builtin_ia32_xsaves64, __builtin_ia32_xrstors64,
 330         __builtin_ia32_xsavec64.
 331         (ix86_init_mmx_sse_builtins): Add __builtin_ia32_clflushopt.
 332         (ix86_expand_builtin): Handle new builtins.
 333         * config/i386/i386.h (TARGET_CLFLUSHOPT) Define.
 334         (TARGET_CLFLUSHOPT_P): Ditto.
 335         (TARGET_XSAVEC): Ditto.
 336         (TARGET_XSAVEC_P): Ditto.
 337         (TARGET_XSAVES): Ditto.
 338         (TARGET_XSAVES_P): Ditto.
 339         * config/i386/i386.md (ANY_XSAVE): Add UNSPECV_XSAVEC, UNSPECV_XSAVES.
 340         (ANY_XSAVE64)" Add UNSPECV_XSAVEC64, UNSPECV_XSAVES64.
 341         (attr xsave): Add xsavec, xsavec64, xsaves, xsaves64.
 342         (ANY_XRSTOR): New.
 343         (ANY_XRSTOR64): Ditto.
 344         (xrstor): Ditto.
 345         (xrstor): Change into <xrstor>.
 346         (xrstor_rex64): Change into <xrstor>_rex64.
 347         (xrstor64): Change into <xrstor>64
 348         (clflushopt): New.
 349         * config/i386/i386.opt (mclflushopt): New.
 350         (mxsavec): Ditto.
 351         (mxsaves): Ditto.
 352         * config/i386/x86intrin.h: Add clflushoptintrin.h, xsavesintrin.h,
 353         xsavecintrin.h.
 354         * doc/invoke.texi: Document new options.
 355
 356 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
 357
 358         PR rtl-optimization/60866
 359         * sel-sched-ir (sel_init_new_insn): New parameter old_seqno.
 360         Default it to -1.  Pass it down to init_simplejump_data.
 361         (init_simplejump_data): New parameter old_seqno.  Pass it down
 362         to get_seqno_for_a_jump.
 363         (get_seqno_for_a_jump): New parameter old_seqno.  Use it for
 364         initializing new jump seqno as a last resort.  Add comment.
 365         (sel_redirect_edge_and_branch): Save old seqno of the conditional
 366         jump and pass it down to sel_init_new_insn.
 367         (sel_redirect_edge_and_branch_force): Likewise.
 368
 369 2014-05-14  Georg-Johann Lay  <avr@gjlay.de>
 370
 371         * config/avr/avr.h (REG_CLASS_CONTENTS): Use unsigned suffix for
 372         shifted values to avoid build warning.
 373
 374 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
 375
 376         * cfgcleanup.c (try_forward_edges): Use location_t for locations.
 377         * cfgrtl.c (rtl_merge_blocks): Fix comment.
 378         (cfg_layout_merge_blocks): Likewise.
 379         * except.c (emit_to_new_bb_before): Remove prev_bb local variable.
 380
 381 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
 382
 383         PR rtl-optimization/60901
 384         * config/i386/i386.c (ix86_dependencies_evaluation_hook): Check that
 385         bb predecessor belongs to the same scheduling region.  Adjust comment.
 386
 387 2014-05-13  Peter Bergner  <bergner@vnet.ibm.com>
 388
 389         * doc/sourcebuild.texi: (dfp_hw): Document.
 390         (p8vector_hw): Likewise.
 391         (powerpc_eabi_ok): Likewise.
 392         (powerpc_elfv2): Likewise.
 393         (powerpc_htm_ok): Likewise.
 394         (ppc_recip_hw): Likewise.
 395         (vsx_hw): Likewise.
 396
 397 2014-05-13  Cary Coutant  <ccoutant@google.com>
 398
 399         * opts.c (finish_options): Use -ggnu-pubnames with -gsplit-dwarf.
 400
 401 2014-05-13  David Malcolm  <dmalcolm@redhat.com>
 402
 403         * gengtype-parse.c (require3): Eliminate in favor of...
 404         (require4): New.
 405         (require_template_declaration): Update to support optional single *
 406         on a type.
 407
 408         * gengtype.c (get_ultimate_base_class): Add a non-const overload.
 409         (create_user_defined_type): Handle a single level of explicit
 410         pointerness within template arguments.
 411         (struct write_types_data): Add field "kind".
 412         (filter_type_name): Handle "*" character.
 413         (write_user_func_for_structure_ptr): Require a write_types_data
 414         rather than just a prefix string, so that we can look up the kind
 415         of the wtd and use it as an index into wrote_user_func_for_ptr,
 416         ensuring that such functions are written at most once.  Support
 417         subclasses by invoking the marking function of the ultimate base
 418         class.
 419         (write_user_func_for_structure_body): Require a write_types_data
 420         rather than just a prefix string, so that we can pass this to
 421         write_user_func_for_structure_ptr.
 422         (write_func_for_structure): Likewise.
 423         (ggc_wtd): Add initializer of new "kind" field.
 424         (pch_wtd): Likewise.
 425
 426         * gengtype.h (enum write_types_kinds): New.
 427         (struct type): Add field wrote_user_func_for_ptr to the "s"
 428         union member.
 429
 430 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
 431
 432         * fold-const.c (optimize_bit_field_compare): Use wi:: operations
 433         instead of const_binop.
 434         (fold_binary_loc): Likewise.
 435
 436 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
 437
 438         * tree-dfa.h (get_addr_base_and_unit_offset_1): Update array index
 439         calculation to match get_ref_base_and_extent.
 440
 441 2014-05-13  Catherine Moore  <clm@codesourcery.com>
 442             Sandra Loosemore  <sandra@codesourcery.com>
 443
 444         * configure.ac: Fix assembly for explicit JALR relocation check.
 445         * configure: Regenerate.
 446
 447 2014-05-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 448
 449         * config/arm/arm.c (neon_itype): Remove NEON_RESULTPAIR.
 450         (arm_init_neon_builtins): Remove handling of NEON_RESULTPAIR.
 451         Remove associated type declarations and initialisations.
 452         (arm_expand_neon_builtin): Likewise.
 453         (neon_emit_pair_result_insn): Delete.
 454         * config/arm/arm_neon_builtins (vtrn, vzip, vuzp): Delete.
 455         * config/arm/neon.md (neon_vtrn<mode>): Delete.
 456         (neon_vzip<mode>): Likewise.
 457         (neon_vuzp<mode>): Likewise.
 458
 459 2014-05-13  Richard Biener  <rguenther@suse.de>
 460
 461         PR ipa/60973
 462         * tree-inline.c (remap_gimple_stmt): Clear tail call flag,
 463         it needs revisiting whether the call still may be tail-called.
 464
 465 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 466
 467         * rtl.def (SYMBOL_REF): Remove middle "0" field.
 468         * rtl.h (block_symbol): Reduce number of fields to 2.
 469         (rtx_def): Add u2.symbol_ref_flags.
 470         (SYMBOL_REF_FLAGS): Use it.
 471         (SYMBOL_REF_DATA, SET_SYMBOL_REF_DECL, SYMBOL_REF_DECL)
 472         (SET_SYMBOL_REF_CONSTANT, SYMBOL_REF_CONSTANT): Lower index.
 473         * gengtype.c (adjust_field_rtx_def): Remove SYMBOL_REF_FLAGS handling.
 474         Lower index of SYMBOL_REF_DATA.
 475         * print-rtl.c (print_rtx): Lower index for SYMBOL_REF_DATA.
 476         Print SYMBOL_REF_FLAGS at the same time.
 477         * genattrtab.c (attr_rtx_1): Only initialize 1 "0" SYMBOL_REF field.
 478
 479 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 480
 481         * rtl.def (VAR_LOCATION): Remove "i" field.
 482         * rtl.h (rtx_def): Add u2.var_location_status.
 483         (PAT_VAR_LOCATION_STATUS): Use it.
 484         (gen_rtx_VAR_LOCATION): Declare.
 485         * gengenrtl.c (excluded_rtx): Add VAR_LOCATION.
 486         * emit-rtl.c (gen_rtx_VAR_LOCATION): New function.
 487         * var-tracking.c (emit_note_insn_var_location): Remove casts.
 488
 489 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 490
 491         * rtl.def (scratch): Fix outdated comment and remove "0" field.
 492         * gengtype.c (adjust_field_rtx_def): Update accordingly.
 493
 494 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 495
 496         * rtl.def (DEBUG_INSN, INSN, JUMP_INSN, CALL_INSN, JUMP_TABLE_DATA)
 497         (BARRIER, CODE_LABEL, NOTE): Remove first "i" field.
 498         * rtl.h (rtx_def): Add insn_uid to u2 field.
 499         (RTX_FLAG_CHECK8): Delete in favor of...
 500         (RTL_INSN_CHAIN_FLAG_CHECK): ...this new macro.
 501         (INSN_DELETED_P): Update accordingly.
 502         (INSN_UID): Use u2.insn_uid.
 503         (INSN_CHAIN_CODE_P): Define.
 504         (PREV_INSN, NEXT_INSN, BLOCK_FOR_INSN, PATTERN, INSN_LOCATION)
 505         (INSN_CODE, REG_NOTES, CALL_INSN_FUNCTION_USAGE, CODE_LABEL_NUMBER)
 506         (NOTE_DATA, NOTE_DELETED_LABEL_NAME, NOTE_BLOCK, NOTE_EH_HANDLER)
 507         (NOTE_BASIC_BLOCK, NOTE_VAR_LOCATION, NOTE_CFI, NOTE_LABEL_NUMBER)
 508         (NOTE_KIND, LABEL_NAME, LABEL_NUSES, JUMP_LABEL, LABEL_REFS): Lower
 509         indices accordingly.
 510         * print-rtl.c (print_rtx): Print INSN_UIDs before the main loop.
 511         Update indices for insn-chain rtxes.
 512         * gengtype.c (gen_rtx_next): Adjust test for insn-chain rtxes.
 513         (adjust_field_rtx_def): Lower '0' indices for all insn-chain rtxes.
 514         * emit-rtl.c (gen_label_rtx): Update gen_rtx_LABEL call.
 515         * caller-save.c (init_caller_save): Update gen_rtx_INSN calls.
 516         * combine.c (try_combine): Likewise.
 517         * ira.c (setup_prohibited_mode_move_regs): Likewise.
 518
 519 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 520
 521         * rtl.def (REG): Remove middle field.
 522         * rtl.h (rtx_def): Add orignal_regno to u2.
 523         (ORIGINAL_REGNO): Use it instead of field 1.
 524         (REG_ATTRS): Lower field index accordingly.
 525         * gengtype.c (adjust_field_rtx_def): Remove handling of
 526         ORIGINAL_REGNO.  Move REG_ATTRS index down.
 527         * print-rtl.c (print_rtx): Move ORIGINAL_REGNO handling to the
 528         code that prints the REGNO.
 529
 530 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 531
 532         * print-rtl.c (print_rtx): Guard whole '0' block with ifndef
 533         GENERATOR_FILE.
 534
 535 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 536
 537         * rtl.h (rtx_def): Mark u2 as GTY ((skip)).
 538
 539 2014-05-13  Bin Cheng  <bin.cheng@arm.com>
 540
 541         * tree-ssa-loop-ivopts.c (contain_complex_addr_expr): New.
 542         (alloc_iv): Lower base expressions containing ADDR_EXPR.
 543
 544 2014-05-13  Ian Bolton  <ian.bolton@arm.com>
 545
 546         * config/aarch64/aarch64-protos.h
 547         (aarch64_hard_regno_caller_save_mode): New prototype.
 548         * config/aarch64/aarch64.c (aarch64_hard_regno_caller_save_mode):
 549         New function.
 550         * config/aarch64/aarch64.h (HARD_REGNO_CALLER_SAVE_MODE): New macro.
 551
 552 2014-05-13  Christian Bruel  <christian.bruel@st.com>
 553
 554         * target.def (mode_switching): New hook vector.
 555         (mode_emit, mode_needed, mode_after, mode_entry): New hooks.
 556         (mode_exit, modepriority_to_mode): Likewise.
 557         * mode-switching.c (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Hookify.
 558         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 559         * target.h: Include tm.h and hard-reg-set.h.
 560         * doc/tm.texi.in (EMIT_MODE_SET, MODE_NEEDED, MODE_AFTER, MODE_ENTRY)
 561         (MODE_EXIT, MODE_PRIORITY_TO_MODE): Delete and hookify.
 562         * doc/tm.texi Regenerate.
 563         * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 564         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 565         * config/sh/sh.c (sh_emit_mode_set, sh_mode_priority): Hookify.
 566         (sh_mode_needed, sh_mode_after, sh_mode_entry, sh_mode_exit): Likewise.
 567         * config/i386/i386.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 568         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 569         * config/i386/i386-protos.h (ix86_mode_needed, ix86_mode_after)
 570         (ix86_mode_entrym, ix86_emit_mode_set): Remove external declaration.
 571         * config/i386/i386.c (ix86_mode_needed, ix86_mode_after, ix86_mode_exit,
 572         (ix86_mode_entry, ix86_mode_priority, ix86_emit_mode_set): Hookify.
 573         * config/epiphany/epiphany.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY):
 574         Delete
 575         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 576         * config/epiphany/epiphany-protos.h (epiphany_mode_needed)
 577         (emit_set_fp_mode, epiphany_mode_entry_exit, epiphany_mode_after)
 578         (epiphany_mode_priority_to_mode): Remove declaration.
 579         * config/epiphany/epiphany.c (emit_set_fp_mode): Hookify.
 580         (epiphany_mode_needed, epiphany_mode_priority_to_mode): Likewise.
 581         (epiphany_mode_entry, epiphany_mode_exit, epiphany_mode_after):
 582         Likewise.
 583         (epiphany_mode_priority_to_mode): Change priority type. Hookify.
 584         (epiphany_mode_needed, epiphany_mode_entry_exit): Hookify.
 585         (epiphany_mode_after, epiphany_mode_entry, emit_set_fp_mode): Hookify.
 586
 587 2014-05-13  Jakub Jelinek  <jakub@redhat.com>
 588
 589         PR target/61060
 590         * config/i386/i386.c (ix86_expand_set_or_movmem): If count_exp
 591         is const0_rtx, return immediately.  Don't test count == 0 when
 592         it is always true.
 593
 594 2014-05-13  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 595
 596         * Makefile.in: add shrink-wrap.o.
 597         * config/i386/i386.c: include "shrink-wrap.h"
 598         * function.c: Likewise.
 599         (requires_stack_frame_p, next_block_for_reg,
 600         move_insn_for_shrink_wrap, prepare_shrink_wrap,
 601         dup_block_and_redirect): Move to shrink-wrap.c
 602         (thread_prologue_and_epilogue_insns): Extract three code segments
 603         as functions in shrink-wrap.c
 604         * function.h: Move #ifdef HAVE_simple_return ... #endif block to
 605         shrink-wrap.h
 606         * shrink-wrap.c: New file.
 607         * shrink-wrap.h: New file.
 608
 609 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
 610
 611         * doc/extend.texi: Reflect current numbers of pragmas.  Remove
 612         reference to Solaris.
 613
 614 2014-05-12  Mike Stump  <mikestump@comcast.net>
 615
 616         PR other/31778
 617         * genattrtab.c (filename): Add.
 618         (convert_set_attr_alternative): Improve error message.
 619         (check_defs): Restore read_md_filename for error messages.
 620         (gen_insn): Save filename.
 621
 622 2014-05-12  Dimitris Papavasiliou  <dpapavas@gmail.com>
 623
 624         * doc/invoke.texi: Document new switches -Wno-shadow-ivar,
 625         -fno-local-ivars and -fivar-visibility.
 626         * c-family/c.opt: Make -Wshadow also implicitly enable
 627         -Wshadow-ivar.
 628
 629 2014-05-12  David Wohlferd <dw@LimeGreenSocks.com>
 630
 631         * doc/tm.texi: Remove reference to deleted macro.
 632         * doc/tm.texi.in: Likewise.
 633
 634 2014-05-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
 635
 636         PR target/60991
 637         * config/avr/avr.c (avr_out_store_psi): Use correct constant
 638         to restore Y.
 639
 640 2014-05-12  Georg-Johann Lay  <avr@gjlay.de>
 641
 642         * config/arm/arm.h (License): Add GCC Runtime Library Exception.
 643         * config/arm/aout.h (License): Same.
 644         * config/arm/bpabi.h (License): Same.
 645         * config/arm/elf.h (License): Same.
 646         * config/arm/linux-elf.h (License): Same.
 647         * config/arm/linux-gas.h (License): Same.
 648         * config/arm/netbsd-elf.h (License): Same.
 649         * config/arm/uclinux-eabi.h (License): Same.
 650         * config/arm/uclinux-elf.h (License): Same.
 651         * config/arm/vxworks.h (License): Same.
 652
 653 2014-05-11  Jakub Jelinek  <jakub@redhat.com>
 654
 655         * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
 656         * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
 657         number of operands to 3.
 658         (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
 659         * tree-nested.c (convert_nonlocal_omp_clauses,
 660         convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
 661         * gimplify.c (gimplify_scan_omp_clauses): Handle
 662         OMP_CLAUSE_LINEAR_STMT.
 663         * omp-low.c (lower_rec_input_clauses): Fix typo.
 664         (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
 665         cast between Fortran boolean_type_node and C _Bool if
 666         needed.
 667
 668 2014-05-11  Richard Sandiford  <rdsandiford@googlemail.com>
 669
 670         PR tree-optimization/61136
 671         * wide-int.h (multiple_of_p): Define a version that doesn't return
 672         the quotient.
 673         * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
 674         integer_zerop/const_binop pair.
 675         (multiple_of_p): Likewise, converting both operands to widest_int
 676         precision.
 677
 678 2014-05-09  Teresa Johnson  <tejohnson@google.com>
 679
 680         * cgraphunit.c (analyze_functions): Use correct dump file.
 681
 682 2014-05-09  Florian Weimer  <fweimer@redhat.com>
 683
 684         * cfgexpand.c (stack_protect_decl_p): New function, extracted from
 685         expand_used_vars.
 686         (stack_protect_return_slot_p): New function.
 687         (expand_used_vars): Call stack_protect_decl_p and
 688         stack_protect_return_slot_p for -fstack-protector-strong.
 689
 690 2014-05-09  Kenneth Zadeck  <zadeck@naturalbridge.com>
 691
 692         PR middle-end/61111
 693         * fold-const.c (fold_binary_loc): Changed width of mask.
 694
 695 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 696
 697         * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
 698         unsigned int initializers for regno_in, regno_out.
 699
 700 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 701
 702         PR target/61055
 703         * config/avr/avr.md (cc): Add new attribute set_vzn.
 704         (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
 705         Set cc insn attribute to set_vzn instead of set_zn for alternatives
 706         with INC, DEC or NEG.
 707         * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
 708         (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
 709         INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
 710
 711 2014-05-09  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 712
 713         Revert:
 714         2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 715
 716         * wide-int.cc (UTItype): Define.
 717         (UDWtype): Define for appropriate W_TYPE_SIZE.
 718
 719 2014-05-09  Richard Biener  <rguenther@suse.de>
 720
 721         * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
 722         * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
 723         (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
 724         (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
 725         ssa_propagate): Adjust.
 726
 727 2014-05-08  Jeff Law  <law@redhat.com>
 728
 729         PR tree-optimization/61009
 730         * tree-ssa-threadedge.c (thread_through_normal_block): Return a
 731         tri-state rather than a boolean.  When a block is too big to
 732         thread through, inform caller via negative return value.
 733         (thread_across_edge): If a block was too big for normal threading,
 734         then it's too big for a joiner too, so remove temporary equivalences
 735         and return immediately.
 736
 737 2014-05-08  Manuel López-Ibáñez  <manu@gcc.gnu.org>
 738             Matthias Klose  <doko@ubuntu.com>
 739
 740         PR driver/61106
 741         * optc-gen.awk: Fix option handling for -Wunused-parameter.
 742
 743 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 744
 745         PR target/59952
 746         * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
 747
 748 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 749
 750         PR target/61092
 751         * config/alpha/alpha.c: Include gimple-iterator.h.
 752         (alpha_gimple_fold_builtin): New function.  Move
 753         ALPHA_BUILTIN_UMULH folding from ...
 754         (alpha_fold_builtin): ... here.
 755         (TARGET_GIMPLE_FOLD_BUILTIN): New define.
 756
 757 2014-05-08  Wei Mi  <wmi@google.com>
 758
 759         PR target/58066
 760         * config/i386/i386.c (ix86_compute_frame_layout): Update
 761         preferred_stack_boundary for call, expanded from tls descriptor.
 762         * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
 763         to depend on SP register.
 764         (*tls_local_dynamic_base_32_gnu): Ditto.
 765         (*tls_local_dynamic_32_once): Ditto.
 766         (tls_global_dynamic_64_<mode>): Set
 767         ix86_tls_descriptor_calls_expanded_in_cfun.
 768         (tls_local_dynamic_base_64_<mode>): Ditto.
 769         (tls_global_dynamic_32): Set
 770         ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
 771         to depend on SP register.
 772         (tls_local_dynamic_base_32): Ditto.
 773
 774 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 775
 776         * config/arm/arm_neon.h: Update comment.
 777         * config/arm/neon-docgen.ml: Delete.
 778         * config/arm/neon-gen.ml: Delete.
 779         * doc/arm-neon-intrinsics.texi: Update comment.
 780
 781 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 782
 783         * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
 784         and v4sf versions.
 785         (vand, vorr, veor, vorn, vbic): Remove.
 786         * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
 787         iterator.
 788         (neon_vsub_unspec): Likewise.
 789         (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
 790
 791 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 792
 793         * config/arm/arm_neon.h (vadd_s8): GNU C implementation
 794         (vadd_s16): Likewise.
 795         (vadd_s32): Likewise.
 796         (vadd_f32): Likewise.
 797         (vadd_u8): Likewise.
 798         (vadd_u16): Likewise.
 799         (vadd_u32): Likewise.
 800         (vadd_s64): Likewise.
 801         (vadd_u64): Likewise.
 802         (vaddq_s8): Likewise.
 803         (vaddq_s16): Likewise.
 804         (vaddq_s32): Likewise.
 805         (vaddq_s64): Likewise.
 806         (vaddq_f32): Likewise.
 807         (vaddq_u8): Likewise.
 808         (vaddq_u16): Likewise.
 809         (vaddq_u32): Likewise.
 810         (vaddq_u64): Likewise.
 811         (vmul_s8): Likewise.
 812         (vmul_s16): Likewise.
 813         (vmul_s32): Likewise.
 814         (vmul_f32): Likewise.
 815         (vmul_u8): Likewise.
 816         (vmul_u16): Likewise.
 817         (vmul_u32): Likewise.
 818         (vmul_p8): Likewise.
 819         (vmulq_s8): Likewise.
 820         (vmulq_s16): Likewise.
 821         (vmulq_s32): Likewise.
 822         (vmulq_f32): Likewise.
 823         (vmulq_u8): Likewise.
 824         (vmulq_u16): Likewise.
 825         (vmulq_u32): Likewise.
 826         (vsub_s8): Likewise.
 827         (vsub_s16): Likewise.
 828         (vsub_s32): Likewise.
 829         (vsub_f32): Likewise.
 830         (vsub_u8): Likewise.
 831         (vsub_u16): Likewise.
 832         (vsub_u32): Likewise.
 833         (vsub_s64): Likewise.
 834         (vsub_u64): Likewise.
 835         (vsubq_s8): Likewise.
 836         (vsubq_s16): Likewise.
 837         (vsubq_s32): Likewise.
 838         (vsubq_s64): Likewise.
 839         (vsubq_f32): Likewise.
 840         (vsubq_u8): Likewise.
 841         (vsubq_u16): Likewise.
 842         (vsubq_u32): Likewise.
 843         (vsubq_u64): Likewise.
 844         (vand_s8): Likewise.
 845         (vand_s16): Likewise.
 846         (vand_s32): Likewise.
 847         (vand_u8): Likewise.
 848         (vand_u16): Likewise.
 849         (vand_u32): Likewise.
 850         (vand_s64): Likewise.
 851         (vand_u64): Likewise.
 852         (vandq_s8): Likewise.
 853         (vandq_s16): Likewise.
 854         (vandq_s32): Likewise.
 855         (vandq_s64): Likewise.
 856         (vandq_u8): Likewise.
 857         (vandq_u16): Likewise.
 858         (vandq_u32): Likewise.
 859         (vandq_u64): Likewise.
 860         (vorr_s8): Likewise.
 861         (vorr_s16): Likewise.
 862         (vorr_s32): Likewise.
 863         (vorr_u8): Likewise.
 864         (vorr_u16): Likewise.
 865         (vorr_u32): Likewise.
 866         (vorr_s64): Likewise.
 867         (vorr_u64): Likewise.
 868         (vorrq_s8): Likewise.
 869         (vorrq_s16): Likewise.
 870         (vorrq_s32): Likewise.
 871         (vorrq_s64): Likewise.
 872         (vorrq_u8): Likewise.
 873         (vorrq_u16): Likewise.
 874         (vorrq_u32): Likewise.
 875         (vorrq_u64): Likewise.
 876         (veor_s8): Likewise.
 877         (veor_s16): Likewise.
 878         (veor_s32): Likewise.
 879         (veor_u8): Likewise.
 880         (veor_u16): Likewise.
 881         (veor_u32): Likewise.
 882         (veor_s64): Likewise.
 883         (veor_u64): Likewise.
 884         (veorq_s8): Likewise.
 885         (veorq_s16): Likewise.
 886         (veorq_s32): Likewise.
 887         (veorq_s64): Likewise.
 888         (veorq_u8): Likewise.
 889         (veorq_u16): Likewise.
 890         (veorq_u32): Likewise.
 891         (veorq_u64): Likewise.
 892         (vbic_s8): Likewise.
 893         (vbic_s16): Likewise.
 894         (vbic_s32): Likewise.
 895         (vbic_u8): Likewise.
 896         (vbic_u16): Likewise.
 897         (vbic_u32): Likewise.
 898         (vbic_s64): Likewise.
 899         (vbic_u64): Likewise.
 900         (vbicq_s8): Likewise.
 901         (vbicq_s16): Likewise.
 902         (vbicq_s32): Likewise.
 903         (vbicq_s64): Likewise.
 904         (vbicq_u8): Likewise.
 905         (vbicq_u16): Likewise.
 906         (vbicq_u32): Likewise.
 907         (vbicq_u64): Likewise.
 908         (vorn_s8): Likewise.
 909         (vorn_s16): Likewise.
 910         (vorn_s32): Likewise.
 911         (vorn_u8): Likewise.
 912         (vorn_u16): Likewise.
 913         (vorn_u32): Likewise.
 914         (vorn_s64): Likewise.
 915         (vorn_u64): Likewise.
 916         (vornq_s8): Likewise.
 917         (vornq_s16): Likewise.
 918         (vornq_s32): Likewise.
 919         (vornq_s64): Likewise.
 920         (vornq_u8): Likewise.
 921         (vornq_u16): Likewise.
 922         (vornq_u32): Likewise.
 923         (vornq_u64): Likewise.
 924
 925 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 926
 927         * wide-int.cc (UTItype): Define.
 928         (UDWtype): Define for appropriate W_TYPE_SIZE.
 929
 930 2014-05-08  Marc Glisse  <marc.glisse@inria.fr>
 931
 932         PR tree-optimization/59100
 933         * tree-ssa-phiopt.c: Include tree-inline.h.
 934         (neutral_element_p, absorbing_element_p): New functions.
 935         (value_replacement): Handle conditional binary operations with a
 936         neutral or absorbing element.
 937
 938 2014-05-08  Richard Biener  <rguenther@suse.de>
 939
 940         * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
 941         folding the expression.
 942         (valueize_expr): Remove.
 943         (visit_reference_op_load): Do not valueize the result of
 944         vn_get_expr_for.
 945         (simplify_binary_expression): Likewise.
 946         (simplify_unary_expression): Likewise.
 947
 948 2014-05-08  Richard Biener  <rguenther@suse.de>
 949
 950         * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
 951         looking at TYPE_ARG_TYPES.
 952
 953 2014-05-08  Richard Biener  <rguenther@suse.de>
 954
 955         * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
 956         pointer propagation special-case.
 957
 958 2014-05-08  Bin Cheng  <bin.cheng@arm.com>
 959
 960         * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
 961         core part of address expressions.
 962
 963 2014-05-08  Alan Modra  <amodra@gmail.com>
 964
 965         PR target/60737
 966         * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
 967         loads and stores when -mno-strict-align at any alignment.
 968         (expand_block_clear): Similarly.  Also correct calculation of
 969         instruction count.
 970
 971 2014-05-07  Thomas Preud'homme  <thomas.preudhomme@arm.com>
 972
 973         PR middle-end/39246
 974         * tree-complex.c (expand_complex_move): Keep line info when expanding
 975         complex move.
 976         * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
 977         of complex expression. Use new argument to display correct location
 978         for values coming from phi statement.
 979         (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
 980         (warn_uninitialized_phi): Pass location of phi argument to
 981         warn_uninit.
 982         * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
 983         COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
 984
 985 2014-05-07  Segher Boessenkool  <segher@kernel.crashing.org>
 986
 987         * config/rs6000/predicates.md (indexed_address_mem): New.
 988         * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
 989         load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
 990         fpstore_ux, fpstore_u.
 991         (sign_extend, indexed, update): New.
 992         (cell_micro): Adjust.
 993         (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
 994         *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
 995         *movsi_internal1, *movsi_internal1_single, *movhi_internal,
 996         *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
 997         *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
 998         *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
 999         *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
1000         *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
1001         *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
1002         *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
1003         *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
1004         *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
1005         *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
1006         *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
1007         *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
1008
1009         * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
1010         * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
1011         *vsx_extract_<mode>_store): Adjust.
1012         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
1013         is_cracked_insn, insn_must_be_first_in_group,
1014         insn_must_be_last_in_group): Adjust.
1015
1016         * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
1017         Adjust.
1018         * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
1019         ppc440-fpstore): Adjust.
1020         * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
1021         ppc476-fpstore): Adjust.
1022         * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
1023         ppc601-fpstore): Adjust.
1024         * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
1025         Adjust.
1026         * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
1027         Adjust.
1028         * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
1029         ppc7450-fpstore): Adjust.
1030         * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
1031         * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
1032         * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
1033         Adjust.
1034         * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
1035         cell-fpload, cell-fpload-update, cell-store, cell-store-update,
1036         cell-fpstore, cell-fpstore-update): Adjust.
1037         * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
1038         ppce300c3_store, ppce300c3_fpstore): Adjust.
1039         * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
1040         e500mc_fpstore): Adjust.
1041         * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
1042         e500mc64_store, e500mc64_fpstore): Adjust.
1043         * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
1044         e5500_fpstore): Adjust.
1045         * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
1046         e6500_fpstore): Adjust.
1047         * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
1048         Adjust.
1049         * config/rs6000/power4.md (power4-load, power4-load-ext,
1050         power4-load-ext-update, power4-load-ext-update-indexed,
1051         power4-load-update-indexed, power4-load-update, power4-fpload,
1052         power4-fpload-update, power4-store, power4-store-update,
1053         power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
1054         Adjust.
1055         * config/rs6000/power5.md (power5-load, power5-load-ext,
1056         power5-load-ext-update, power5-load-ext-update-indexed,
1057         power5-load-update-indexed, power5-load-update, power5-fpload,
1058         power5-fpload-update, power5-store, power5-store-update,
1059         power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
1060         Adjust.
1061         * config/rs6000/power6.md (power6-load, power6-load-ext,
1062         power6-load-update, power6-load-update-indexed,
1063         power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
1064         power6-fpload-update, power6-store, power6-store-update,
1065         power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
1066         Adjust.
1067         * config/rs6000/power7.md (power7-load, power7-load-ext,
1068         power7-load-update, power7-load-update-indexed,
1069         power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
1070         power7-fpload-update, power7-store, power7-store-update,
1071         power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
1072         Adjust.
1073         * config/rs6000/power8.md (power8-load, power8-load-update,
1074         power8-load-ext, power8-load-ext-update, power8-fpload,
1075         power8-fpload-update, power8-store, power8-store-update-indexed,
1076         power8-fpstore, power8-fpstore-update): Adjust.
1077         * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
1078         Adjust.
1079         * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
1080         titan_lsu_store, titan_lsu_fpstore): Adjust.
1081         * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
1082
1083 2014-05-07  Oleg Endo  <olegendo@gcc.gnu.org>
1084
1085         PR target/60884
1086         * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
1087         unrolled byte insns.  Emit address increments after move insns.
1088
1089 2014-05-07  David Malcolm  <dmalcolm@redhat.com>
1090
1091         * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
1092         const_gimple, rather than a gimple.
1093         (gimple_call_builtin_p): Likewise, for the three variants.
1094
1095         * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
1096         (gimple_call_builtin_p): Likewise, for the three variants.
1097
1098 2014-05-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1099
1100         PR tree-optimization/61095
1101         * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
1102
1103 2014-05-07  Richard Biener  <rguenther@suse.de>
1104
1105         PR tree-optimization/61034
1106         * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
1107         (maybe_skip_until): Use translate to take into account
1108         lattices when trying to do disambiguations.
1109         (get_continuation_for_phi_1): Likewise.
1110         (get_continuation_for_phi): Adjust for added translate arguments.
1111         (walk_non_aliased_vuses): Likewise.
1112         * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
1113         (walk_non_aliased_vuses): Likewise.
1114         (call_may_clobber_ref_p_1): Declare.
1115         * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
1116         calls.  Stop early if we are only supposed to disambiguate.
1117         * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
1118
1119 2014-05-07  Joern Rennecke  <joern.rennecke@embecosm.com>
1120
1121         * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
1122         Emit an error when the function has arguments.
1123
1124 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
1125
1126         * cfgloop.h (unswitch_loops): Remove.
1127         * doc/passes.texi: Remove references to loop-unswitch.c
1128         * timevar.def (TV_LOOP_UNSWITCH): Remove.
1129
1130 2014-05-07  Evgeny Stupachenko  <evstupac@gmail.com>
1131
1132         * tree-vect-data-refs.c (vect_grouped_load_supported): New
1133         check for loads group of length 3.
1134         (vect_permute_load_chain): New permutations for loads group of
1135         length 3.
1136         * tree-vect-stmts.c (vect_model_load_cost): Change cost
1137         of vec_perm_shuffle for the new permutations.
1138
1139 2014-05-07  Alan Lawrence  <alan.lawrence@arm.com>
1140
1141         * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
1142         vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
1143         vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
1144         vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
1145         vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
1146         vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
1147         vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
1148         vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
1149
1150 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
1151
1152         * loop-unswitch.c: Delete.
1153
1154 2014-05-07  Richard Biener  <rguenther@suse.de>
1155
1156         * config.gcc: Always set need_64bit_hwint to yes.
1157
1158 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
1159
1160         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
1161         of using optimize_size.
1162
1163 2014-05-06  Mike Stump  <mikestump@comcast.net>
1164
1165         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
1166
1167 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
1168
1169         * config/i386/sse.md (*mov<mode>_internal)
1170         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
1171         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
1172         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
1173         (*<code><mode>3, *andnot<mode>3<mask_name>)
1174         (<mask_codefor><code><mode>3<mask_name>): Only consider
1175         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
1176
1177 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
1178
1179         Revert:
1180         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1181
1182         * lra-constraints.c (valid_address_p): Move earlier in file.
1183         Add a constraint argument to the address_info version.
1184         (satisfies_memory_constraint_p): New function.
1185         (satisfies_address_constraint_p): Likewise.
1186         (process_alt_operands, curr_insn_transform): Use them.
1187         (process_address): Pass the constraint to valid_address_p when
1188         checking address operands.
1189
1190 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
1191
1192         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
1193         to their respective blocks.  Fix inadvertent use of "node".
1194
1195 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
1196
1197         * emit-rtl.c (init_derived_machine_modes): New functionm, split
1198         out from...
1199         (init_emit_once): ...here.
1200         * rtl.h (init_derived_machine_modes): Declare.
1201         * toplev.c (do_compile): Call it even if no_backend.
1202
1203 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
1204             Mike Stump  <mikestump@comcast.net>
1205             Richard Sandiford  <rdsandiford@googlemail.com>
1206             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1207
1208         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
1209         (rtx_equal_for_memref_p): Update comment.
1210         (adjust_offset_for_component_ref): Use wide-int interfaces.
1211         * builtins.c (get_object_alignment_2): Likewise.
1212         (c_readstr): Likewise.
1213         (target_char_cast): Add comment.
1214         (determine_block_size): Use wide-int interfaces.
1215         (expand_builtin_signbit): Likewise.
1216         (fold_builtin_int_roundingfn): Likewise.
1217         (fold_builtin_bitop): Likewise.
1218         (fold_builtin_bswap): Likewise.
1219         (fold_builtin_logarithm): Use signop.
1220         (fold_builtin_pow): Likewise.
1221         (fold_builtin_memory_op): Use wide-int interfaces.
1222         (fold_builtin_object_size): Likewise.
1223         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
1224         nb_iterations_estimate.
1225         (record_niter_bound): Use wide-int interfaces.
1226         (get_estimated_loop_iterations_int): Likewise.
1227         (get_estimated_loop_iterations): Likewise.
1228         (get_max_loop_iterations): Likewise.
1229         * cfgloop.h: Include wide-int.h.
1230         (struct nb_iter_bound): Change bound to widest_int.
1231         (struct loop): Change nb_iterations_upper_bound and
1232         nb_iterations_estimate to widest_int.
1233         (record_niter_bound): Switch to use widest_int.
1234         (get_estimated_loop_iterations): Likewise.
1235         (get_max_loop_iterations): Likewise.
1236         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
1237         update for wide-int.
1238         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
1239         * combine.c (try_combine): Likewise.
1240         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
1241         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
1242         interfaces.
1243         (aarch64_float_const_representable_p): Likewise.
1244         * config/arc/arc.c: Include wide-int.h.
1245         (arc_can_use_doloop_p): Use wide-int interfaces.
1246         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1247         (vfp3_const_double_index): Likewise.
1248         * config/avr/avr.c (avr_out_round): Likewise.
1249         (avr_fold_builtin): Likewise.
1250         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1251         (bfin_can_use_doloop_p): Likewise.
1252         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1253         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
1254         * config/i386/i386.c: Include wide-int.h.
1255         (ix86_data_alignment): Use wide-int interfaces.
1256         (ix86_local_alignment): Likewise.
1257         (ix86_emit_swsqrtsf): Update real_from_integer.
1258         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
1259         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1260         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
1261         (zero_constant): Likewise.
1262         (input_operand): Likewise.
1263         (splat_input_operand): Likewise.
1264         (non_logical_cint_operand): Change const_double to const_wide_int.
1265         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
1266         (easy_altivec_constant): Remove comment.
1267         (paired_expand_vector_init): Use CONSTANT_P.
1268         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
1269         (rs6000_emit_move): Update checks.
1270         (rs6000_aggregate_candidate): Use wide-int interfaces.
1271         (rs6000_expand_ternop_builtin): Likewise.
1272         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
1273         (rs6000_assemble_integer): Likewise.
1274         (rs6000_hash_constant): Likewise.
1275         (output_toc): Likewise.
1276         (rs6000_rtx_costs): Likewise.
1277         (rs6000_emit_swrsqrt); Update call to real_from_integer.
1278         * config/rs6000/rs6000-c.c: Include wide-int.h.
1279         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
1280         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
1281         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
1282         Handle CONST_WIDE_INT.
1283         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
1284         Use tree_fits_uhwi_p.
1285         * config/sparc/sparc.c: Include wide-int.h.
1286         (sparc_fold_builtin): Use wide-int interfaces.
1287         * config/vax/vax.c: Include wide-int.h.
1288         (vax_float_literal): Use real_from_integer.
1289         * coretypes.h (struct hwivec_def): New.
1290         (hwivec): New.
1291         (const_hwivec): New.
1292         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
1293         (equiv_constant): Handle CONST_WIDE_INT.
1294         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
1295         (cselib_hash_rtx): Handle CONST_WIDE_INT.
1296         * dbxout.c (stabstr_U): Use wide-int interfaces.
1297         (dbxout_type): Update to use cst_fits_shwi_p.
1298         * defaults.h (LOG2_BITS_PER_UNIT): Define.
1299         (TARGET_SUPPORTS_WIDE_INT): Add default.
1300         * dfp.c: Include wide-int.h.
1301         (decimal_real_to_integer2): Use wide-int interfaces and rename to
1302         decimal_real_to_integer.
1303         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
1304         decimal_real_to_integer.
1305         * doc/generic.texi (Constant expressions): Update for wide_int.
1306         * doc/rtl.texi (const_double): Likewise.
1307         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
1308         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
1309         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
1310         (REAL_VALUE_FROM_INT): Remove.
1311         (TARGET_SUPPORTS_WIDE_INT): New.
1312         * doc/tm.texi: Regenerate.
1313         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
1314         * double-int.h: Include wide-int.h.
1315         (struct wi::int_traits): New.
1316         * dwarf2out.c (get_full_len): New.
1317         (dw_val_equal_p): Add case dw_val_class_wide_int.
1318         (size_of_loc_descr): Likewise.
1319         (output_loc_operands): Likewise.
1320         (insert_double): Remove.
1321         (insert_wide_int): New.
1322         (add_AT_wide): New.
1323         (print_die): Add case dw_val_class_wide_int.
1324         (attr_checksum): Likewise.
1325         (attr_checksum_ordered): Likewise.
1326         (same_dw_val_p): Likewise.
1327         (size_of_die): Likewise.
1328         (value_format): Likewise.
1329         (output_die): Likewise.
1330         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
1331         Use wide-int.
1332         (clz_loc_descriptor): Use wide-int interfaces.
1333         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
1334         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
1335         (round_up_to_align): Use wide-int interfaces.
1336         (field_byte_offset): Likewise.
1337         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
1338         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
1339         CONST_DOUBLE handling.  Use wide-int interfaces.
1340         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
1341         (gen_enumeration_type_die): Use add_AT_wide.
1342         (hash_loc_operands): Add case dw_val_class_wide_int.
1343         (compare_loc_operands): Likewise.
1344         * dwarf2out.h: Include wide-int.h.
1345         (wide_int_ptr): New.
1346         (enum dw_val_class): Add dw_val_class_wide_int.
1347         (struct dw_val_struct): Add val_wide.
1348         * emit-rtl.c (const_wide_int_htab): New.
1349         (const_wide_int_htab_hash): New.
1350         (const_wide_int_htab_eq): New.
1351         (lookup_const_wide_int): New.
1352         (const_double_htab_hash): Use wide-int interfaces.
1353         (const_double_htab_eq): Likewise.
1354         (rtx_to_double_int): Conditionally compile for wide-int.
1355         (immed_double_int_const): Rename to immed_wide_int_const and
1356         update for wide-int.
1357         (immed_double_const): Conditionally compile for wide-int.
1358         (init_emit_once): Use wide-int interfaces.
1359         * explow.c (plus_constant): Likewise.
1360         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
1361         (lshift_value): Use wide-int interfaces.
1362         (expand_mult): Likewise.
1363         (choose_multiplier): Likewise.
1364         (expand_smod_pow2): Likewise.
1365         (make_tree): Likewise.
1366         * expr.c (convert_modes): Consolidate handling of constants.
1367         Use wide-int interfaces.
1368         (emit_group_load_1): Add note.
1369         (store_expr): Update comment.
1370         (get_inner_reference): Use wide-int interfaces.
1371         (expand_constructor): Update comment.
1372         (expand_expr_real_2): Use wide-int interfaces.
1373         (expand_expr_real_1): Likewise.
1374         (reduce_to_bit_field_precision): Likewise.
1375         (const_vector_from_tree): Likewise.
1376         * final.c: Include wide-int-print.h.
1377         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
1378         * fixed-value.c: Include wide-int.h.
1379         (fixed_from_string): Use wide-int interfaces.
1380         (fixed_to_decimal): Likewise.
1381         (fixed_convert_from_real): Likewise.
1382         (real_convert_from_fixed): Likewise.
1383         * fold-const.h (mem_ref_offset): Return an offset_int.
1384         (div_if_zero_remainder): Remove code parameter.
1385         * fold-const.c (div_if_zero_remainder): Remove code parameter.
1386         Use wide-int interfaces.
1387         (may_negate_without_overflow_p): Use wide-int interfaces.
1388         (negate_expr_p): Likewise.
1389         (fold_negate_expr): Likewise.
1390         (int_const_binop_1): Likewise.
1391         (const_binop): Likewise.
1392         (fold_convert_const_int_from_int): Likewise.
1393         (fold_convert_const_int_from_real): Likewise.
1394         (fold_convert_const_int_from_fixed): Likewise.
1395         (fold_convert_const_fixed_from_int): Likewise.
1396         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
1397         (sign_bit_p): Use wide-int interfaces.
1398         (make_range_step): Likewise.
1399         (build_range_check): Likewise.  Pass an integer of the correct type
1400         instead of using integer_one_node.
1401         (range_predecessor): Pass an integer of the correct type instead
1402         of using integer_one_node.
1403         (range_successor): Likewise.
1404         (merge_ranges): Likewise.
1405         (unextend): Use wide-int interfaces.
1406         (extract_muldiv_1): Likewise.
1407         (fold_div_compare): Likewise.
1408         (fold_single_bit_test): Likewise.
1409         (fold_sign_changed_comparison): Likewise.
1410         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
1411         (fold_plusminus_mult_expr): Use wide-int interfaces.
1412         (native_encode_int): Likewise.
1413         (native_interpret_int): Likewise.
1414         (fold_unary_loc): Likewise.
1415         (pointer_may_wrap_p): Likewise.
1416         (size_low_cst): Likewise.
1417         (mask_with_tz): Likewise.
1418         (fold_binary_loc): Likewise.
1419         (fold_ternary_loc): Likewise.
1420         (multiple_of_p): Likewise.
1421         (tree_call_nonnegative_warnv_p): Update calls to
1422         tree_int_cst_min_precision and real_from_integer.
1423         (fold_negate_const): Use wide-int interfaces.
1424         (fold_abs_const): Likewise.
1425         (fold_relational_const): Use tree_int_cst_lt.
1426         (round_up_loc): Use wide-int interfaces.
1427         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
1428         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
1429         * gengtype.c: Remove include of double-int.h.
1430         (do_typedef): Use wide-int interfaces.
1431         (open_base_files): Add wide-int.h.
1432         (main): Add offset_int and widest_int typedefs.
1433         * gengtype-lex.l: Handle "^".
1434         (CXX_KEYWORD): Add "static".
1435         * gengtype-parse.c (require3): New.
1436         (require_template_declaration): Handle constant template arguments
1437         and nested templates.
1438         * gengtype-state.c: Don't include "double-int.h".
1439         * genpreds.c (write_one_predicate_function): Update comment.
1440         (write_tm_constrs_h): Add check for hval and lval use in
1441         CONST_WIDE_INT.
1442         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
1443         (add_to_sequence): Likewise.
1444         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
1445         and const_double_operand.
1446         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
1447         interfaces.
1448         * gimple-fold.c (get_base_constructor): Likewise.
1449         (fold_array_ctor_reference): Likewise.
1450         (fold_nonarray_ctor_reference): Likewise.
1451         (fold_const_aggregate_ref_1): Likewise.
1452         (gimple_val_nonnegative_real_p): Likewise.
1453         (gimple_fold_indirect_ref): Likewise.
1454         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
1455         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
1456         (struct slsr_cand_d): Change index to be widest_int.
1457         (struct incr_info_d): Change incr to be widest_int.
1458         (alloc_cand_and_find_basis): Use wide-int interfaces.
1459         (slsr_process_phi): Likewise.
1460         (backtrace_base_for_ref): Likewise.  Return a widest_int.
1461         (restructure_reference): Take a widest_int instead of a double_int.
1462         (slsr_process_ref): Use wide-int interfaces.
1463         (create_mul_ssa_cand): Likewise.
1464         (create_mul_imm_cand): Likewise.
1465         (create_add_ssa_cand): Likewise.
1466         (create_add_imm_cand): Take a widest_int instead of a double_int.
1467         (slsr_process_add): Use wide-int interfaces.
1468         (slsr_process_cast): Likewise.
1469         (slsr_process_copy): Likewise.
1470         (dump_candidate): Likewise.
1471         (dump_incr_vec): Likewise.
1472         (replace_ref): Likewise.
1473         (cand_increment): Likewise.  Return a widest_int.
1474         (cand_abs_increment): Likewise.
1475         (replace_mult_candidate): Take a widest_int instead of a double_int.
1476         (replace_unconditional_candidate): Use wide-int interfaces.
1477         (incr_vec_index): Take a widest_int instead of a double_int.
1478         (create_add_on_incoming_edge): Likewise.
1479         (create_phi_basis): Use wide-int interfaces.
1480         (replace_conditional_candidate): Likewise.
1481         (record_increment): Take a widest_int instead of a double_int.
1482         (record_phi_increments): Use wide-int interfaces.
1483         (phi_incr_cost): Take a widest_int instead of a double_int.
1484         (lowest_cost_path): Likewise.
1485         (total_savings): Likewise.
1486         (analyze_increments): Use wide-int interfaces.
1487         (ncd_with_phi): Take a widest_int instead of a double_int.
1488         (ncd_of_cand_and_phis): Likewise.
1489         (nearest_common_dominator_for_cands): Likewise.
1490         (insert_initializers): Use wide-int interfaces.
1491         (all_phi_incrs_profitable): Likewise.
1492         (replace_one_candidate): Likewise.
1493         (replace_profitable_candidates): Likewise.
1494         * godump.c: Include wide-int-print.h.
1495         (go_output_typedef): Use wide-int interfaces.
1496         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
1497         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1498         (build_loop_iteration_domains): Likewise.
1499         * hooks.h: Include wide-int.h rather than double-int.h.
1500         (hook_bool_dint_dint_uint_bool_true): Delete.
1501         (hook_bool_wint_wint_uint_bool_true): Declare.
1502         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
1503         (hook_bool_wint_wint_uint_bool_true): New.
1504         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
1505         interfaces.
1506         (ubsan_expand_si_overflow_mul_check): Likewise.
1507         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
1508         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1509         (get_ancestor_addr_info): Likewise.
1510         (ipa_modify_call_arguments): Likewise.
1511         * loop-doloop.c (doloop_modify): Likewise.
1512         (doloop_optimize): Likewise.
1513         * loop-iv.c (iv_number_of_iterations): Likewise.
1514         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
1515         (unroll_loop_constant_iterations): Likewise.
1516         (decide_unroll_runtime_iterations): Likewise.
1517         (unroll_loop_runtime_iterations): Likewise.
1518         (decide_peel_simple): Likewise.
1519         (decide_unroll_stupid): Likewise.
1520         * lto-streamer-in.c (streamer_read_wi): Add.
1521         (input_cfg): Use wide-int interfaces.
1522         (lto_input_tree_1): Likewise.
1523         * lto-streamer-out.c (streamer_write_wi): Add.
1524         (hash_tree): Use wide-int interfaces.
1525         (output_cfg): Likewise.
1526         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
1527         (GTFILES): Add wide-int.h and signop.h.
1528         (TAGS): Look for .cc files too.
1529         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
1530         * optabs.c (expand_subword_shift): Likewise.
1531         (expand_doubleword_shift): Likewise.
1532         (expand_absneg_bit): Likewise.
1533         (expand_copysign_absneg): Likewise.
1534         (expand_copysign_bit): Likewise.
1535         * postreload.c (reload_cse_simplify_set): Likewise.
1536         * predict.c (predict_iv_comparison): Likewise.
1537         * pretty-print.h: Include wide-int-print.h.
1538         (pp_wide_int) New.
1539         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
1540         * print-tree.c: Include wide-int-print.h.
1541         (print_node_brief): Use wide-int interfaces.
1542         (print_node): Likewise.
1543         * read-rtl.c (validate_const_wide_int): New.
1544         (read_rtx_code): Add CONST_WIDE_INT case.
1545         * real.c: Include wide-int.h.
1546         (real_to_integer2): Delete.
1547         (real_to_integer): New function, returning a wide_int.
1548         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1549         (ten_to_ptwo): Update call to real_from_integer.
1550         (real_digit): Likewise.
1551         * real.h: Include signop.h, wide-int.h and insn-modes.h.
1552         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
1553         (REAL_VALUE_TO_INT): Delete.
1554         (real_to_integer): Declare a wide-int form.
1555         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1556         * recog.c (const_int_operand): Improve comment.
1557         (const_scalar_int_operand): New.
1558         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
1559         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
1560         (split_double): Likewise.
1561         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
1562         (rtx_size): Likewise.
1563         (rtx_alloc_stat_v): New.
1564         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
1565         (cwi_output_hex): New.
1566         (iterative_hash_rtx): Handle CONST_WIDE_INT.
1567         (cwi_check_failed_bounds): New.
1568         * rtl.def (CONST_WIDE_INT): New.
1569         * rtl.h: Include <utility> and wide-int.h.
1570         (struct hwivec_def): New.
1571         (CWI_GET_NUM_ELEM): New.
1572         (CWI_PUT_NUM_ELEM): New.
1573         (struct rtx_def): Add num_elem and hwiv.
1574         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
1575         (CASE_CONST_UNIQUE): Likewise.
1576         (CASE_CONST_ANY): Likewise.
1577         (CONST_SCALAR_INT_P): Likewise.
1578         (CONST_WIDE_INT_P): New.
1579         (CWI_ELT): New.
1580         (HWIVEC_CHECK): New.
1581         (cwi_check_failed_bounds): New.
1582         (CWI_ELT): New.
1583         (HWIVEC_CHECK): New.
1584         (CONST_WIDE_INT_VEC) New.
1585         (CONST_WIDE_INT_NUNITS) New.
1586         (CONST_WIDE_INT_ELT) New.
1587         (rtx_mode_t): New type.
1588         (wi::int_traits <rtx_mode_t>): New.
1589         (wi::shwi): New.
1590         (wi::min_value): New.
1591         (wi::max_value): New.
1592         (rtx_alloc_v) New.
1593         (const_wide_int_alloc): New.
1594         (immed_wide_int_const): New.
1595         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
1596         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
1597         * signop.h: New file.
1598         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
1599         (simplify_const_unary_operation): Use wide-int interfaces.
1600         (simplify_binary_operation_1): Likewise.
1601         (simplify_const_binary_operation): Likewise.
1602         (simplify_const_relational_operation): Likewise.
1603         (simplify_immed_subreg): Likewise.
1604         * stmt.c (expand_case): Likewise.
1605         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
1606         signop rather than a bool.
1607         * stor-layout.c (layout_type): Use wide-int interfaces.
1608         (initialize_sizetypes): Update calls to
1609         set_min_and_max_values_for_integral_type.
1610         (set_min_and_max_values_for_integral_type): Take a signop rather
1611         than a bool.  Use wide-int interfaces.
1612         (fixup_signed_type): Update accordingly.  Remove
1613         HOST_BITS_PER_DOUBLE_INT limit.
1614         (fixup_unsigned_type): Likewise.
1615         * system.h (STATIC_CONSTANT_P): New.
1616         (STATIC_ASSERT): New.
1617         * target.def (can_use_doloop_p): Take widest_ints rather than
1618         double_ints.
1619         * target.h: Include wide-int.h rather than double-int.h.
1620         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
1621         than double_ints.
1622         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
1623         rather than INT_CST_LT_UNSIGNED.
1624         (can_use_doloop_if_innermost): Take widest_ints rather than
1625         double_ints.
1626         * tree-affine.c: Include wide-int-print.h.
1627         (double_int_ext_for_comb): Delete.
1628         (wide_int_ext_for_comb): New.
1629         (aff_combination_zero): Use wide-int interfaces.
1630         (aff_combination_const): Take a widest_int instead of a double_int.
1631         (aff_combination_elt): Use wide-int interfaces.
1632         (aff_combination_scale): Take a widest_int instead of a double_int.
1633         (aff_combination_add_elt): Likewise.
1634         (aff_combination_add_cst): Likewise.
1635         (aff_combination_add): Use wide-int interfaces.
1636         (aff_combination_convert): Likewise.
1637         (tree_to_aff_combination): Likewise.
1638         (add_elt_to_tree): Take a widest_int instead of a double_int.
1639         (aff_combination_to_tree): Use wide-int interfaces.
1640         (aff_combination_remove_elt): Likewise.
1641         (aff_combination_add_product): Take a widest_int instead of
1642         a double_int.
1643         (aff_combination_mult): Use wide-int interfaces.
1644         (aff_combination_expand): Likewise.
1645         (double_int_constant_multiple_p): Delete.
1646         (wide_int_constant_multiple_p): New.
1647         (aff_combination_constant_multiple_p): Take a widest_int pointer
1648         instead of a double_int pointer.
1649         (print_aff): Use wide-int interfaces.
1650         (get_inner_reference_aff): Take a widest_int pointer
1651         instead of a double_int pointer.
1652         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
1653         * tree-affine.h: Include wide-int.h.
1654         (struct aff_comb_elt): Change type of coef to widest_int.
1655         (struct affine_tree_combination): Change type of offset to widest_int.
1656         (double_int_ext_for_comb): Delete.
1657         (wide_int_ext_for_comb): New.
1658         (aff_combination_const): Use widest_int instead of double_int.
1659         (aff_combination_scale): Likewise.
1660         (aff_combination_add_elt): Likewise.
1661         (aff_combination_constant_multiple_p): Likewise.
1662         (get_inner_reference_aff): Likewise.
1663         (aff_comb_cannot_overlap_p): Likewise.
1664         (aff_combination_zero_p): Use wide-int interfaces.
1665         * tree.c: Include tree.h.
1666         (init_ttree): Use make_int_cst.
1667         (tree_code_size): Removed code for INTEGER_CST case.
1668         (tree_size): Add INTEGER_CST case.
1669         (make_node_stat): Update comment.
1670         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
1671         (build_int_cst_type): Use wide-int interfaces.
1672         (double_int_to_tree): Likewise.
1673         (double_int_fits_to_tree_p): Delete.
1674         (force_fit_type_double): Delete.
1675         (force_fit_type): New.
1676         (int_cst_hash_hash): Use wide-int interfaces.
1677         (int_cst_hash_eq): Likewise.
1678         (build_int_cst_wide): Delete.
1679         (wide_int_to_tree): New.
1680         (cache_integer_cst): Use wide-int interfaces.
1681         (build_low_bits_mask): Likewise.
1682         (cst_and_fits_in_hwi): Likewise.
1683         (real_value_from_int_cst): Likewise.
1684         (make_int_cst_stat): New.
1685         (integer_zerop): Use wide_int interfaces.
1686         (integer_onep): Likewise.
1687         (integer_all_onesp): Likewise.
1688         (integer_pow2p): Likewise.
1689         (integer_nonzerop): Likewise.
1690         (tree_log2): Likewise.
1691         (tree_floor_log2): Likewise.
1692         (tree_ctz): Likewise.
1693         (int_size_in_bytes): Likewise.
1694         (mem_ref_offset): Return an offset_int rather than a double_int.
1695         (build_type_attribute_qual_variant): Use wide_int interfaces.
1696         (type_hash_eq): Likewise
1697         (tree_int_cst_equal): Likewise.
1698         (tree_int_cst_lt): Delete.
1699         (tree_int_cst_compare): Likewise.
1700         (tree_fits_shwi_p): Use wide_int interfaces.
1701         (tree_fits_uhwi_p): Likewise.
1702         (tree_int_cst_sign_bit): Likewise.
1703         (tree_int_cst_sgn): Likewise.
1704         (tree_int_cst_min_precision): Take a signop rather than a bool.
1705         (simple_cst_equal): Use wide_int interfaces.
1706         (compare_tree_int): Likewise.
1707         (iterative_hash_expr): Likewise.
1708         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
1709         INT_CST_LT.
1710         (get_type_static_bounds): Use wide_int interfaces.
1711         (tree_int_cst_elt_check_failed): New.
1712         (build_common_tree_nodes): Reordered to set prec before filling in
1713         value.
1714         (int_cst_value): Check cst_and_fits_in_hwi.
1715         (widest_int_cst_value): Use wide_int interfaces.
1716         (upper_bound_in_type): Likewise.
1717         (lower_bound_in_type): Likewise.
1718         (num_ending_zeros): Likewise.
1719         (drop_tree_overflow): Likewise.
1720         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
1721         (gen_conditions_for_pow_cst_base): Likewise.
1722         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
1723         (group_case_labels_stmt): Use wide-int interfaces.
1724         (verify_gimple_assign_binary): Likewise.
1725         (print_loop): Likewise.
1726         * tree-chrec.c (tree_fold_binomial): Likewise.
1727         * tree-core.h (struct tree_base): Add int_length.
1728         (struct tree_int_cst): Change rep of value.
1729         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
1730         (dr_may_alias_p): Likewise.
1731         (max_stmt_executions_tree): Likewise.
1732         * tree.def (INTEGER_CST): Update comment.
1733         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
1734         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
1735         * tree-dump.c: Include wide-int.h and wide-int-print.h.
1736         (dequeue_and_dump): Use wide-int interfaces.
1737         * tree.h: Include wide-int.h.
1738         (NULL_TREE): Moved to earlier loc in file.
1739         (TREE_INT_CST_ELT_CHECK): New.
1740         (tree_int_cst_elt_check_failed): New.
1741         (TYPE_SIGN): New.
1742         (TREE_INT_CST): Delete.
1743         (TREE_INT_CST_LOW): Use wide-int interfaces.
1744         (TREE_INT_CST_HIGH): Delete.
1745         (TREE_INT_CST_NUNITS): New.
1746         (TREE_INT_CST_EXT_NUNITS): Likewise.
1747         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
1748         (TREE_INT_CST_ELT): Likewise.
1749         (INT_CST_LT): Delete.
1750         (tree_int_cst_elt_check): New (two forms).
1751         (type_code_size): Update comment.
1752         (make_int_cst_stat, make_int_cst): New.
1753         (tree_to_double_int): Delete.
1754         (double_int_fits_to_tree_p): Delete.
1755         (force_fit_type_double): Delete.
1756         (build_int_cstu): Replace with out-of-line function.
1757         (build_int_cst_wide): Delete.
1758         (tree_int_cst_lt): Define inline.
1759         (tree_int_cst_le): New.
1760         (tree_int_cst_compare): Define inline.
1761         (tree_int_cst_min_precision): Take a signop rather than a bool.
1762         (wi::int_traits <const_tree>): New.
1763         (wi::int_traits <tree>): New.
1764         (wi::extended_tree): New.
1765         (wi::int_traits <wi::extended_tree>): New.
1766         (wi::to_widest): New.
1767         (wi::to_offset): New.
1768         (wi::fits_to_tree_p): New.
1769         (wi::min_value): New.
1770         (wi::max_value): New.
1771         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
1772         (copy_tree_body_r): Likewise.
1773         * tree-object-size.c (compute_object_offset): Likewise.
1774         (addr_object_size): Likewise.
1775         * tree-predcom.c: Include wide-int-print.h.
1776         (struct dref_d): Change type of offset to widest_int.
1777         (dump_dref): Call wide-int printer.
1778         (aff_combination_dr_offset): Use wide-int interfaces.
1779         (determine_offset): Take a widest_int pointer rather than a
1780         double_int pointer.
1781         (split_data_refs_to_components): Use wide-int interfaces.
1782         (suitable_component_p): Likewise.
1783         (order_drefs): Likewise.
1784         (add_ref_to_chain): Likewise.
1785         (valid_initializer_p): Likewise.
1786         (determine_roots_comp): Likewise.
1787         * tree-pretty-print.c: Include wide-int-print.h.
1788         (dump_generic_node): Use wide-int interfaces.
1789         * tree-sra.c (sra_ipa_modify_expr): Likewise.
1790         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1791         (move_fixed_address_to_symbol): Likewise.
1792         (move_hint_to_base): Likewise.
1793         (move_pointer_to_base): Likewise.
1794         (move_variant_to_index): Likewise.
1795         (most_expensive_mult_to_index): Likewise.
1796         (addr_to_parts): Likewise.
1797         (copy_ref_info): Likewise.
1798         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1799         (indirect_refs_may_alias_p): Likewise.
1800         (stmt_kills_ref_p_1): Likewise.
1801         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
1802         * tree-ssa-ccp.c: Update comment at top of file.  Include
1803         wide-int-print.h.
1804         (struct prop_value_d): Change type of mask to widest_int.
1805         (extend_mask): New function.
1806         (dump_lattice_value): Use wide-int interfaces.
1807         (get_default_value): Likewise.
1808         (set_constant_value): Likewise.
1809         (set_value_varying): Likewise.
1810         (valid_lattice_transition): Likewise.
1811         (set_lattice_value): Likewise.
1812         (value_to_double_int): Delete.
1813         (value_to_wide_int): New.
1814         (get_value_from_alignment): Use wide-int interfaces.
1815         (get_value_for_expr): Likewise.
1816         (do_dbg_cnt): Likewise.
1817         (ccp_finalize): Likewise.
1818         (ccp_lattice_meet): Likewise.
1819         (bit_value_unop_1): Use widest_ints rather than double_ints.
1820         (bit_value_binop_1): Likewise.
1821         (bit_value_unop): Use wide-int interfaces.
1822         (bit_value_binop): Likewise.
1823         (bit_value_assume_aligned): Likewise.
1824         (evaluate_stmt): Likewise.
1825         (ccp_fold_stmt): Likewise.
1826         (visit_cond_stmt): Likewise.
1827         (ccp_visit_stmt): Likewise.
1828         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1829         (constant_pointer_difference): Likewise.
1830         (associate_pointerplus): Likewise.
1831         (combine_conversions): Likewise.
1832         * tree-ssa-loop.h: Include wide-int.h.
1833         (struct tree_niter_desc): Change type of max to widest_int.
1834         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
1835         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
1836         (remove_redundant_iv_tests): Likewise.
1837         (canonicalize_loop_induction_variables): Likewise.
1838         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
1839         (constant_multiple_of): Take a widest_int pointer instead of
1840         a double_int pointer.
1841         (get_computation_aff): Use wide-int interfaces.
1842         (ptr_difference_cost): Likewise.
1843         (difference_cost): Likewise.
1844         (get_loop_invariant_expr_id): Likewise.
1845         (get_computation_cost_at): Likewise.
1846         (iv_elimination_compare_lt): Likewise.
1847         (may_eliminate_iv): Likewise.
1848         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
1849         instead of double_int.
1850         (max_loop_iterations): Likewise.
1851         (max_stmt_executions): Likewise.
1852         (estimated_stmt_executions): Likewise.
1853         * tree-ssa-loop-niter.c: Include wide-int-print.h.
1854         (split_to_var_and_offset): Use wide-int interfaces.
1855         (determine_value_range): Likewise.
1856         (bound_difference_of_offsetted_base): Likewise.
1857         (bounds_add): Take a widest_int instead of a double_int.
1858         (number_of_iterations_ne_max): Use wide-int interfaces.
1859         (number_of_iterations_ne): Likewise.
1860         (number_of_iterations_lt_to_ne): Likewise.
1861         (assert_loop_rolls_lt): Likewise.
1862         (number_of_iterations_lt): Likewise.
1863         (number_of_iterations_le): Likewise.
1864         (number_of_iterations_cond): Likewise.
1865         (number_of_iterations_exit): Likewise.
1866         (finite_loop_p): Likewise.
1867         (derive_constant_upper_bound_assign): Likewise.
1868         (derive_constant_upper_bound): Return a widest_int.
1869         (derive_constant_upper_bound_ops): Likewise.
1870         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
1871         (record_estimate): Take a widest_int rather than a double_int.
1872         (record_nonwrapping_iv): Use wide-int interfaces.
1873         (double_int_cmp): Delete.
1874         (wide_int_cmp): New.
1875         (bound_index): Take a widest_int rather than a double_int.
1876         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
1877         (maybe_lower_iteration_bound): Likewise.
1878         (estimate_numbers_of_iterations_loop): Likewise.
1879         (estimated_loop_iterations): Take a widest_int pointer than than
1880         a double_int pointer.
1881         (estimated_loop_iterations_int): Use wide-int interfaces.
1882         (max_loop_iterations): Take a widest_int pointer than than
1883         a double_int pointer.
1884         (max_loop_iterations_int): Use wide-int interfaces.
1885         (max_stmt_executions): Take a widest_int pointer than than
1886         a double_int pointer.
1887         (estimated_stmt_executions): Likewise.
1888         (n_of_executions_at_most): Use wide-int interfaces.
1889         (scev_probably_wraps_p): Likewise.
1890         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
1891         to real_to_integer.
1892         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
1893         interfaces.
1894         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
1895         double_ints.  Adjust for trailing_wide_ints <3> representation.
1896         (set_nonzero_bits): Likewise.
1897         (get_range_info): Return wide_ints rather than double_ints.
1898         Adjust for trailing_wide_ints <3> representation.
1899         (get_nonzero_bits): Likewise.
1900         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
1901         representation.
1902         * tree-ssanames.h (struct range_info_def): Replace min, max and
1903         nonzero_bits with a trailing_wide_ints <3>.
1904         (set_range_info): Use wide_int_refs rather than double_ints.
1905         (set_nonzero_bits): Likewise.
1906         (get_range_info): Return wide_ints rather than double_ints.
1907         (get_nonzero_bits): Likewise.
1908         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
1909         * tree-ssa-pre.c (phi_translate_1): Likewise.
1910         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
1911         (acceptable_pow_call): Likewise.
1912         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
1913         interfaces.
1914         (vn_reference_fold_indirect): Likewise.
1915         (vn_reference_maybe_forwprop_address): Likewise.
1916         (valueize_refs_1): Likewise.
1917         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1918         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
1919         tree_int_cst_lt and tree_int_cst_le.
1920         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
1921         interfaces.
1922         (streamer_alloc_tree): Likewise.
1923         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
1924         (streamer_write_tree_header): Likewise.
1925         (streamer_write_integer_cst): Likewise.
1926         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
1927         (build_constructors): Likewise.
1928         (array_value_type): Likewise.
1929         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
1930         (vect_check_gather): Likewise.
1931         * tree-vect-generic.c (build_replicated_const): Likewise.
1932         (expand_vector_divmod): Likewise.
1933         * tree-vect-loop.c (vect_transform_loop): Likewise.
1934         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
1935         (vect_do_peeling_for_alignment): Likewise.
1936         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
1937         * tree-vrp.c: Include wide-int.h.
1938         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
1939         (extract_range_from_assert): Use wide-int interfaces.
1940         (vrp_int_const_binop): Likewise.
1941         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
1942         double_int pointers.
1943         (ranges_from_anti_range): Use wide-int interfaces.
1944         (quad_int_cmp): Delete.
1945         (quad_int_pair_sort): Likewise.
1946         (extract_range_from_binary_expr_1): Use wide-int interfaces.
1947         (extract_range_from_unary_expr_1): Likewise.
1948         (adjust_range_with_scev): Likewise.
1949         (masked_increment): Take and return wide_ints rather than double_ints.
1950         (register_edge_assert_for_2): Use wide-int interfaces.
1951         (check_array_ref): Likewise.
1952         (search_for_addr_array): Likewise.
1953         (maybe_set_nonzero_bits): Likewise.
1954         (union_ranges): Pass an integer of the correct type instead of
1955         using integer_one_node.
1956         (intersect_ranges): Likewise.
1957         (simplify_truth_ops_using_ranges): Likewise.
1958         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
1959         (range_fits_type_p): Likewise.
1960         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
1961         a bool.
1962         (simplify_conversion_using_ranges): Use wide-int interfaces.
1963         (simplify_float_conversion_using_ranges): Likewise.
1964         (vrp_finalize): Likewise.
1965         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
1966         (gimple_stringops_transform): Likewise.
1967         * varasm.c (decode_addr_const): Likewise.
1968         (const_hash_1): Likewise.
1969         (const_rtx_hash_1): Likewise
1970         (output_constant): Likewise.
1971         (array_size_for_constructor): Likewise.
1972         (output_constructor_regular_field): Likewise.
1973         (output_constructor_bitfield): Likewise.
1974         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
1975         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
1976         GENERATOR_FILEs.
1977         * gencheck.c: Define BITS_PER_UNIT.
1978         * wide-int.cc: New.
1979         * wide-int.h: New.
1980         * wide-int-print.cc: New.
1981         * wide-int-print.h: New.
1982
1983 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1984
1985         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
1986
1987 2014-05-06  Richard Biener  <rguenther@suse.de>
1988
1989         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
1990         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
1991         (TODO_verify_all): Adjust.
1992         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
1993         TODO_verify_stmts and TODO_verify_rtl_sharing.
1994         * bb-reorder.c: Likewise.
1995         * cfgexpand.c: Likewise.
1996         * cprop.c: Likewise.
1997         * cse.c: Likewise.
1998         * function.c: Likewise.
1999         * fwprop.c: Likewise.
2000         * gcse.c: Likewise.
2001         * gimple-ssa-isolate-paths.c: Likewise.
2002         * gimple-ssa-strength-reduction.c: Likewise.
2003         * ipa-split.c: Likewise.
2004         * loop-init.c: Likewise.
2005         * loop-unroll.c: Likewise.
2006         * lower-subreg.c: Likewise.
2007         * modulo-sched.c: Likewise.
2008         * postreload-gcse.c: Likewise.
2009         * predict.c: Likewise.
2010         * recog.c: Likewise.
2011         * sched-rgn.c: Likewise.
2012         * store-motion.c: Likewise.
2013         * tracer.c: Likewise.
2014         * trans-mem.c: Likewise.
2015         * tree-call-cdce.c: Likewise.
2016         * tree-cfg.c: Likewise.
2017         * tree-cfgcleanup.c: Likewise.
2018         * tree-complex.c: Likewise.
2019         * tree-eh.c: Likewise.
2020         * tree-emutls.c: Likewise.
2021         * tree-if-conv.c: Likewise.
2022         * tree-into-ssa.c: Likewise.
2023         * tree-loop-distribution.c: Likewise.
2024         * tree-object-size.c: Likewise.
2025         * tree-parloops.c: Likewise.
2026         * tree-pass.h: Likewise.
2027         * tree-sra.c: Likewise.
2028         * tree-ssa-ccp.c: Likewise.
2029         * tree-ssa-copy.c: Likewise.
2030         * tree-ssa-copyrename.c: Likewise.
2031         * tree-ssa-dce.c: Likewise.
2032         * tree-ssa-dom.c: Likewise.
2033         * tree-ssa-dse.c: Likewise.
2034         * tree-ssa-forwprop.c: Likewise.
2035         * tree-ssa-ifcombine.c: Likewise.
2036         * tree-ssa-loop-ch.c: Likewise.
2037         * tree-ssa-loop-ivcanon.c: Likewise.
2038         * tree-ssa-loop.c: Likewise.
2039         * tree-ssa-math-opts.c: Likewise.
2040         * tree-ssa-phiopt.c: Likewise.
2041         * tree-ssa-phiprop.c: Likewise.
2042         * tree-ssa-pre.c: Likewise.
2043         * tree-ssa-reassoc.c: Likewise.
2044         * tree-ssa-sink.c: Likewise.
2045         * tree-ssa-strlen.c: Likewise.
2046         * tree-ssa-tail-merge.c: Likewise.
2047         * tree-ssa-uncprop.c: Likewise.
2048         * tree-switch-conversion.c: Likewise.
2049         * tree-tailcall.c: Likewise.
2050         * tree-vect-generic.c: Likewise.
2051         * tree-vectorizer.c: Likewise.
2052         * tree-vrp.c: Likewise.
2053         * tsan.c: Likewise.
2054         * var-tracking.c: Likewise.
2055         * bt-load.c: Likewise.
2056         * cfgcleanup.c: Likewise.
2057         * combine-stack-adj.c: Likewise.
2058         * combine.c: Likewise.
2059         * compare-elim.c: Likewise.
2060         * config/epiphany/resolve-sw-modes.c: Likewise.
2061         * config/i386/i386.c: Likewise.
2062         * config/mips/mips.c: Likewise.
2063         * config/s390/s390.c: Likewise.
2064         * config/sh/sh_treg_combine.cc: Likewise.
2065         * config/sparc/sparc.c: Likewise.
2066         * dce.c: Likewise.
2067         * dse.c: Likewise.
2068         * final.c: Likewise.
2069         * ifcvt.c: Likewise.
2070         * mode-switching.c: Likewise.
2071         * passes.c: Likewise.
2072         * postreload.c: Likewise.
2073         * ree.c: Likewise.
2074         * reg-stack.c: Likewise.
2075         * regcprop.c: Likewise.
2076         * regrename.c: Likewise.
2077         * web.c: Likewise.
2078
2079 2014-05-06  Richard Biener  <rguenther@suse.de>
2080
2081         PR middle-end/61070
2082         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
2083         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
2084
2085 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
2086
2087         PR ipa/60965
2088         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
2089
2090 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
2091             Tom de Vries  <tom@codesourcery.com>
2092
2093         * target.def (call_fusage_contains_non_callee_clobbers): New
2094         DEFHOOKPOD.
2095         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
2096         Hooks to @menu.
2097         (@node Miscellaneous Register Hooks): New node.
2098         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
2099         * doc/tm.texi: Regenerate.
2100
2101 2014-05-05  Marek Polacek  <polacek@redhat.com>
2102
2103         PR driver/61065
2104         * opts.c (common_handle_option): Call error_at instead of warning_at.
2105
2106 2014-05-05  Richard Biener  <rguenther@suse.de>
2107
2108         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
2109         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
2110         under the TODO_verify_il umbrella.
2111
2112 2014-05-05  Richard Biener  <rguenther@suse.de>
2113
2114         * passes.c (execute_function_todo): Move TODO_verify_flow under
2115         the TODO_verify_ul umbrella.
2116
2117 2014-05-05  Richard Biener  <rguenther@suse.de>
2118
2119         PR middle-end/61010
2120         * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
2121         X & CST away from a CST that is the mask of a mode.
2122
2123 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2124
2125         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
2126         int argument to enum machine_mode.
2127         (picochip_class_max_nregs): Ditto.
2128         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
2129         (picochip_class_max_nregs): Ditto.
2130
2131 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2132
2133         * target.def: Add new target hook.
2134         * doc/tm.texi: Regenerate.
2135         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
2136         * targhooks.c (default_keep_leaf_when_profiled): New function.
2137
2138         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
2139         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
2140
2141 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
2142
2143         PR tree-optimization/60363
2144         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
2145         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
2146         (update_destination_phis): New parameter.
2147         (create_edge_and_update_destination_phis): Ditto.
2148         (ssa_fix_duplicate_block_edges): Pass new arguments.
2149         (thread_single_edge): Ditto.
2150
2151 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
2152
2153         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
2154         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
2155         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
2156         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
2157         Use RS6000_BTM_HARD_FLOAT.
2158         (BU_MISC_2): Likewise.
2159         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
2160         RS6000_BTM_HARD_FLOAT.
2161         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
2162         is explicitly used.
2163         (rs6000_invalid_builtin): Add hard floating builtin support.
2164         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
2165         hard float builtins.
2166         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
2167
2168 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2169
2170         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
2171         Add missing function* argument.
2172
2173 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
2174
2175         * lra-constraints.c (valid_address_p): Move earlier in file.
2176         Add a constraint argument to the address_info version.
2177         (satisfies_memory_constraint_p): New function.
2178         (satisfies_address_constraint_p): Likewise.
2179         (process_alt_operands, curr_insn_transform): Use them.
2180         (process_address): Pass the constraint to valid_address_p when
2181         checking address operands.
2182
2183 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
2184
2185         * config/mips/mips.c (mips_isa_rev): New variable.
2186         (mips_set_architecture): Set it.
2187         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
2188         from mips_isa_rev.
2189         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
2190         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
2191         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
2192         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
2193         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
2194         conditions in terms of mips_isa_rev.
2195         (mips_isa_rev): Declare.
2196
2197 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2198
2199         * config/sh/sh-mem.cc: Use tabs instead of spaces.
2200         (prob_unlikely, prob_likely): Make variables const.
2201
2202 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
2203
2204         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
2205
2206 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2207
2208         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
2209
2210 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2211
2212         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
2213         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
2214         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
2215         functions.
2216         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
2217         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
2218         sh_pass_in_reg_p.
2219         Replace usage of ROUND_REG with sh_round_reg.
2220         Use CEIL instead of ROUND_ADVANCE.
2221
2222 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2223
2224         PR target/61026
2225         * config/sh/sh.c: Include stdlib headers before everything else.
2226
2227 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
2228
2229         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
2230         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
2231         (gimplify_adjust_omp_clauses): Simd region is never
2232         directly nested in combined parallel.  Instead, for linear
2233         with copyin/copyout, if in combined for simd loop, make decl
2234         firstprivate/lastprivate on OMP_FOR.
2235         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
2236         expand_omp_for_static_chunk): When setting endvar, also set
2237         fd->loop.v to the same value.
2238
2239 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
2240
2241         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
2242
2243 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
2244
2245         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
2246         expression.
2247
2248 2014-05-02  Marek Polacek  <polacek@redhat.com>
2249
2250         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
2251
2252 2014-05-02  Kito Cheng  <kito@0xlab.org>
2253
2254         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
2255         to a C expression marco.
2256         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
2257         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
2258         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
2259         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
2260         HONOR_REG_ALLOC_ORDER.
2261         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
2262
2263 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2264
2265         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
2266
2267 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2268
2269         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
2270
2271 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
2272
2273         * tree-if-conv.c (is_cond_scalar_reduction): New function.
2274         (convert_scalar_cond_reduction): Likewise.
2275         (predicate_scalar_phi): Add recognition and transformation
2276         of simple conditioanl reduction to be vectorizable.
2277
2278 2014-05-01  Marek Polacek  <polacek@redhat.com>
2279
2280         PR c/43245
2281         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
2282
2283 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
2284
2285         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
2286         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
2287         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
2288         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
2289         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
2290         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
2291         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
2292         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
2293
2294 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2295
2296         * config/arc/arc.opt (mlra): Move comment above option name
2297         to avoid mis-parsing as language options.
2298
2299 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2300
2301         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
2302         * config/sol2.h: ... here.
2303         * config/sol2-10.h: Remove.
2304
2305         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
2306         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
2307         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
2308         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
2309         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
2310         * config/sol2.h: ... here.
2311         (SECTION_NAME_FORMAT): Don't redefine.
2312         (STARTFILE_ARCH32_SPEC): Rename to ...
2313         (STARTFILE_ARCH_SPEC): ... this.
2314         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
2315         * config/sparc/sol2.h: ... here.
2316         (SECTION_NAME_FORMAT): Don't undef.
2317         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
2318         (SUBTARGET_EXTRA_SPECS): Remove.
2319         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
2320
2321         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
2322         (MD_STARTFILE_PREFIX): Remove.
2323         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
2324         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
2325         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
2326         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
2327         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
2328         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
2329         * config/i386/sol2.h: ... here.
2330         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
2331         * config/i386/sol2-bi.h: Remove.
2332         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
2333         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
2334
2335         * config/i386/t-sol2-64: Rename to ...
2336         * config/i386/t-sol2: ... this.
2337         * config/sparc/t-sol2-64: Rename to ...
2338         * config/sparc/t-sol2: ... this.
2339
2340         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
2341         sol2_tm_file_head, sol2_tm_file_tail.
2342         Include ${cpu_type}/sol2.h before sol2.h.
2343         Remove sol2-10.h.
2344         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
2345         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
2346         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
2347         Reflect i386/t-sol2-64 renaming.
2348         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
2349         Reflect sparc/t-sol2-64 renaming.
2350
2351 2014-04-30  Richard Biener  <rguenther@suse.de>
2352
2353         * passes.c (execute_function_todo): Move TODO_verify_stmts
2354         and TODO_verify_ssa under the TODO_verify_il umbrella.
2355         * tree-ssa.h (verify_ssa): Adjust prototype.
2356         * tree-ssa.c (verify_ssa): Add parameter to tell whether
2357         we should verify SSA operands.
2358         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
2359         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
2360         whether we should verify whether not throwing stmts have EH info.
2361         * graphite-scop-detection.c (create_sese_edges): Adjust.
2362         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
2363         * tree-eh.c (lower_try_finally_switch): Do not add the
2364         default case label twice.
2365
2366 2014-04-30  Marek Polacek  <polacek@redhat.com>
2367
2368         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
2369         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
2370         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
2371         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
2372
2373 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
2374
2375         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
2376         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
2377         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
2378         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
2379         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
2380         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
2381         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
2382         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
2383
2384 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
2385
2386         * tree-cfg.c (dump_function_to_file): Dump the return type of
2387         functions, in a line to itself before the function body, mimicking
2388         the layout of a C function.
2389
2390 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
2391
2392         PR tree-optimization/60971
2393         * tree-tailcall.c (process_assignment): Reject conversions which
2394         reduce precision.
2395
2396 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
2397
2398         * calls.c (initialize_argument_information): Always treat
2399         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
2400         (expand_call): Likewise.
2401         (emit_library_call_calue_1): Likewise.
2402         * expr.c (PUSH_ARGS_REVERSED): Do not define.
2403         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
2404         code accordingly.
2405
2406 2014-04-29  Nick Clifton  <nickc@redhat.com>
2407
2408         * config/msp430/msp430.md (umulsidi): Fix typo.
2409         (mulhisi3): Enable even inside interrupt handlers.
2410         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
2411         bigger return address pushed in large mode.
2412
2413 2014-04-29  Nick Clifton  <nickc@redhat.com>
2414
2415         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
2416         (arc_init_reg_tables): Use a machine_mode enum to iterate over
2417         available modes.
2418         * config/m32r/m32r.c (init_reg_tables): Likewise.
2419         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
2420         enum to hold the modes.
2421
2422 2014-04-29  Richard Biener  <rguenther@suse.de>
2423
2424         * dominance.c (free_dominance_info): Add overload with
2425         function parameter.
2426         (dom_info_state): Likewise.
2427         (dom_info_available_p): Likewise.
2428         * basic-block.h (free_dominance_info, dom_info_state,
2429         dom_info_available_p): Declare overloads.
2430         * passes.c (execute_function_todo): Verify that verifiers
2431         don't change dominator info state.  Drop dominator info
2432         for IPA pass invocations.
2433         * cgraph.c (release_function_body): Restore asserts that
2434         dominator information is released.
2435
2436 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
2437
2438         * doc/invoke.texi: Fix typo.
2439         * tree-vrp.c: Fix typos.
2440         * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
2441
2442 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2443
2444         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
2445
2446 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2447
2448         * config/aarch64/aarch64-builtins.c
2449         (aarch64_types_storestruct_lane_qualifiers): New.
2450         (TYPES_STORESTRUCT_LANE): Likewise.
2451         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
2452         (st3_lane): Likewise.
2453         (st4_lane): Likewise.
2454         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
2455         (vec_store_lanesci_lane<mode>): Likewise.
2456         (vec_store_lanesxi_lane<mode>): Likewise.
2457         (aarch64_st2_lane<VQ:mode>): Likewise.
2458         (aarch64_st3_lane<VQ:mode>): Likewise.
2459         (aarch64_st4_lane<VQ:mode>): Likewise.
2460         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
2461         * config/aarch64/arm_neon.h
2462         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
2463         use new macro arguments.
2464         (__ST3_LANE_FUNC): Likewise.
2465         (__ST4_LANE_FUNC): Likewise.
2466         * config/aarch64/iterators.md (V_TWO_ELEM): New.
2467         (V_THREE_ELEM): Likewise.
2468         (V_FOUR_ELEM): Likewise.
2469
2470 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
2471
2472         * doc/gimple.texi: Replace the description of the now-defunct
2473         union gimple_statement_d with a diagram showing the
2474         gimple_statement_base class hierarchy and its relationships to
2475         the GSS_ and GIMPLE_ enums.
2476
2477 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2478
2479         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
2480         * config/aarch64/aarch64.c
2481         (aarch64_cannot_change_mode_class): Weaken conditions.
2482         (aarch64_modes_tieable_p): New.
2483         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
2484
2485 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
2486
2487         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2488         (loadsync_<mode>): Change mode.
2489         (load_quadpti, store_quadpti): New.
2490         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2491         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2492
2493 2014-04-28  Martin Jambor  <mjambor@suse.cz>
2494
2495         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
2496         same alias type as the original statement.
2497         (subreplacement_assignment_data): New type.
2498         (handle_unscalarized_data_in_subtree): New type of parameter,
2499         generate new memory accesses with same alias type as the original
2500         statement.
2501         (load_assign_lhs_subreplacements): Likewise.
2502         (sra_modify_constructor_assign): Generate new memory accesses with
2503         same alias type as the original statement.
2504
2505 2014-04-28  Richard Biener  <rguenther@suse.de>
2506
2507         * tree-pass.h (TODO_verify_il): Define.
2508         (TODO_verify_all): Complete properly.
2509         * passes.c (execute_function_todo): Move existing loop-closed
2510         SSA verification under TODO_verify_il.
2511         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
2512         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
2513         Fix tree sharing issue.
2514
2515 2014-04-28  Richard Biener  <rguenther@suse.de>
2516
2517         PR middle-end/60092
2518         * builtins.def (DEF_C11_BUILTIN): Add.
2519         (BUILT_IN_ALIGNED_ALLOC): Likewise.
2520         * coretypes.h (enum function_class): Add function_c11_misc.
2521         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
2522         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
2523         (call_may_clobber_ref_p_1): Likewise.
2524         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2525         (mark_all_reaching_defs_necessary_1): Likewise.
2526         (propagate_necessity): Likewise.
2527         (eliminate_unnecessary_stmts): Likewise.
2528         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
2529
2530 2014-04-28  Richard Biener  <rguenther@suse.de>
2531
2532         * tree-vrp.c (vrp_var_may_overflow): Remove.
2533         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
2534         with overflow immediately bump to one before that value and
2535         let iteration figure out overflow status.
2536
2537 2014-04-28  Richard Biener  <rguenther@suse.de>
2538
2539         * configure.ac: Do valgrind header checks unconditionally.
2540         Add --enable-valgrind-annotations.
2541         * system.h: Guard valgrind header inclusion with
2542         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
2543         * alloc-pool.c (pool_alloc, pool_free): Use
2544         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
2545         to guard possibly dead code.
2546         * config.in: Regenerated.
2547         * configure: Likewise.
2548
2549 2014-04-28  Jeff Law  <law@redhat.com>
2550
2551         PR tree-optimization/60902
2552         * tree-ssa-threadedge.c
2553         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
2554         over real defs when invalidating outputs from statements that do not
2555         produce useful outputs for threading.
2556
2557 2014-04-28  Richard Biener  <rguenther@suse.de>
2558
2559         PR tree-optimization/60979
2560         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
2561         SCOPs that end in a block with a successor with abnormal
2562         predecessors.
2563
2564 2014-04-28  Richard Biener  <rguenther@suse.de>
2565
2566         * tree-pass.h (execute_pass_list): Adjust prototype.
2567         * passes.c (pass_manager::execute_early_local_passes): Adjust.
2568         (do_per_function): Change callback signature, push all actual
2569         work to the callbals.
2570         (do_per_function_toporder): Likewise.
2571         (execute_function_dump): Adjust.
2572         (execute_function_todo): Likewise.
2573         (clear_last_verified): Likewise.
2574         (verify_curr_properties): Likewise.
2575         (update_properties_after_pass): Likewise.
2576         (execute_pass_list_1): Split out from ...
2577         (execute_pass_list): ... here.  Adjust.
2578         (execute_ipa_pass_list): Likewise.
2579         * cgraphunit.c (cgraph_add_new_function): Adjust.
2580         (analyze_function): Likewise.
2581         (expand_function): Likewise.
2582         * cgraph.c (release_function_body): Free dominance info
2583         here instead of asserting it was magically freed elsewhere.
2584
2585 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
2586
2587         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
2588         * configure: Regenerate.
2589         * config/sparc/sparc.opt (muser-mode): New option.
2590         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
2591         for LEON3.
2592         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
2593         * doc/invoke.texi (SPARC options): Document -muser-mode.
2594
2595 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
2596
2597         * cselib.c (find_slot_memmode): Delete.
2598         (cselib_hasher): Change compare_type to a struct.
2599         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
2600         constants.
2601         (preserve_constants_and_equivs): Adjust for new compare_type.
2602         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
2603         (wrap_constant): Delete.
2604         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
2605
2606 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
2607
2608         * doc/install.texi (Building with profile feedback): Remove
2609         outdated sentence.
2610
2611 2014-04-26  Tom de Vries  <tom@codesourcery.com>
2612
2613         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
2614         array accesses.
2615
2616 2014-04-25  Cary Coutant  <ccoutant@google.com>
2617
2618         PR debug/60929
2619         * dwarf2out.c (should_move_die_to_comdat): A type definition
2620         can contain a subprogram definition, but don't move it to a
2621         comdat unit.
2622         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
2623         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
2624         from original DIE.
2625         (clone_tree_hash): Rename to...
2626         (clone_tree_partial): ...this; change callers.  Copy
2627         DW_TAG_subprogram DIEs as declarations.
2628         (copy_decls_walk): Don't copy children of a declaration into a
2629         type unit.
2630
2631 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
2632
2633         PR target/60969
2634         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
2635         alternative 12.
2636
2637 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
2638
2639         * config/arm/predicates.md (call_insn_operand): Add long_call check.
2640         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
2641         reg for long_call.
2642         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
2643         restriction.
2644
2645 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2646
2647         * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
2648
2649 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2650
2651         PR tree-optimization/60930
2652         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
2653         creating a multiply candidate by folding two constant
2654         multiplicands when the result overflows.
2655
2656 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
2657
2658         PR tree-optimization/60960
2659         * tree-vect-generic.c (expand_vector_operation): Only call
2660         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
2661
2662 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2663
2664         * expr.c (clobber_reg_mode): New function.
2665         * expr.h (clobber_reg): New function.
2666
2667 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2668
2669         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
2670         clobbers.
2671
2672 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
2673             Tom de Vries  <tom@codesourcery.com>
2674
2675         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
2676         handle.
2677         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
2678         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
2679         new argument to find_all_hard_reg_sets call.
2680
2681 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2682
2683         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
2684         Use HOST_WIDE_INT_C for mask literal.
2685         (aarch_rev16_shleft_mask_imm_p): Likewise.
2686
2687 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
2688
2689         PR target/60941
2690         * config/sparc/sparc.md (ashlsi3_extend): Delete.
2691
2692 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
2693
2694         PR preprocessor/56540
2695         * config/i386/i386-c.c (ix86_target_macros): Define
2696         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
2697
2698 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2699
2700         * configure.ac (tga_func): Remove.
2701         (LIB_TLS_SPEC): Remove.
2702         * configure: Regenerate.
2703         * config.in: Regenerate.
2704         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
2705
2706 2014-04-25  Richard Biener  <rguenther@suse.de>
2707
2708         PR ipa/60912
2709         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
2710         call stmt use/clobber sets during stmt walk instead of
2711         walking the possibly incomplete set of caller edges.
2712
2713 2014-04-25  Richard Biener  <rguenther@suse.de>
2714
2715         PR ipa/60911
2716         * passes.c (apply_ipa_transforms): Inline into only caller ...
2717         (execute_one_pass): ... here.  Properly bring in function
2718         bodies for nodes we want to apply IPA transforms to.
2719
2720 2014-04-24  Cong Hou  <congh@google.com>
2721
2722         PR tree-optimization/60896
2723         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
2724         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
2725         (vect_mark_pattern_stmts): Set the def type of all statements in
2726         PATTERN_DEF_SEQ as vect_internal_def.
2727
2728 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
2729
2730         * doc/extend.texi (PowerPC Built-in Functions): Document new
2731         powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
2732         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
2733
2734         * config/rs6000/predicates.md (const_0_to_3_operand): New
2735         predicate to match 0..3 integer constants.
2736
2737         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
2738         to support adding miscellaneous builtin functions.
2739         (BU_DFP_MISC_2): Likewise.
2740         (BU_P7_MISC_1): Likewise.
2741         (BU_P7_MISC_2): Likewise.
2742         (BU_P8V_MISC_3): Likewise.
2743         (BU_MISC_1): Likewise.
2744         (BU_MISC_2): Likewise.
2745         (DIVWE): Add extended divide builtin functions.
2746         (DIVWEO): Likewise.
2747         (DIVWEU): Likewise.
2748         (DIVWEUO): Likewise.
2749         (DIVDE): Likewise.
2750         (DIVDEO): Likewise.
2751         (DIVDEU): Likewise.
2752         (DIVDEUO): Likewise.
2753         (DXEX): Add decimal floating-point builtin functions.
2754         (DXEXQ): Likewise.
2755         (DDEDPD): Likewise.
2756         (DDEDPDQ): Likewise.
2757         (DENBCD): Likewise.
2758         (DENBCDQ): Likewise.
2759         (DIEX): Likewise.
2760         (DIEXQ): Likewise.
2761         (DSCLI): Likewise.
2762         (DSCLIQ): Likewise.
2763         (DSCRI): Likewise.
2764         (DSCRIQ): Likewise.
2765         (CDTBCD): Add new BCD builtin functions.
2766         (CBCDTD): Likewise.
2767         (ADDG6S): Likewise.
2768         (BCDADD): Likewise.
2769         (BCDADD_LT): Likewise.
2770         (BCDADD_EQ): Likewise.
2771         (BCDADD_GT): Likewise.
2772         (BCDADD_OV): Likewise.
2773         (BCDSUB): Likewise.
2774         (BCDSUB_LT): Likewise.
2775         (BCDSUB_EQ): Likewise.
2776         (BCDSUB_GT): Likewise.
2777         (BCDSUB_OV): Likewise.
2778         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
2779         (UNPACK_TD): Likewise.
2780         (PACK_TF): Likewise.
2781         (UNPACK_TF): Likewise.
2782         (UNPACK_TF_0): Likewise.
2783         (UNPACK_TF_1): Likewise.
2784         (PACK_V1TI): Likewise.
2785         (UNPACK_V1TI): Likewise.
2786
2787         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2788         support for decimal floating point builtin functions.
2789         (rs6000_expand_ternop_builtin): Add checks for the new builtin
2790         functions that take constant arguments.
2791         (rs6000_invalid_builtin): Add decimal floating point builtin support.
2792         (rs6000_init_builtins): Setup long double, _Decimal64, and
2793         _Decimal128 types for new builtin functions.
2794         (builtin_function_type): Set the unsigned flags appropriately for
2795         the new builtin functions.
2796         (rs6000_opt_masks): Add support for decimal floating point builtin
2797         functions.
2798
2799         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
2800         floating point builtin functions.
2801         (RS6000_BTM_COMMON): Likewise.
2802         (RS6000_BTI_long_double): Likewise.
2803         (RS6000_BTI_dfloat64): Likewise.
2804         (RS6000_BTI_dfloat128): Likewise.
2805         (long_double_type_internal_node): Likewise.
2806         (dfloat64_type_internal_node): Likewise.
2807         (dfloat128_type_internal_node): Likewise.
2808
2809         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
2810         2.07 bcd arithmetic instructions.
2811         (UNSPEC_BCDSUB): Likewise.
2812         (UNSPEC_BCD_OVERFLOW): Likewise.
2813         (UNSPEC_BCD_ADD_SUB): Likewise.
2814         (bcd_add_sub): Likewise.
2815         (BCD_TEST): Likewise.
2816         (bcd<bcd_add_sub>): Likewise.
2817         (bcd<bcd_add_sub>_test): Likewise.
2818         (bcd<bcd_add_sub>_test2): Likewise.
2819         (bcd<bcd_add_sub>_<code>): Likewise.
2820         (peephole2 for combined bcd ops): Likewise.
2821
2822         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
2823         decimal floating point builtin functions.
2824         (UNSPEC_DENBCD): Likewise.
2825         (UNSPEC_DXEX): Likewise.
2826         (UNSPEC_DIEX): Likewise.
2827         (UNSPEC_DSCLI): Likewise.
2828         (UNSPEC_DSCRI): Likewise.
2829         (D64_D128): Likewise.
2830         (dfp_suffix): Likewise.
2831         (dfp_ddedpd_<mode>): Likewise.
2832         (dfp_denbcd_<mode>): Likewise.
2833         (dfp_dxex_<mode>): Likewise.
2834         (dfp_diex_<mode>): Likewise.
2835         (dfp_dscli_<mode>): Likewise.
2836         (dfp_dscri_<mode>): Likewise.
2837
2838         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
2839         builtin functions.
2840         (UNSPEC_CDTBCD): Likewise.
2841         (UNSPEC_CBCDTD): Likewise.
2842         (UNSPEC_DIVE): Add support for new extended divide builtin functions.
2843         (UNSPEC_DIVEO): Likewise.
2844         (UNSPEC_DIVEU): Likewise.
2845         (UNSPEC_DIVEUO): Likewise.
2846         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
2847         pack/unpack 128-bit types.
2848         (UNSPEC_PACK_128BIT): Likewise.
2849         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
2850         (udiv<mode>3): Use idiv_ldiv mode attribute.
2851         (div<mode>3): Likewise.
2852         (addg6s): Add new BCD builtin functions.
2853         (cdtbcd): Likewise.
2854         (cbcdtd): Likewise.
2855         (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
2856         (div_extend): Likewise.
2857         (div<div_extend>_<mode>"): Likewise.
2858         (FP128_64): Add support for new builtin functions to pack/unpack
2859         128-bit types.
2860         (unpack<mode>): Likewise.
2861         (unpacktf_0): Likewise.
2862         (unpacktf_1): Likewise.
2863         (unpack<mode>_dm): Likewise.
2864         (unpack<mode>_nodm): Likewise.
2865         (pack<mode>): Likewise.
2866         (unpackv1ti): Likewise.
2867         (packv1ti): Likewise.
2868
2869 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
2870
2871         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
2872         is disabled.
2873
2874 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
2875
2876         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
2877         * gimplify.c (omp_is_private): Change last argument's type to int.
2878         Only diagnose lastprivate if the simd argument is 1, only diagnose
2879         linear if the simd argument is 2.
2880         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
2881         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
2882         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
2883         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
2884         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
2885         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2886         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
2887         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2888         * tree-nested.c (convert_nonlocal_omp_clauses,
2889         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
2890
2891 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
2892
2893         PR target/60822
2894         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
2895         operand 1.
2896
2897 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
2898
2899         * flag-types.h (enum ivar_visibility): Add.
2900
2901 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
2902
2903         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
2904         function * argument.
2905
2906 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
2907
2908         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
2909
2910 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2911             Tom de Vries  <tom@codesourcery.com>
2912
2913         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
2914         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
2915         reg-note.
2916         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
2917         * emit-rtl.c (try_split): Same.
2918
2919 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2920             Tom de Vries  <tom@codesourcery.com>
2921
2922         * common.opt (fuse-caller-save): New option.
2923
2924 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
2925
2926         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
2927         elements for big-endian.
2928
2929 2014-04-24  Richard Biener  <rguenther@suse.de>
2930
2931         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
2932         during TER and instead use the sepops interface for expanding
2933         non-GIMPLE_SINGLE_RHS.
2934
2935 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2936
2937         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
2938         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
2939
2940 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2941
2942         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
2943         assembler 64-bit option.
2944         * configure: Regenerate.
2945
2946 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2947
2948         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
2949         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
2950         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
2951         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
2952         (TARGET_CRYPTO): Take TARGET_SIMD into account.
2953
2954 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2955
2956         * config/aarch64/aarch64-builtins.c
2957         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
2958         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
2959         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
2960         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
2961         builtins.
2962         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
2963         (Vrevsuff): New mode attribute.
2964
2965 2014-04-24  Terry Guo  <terry.guo@arm.com>
2966
2967         * config/arm/arm.h (machine_function): Define variable
2968         after_arm_reorg here.
2969         * config/arm/arm.c (after_arm_reorg): Remove the definition.
2970         (arm_split_constant): Update the way to access variable
2971         after_arm_reorg.
2972         (arm_reorg): Ditto.
2973         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
2974
2975 2014-04-23  Tom de Vries  <tom@codesourcery.com>
2976
2977         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
2978
2979 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
2980
2981         * is-a.h: Update comments to reflect the following changes to the
2982         "pointerness" of the API, making the template parameter match the
2983         return type, allowing use of is-a.h with typedefs of pointers.
2984         (is_a_helper::cast): Return a T rather then a pointer to a T, so
2985         that the return type matches the parameter to the is_a_helper.
2986         (as_a): Likewise.
2987         (dyn_cast): Likewise.
2988
2989         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
2990         pointer from the is-a.h API.
2991
2992         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
2993         (is_a_helper <cgraph_node *>::test): ...this, matching change to
2994         is-a.h API.
2995         (is_a_helper <varpool_node>::test): Likewise, convert to...
2996         (is_a_helper <varpool_node *>::test): ...this.
2997
2998         (varpool_first_variable): Update for removal of implicit pointer
2999         from the is-a.h API.
3000         (varpool_next_variable): Likewise.
3001         (varpool_first_static_initializer): Likewise.
3002         (varpool_next_static_initializer): Likewise.
3003         (varpool_first_defined_variable): Likewise.
3004         (varpool_next_defined_variable): Likewise.
3005         (cgraph_first_defined_function): Likewise.
3006         (cgraph_next_defined_function): Likewise.
3007         (cgraph_first_function): Likewise.
3008         (cgraph_next_function): Likewise.
3009         (cgraph_first_function_with_gimple_body): Likewise.
3010         (cgraph_next_function_with_gimple_body): Likewise.
3011         (cgraph_alias_target): Likewise.
3012         (varpool_alias_target): Likewise.
3013         (cgraph_function_or_thunk_node): Likewise.
3014         (varpool_variable_node): Likewise.
3015         (symtab_real_symbol_p): Likewise.
3016         * cgraphunit.c (referred_to_p): Likewise.
3017         (analyze_functions): Likewise.
3018         (handle_alias_pairs): Likewise.
3019         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
3020         * gimple-ssa.h (gimple_vuse_op): Likewise.
3021         (gimple_vdef_op): Likewise.
3022         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
3023         * gimple.c (gimple_build_asm_1): Likewise.
3024         (gimple_build_try): Likewise.
3025         (gimple_build_resx): Likewise.
3026         (gimple_build_eh_dispatch): Likewise.
3027         (gimple_build_omp_for): Likewise.
3028         (gimple_omp_for_set_clauses): Likewise.
3029
3030         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
3031         (is_a_helper <gimple_statement_asm *>::test): ...this.
3032         (is_a_helper <gimple_statement_bind>::test): Convert to...
3033         (is_a_helper <gimple_statement_bind *>::test): ...this.
3034         (is_a_helper <gimple_statement_call>::test): Convert to...
3035         (is_a_helper <gimple_statement_call *>::test): ...this.
3036         (is_a_helper <gimple_statement_catch>::test): Convert to...
3037         (is_a_helper <gimple_statement_catch *>::test): ...this.
3038         (is_a_helper <gimple_statement_resx>::test): Convert to...
3039         (is_a_helper <gimple_statement_resx *>::test): ...this.
3040         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
3041         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
3042         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
3043         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
3044         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
3045         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
3046         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
3047         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
3048         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
3049         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
3050         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
3051         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
3052         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
3053         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
3054         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
3055         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
3056         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
3057         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
3058         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
3059         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
3060         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
3061         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
3062         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
3063         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
3064         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
3065         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
3066         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
3067         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
3068         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
3069         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
3070         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
3071         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
3072         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
3073         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
3074         (is_a_helper <gimple_statement_phi>::test): Convert to...
3075         (is_a_helper <gimple_statement_phi *>::test): ...this.
3076         (is_a_helper <gimple_statement_transaction>::test): Convert to...
3077         (is_a_helper <gimple_statement_transaction *>::test): ...this.
3078         (is_a_helper <gimple_statement_try>::test): Convert to...
3079         (is_a_helper <gimple_statement_try *>::test): ...this.
3080         (is_a_helper <gimple_statement_wce>::test): Convert to...
3081         (is_a_helper <gimple_statement_wce *>::test): ...this.
3082         (is_a_helper <const gimple_statement_asm>::test): Convert to...
3083         (is_a_helper <const gimple_statement_asm *>::test): ...this.
3084         (is_a_helper <const gimple_statement_bind>::test): Convert to...
3085         (is_a_helper <const gimple_statement_bind *>::test): ...this.
3086         (is_a_helper <const gimple_statement_call>::test): Convert to...
3087         (is_a_helper <const gimple_statement_call *>::test): ...this.
3088         (is_a_helper <const gimple_statement_catch>::test): Convert to...
3089         (is_a_helper <const gimple_statement_catch *>::test): ...this.
3090         (is_a_helper <const gimple_statement_resx>::test): Convert to...
3091         (is_a_helper <const gimple_statement_resx *>::test): ...this.
3092         (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
3093         (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
3094         (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
3095         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
3096         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
3097         Convert to...
3098         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
3099         ...this.
3100         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
3101         Convert to...
3102         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
3103         ...this.
3104         (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
3105         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
3106         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
3107         to...
3108         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
3109         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
3110         to...
3111         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
3112         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
3113         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
3114         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
3115         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
3116         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
3117         to...
3118         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
3119         (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
3120         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
3121         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
3122         to...
3123         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
3124         (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
3125         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
3126         (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
3127         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
3128         (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
3129         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
3130         (is_a_helper <const gimple_statement_phi>::test): Convert to...
3131         (is_a_helper <const gimple_statement_phi *>::test): ...this.
3132         (is_a_helper <const gimple_statement_transaction>::test): Convert to...
3133         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
3134         (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
3135         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
3136         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
3137         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
3138         (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
3139         to...
3140         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
3141         ...this.
3142         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
3143         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
3144
3145         (gimple_use_ops): Update for removal of implicit pointer from the
3146         is-a.h API.
3147         (gimple_set_use_ops): Likewise.
3148         (gimple_vuse): Likewise.
3149         (gimple_vdef): Likewise.
3150         (gimple_vuse_ptr): Likewise.
3151         (gimple_vdef_ptr): Likewise.
3152         (gimple_set_vuse): Likewise.
3153         (gimple_set_vdef): Likewise.
3154         (gimple_omp_return_set_lhs): Likewise.
3155         (gimple_omp_return_lhs): Likewise.
3156         (gimple_omp_return_lhs_ptr): Likewise.
3157         (gimple_call_fntype): Likewise.
3158         (gimple_call_set_fntype): Likewise.
3159         (gimple_call_set_internal_fn): Likewise.
3160         (gimple_call_use_set): Likewise.
3161         (gimple_call_clobber_set): Likewise.
3162         (gimple_bind_vars): Likewise.
3163         (gimple_bind_set_vars): Likewise.
3164         (gimple_bind_body_ptr): Likewise.
3165         (gimple_bind_set_body): Likewise.
3166         (gimple_bind_add_stmt): Likewise.
3167         (gimple_bind_block): Likewise.
3168         (gimple_bind_set_block): Likewise.
3169         (gimple_asm_ninputs): Likewise.
3170         (gimple_asm_noutputs): Likewise.
3171         (gimple_asm_nclobbers): Likewise.
3172         (gimple_asm_nlabels): Likewise.
3173         (gimple_asm_input_op): Likewise.
3174         (gimple_asm_input_op_ptr): Likewise.
3175         (gimple_asm_output_op): Likewise.
3176         (gimple_asm_output_op_ptr): Likewise.
3177         (gimple_asm_set_output_op): Likewise.
3178         (gimple_asm_clobber_op): Likewise.
3179         (gimple_asm_set_clobber_op): Likewise.
3180         (gimple_asm_label_op): Likewise.
3181         (gimple_asm_set_label_op): Likewise.
3182         (gimple_asm_string): Likewise.
3183         (gimple_catch_types): Likewise.
3184         (gimple_catch_types_ptr): Likewise.
3185         (gimple_catch_handler_ptr): Likewise.
3186         (gimple_catch_set_types): Likewise.
3187         (gimple_catch_set_handler): Likewise.
3188         (gimple_eh_filter_types): Likewise.
3189         (gimple_eh_filter_types_ptr): Likewise.
3190         (gimple_eh_filter_failure_ptr): Likewise.
3191         (gimple_eh_filter_set_types): Likewise.
3192         (gimple_eh_filter_set_failure): Likewise.
3193         (gimple_eh_must_not_throw_fndecl): Likewise.
3194         (gimple_eh_must_not_throw_set_fndecl): Likewise.
3195         (gimple_eh_else_n_body_ptr): Likewise.
3196         (gimple_eh_else_e_body_ptr): Likewise.
3197         (gimple_eh_else_set_n_body): Likewise.
3198         (gimple_eh_else_set_e_body): Likewise.
3199         (gimple_try_eval_ptr): Likewise.
3200         (gimple_try_cleanup_ptr): Likewise.
3201         (gimple_try_set_eval): Likewise.
3202         (gimple_try_set_cleanup): Likewise.
3203         (gimple_wce_cleanup_ptr): Likewise.
3204         (gimple_wce_set_cleanup): Likewise.
3205         (gimple_phi_capacity): Likewise.
3206         (gimple_phi_num_args): Likewise.
3207         (gimple_phi_result): Likewise.
3208         (gimple_phi_result_ptr): Likewise.
3209         (gimple_phi_set_result): Likewise.
3210         (gimple_phi_arg): Likewise.
3211         (gimple_phi_set_arg): Likewise.
3212         (gimple_resx_region): Likewise.
3213         (gimple_resx_set_region): Likewise.
3214         (gimple_eh_dispatch_region): Likewise.
3215         (gimple_eh_dispatch_set_region): Likewise.
3216         (gimple_omp_critical_name): Likewise.
3217         (gimple_omp_critical_name_ptr): Likewise.
3218         (gimple_omp_critical_set_name): Likewise.
3219         (gimple_omp_for_clauses): Likewise.
3220         (gimple_omp_for_clauses_ptr): Likewise.
3221         (gimple_omp_for_set_clauses): Likewise.
3222         (gimple_omp_for_collapse): Likewise.
3223         (gimple_omp_for_index): Likewise.
3224         (gimple_omp_for_index_ptr): Likewise.
3225         (gimple_omp_for_set_index): Likewise.
3226         (gimple_omp_for_initial): Likewise.
3227         (gimple_omp_for_initial_ptr): Likewise.
3228         (gimple_omp_for_set_initial): Likewise.
3229         (gimple_omp_for_final): Likewise.
3230         (gimple_omp_for_final_ptr): Likewise.
3231         (gimple_omp_for_set_final): Likewise.
3232         (gimple_omp_for_incr): Likewise.
3233         (gimple_omp_for_incr_ptr): Likewise.
3234         (gimple_omp_for_set_incr): Likewise.
3235         (gimple_omp_for_pre_body_ptr): Likewise.
3236         (gimple_omp_for_set_pre_body): Likewise.
3237         (gimple_omp_parallel_clauses): Likewise.
3238         (gimple_omp_parallel_clauses_ptr): Likewise.
3239         (gimple_omp_parallel_set_clauses): Likewise.
3240         (gimple_omp_parallel_child_fn): Likewise.
3241         (gimple_omp_parallel_child_fn_ptr): Likewise.
3242         (gimple_omp_parallel_set_child_fn): Likewise.
3243         (gimple_omp_parallel_data_arg): Likewise.
3244         (gimple_omp_parallel_data_arg_ptr): Likewise.
3245         (gimple_omp_parallel_set_data_arg): Likewise.
3246         (gimple_omp_task_clauses): Likewise.
3247         (gimple_omp_task_clauses_ptr): Likewise.
3248         (gimple_omp_task_set_clauses): Likewise.
3249         (gimple_omp_task_child_fn): Likewise.
3250         (gimple_omp_task_child_fn_ptr): Likewise.
3251         (gimple_omp_task_set_child_fn): Likewise.
3252         (gimple_omp_task_data_arg): Likewise.
3253         (gimple_omp_task_data_arg_ptr): Likewise.
3254         (gimple_omp_task_set_data_arg): Likewise.
3255         (gimple_omp_taskreg_clauses): Likewise.
3256         (gimple_omp_taskreg_clauses_ptr): Likewise.
3257         (gimple_omp_taskreg_set_clauses): Likewise.
3258         (gimple_omp_taskreg_child_fn): Likewise.
3259         (gimple_omp_taskreg_child_fn_ptr): Likewise.
3260         (gimple_omp_taskreg_set_child_fn): Likewise.
3261         (gimple_omp_taskreg_data_arg): Likewise.
3262         (gimple_omp_taskreg_data_arg_ptr): Likewise.
3263         (gimple_omp_taskreg_set_data_arg): Likewise.
3264         (gimple_omp_task_copy_fn): Likewise.
3265         (gimple_omp_task_copy_fn_ptr): Likewise.
3266         (gimple_omp_task_set_copy_fn): Likewise.
3267         (gimple_omp_task_arg_size): Likewise.
3268         (gimple_omp_task_arg_size_ptr): Likewise.
3269         (gimple_omp_task_set_arg_size): Likewise.
3270         (gimple_omp_task_arg_align): Likewise.
3271         (gimple_omp_task_arg_align_ptr): Likewise.
3272         (gimple_omp_task_set_arg_align): Likewise.
3273         (gimple_omp_single_clauses): Likewise.
3274         (gimple_omp_single_clauses_ptr): Likewise.
3275         (gimple_omp_single_set_clauses): Likewise.
3276         (gimple_omp_target_clauses): Likewise.
3277         (gimple_omp_target_clauses_ptr): Likewise.
3278         (gimple_omp_target_set_clauses): Likewise.
3279         (gimple_omp_target_child_fn): Likewise.
3280         (gimple_omp_target_child_fn_ptr): Likewise.
3281         (gimple_omp_target_set_child_fn): Likewise.
3282         (gimple_omp_target_data_arg): Likewise.
3283         (gimple_omp_target_data_arg_ptr): Likewise.
3284         (gimple_omp_target_set_data_arg): Likewise.
3285         (gimple_omp_teams_clauses): Likewise.
3286         (gimple_omp_teams_clauses_ptr): Likewise.
3287         (gimple_omp_teams_set_clauses): Likewise.
3288         (gimple_omp_sections_clauses): Likewise.
3289         (gimple_omp_sections_clauses_ptr): Likewise.
3290         (gimple_omp_sections_set_clauses): Likewise.
3291         (gimple_omp_sections_control): Likewise.
3292         (gimple_omp_sections_control_ptr): Likewise.
3293         (gimple_omp_sections_set_control): Likewise.
3294         (gimple_omp_for_set_cond): Likewise.
3295         (gimple_omp_for_cond): Likewise.
3296         (gimple_omp_atomic_store_set_val): Likewise.
3297         (gimple_omp_atomic_store_val): Likewise.
3298         (gimple_omp_atomic_store_val_ptr): Likewise.
3299         (gimple_omp_atomic_load_set_lhs): Likewise.
3300         (gimple_omp_atomic_load_lhs): Likewise.
3301         (gimple_omp_atomic_load_lhs_ptr): Likewise.
3302         (gimple_omp_atomic_load_set_rhs): Likewise.
3303         (gimple_omp_atomic_load_rhs): Likewise.
3304         (gimple_omp_atomic_load_rhs_ptr): Likewise.
3305         (gimple_omp_continue_control_def): Likewise.
3306         (gimple_omp_continue_control_def_ptr): Likewise.
3307         (gimple_omp_continue_set_control_def): Likewise.
3308         (gimple_omp_continue_control_use): Likewise.
3309         (gimple_omp_continue_control_use_ptr): Likewise.
3310         (gimple_omp_continue_set_control_use): Likewise.
3311         (gimple_transaction_body_ptr): Likewise.
3312         (gimple_transaction_label): Likewise.
3313         (gimple_transaction_label_ptr): Likewise.
3314         (gimple_transaction_set_body): Likewise.
3315         (gimple_transaction_set_label): Likewise.
3316
3317         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
3318         * ipa-inline-analysis.c (inline_write_summary): Likewise.
3319         * ipa-ref.c (ipa_record_reference): Likewise.
3320         * ipa-reference.c (analyze_function): Likewise.
3321         (ipa_reference_write_optimization_summary): Likewise.
3322         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
3323         (address_taken_from_non_vtable_p): Likewise.
3324         (comdat_can_be_unshared_p_1): Likewise.
3325         * lto-cgraph.c (lto_output_ref): Likewise.
3326         (add_references): Likewise.
3327         (compute_ltrans_boundary): Likewise.
3328         (output_symtab): Likewise.
3329         (input_ref): Likewise.
3330         (input_cgraph_1): Likewise.
3331         (output_cgraph_opt_summary): Likewise.
3332         * lto-streamer-out.c (lto_output): Likewise.
3333         (output_symbol_p): Likewise.
3334         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
3335         (lsei_start_function_in_partition): Likewise.
3336         (lsei_next_variable_in_partition): Likewise.
3337         (lsei_start_variable_in_partition): Likewise.
3338         * symtab.c (insert_to_assembler_name_hash): Likewise.
3339         (unlink_from_assembler_name_hash): Likewise.
3340         (symtab_unregister_node): Likewise.
3341         (symtab_remove_node): Likewise.
3342         (dump_symtab_node): Likewise.
3343         (verify_symtab_base): Likewise.
3344         (verify_symtab_node): Likewise.
3345         (symtab_make_decl_local): Likewise.
3346         (symtab_alias_ultimate_target): Likewise.
3347         (symtab_resolve_alias): Likewise.
3348         (symtab_get_symbol_partitioning_class): Likewise.
3349         * tree-phinodes.c (allocate_phi_node): Likewise.
3350         (reserve_phi_args_for_new_edge): Likewise.
3351         (remove_phi_args): Likewise.
3352         * varpool.c (varpool_node_for_asm): Likewise.
3353         (varpool_remove_unreferenced_decls): Likewise.
3354
3355 2014-04-23  Jeff Law  <law@redhat.com>
3356
3357         PR tree-optimization/60902
3358         * tree-ssa-threadedge.c
3359         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
3360         invalidate outputs from statements that do not produce useful
3361         outputs for threading.
3362
3363 2014-04-23  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3364
3365         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
3366         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
3367         machine descriptions for Stack Smashing Protector.
3368
3369 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
3370
3371         * aarch64.md (<optab>_rol<mode>3): New pattern.
3372         (<optab>_rolsi3_uxtw): Likewise.
3373         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
3374
3375 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
3376
3377         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
3378         (arm_cortex_a12_tune): Likewise.
3379
3380 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3381
3382         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
3383
3384 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3385
3386         * config/arm/arm.md (arm_rev16si2): New pattern.
3387         (arm_rev16si2_alt): Likewise.
3388         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
3389
3390 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3391
3392         * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
3393         (rev16<mode>2_alt): Likewise.
3394         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
3395         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
3396         (aarch_rev16_shleft_mask_imm_p): Likewise.
3397         (aarch_rev16_p_1): Likewise.
3398         (aarch_rev16_p): Likewise.
3399         * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
3400         (aarch_rev16_shright_mask_imm_p): Likewise.
3401         (aarch_rev16_shleft_mask_imm_p): Likewise.
3402
3403 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3404
3405         * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
3406         * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
3407         rev cost.
3408         (cortex_a53_extra_costs): Likewise.
3409         (cortex_a57_extra_costs): Likewise.
3410         * config/arm/arm.c (cortexa9_extra_costs): Likewise.
3411         (cortexa7_extra_costs): Likewise.
3412         (cortexa8_extra_costs): Likewise.
3413         (cortexa12_extra_costs): Likewise.
3414         (cortexa15_extra_costs): Likewise.
3415         (v7m_extra_costs): Likewise.
3416         (arm_new_rtx_costs): Handle BSWAP.
3417
3418 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3419
3420         * config/arm/arm.c (cortexa8_extra_costs): New table.
3421         (arm_cortex_a8_tune): New tuning struct.
3422         * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
3423
3424 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3425
3426         * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
3427
3428 2014-04-23  Richard Biener  <rguenther@suse.de>
3429
3430         * Makefile.in (OBJS): Remove loop-unswitch.o.
3431         * tree-pass.h (make_pass_rtl_unswitch): Remove.
3432         * passes.def (pass_rtl_unswitch): Likewise.
3433         * loop-init.c (gate_rtl_unswitch): Likewise.
3434         (rtl_unswitch): Likewise.
3435         (pass_data_rtl_unswitch): Likewise.
3436         (pass_rtl_unswitch): Likewise.
3437         (make_pass_rtl_unswitch): Likewise.
3438         * rtl.h (reversed_condition): Likewise.
3439         (compare_and_jump_seq): Likewise.
3440         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
3441         and make static.
3442         * loop-unroll.c (compare_and_jump_seq): Likewise.
3443
3444 2014-04-23  Richard Biener  <rguenther@suse.de>
3445
3446         PR tree-optimization/60903
3447         * tree-ssa-loop-im.c (analyze_memory_references): Remove
3448         commented code block.
3449         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
3450         loop flags to newly created BBs and edges.
3451
3452 2014-04-23  Nick Clifton  <nickc@redhat.com>
3453
3454         * config/msp430/msp430.c (msp430_handle_option): Move function
3455         to msp430-common.c
3456         (msp430_option_override): Simplify mcu and mcpu option handling.
3457         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
3458         support for -mhwmult command line option.
3459         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
3460         -mhwmult command line option.
3461         (msp430_hwmult_enabled): Delete.
3462         (msp43o_output_labelref): Add support for -mhwmult command line option.
3463         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
3464         (umulsidi3): Likewise.
3465         * config/msp430/msp430.opt (mmcu): Add Report attribute.
3466         (mcpu, mlarge, msmall): Likewise.
3467         (mhwmult): New option.
3468         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
3469         prototype.
3470         (msp430_is_f5_mcu): Remove prototype.
3471         (msp430_use_f5_series_hwmult): Add prototype.
3472         * config/msp430/msp430-opts.h: New file.
3473         * common/config/msp430: New directory.
3474         * common/config/msp430/msp430-common.c: New file.
3475         * config.gcc (msp430): Remove target_has_targetm_common.
3476         * doc/invoke.texi: Document -mhwmult command line option.
3477
3478 2014-04-23  Nick Clifton  <nickc@redhat.com>
3479
3480         * config/i386/cygwin.h (ENDFILE_SPEC): Include
3481         default-manifest.o if it can be found in the search path.
3482         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
3483
3484 2014-04-23  Terry Guo  <terry.guo@arm.com>
3485
3486         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
3487
3488 2014-04-23  Richard Biener  <rguenther@suse.de>
3489
3490         PR middle-end/60895
3491         * tree-inline.c (declare_return_variable): Use mark_addressable.
3492
3493 2014-04-23  Richard Biener  <rguenther@suse.de>
3494
3495         PR middle-end/60891
3496         * loop-init.c (loop_optimizer_init): Make sure to apply
3497         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
3498
3499 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3500
3501         PR sanitizer/60275
3502         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
3503         New options.
3504         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
3505         if flag_sanitize_undefined_trap_on_error.
3506         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
3507         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
3508         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
3509         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3510         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
3511         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
3512         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
3513         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
3514         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
3515         * ubsan.c (ubsan_instrument_unreachable): Return
3516         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
3517         (ubsan_expand_null_ifn): Emit __builtin_trap ()
3518         if flag_sanitize_undefined_trap_on_error and
3519         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
3520         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
3521         instrument_bool_enum_load): Emit __builtin_trap () if
3522         flag_sanitize_undefined_trap_on_error and
3523         __builtin_handle_*_abort () if !flag_sanitize_recover.
3524         * doc/invoke.texi (-fsanitize-recover,
3525         -fsanitize-undefined-trap-on-error): Document.
3526
3527 2014-04-22  Christian Bruel  <christian.bruel@st.com>
3528
3529         * config/sh/sh.md (mov<mode>): Replace movQIHI.
3530          Force immediates to SImode.
3531
3532 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
3533
3534         * config/nios2/nios2.md (UNSPEC_ROUND): New.
3535         (lroundsfsi2): New.
3536         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
3537         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
3538         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
3539         (nios2_fpu_insn): Add entry for round.
3540         (N2FPU_NO_ERRNO_P): Define.
3541         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
3542         flag_errno_math.
3543         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
3544
3545 2014-04-22  Richard Henderson  <rth@redhat.com>
3546
3547         * config/aarch64/aarch64 (addti3, subti3): New expanders.
3548         (add<GPI>3_compare0): Remove leading * from name.
3549         (add<GPI>3_carryin): Likewise.
3550         (sub<GPI>3_compare0): Likewise.
3551         (sub<GPI>3_carryin): Likewise.
3552         (<su_optab>mulditi3): New expander.
3553         (multi3): New expander.
3554         (madd<GPI>): Remove leading * from name.
3555
3556 2014-04-22  Martin Jambor  <mjambor@suse.cz>
3557
3558         * cgraphclones.c (cgraph_function_versioning): Copy
3559         ipa_transforms_to_apply instead of asserting it is empty.
3560
3561 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
3562
3563         PR target/60868
3564         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
3565         on count_exp to get mode.
3566
3567 2014-04-22  Andrew Pinski  <apinski@cavium.com>
3568
3569         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3570         Handle TLS for ILP32.
3571         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
3572         (tlsie_small_<mode>): this and handle PTR.
3573         (tlsie_small_sidi): New pattern.
3574         (tlsle_small): Change to an expand to handle ILP32.
3575         (tlsle_small_<mode>): New pattern.
3576         (tlsdesc_small): Rename to ...
3577         (tlsdesc_small_<mode>): this and handle PTR.
3578
3579 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3580
3581         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
3582
3583 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3584
3585         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3586         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
3587         (aarch64_types_signed_poly_qualifiers): Likewise.
3588         (aarch64_types_unsigned_signed_qualifiers): Likewise.
3589         (aarch64_types_poly_signed_qualifiers): Likewise.
3590         (TYPES_REINTERP_SS): Type macro added.
3591         (TYPES_REINTERP_SU): Likewise.
3592         (TYPES_REINTERP_SP): Likewise.
3593         (TYPES_REINTERP_US): Likewise.
3594         (TYPES_REINTERP_PS): Likewise.
3595         (aarch64_fold_builtin): New expression folding added.
3596         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
3597         Declarations removed.
3598         (REINTERP_SS): Declarations added.
3599         (REINTERP_US): Likewise.
3600         (REINTERP_PS): Likewise.
3601         (REINTERP_SU): Likewise.
3602         (REINTERP_SP): Likewise.
3603         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
3604         (vreinterpretq_p8_f64): Likewise.
3605         (vreinterpret_p16_f64): Likewise.
3606         (vreinterpretq_p16_f64): Likewise.
3607         (vreinterpret_f32_f64): Likewise.
3608         (vreinterpretq_f32_f64): Likewise.
3609         (vreinterpret_f64_f32): Likewise.
3610         (vreinterpret_f64_p8): Likewise.
3611         (vreinterpret_f64_p16): Likewise.
3612         (vreinterpret_f64_s8): Likewise.
3613         (vreinterpret_f64_s16): Likewise.
3614         (vreinterpret_f64_s32): Likewise.
3615         (vreinterpret_f64_s64): Likewise.
3616         (vreinterpret_f64_u8): Likewise.
3617         (vreinterpret_f64_u16): Likewise.
3618         (vreinterpret_f64_u32): Likewise.
3619         (vreinterpret_f64_u64): Likewise.
3620         (vreinterpretq_f64_f32): Likewise.
3621         (vreinterpretq_f64_p8): Likewise.
3622         (vreinterpretq_f64_p16): Likewise.
3623         (vreinterpretq_f64_s8): Likewise.
3624         (vreinterpretq_f64_s16): Likewise.
3625         (vreinterpretq_f64_s32): Likewise.
3626         (vreinterpretq_f64_s64): Likewise.
3627         (vreinterpretq_f64_u8): Likewise.
3628         (vreinterpretq_f64_u16): Likewise.
3629         (vreinterpretq_f64_u32): Likewise.
3630         (vreinterpretq_f64_u64): Likewise.
3631         (vreinterpret_s64_f64): Likewise.
3632         (vreinterpretq_s64_f64): Likewise.
3633         (vreinterpret_u64_f64): Likewise.
3634         (vreinterpretq_u64_f64): Likewise.
3635         (vreinterpret_s8_f64): Likewise.
3636         (vreinterpretq_s8_f64): Likewise.
3637         (vreinterpret_s16_f64): Likewise.
3638         (vreinterpretq_s16_f64): Likewise.
3639         (vreinterpret_s32_f64): Likewise.
3640         (vreinterpretq_s32_f64): Likewise.
3641         (vreinterpret_u8_f64): Likewise.
3642         (vreinterpretq_u8_f64): Likewise.
3643         (vreinterpret_u16_f64): Likewise.
3644         (vreinterpretq_u16_f64): Likewise.
3645         (vreinterpret_u32_f64): Likewise.
3646         (vreinterpretq_u32_f64): Likewise.
3647
3648 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3649
3650         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3651         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
3652         (vreinterpret_p8_s8): Likewise.
3653         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
3654         (vreinterpret_p8_s16): Likewise.
3655         (vreinterpret_p8_s32): Likewise.
3656         (vreinterpret_p8_s64): Likewise.
3657         (vreinterpret_p8_f32): Likewise.
3658         (vreinterpret_p8_u8): Likewise.
3659         (vreinterpret_p8_u16): Likewise.
3660         (vreinterpret_p8_u32): Likewise.
3661         (vreinterpret_p8_u64): Likewise.
3662         (vreinterpret_p8_p16): Likewise.
3663         (vreinterpretq_p8_s8): Likewise.
3664         (vreinterpretq_p8_s16): Likewise.
3665         (vreinterpretq_p8_s32): Likewise.
3666         (vreinterpretq_p8_s64): Likewise.
3667         (vreinterpretq_p8_f32): Likewise.
3668         (vreinterpretq_p8_u8): Likewise.
3669         (vreinterpretq_p8_u16): Likewise.
3670         (vreinterpretq_p8_u32): Likewise.
3671         (vreinterpretq_p8_u64): Likewise.
3672         (vreinterpretq_p8_p16): Likewise.
3673         (vreinterpret_p16_s8): Likewise.
3674         (vreinterpret_p16_s16): Likewise.
3675         (vreinterpret_p16_s32): Likewise.
3676         (vreinterpret_p16_s64): Likewise.
3677         (vreinterpret_p16_f32): Likewise.
3678         (vreinterpret_p16_u8): Likewise.
3679         (vreinterpret_p16_u16): Likewise.
3680         (vreinterpret_p16_u32): Likewise.
3681         (vreinterpret_p16_u64): Likewise.
3682         (vreinterpret_p16_p8): Likewise.
3683         (vreinterpretq_p16_s8): Likewise.
3684         (vreinterpretq_p16_s16): Likewise.
3685         (vreinterpretq_p16_s32): Likewise.
3686         (vreinterpretq_p16_s64): Likewise.
3687         (vreinterpretq_p16_f32): Likewise.
3688         (vreinterpretq_p16_u8): Likewise.
3689         (vreinterpretq_p16_u16): Likewise.
3690         (vreinterpretq_p16_u32): Likewise.
3691         (vreinterpretq_p16_u64): Likewise.
3692         (vreinterpretq_p16_p8): Likewise.
3693         (vreinterpret_f32_s8): Likewise.
3694         (vreinterpret_f32_s16): Likewise.
3695         (vreinterpret_f32_s32): Likewise.
3696         (vreinterpret_f32_s64): Likewise.
3697         (vreinterpret_f32_u8): Likewise.
3698         (vreinterpret_f32_u16): Likewise.
3699         (vreinterpret_f32_u32): Likewise.
3700         (vreinterpret_f32_u64): Likewise.
3701         (vreinterpret_f32_p8): Likewise.
3702         (vreinterpret_f32_p16): Likewise.
3703         (vreinterpretq_f32_s8): Likewise.
3704         (vreinterpretq_f32_s16): Likewise.
3705         (vreinterpretq_f32_s32): Likewise.
3706         (vreinterpretq_f32_s64): Likewise.
3707         (vreinterpretq_f32_u8): Likewise.
3708         (vreinterpretq_f32_u16): Likewise.
3709         (vreinterpretq_f32_u32): Likewise.
3710         (vreinterpretq_f32_u64): Likewise.
3711         (vreinterpretq_f32_p8): Likewise.
3712         (vreinterpretq_f32_p16): Likewise.
3713         (vreinterpret_s64_s8): Likewise.
3714         (vreinterpret_s64_s16): Likewise.
3715         (vreinterpret_s64_s32): Likewise.
3716         (vreinterpret_s64_f32): Likewise.
3717         (vreinterpret_s64_u8): Likewise.
3718         (vreinterpret_s64_u16): Likewise.
3719         (vreinterpret_s64_u32): Likewise.
3720         (vreinterpret_s64_u64): Likewise.
3721         (vreinterpret_s64_p8): Likewise.
3722         (vreinterpret_s64_p16): Likewise.
3723         (vreinterpretq_s64_s8): Likewise.
3724         (vreinterpretq_s64_s16): Likewise.
3725         (vreinterpretq_s64_s32): Likewise.
3726         (vreinterpretq_s64_f32): Likewise.
3727         (vreinterpretq_s64_u8): Likewise.
3728         (vreinterpretq_s64_u16): Likewise.
3729         (vreinterpretq_s64_u32): Likewise.
3730         (vreinterpretq_s64_u64): Likewise.
3731         (vreinterpretq_s64_p8): Likewise.
3732         (vreinterpretq_s64_p16): Likewise.
3733         (vreinterpret_u64_s8): Likewise.
3734         (vreinterpret_u64_s16): Likewise.
3735         (vreinterpret_u64_s32): Likewise.
3736         (vreinterpret_u64_s64): Likewise.
3737         (vreinterpret_u64_f32): Likewise.
3738         (vreinterpret_u64_u8): Likewise.
3739         (vreinterpret_u64_u16): Likewise.
3740         (vreinterpret_u64_u32): Likewise.
3741         (vreinterpret_u64_p8): Likewise.
3742         (vreinterpret_u64_p16): Likewise.
3743         (vreinterpretq_u64_s8): Likewise.
3744         (vreinterpretq_u64_s16): Likewise.
3745         (vreinterpretq_u64_s32): Likewise.
3746         (vreinterpretq_u64_s64): Likewise.
3747         (vreinterpretq_u64_f32): Likewise.
3748         (vreinterpretq_u64_u8): Likewise.
3749         (vreinterpretq_u64_u16): Likewise.
3750         (vreinterpretq_u64_u32): Likewise.
3751         (vreinterpretq_u64_p8): Likewise.
3752         (vreinterpretq_u64_p16): Likewise.
3753         (vreinterpret_s8_s16): Likewise.
3754         (vreinterpret_s8_s32): Likewise.
3755         (vreinterpret_s8_s64): Likewise.
3756         (vreinterpret_s8_f32): Likewise.
3757         (vreinterpret_s8_u8): Likewise.
3758         (vreinterpret_s8_u16): Likewise.
3759         (vreinterpret_s8_u32): Likewise.
3760         (vreinterpret_s8_u64): Likewise.
3761         (vreinterpret_s8_p8): Likewise.
3762         (vreinterpret_s8_p16): Likewise.
3763         (vreinterpretq_s8_s16): Likewise.
3764         (vreinterpretq_s8_s32): Likewise.
3765         (vreinterpretq_s8_s64): Likewise.
3766         (vreinterpretq_s8_f32): Likewise.
3767         (vreinterpretq_s8_u8): Likewise.
3768         (vreinterpretq_s8_u16): Likewise.
3769         (vreinterpretq_s8_u32): Likewise.
3770         (vreinterpretq_s8_u64): Likewise.
3771         (vreinterpretq_s8_p8): Likewise.
3772         (vreinterpretq_s8_p16): Likewise.
3773         (vreinterpret_s16_s8): Likewise.
3774         (vreinterpret_s16_s32): Likewise.
3775         (vreinterpret_s16_s64): Likewise.
3776         (vreinterpret_s16_f32): Likewise.
3777         (vreinterpret_s16_u8): Likewise.
3778         (vreinterpret_s16_u16): Likewise.
3779         (vreinterpret_s16_u32): Likewise.
3780         (vreinterpret_s16_u64): Likewise.
3781         (vreinterpret_s16_p8): Likewise.
3782         (vreinterpret_s16_p16): Likewise.
3783         (vreinterpretq_s16_s8): Likewise.
3784         (vreinterpretq_s16_s32): Likewise.
3785         (vreinterpretq_s16_s64): Likewise.
3786         (vreinterpretq_s16_f32): Likewise.
3787         (vreinterpretq_s16_u8): Likewise.
3788         (vreinterpretq_s16_u16): Likewise.
3789         (vreinterpretq_s16_u32): Likewise.
3790         (vreinterpretq_s16_u64): Likewise.
3791         (vreinterpretq_s16_p8): Likewise.
3792         (vreinterpretq_s16_p16): Likewise.
3793         (vreinterpret_s32_s8): Likewise.
3794         (vreinterpret_s32_s16): Likewise.
3795         (vreinterpret_s32_s64): Likewise.
3796         (vreinterpret_s32_f32): Likewise.
3797         (vreinterpret_s32_u8): Likewise.
3798         (vreinterpret_s32_u16): Likewise.
3799         (vreinterpret_s32_u32): Likewise.
3800         (vreinterpret_s32_u64): Likewise.
3801         (vreinterpret_s32_p8): Likewise.
3802         (vreinterpret_s32_p16): Likewise.
3803         (vreinterpretq_s32_s8): Likewise.
3804         (vreinterpretq_s32_s16): Likewise.
3805         (vreinterpretq_s32_s64): Likewise.
3806         (vreinterpretq_s32_f32): Likewise.
3807         (vreinterpretq_s32_u8): Likewise.
3808         (vreinterpretq_s32_u16): Likewise.
3809         (vreinterpretq_s32_u32): Likewise.
3810         (vreinterpretq_s32_u64): Likewise.
3811         (vreinterpretq_s32_p8): Likewise.
3812         (vreinterpretq_s32_p16): Likewise.
3813         (vreinterpret_u8_s8): Likewise.
3814         (vreinterpret_u8_s16): Likewise.
3815         (vreinterpret_u8_s32): Likewise.
3816         (vreinterpret_u8_s64): Likewise.
3817         (vreinterpret_u8_f32): Likewise.
3818         (vreinterpret_u8_u16): Likewise.
3819         (vreinterpret_u8_u32): Likewise.
3820         (vreinterpret_u8_u64): Likewise.
3821         (vreinterpret_u8_p8): Likewise.
3822         (vreinterpret_u8_p16): Likewise.
3823         (vreinterpretq_u8_s8): Likewise.
3824         (vreinterpretq_u8_s16): Likewise.
3825         (vreinterpretq_u8_s32): Likewise.
3826         (vreinterpretq_u8_s64): Likewise.
3827         (vreinterpretq_u8_f32): Likewise.
3828         (vreinterpretq_u8_u16): Likewise.
3829         (vreinterpretq_u8_u32): Likewise.
3830         (vreinterpretq_u8_u64): Likewise.
3831         (vreinterpretq_u8_p8): Likewise.
3832         (vreinterpretq_u8_p16): Likewise.
3833         (vreinterpret_u16_s8): Likewise.
3834         (vreinterpret_u16_s16): Likewise.
3835         (vreinterpret_u16_s32): Likewise.
3836         (vreinterpret_u16_s64): Likewise.
3837         (vreinterpret_u16_f32): Likewise.
3838         (vreinterpret_u16_u8): Likewise.
3839         (vreinterpret_u16_u32): Likewise.
3840         (vreinterpret_u16_u64): Likewise.
3841         (vreinterpret_u16_p8): Likewise.
3842         (vreinterpret_u16_p16): Likewise.
3843         (vreinterpretq_u16_s8): Likewise.
3844         (vreinterpretq_u16_s16): Likewise.
3845         (vreinterpretq_u16_s32): Likewise.
3846         (vreinterpretq_u16_s64): Likewise.
3847         (vreinterpretq_u16_f32): Likewise.
3848         (vreinterpretq_u16_u8): Likewise.
3849         (vreinterpretq_u16_u32): Likewise.
3850         (vreinterpretq_u16_u64): Likewise.
3851         (vreinterpretq_u16_p8): Likewise.
3852         (vreinterpretq_u16_p16): Likewise.
3853         (vreinterpret_u32_s8): Likewise.
3854         (vreinterpret_u32_s16): Likewise.
3855         (vreinterpret_u32_s32): Likewise.
3856         (vreinterpret_u32_s64): Likewise.
3857         (vreinterpret_u32_f32): Likewise.
3858         (vreinterpret_u32_u8): Likewise.
3859         (vreinterpret_u32_u16): Likewise.
3860         (vreinterpret_u32_u64): Likewise.
3861         (vreinterpret_u32_p8): Likewise.
3862         (vreinterpret_u32_p16): Likewise.
3863         (vreinterpretq_u32_s8): Likewise.
3864         (vreinterpretq_u32_s16): Likewise.
3865         (vreinterpretq_u32_s32): Likewise.
3866         (vreinterpretq_u32_s64): Likewise.
3867         (vreinterpretq_u32_f32): Likewise.
3868         (vreinterpretq_u32_u8): Likewise.
3869         (vreinterpretq_u32_u16): Likewise.
3870         (vreinterpretq_u32_u64): Likewise.
3871         (vreinterpretq_u32_p8): Likewise.
3872         (vreinterpretq_u32_p16): Likewise.
3873
3874 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3875
3876         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
3877         Pattern extended.
3878         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
3879         (sqabs): Likewise.
3880         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
3881         (vqnegd_s64): Likewise.
3882         (vqabs_s64): Likewise.
3883         (vqabsd_s64): Likewise.
3884
3885 2014-04-22  Richard Henderson  <rth@redhat.com>
3886
3887         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
3888         computation to the top of the loop.
3889
3890 2014-04-22  Renlin  <renlin.li@arm.com>
3891             Jiong Wang  <jiong.wang@arm.com>
3892
3893         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
3894         * config/aarch64/aarch64.c (aarch64_layout_frame)
3895         (aarch64_initial_elimination_offset): Likewise.
3896
3897 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3898
3899         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
3900         Fix indentation.
3901
3902 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
3903
3904         * machmode.h (bitwise_mode_for_mode): Declare.
3905         * stor-layout.h (bitwise_type_for_mode): Likewise.
3906         * stor-layout.c (bitwise_mode_for_mode): New function.
3907         (bitwise_type_for_mode): Likewise.
3908         * builtins.c (fold_builtin_memory_op): Use it instead of
3909         int_mode_for_mode and build_nonstandard_integer_type.
3910
3911 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3912
3913         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
3914         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
3915         (*-*-solaris2*): Simplify.
3916         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
3917         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
3918         *-*-solaris2.9* handling.
3919
3920         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
3921         as bug.
3922         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
3923         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
3924         handling, simplify.
3925         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
3926         * configure: Regenerate.
3927
3928         * config/i386/sol2-9.h: Remove.
3929
3930         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
3931         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
3932         Remove Solaris 9 references.
3933
3934 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
3935
3936         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
3937         (floatuns<GPI:mode><GPF:mode>2): Remove.
3938         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
3939         and floatuns conversions.
3940         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
3941         and floatuns conversions.
3942         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
3943         (w1,w2): New mode attributes for inequal width conversions.
3944
3945 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
3946
3947         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
3948         the output asm format.
3949
3950 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
3951
3952         * config/aarch64/aarch64-simd.md
3953         (aarch64_cm<optab>di): Always split.
3954         (*aarch64_cm<optab>di): New.
3955         (aarch64_cmtstdi): Always split.
3956         (*aarch64_cmtstdi): New.
3957
3958 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3959
3960         PR tree-optimization/60823
3961         * omp-low.c (ipa_simd_modify_function_body): Go through
3962         all SSA_NAMEs and for those refering to vector arguments
3963         which are going to be replaced adjust SSA_NAME_VAR and,
3964         if it is a default definition, change it into a non-default
3965         definition assigned at the beginning of function from new_decl.
3966         (ipa_simd_modify_stmt_ops): Rewritten.
3967         * tree-dfa.c (set_ssa_default_def): When removing default def,
3968         check for NULL loc instead of NULL *loc.
3969
3970 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3971
3972         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
3973         restrictions on core registers for DImode values in Thumb2.
3974
3975 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3976
3977         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
3978         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
3979
3980 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3981
3982         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
3983         (*iordi_notzesidi_di): Likewise.
3984         (*iordi_notsesidi_di): Likewise.
3985
3986 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3987
3988         * config/arm/arm-protos.h (tune_params): New struct members.
3989         * config/arm/arm.c: Initialise tune_params per processor.
3990         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
3991         for speed, based on new tune_params.
3992
3993 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3994
3995         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
3996         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
3997         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
3998         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
3999         * config/aarch64/arm_neon.h (vrnd_f64): Added.
4000         (vrnda_f64): Likewise.
4001         (vrndi_f64): Likewise.
4002         (vrndm_f64): Likewise.
4003         (vrndn_f64): Likewise.
4004         (vrndp_f64): Likewise.
4005         (vrndx_f64): Likewise.
4006
4007 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4008
4009         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
4010         GET_MODE_SIZE argument is enum machine_mode.
4011
4012 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
4013
4014         PR target/60910
4015         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
4016         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
4017
4018 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
4019
4020         PR middle-end/60281
4021         * asan.c (asan_emit_stack_protection): Force the base to align to
4022         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
4023         appropriate bits if STRICT_ALIGNMENT.
4024         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
4025         when asan is on.
4026         (expand_used_vars): Leave a space in the stack frame for alignment
4027         if STRICT_ALIGNMENT.
4028
4029 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
4030
4031         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
4032         than a gimple.
4033         (gimple_store_p): Likewise.
4034         (gimple_assign_load_p): Likewise.
4035         (gimple_assign_cast_p): Likewise.
4036         (gimple_clobber_p): Likewise.
4037
4038         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
4039         rather than a gimple.
4040         (gimple_assign_cast_p): Likewise.
4041
4042 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
4043
4044         PR target/60735
4045         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
4046         If mode is DDmode and TARGET_E500_DOUBLE allow move.
4047
4048         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
4049         more debug information for E500 if -mdebug=reg.
4050
4051 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
4052
4053         PR target/60909
4054         * config/i386/i386.c (ix86_expand_builtin)
4055         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
4056         register for target RTX.
4057         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
4058
4059 2014-04-18  Cong Hou  <congh@google.com>
4060
4061         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
4062         the widen-mult pattern by handling two operands with different sizes,
4063         and operands whose size is smaller than half of the result type.
4064
4065 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
4066
4067         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
4068         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
4069         (do_estimate_edge_time): Compute it.
4070         * ipa-inline.c (want_inline_small_function_p): Bypass
4071         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
4072
4073 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
4074
4075         * ipa-inline.c (spec_rem): New static variable.
4076         (dump_overall_stats): New function.
4077         (dump_inline_stats): New function.
4078
4079 2014-04-18  Richard Henderson  <rth@redhat.com>
4080
4081         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
4082         to GET_MODE_SIZE, not a reg_class_t.
4083
4084 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4085
4086         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
4087         (vsx_xxmrglw_<mode>): Likewise.
4088
4089 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
4090
4091         PR target/60876
4092         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
4093         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
4094         (rs6000_init_hard_regno_mode_ok): Likewise.
4095
4096 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
4097
4098         * ipa-inline.c (inline_small_functions): Account only non-cold
4099         functions.
4100         * doc/invoke.texi (inline-unit-growth): Update documentation.
4101
4102 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
4103
4104         * config/rs6000/rs6000.md (addti3, subti3): New.
4105
4106 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
4107
4108         PR target/60863
4109         * config/i386/i386.c (ix86_expand_clear): Remove outdated
4110         comment.  Check optimize_insn_for_size_p instead of
4111         optimize_insn_for_speed_p.
4112
4113 2014-04-17  Martin Jambor  <mjambor@suse.cz>
4114
4115         * gimple-iterator.c (gsi_start_edge): New function.
4116         * gimple-iterator.h (gsi_start_edge): Declare.
4117         * tree-sra.c (single_non_eh_succ): New function.
4118         (disqualify_ops_if_throwing_stmt): Renamed to
4119         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
4120         having one non-EH successor BB.
4121         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
4122         generate loads into replacements.
4123         (sra_modify_assign): Likewise and and also use the simple path for
4124         such statements.
4125         (sra_modify_function_body): Commit statements on edges.
4126
4127 2014-04-17  Richard Biener  <rguenther@suse.de>
4128
4129         PR middle-end/60849
4130         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
4131         comparison results and add clarifying comment.
4132
4133 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
4134
4135         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
4136         (blank_mode): Initialize it.
4137         (emit_mode_size_inline, emit_mode_nunits_inline,
4138         emit_mode_inner_inline): New functions.
4139         (emit_insn_modes_h): Call them and surround their output with
4140         #if GCC_VERSION >= 4001 ... #endif.
4141         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
4142         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
4143         mode_* arrays if the argument is __builtin_constant_p.
4144         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
4145         is enum machine_mode.
4146
4147 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4148
4149         * passes.c (opt_pass::execute): Adjust.
4150         (pass_manager::execute_pass_mode_switching): Likewise.
4151         (early_local_passes::execute): Likewise.
4152         (execute_one_pass): Pass cfun to the pass's execute method.
4153         * tree-pass.h (opt_pass::execute): Add function * argument.
4154         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
4155         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
4156         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
4157         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4158         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
4159         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
4160         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
4161         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
4162         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
4163         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
4164         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
4165         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
4166         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
4167         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
4168         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
4169         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
4170         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
4171         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
4172         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
4173         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4174         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4175         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4176         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4177         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4178         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4179         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4180         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
4181         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
4182         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
4183         Adjust.
4184
4185 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4186
4187         * passes.c (opt_pass::gate): Take function * argument.
4188         (gate_all_early_local_passes): Merge into
4189         (early_local_passes::gate): this.
4190         (gate_all_early_optimizations): Merge into
4191         (all_early_optimizations::gate): this.
4192         (gate_all_optimizations): Mege into
4193         (all_optimizations::gate): this.
4194         (gate_all_optimizations_g): Merge into
4195         (all_optimizations_g::gate): this.
4196         (gate_rest_of_compilation): Mege into
4197         (rest_of_compilation::gate): this.
4198         (gate_postreload): Merge into
4199         (postreload::gate): this.
4200         (dump_one_pass): Pass cfun to the pass's gate method.
4201         (execute_ipa_summary_passes): Likewise.
4202         (execute_one_pass): Likewise.
4203         (ipa_write_summaries_2): Likewise.
4204         (ipa_write_optimization_summaries_1): Likewise.
4205         (ipa_read_summaries_1): Likewise.
4206         (ipa_read_optimization_summaries_1): Likewise.
4207         (execute_ipa_stmt_fixups): Likewise.
4208         * tree-pass.h (opt_pass::gate): Add function * argument.
4209         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
4210         combine-stack-adj.c, combine.c, compare-elim.c,
4211         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4212         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
4213         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
4214         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
4215         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
4216         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
4217         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
4218         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
4219         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
4220         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
4221         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
4222         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
4223         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
4224         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
4225         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4226         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4227         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4228         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4229         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4230         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4231         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4232         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4233         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
4234         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
4235         var-tracking.c, vtable-verify.c, web.c: Adjust.
4236
4237 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4238
4239         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
4240         * configure: Regenerate.
4241
4242 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4243
4244         * passes.c (dump_one_pass): don't check pass->has_gate.
4245         (execute_ipa_summary_passes): Likewise.
4246         (execute_one_pass): Likewise.
4247         (ipa_write_summaries_2): Likewise.
4248         (ipa_write_optimization_summaries_1): Likewise.
4249         (ipa_read_optimization_summaries_1): Likewise.
4250         (execute_ipa_stmt_fixups): Likewise.
4251         * tree-pass.h (pass_data::has_gate): Remove.
4252         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
4253         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
4254         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
4255         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4256         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
4257         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
4258         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
4259         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
4260         gimple-low.c, gimple-ssa-isolate-paths.c,
4261         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
4262         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
4263         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
4264         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
4265         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
4266         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
4267         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
4268         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
4269         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
4270         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
4271         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
4272         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4273         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4274         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4275         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4276         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4277         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4278         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4279         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4280         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
4281         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
4282         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
4283         Adjust.
4284
4285 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4286
4287         * pass_manager.h (pass_manager::register_dump_files_1): Remove
4288         declaration.
4289         * passes.c (pass_manager::register_dump_files_1): Merge into
4290         (pass_manager::register_dump_files): this, and remove its handling of
4291         properties since the pass always has the properties anyway.
4292         (pass_manager::pass_manager): Adjust.
4293
4294 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4295
4296         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
4297         * passes.c (pass_manager::register_dump_files_1): Remove dead code
4298         dealing with properties.
4299         (pass_manager::register_dump_files): Adjust.
4300
4301 2014-03-20  Mark Wielaard  <mjw@redhat.com>
4302
4303         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
4304         then represent the bound as normal constant value.
4305
4306 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
4307
4308         PR target/60847
4309         Forward port from 4.8 branch
4310         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
4311
4312         * config/i386/bmiintrin.h (_blsi_u32): New.
4313         (_blsi_u64): Ditto.
4314         (_blsr_u32): Ditto.
4315         (_blsr_u64): Ditto.
4316         (_blsmsk_u32): Ditto.
4317         (_blsmsk_u64): Ditto.
4318         (_tzcnt_u32): Ditto.
4319         (_tzcnt_u64): Ditto.
4320
4321 2014-04-17  Kito Cheng  <kito@0xlab.org>
4322
4323         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
4324
4325 2014-04-17  Richard Biener  <rguenther@suse.de>
4326
4327         PR middle-end/60849
4328         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
4329         boolean results for comparisons.
4330
4331 2014-04-17  Richard Biener  <rguenther@suse.de>
4332
4333         PR tree-optimization/60836
4334         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
4335         initial PHI args to be gimple values.
4336
4337 2014-04-17   Richard Biener  <rguenther@suse.de>
4338
4339         PR tree-optimization/60841
4340         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
4341         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
4342         of stmts to SLP build.
4343         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4344         (vect_analyze_slp): Likewise.
4345         (vect_analyze_slp_instance): Likewise.
4346         (vect_build_slp_tree): Limit overall SLP tree growth.
4347         * tree-vectorizer.h (vect_analyze_data_refs,
4348         vect_analyze_slp): Adjust prototypes.
4349
4350 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4351
4352         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
4353         Silvermont.
4354
4355 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4356
4357         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
4358         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
4359         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
4360         for TARGET_SLOW_PSHUFB
4361
4362 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4363
4364         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
4365         * config/i386/i386.c (intel_cost): Ditto.
4366
4367 2014-04-17  Joey Ye  <joey.ye@arm.com>
4368
4369         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
4370
4371 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4372
4373         * opts.c (common_handle_option): Disable -fipa-reference coorectly
4374         with -fuse-profile.
4375
4376 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4377
4378         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
4379         (type_all_derivations_known_p): New predicate.
4380         (type_all_ctors_visible_p): New predicate.
4381         (type_possibly_instantiated_p): New predicate.
4382         (get_odr_type): Compute all_derivations_known.
4383         (dump_odr_type): Dump the flag.
4384         (maybe_record_type): Cleanup.
4385         (record_target_from_binfo): Add bases_to_consider array;
4386         record bases for types w/o instances and skip CXX destructor.
4387         (possible_polymorphic_call_targets_1): Add bases_to_consider
4388         and consider_construction parameters; check if type may have instance.
4389         (get_polymorphic_call_info): Set maybe_in_construction to true
4390         when we know nothing.
4391         (record_targets_from_bases): Skip CXX destructors; they are
4392         never called for types in construction.
4393         (possible_polymorphic_call_targets): Do not record target when
4394         type may not have instance.
4395
4396 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4397
4398         PR ipa/60854
4399         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
4400         external aliases alive, too.
4401
4402 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
4403
4404         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
4405         definition.
4406
4407 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
4408
4409         * final.c (compute_alignments): Do not apply loop alignment to a block
4410         falling through to the exit.
4411
4412 2014-04-16  Catherine Moore  <clm@codesourcery.com>
4413
4414         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
4415         Adjust constraints for microMIPS store patterns.
4416
4417 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
4418
4419         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
4420
4421 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
4422
4423         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
4424         (append_use): Run at -O0.
4425         (append_vdef): Likewise.
4426         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
4427         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
4428
4429 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
4430
4431         PR tree-optimization/60844
4432         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
4433         (propagate_op_to_single_use, remove_visited_stmt_chain,
4434         linearize_expr, repropagate_negates, reassociate_bb): Use it
4435         instead of gsi_remove.
4436
4437 2014-04-16  Martin Jambor  <mjambor@suse.cz>
4438
4439         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
4440         ipa_transforms_to_apply.
4441         (cgraph_function_versioning): Assert that old_node has empty
4442         ipa_transforms_to_apply.
4443         * trans-mem.c (ipa_tm_create_version): Likewise.
4444         * tree-inline.c (tree_function_versioning): Do not duplicate
4445         ipa_transforms_to_apply.
4446
4447 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4448
4449         PR target/60817
4450         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
4451         x86_64-*-* cases.
4452         Pass necessary as flags on 64-bit Solaris/x86.
4453         Use lowercase relocs for x86_64-*-*.
4454         * configure: Regenerate.
4455
4456 2014-04-15  Jan Hubicka  <jh@suse.cz>
4457
4458         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
4459         (maybe_record_node, likely_target_p): Use it.
4460
4461 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4462
4463         PR target/60839
4464         Revert following patch
4465
4466         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4467
4468         PR target/60735
4469         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4470         software floating point or no floating point registers, do not
4471         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4472         in GPRs that occurs after we tested for GPRs that would never be
4473         true.
4474
4475         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4476         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4477         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4478         specifically allow DDmode, since that does not use the SPE SIMD
4479         instructions.
4480
4481 2014-03-21  Mark Wielaard  <mjw@redhat.com>
4482
4483         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
4484         as unsigned or int depending on type and value used.
4485
4486 2014-04-15  Richard Biener  <rguenther@suse.de>
4487
4488         PR rtl-optimization/56965
4489         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
4490         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
4491         ... here.
4492         * alias.c (true_dependence_1): Do not call
4493         nonoverlapping_component_refs_p.
4494         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
4495         nonoverlapping_component_refs_p.
4496         (indirect_refs_may_alias_p): Likewise.
4497
4498 2014-04-15  Teresa Johnson  <tejohnson@google.com>
4499
4500         * cfg.c (dump_bb_info): Fix flags check.
4501         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
4502
4503 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4504
4505         PR rtl-optimization/60663
4506         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
4507         avoid 0 cost.
4508
4509 2014-04-15  Richard Biener  <rguenther@suse.de>
4510
4511         * lto-streamer.h (LTO_major_version): Bump to 4.
4512
4513 2014-04-15  Richard Biener  <rguenther@suse.de>
4514
4515         * common.opt (lto_partition_model): New enum.
4516         (flto-partition=): Merge separate options with a single with argument,
4517         add -flto-partition=one support.
4518         * flag-types.h (enum lto_partition_model): Declare.
4519         * opts.c (finish_options): Remove duplicate -flto-partition=
4520         option check.
4521         * lto-wrapper.c (run_gcc): Adjust.
4522
4523 2014-04-15  Richard Biener  <rguenther@suse.de>
4524
4525         * alias.c (ncr_compar): New function.
4526         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
4527
4528 2014-04-15  Richard Biener  <rguenther@suse.de>
4529
4530         * alias.c (record_component_aliases): Do not walk BINFOs.
4531
4532 2014-04-15  Richard Biener  <rguenther@suse.de>
4533
4534         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4535         Add struct function argument and adjust.
4536         (find_func_aliases_for_call): Likewise.
4537         (find_func_aliases): Likewise.
4538         (find_func_clobbers): Likewise.
4539         (intra_create_variable_infos): Likewise.
4540         (compute_points_to_sets): Likewise.
4541         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
4542
4543 2014-04-15  Richard Biener  <rguenther@suse.de>
4544
4545         * tree.c (iterative_hash_expr): Use enum tree_code_class
4546         to store TREE_CODE_CLASS.
4547         (tree_block): Likewise.
4548         (tree_set_block): Likewise.
4549         * tree.h (fold_build_pointer_plus_loc): Use
4550         convert_to_ptrofftype_loc.
4551
4552 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
4553
4554         PR plugins/59335
4555         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
4556         added in 4.9.
4557
4558 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
4559
4560         * cfgloop.h (struct loop): Move force_vectorize down.
4561         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
4562         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
4563         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
4564         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
4565         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
4566         * tree-core.h (enum annot_expr_kind): Add new kind values.
4567         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
4568         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
4569         kinds.
4570         * tree.def (ANNOTATE_EXPR): Tweak comment.
4571
4572 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4573
4574         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
4575         cxa_pure_virtual).
4576
4577 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
4578
4579         * tree.h (TYPE_IDENTIFIER): Declare.
4580         * tree.c (subrange_type_for_debug_p): Use it.
4581         * godump.c (go_format_type): Likewise.
4582         * dwarf2out.c (is_cxx_auto, modified_type_die,
4583         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
4584         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
4585
4586 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
4587
4588         PR lto/60820
4589         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
4590
4591 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
4592
4593         * config/i386/i386.c (examine_argument): Return bool.  Return true if
4594         parameter should be passed in memory.
4595         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
4596         (construct_container): Update calls to examine_argument.
4597         (function_arg_advance_64): Ditto.
4598         (return_in_memory_32): Merge with ix86_return_in_memory.
4599         (return_in_memory_64): Ditto.
4600         (return_in_memory_ms_64): Ditto.
4601
4602 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4603
4604         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
4605         * coverage.c (coverage_compute_profile_id): Handle externally visible
4606         symbols.
4607
4608 2014-04-14  Martin Jambor  <mjambor@suse.cz>
4609
4610         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
4611         DECL_DISREGARD_INLINE_LIMITS functions.
4612
4613 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4614
4615         PR target/60827
4616         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
4617
4618 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4619
4620         PR target/60827
4621         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
4622         optimize_insn_for_speed_p instead of
4623         optimize_function_for_speed_p.
4624
4625 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
4626
4627         * doc/invoke.texi (free): Document AArch64.
4628
4629 2014-04-14  Richard Biener  <rguenther@suse.de>
4630
4631         PR tree-optimization/60042
4632         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
4633         (insert_into_preds_of_block): Do not prevent PHI insertion
4634         for REFERENCE exprs here ...
4635         (eliminate_dom_walker::before_dom_children): ... but prevent
4636         their use here under similar conditions when applied to the
4637         IL after PRE optimizations.
4638
4639 2014-04-14  Richard Biener  <rguenther@suse.de>
4640
4641         * passes.def: Move early points-to after early SRA.
4642
4643 2014-04-14  Richard Biener  <rguenther@suse.de>
4644
4645         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
4646         check for which sign-changes we allow when forwarding
4647         a converted value into a switch.
4648
4649 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4650
4651         * stor-layout.c (place_field): Finalize non-constant offset for the
4652         field, if any.
4653
4654 2014-04-14  Richard Biener  <rguenther@suse.de>
4655
4656         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
4657         as argument.
4658         (expand_switch_using_bit_tests_p): Likewise.
4659         (process_switch): Compute and pass on speed_p based on the
4660         switch stmt.
4661         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
4662         optimize_bb_for_speed_p.
4663
4664 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4665
4666         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
4667         * function.h (struct function): Rename has_force_vect_loops into
4668         has_force_vectorize_loops.
4669         * lto-streamer-in.c (input_cfg): Adjust for renaming.
4670         (input_struct_function_base): Likewise.
4671         * lto-streamer-out.c (output_cfg): Likewise.
4672         (output_struct_function_base): Likewise.
4673         * omp-low.c (expand_omp_simd): Likewise.
4674         * tree-cfg.c (move_sese_region_to_fn): Likewise.
4675         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4676         (version_loop_for_if_conversion): Likewise.
4677         (tree_if_conversion): Likewise.
4678         (main_tree_if_conversion): Likewise.
4679         (gate_tree_if_conversion): Likewise.
4680         * tree-inline.c (copy_loops): Likewise.
4681         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
4682         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
4683         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
4684         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
4685         * tree-vectorizer.c (vectorize_loops): Likewise.
4686         * tree-vectorizer.h (unlimited_cost_model): Likewise.
4687
4688 2014-04-14  Richard Biener  <rguenther@suse.de>
4689
4690         PR lto/60720
4691         * lto-streamer-out.c (wrap_refs): New function.
4692         (lto_output): Wrap symbol references in global initializes in
4693         type-preserving MEM_REFs.
4694
4695 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4696
4697         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
4698
4699 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4700
4701         * config/sh/sh.md (setmemqi): New expand pattern.
4702         * config/sh/sh.h (CLEAR_RATIO): Define.
4703         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
4704         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
4705
4706 2014-04-14  Richard Biener  <rguenther@suse.de>
4707
4708         PR middle-end/55022
4709         * fold-const.c (negate_expr_p): Don't negate directional rounding
4710         division.
4711         (fold_negate_expr): Likewise.
4712
4713 2014-04-14  Richard Biener  <rguenther@suse.de>
4714
4715         PR tree-optimization/59817
4716         PR tree-optimization/60453
4717         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
4718         recursion to catch all CHRECs in the scalar evolution and restrict
4719         the predicate for the remains appropriately.
4720
4721 2014-04-12  Catherine Moore  <clm@codesourcery.com>
4722
4723         * config/mips/constraints.md: Add new register constraint "kb".
4724         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
4725         (*movhi_internal): Likewise.
4726         (*movqi_internal): Likewise.
4727         * config/mips/mips.h (M16_STORE_REGS): New register class.
4728         (REG_CLASS_NAMES): Add M16_STORE_REGS.
4729         (REG_CLASS_CONTENTS): Likewise.
4730         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
4731
4732 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4733
4734         PR c/60194
4735         * doc/invoke.texi (-Wformat-signedness): Document it.
4736         (Wformat=2): Mention that this enables -Wformat-signedness.
4737
4738 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4739
4740         * common/config/epiphany/epiphany-common.c
4741         (epiphany_option_optimization_table): Enable section anchors by
4742         default at -O1 or higher.
4743         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
4744         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
4745         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
4746         carries no extra cost.
4747         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
4748         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
4749         * config/epiphany/predicates.md (memclob_operand): New predicate.
4750         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
4751         Use memclob_operand predicate and X constraint for operand 3.
4752
4753 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4754
4755         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
4756         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
4757         its operands.
4758
4759 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4760
4761         PR rtl-optimization/60651
4762         * mode-switching.c (optimize_mode_switching): Make sure to emit
4763         sets of a lower numbered entity before sets of a higher numbered
4764         entity to a mode of the same or lower priority.
4765         When creating a seginfo for a basic block that starts with a code
4766         label, move the insertion point past the code label.
4767         (new_seginfo): Document and enforce requirement that
4768         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
4769         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
4770         * doc/tm.texi: Regenerate.
4771
4772 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4773
4774         PR target/60811
4775         * config/arc/arc.c (arc_save_restore): Fix assert typo.
4776
4777 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
4778
4779         * BASE-VER: Set to 4.10.0.
4780
4781 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4782
4783         PR other/59055
4784         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
4785         * doc/gcc.texi (Service): Update description in the @menu
4786         * doc/invoke.texi (Option Summary): Remove misplaced and
4787         duplicated @menu.
4788
4789 2014-04-11  Steve Ellcey  <sellcey@mips.com>
4790             Jakub Jelinek  <jakub@redhat.com>
4791
4792         PR middle-end/60556
4793         * expr.c (convert_move): Use emit_store_flag_force instead of
4794         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
4795         argument to it.
4796
4797 2014-04-11  Richard Biener  <rguenther@suse.de>
4798
4799         PR middle-end/60797
4800         * varasm.c (assemble_alias): Avoid endless error reporting
4801         recursion by setting TREE_ASM_WRITTEN.
4802
4803 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4804
4805         * config/s390/s390.md: Add a splitter for NOT rtx.
4806
4807 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
4808
4809         PR rtl-optimization/60663
4810         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
4811
4812 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
4813             Jakub Jelinek  <jakub@redhat.com>
4814
4815         PR lto/60567
4816         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
4817         flag from decl_node to node.
4818
4819 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4820
4821         PR debug/60655
4822         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
4823         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
4824         ameliorating the cases where it can be.
4825
4826 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
4827
4828         Revert
4829         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4830
4831         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4832         (loadsync_<mode>): Change mode.
4833         (load_quadpti, store_quadpti): New.
4834         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4835         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4836         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
4837
4838 2014-04-09  Cong Hou  <congh@google.com>
4839
4840         PR testsuite/60773
4841         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
4842         documentation.
4843
4844 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4845
4846         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
4847         instead of vnor to exploit possible fusion opportunity in the
4848         future.
4849         (altivec_expand_vec_perm_const_le): Likewise.
4850
4851 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4852
4853         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4854         (loadsync_<mode>): Change mode.
4855         (load_quadpti, store_quadpti): New.
4856         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4857         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4858
4859 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
4860
4861         PR target/60763
4862         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
4863         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
4864         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
4865
4866 2014-04-08  Richard Biener  <rguenther@suse.de>
4867
4868         PR middle-end/60706
4869         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
4870         a 64bit widest int print double-int similar to on HWI64 hosts.
4871
4872 2014-04-08  Richard Biener  <rguenther@suse.de>
4873
4874         PR tree-optimization/60785
4875         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
4876         default defs properly.
4877
4878 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
4879
4880         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
4881         (Weffc++): Likewise.
4882
4883 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
4884
4885         * ipa-devirt.c (maybe_record_node): When node is not recorded,
4886         set completep to false rather than true.
4887
4888 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
4889
4890         PR target/60504
4891         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
4892         ARM_TARGET2_DWARF_FORMAT.
4893
4894 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
4895
4896         PR target/60609
4897         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
4898         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
4899         ADDR_DIFF_VEC.
4900
4901 2014-04-07  Richard Biener  <rguenther@suse.de>
4902
4903         PR tree-optimization/60766
4904         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
4905         (may_eliminate_iv): Convert cand_value_at result to desired type.
4906
4907 2014-04-07  Jason Merrill  <jason@redhat.com>
4908
4909         PR c++/60731
4910         * common.opt (-fno-gnu-unique): Add.
4911         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
4912
4913 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4914
4915         * haifa-sched.c: Fix outdated function reference and minor
4916         grammar errors in introductory comment.
4917
4918 2014-04-07  Richard Biener  <rguenther@suse.de>
4919
4920         PR middle-end/60750
4921         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
4922         for noreturn calls.
4923         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
4924
4925 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
4926
4927         PR debug/55794
4928         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
4929         size accounting for thunks.
4930         (pa_asm_output_mi_thunk): Use final_start_function() and
4931         final_end_function() to output function start and end directives.
4932
4933 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
4934
4935         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
4936         device specific ISA/ feature information. Remove short_sp and
4937         errata_skip ds.  Add avr_device_specific_features enum to have device
4938         specific info.
4939         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
4940         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
4941         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
4942         updated device specific info.
4943         * config/avr/avr-mcus.def: Merge device specific details to
4944         dev_attribute field.
4945         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
4946         errata_skip.
4947         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
4948         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
4949         assembler if RMW isa supported by current device.
4950         * config/avr/genmultilib.awk: Update as device info structure changed.
4951         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
4952
4953 2014-04-04  Cong Hou  <congh@google.com>
4954
4955         PR tree-optimization/60656
4956         * tree-vect-stmts.c (supportable_widening_operation):
4957         Fix a bug that elements in a vector with vect_used_by_reduction
4958         property are incorrectly reordered when the operation on it is not
4959         consistant with the one in reduction operation.
4960
4961 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
4962
4963         PR rtl-optimization/60155
4964         * gcse.c (record_set_data): New function.
4965         (single_set_gcse): New function.
4966         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
4967         (hoist_code): Likewise.
4968         (get_pressure_class_and_nregs): Likewise.
4969
4970 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
4971
4972         * explow.c (probe_stack_range): Emit a final optimization blockage.
4973
4974 2014-04-04  Anthony Green  <green@moxielogic.com>
4975
4976         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
4977         typos.
4978
4979 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
4980
4981         PR ipa/59626
4982         * lto-cgraph.c (input_overwrite_node): Check that partitioning
4983         flags are set only during streaming.
4984         * ipa.c (process_references, walk_polymorphic_call_targets,
4985         symtab_remove_unreachable_nodes): Drop bodies of always inline
4986         after early inlining.
4987         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
4988
4989 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
4990         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4991
4992         PR debug/60655
4993         * dwarf2out.c (const_ok_for_output_1): Reject expressions
4994         containing a NOT.
4995
4996 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4997
4998         PR bootstrap/60743
4999         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
5000         duration.
5001         (cortex_a53_fdivd): Likewise.
5002
5003 2014-04-04  Martin Jambor  <mjambor@suse.cz>
5004
5005         PR ipa/60640
5006         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
5007         Adjust all callers.
5008         * cgraph.c (clone_of_p): Also return true if thunks match.
5009         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
5010         cgraph_function_or_thunk_node and an obsolete comment.
5011         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
5012         file.
5013         (build_function_decl_skip_args): Likewise.
5014         (set_new_clone_decl_and_node_flags): New function.
5015         (duplicate_thunk_for_node): Likewise.
5016         (redirect_edge_duplicating_thunks): Likewise.
5017         (cgraph_clone_node): New parameter args_to_skip, pass it to
5018         redirect_edge_duplicating_thunks which is called instead of
5019         cgraph_redirect_edge_callee.
5020         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
5021         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
5022
5023 2014-04-04  Jeff Law  <law@redhat.com>
5024
5025         PR target/60657
5026         * config/arm/predicates.md (const_int_I_operand): New predicate.
5027         (const_int_M_operand): Similarly.
5028         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
5029         const_int_operand.
5030         (insv_t2, extv_reg, extzv_t2): Likewise.
5031         (load_multiple_with_writeback): Similarly for const_int_I_operand.
5032         (pop_multiple_with_writeback_and_return): Likewise.
5033         (vfp_pop_multiple_with_writeback): Likewise
5034
5035 2014-04-04  Richard Biener  <rguenther@suse.de>
5036
5037         PR ipa/60746
5038         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
5039         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
5040         non-GIMPLE_LABELs.
5041         * gimplify.h (gimple_add_tmp_var_fn): Declare.
5042         * gimplify.c (gimple_add_tmp_var_fn): New function.
5043         * gimple-expr.h (create_tmp_reg_fn): Declare.
5044         * gimple-expr.c (create_tmp_reg_fn): New function.
5045         * gimple-low.c (record_vars_into): Don't change cfun.
5046         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
5047         code generation without cfun.
5048
5049 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
5050
5051         PR bootstrap/60719
5052         * Makefile.in (install-driver): Fix shell scripting.
5053
5054 2014-04-03  Cong Hou  <congh@google.com>
5055
5056         PR tree-optimization/60505
5057         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
5058         threshold of number of iterations below which no vectorization
5059         will be done.
5060         * tree-vect-loop.c (new_loop_vec_info):
5061         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
5062         * tree-vect-loop.c (vect_analyze_loop_operations):
5063         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
5064         * tree-vect-loop.c (vect_transform_loop):
5065         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
5066         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
5067         of iterations of the loop and see if we should build the epilogue.
5068
5069 2014-04-03  Richard Biener  <rguenther@suse.de>
5070
5071         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
5072         (streamer_tree_cache_create): Adjust.
5073         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
5074         to allow optional nodes array.
5075         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
5076         (streamer_tree_cache_append): Likewise.
5077         (streamer_tree_cache_create): Create nodes array optionally
5078         as specified by parameter.
5079         * lto-streamer-out.c (create_output_block): Avoid maintaining
5080         the node array in the writer cache.
5081         (DFS_write_tree): Remove assertion.
5082         (produce_asm_for_decls): Free the out decl state hash table early.
5083         * lto-streamer-in.c (lto_data_in_create): Adjust for
5084         streamer_tree_cache_create prototype change.
5085
5086 2014-04-03  Richard Biener  <rguenther@suse.de>
5087
5088         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
5089         set TREE_CHAIN to NULL_TREE.
5090
5091 2014-04-03  Richard Biener  <rguenther@suse.de>
5092
5093         PR tree-optimization/60740
5094         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
5095         over all GIMPLE_COND operands.
5096
5097 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
5098
5099         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
5100         (Weffc++): Remove Scott's numbering, merge lists and reference
5101         Wnon-virtual-dtor.
5102
5103 2014-04-03  Nick Clifton  <nickc@redhat.com>
5104
5105         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
5106         properly.
5107
5108 2014-04-03  Martin Jambor  <mjambor@suse.cz>
5109
5110         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
5111         mention gcc_unreachable before failing.
5112         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
5113         removed symbols.
5114
5115 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
5116
5117         PR ipa/60659
5118         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
5119         inconsistent code and instead mark the context inconsistent.
5120         (possible_polymorphic_call_targets): For inconsistent contexts
5121         return empty complete list.
5122
5123 2014-04-02  Anthony Green  <green@moxielogic.com>
5124
5125         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
5126         (extendqisi2, extendhisi2): Define.
5127         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
5128         (WCHAR_TYPE): Change to unsigned int.
5129
5130 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5131
5132         PR tree-optimization/60733
5133         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
5134         insertion point for PHI candidates to be the end of the feeding
5135         block for the PHI argument.
5136
5137 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
5138
5139         PR rtl-optimization/60650
5140         * lra-constraints.c (process_alt_operands): Decrease reject for
5141         earlyclobber matching.
5142
5143 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5144
5145         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
5146
5147 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5148
5149         * config/spu/spu.c (pad_bb): Do not crash when the last
5150         insn is CODE_FOR_blockage.
5151
5152 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5153
5154         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
5155         lies outside the target mode.
5156
5157 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5158
5159         PR target/60735
5160         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
5161         software floating point or no floating point registers, do not
5162         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
5163         in GPRs that occurs after we tested for GPRs that would never be
5164         true.
5165
5166         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
5167         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
5168         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
5169         specifically allow DDmode, since that does not use the SPE SIMD
5170         instructions.
5171
5172 2014-04-02  Richard Biener  <rguenther@suse.de>
5173
5174         PR middle-end/60729
5175         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
5176         MODE_INTs.  Properly use negv_optab.
5177         (expand_abs): Likewise.
5178
5179 2014-04-02  Richard Biener  <rguenther@suse.de>
5180
5181         PR bootstrap/60719
5182         * Makefile.in (install-driver): Guard extra installs with special
5183         names properly.
5184
5185 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
5186
5187         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5188         Document vec_vgbbd.
5189
5190 2014-04-01  Richard Henderson  <rth@redhat.com>
5191
5192         PR target/60704
5193         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
5194         alternative enabled before register allocation.
5195
5196 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
5197
5198         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
5199         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
5200         typo.
5201         (nios2_large_got_address): Remove unneeded 'sym' parameter.
5202         (nios2_got_address): Update nios2_large_got_address call site.
5203         (nios2_delegitimize_address): New function.
5204         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
5205         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
5206         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
5207
5208 2014-04-01  Martin Husemann  <martin@duskware.de>
5209
5210         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
5211         for -mabi=32.
5212
5213 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
5214
5215         PR rtl-optimization/60604
5216         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
5217         check from register_operand.
5218         (register_operand): Redefine in terms of general_operand.
5219         (nonmemory_operand): Use register_operand for the non-constant cases.
5220
5221 2014-04-01  Richard Biener  <rguenther@suse.de>
5222
5223         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
5224
5225 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5226
5227         * doc/invoke.texi (mapp-regs): Clarify.
5228
5229 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
5230
5231         * config/i386/avx512fintrin.h (__v32hi): Define type.
5232         (__v64qi): Likewise.
5233         (_mm512_set1_epi8): Define.
5234         (_mm512_set1_epi16): Define.
5235         (_mm512_set4_epi32): Define.
5236         (_mm512_set4_epi64): Define.
5237         (_mm512_set4_pd): Define.
5238         (_mm512_set4_ps): Define.
5239         (_mm512_setr4_epi64): Define.
5240         (_mm512_setr4_epi32): Define.
5241         (_mm512_setr4_pd): Define.
5242         (_mm512_setr4_ps): Define.
5243         (_mm512_setzero_epi32): Define.
5244
5245 2014-03-31  Martin Jambor  <mjambor@suse.cz>
5246
5247         PR middle-end/60647
5248         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
5249         callsite_arguments_match_p.  Updated all callers.  Also check types of
5250         corresponding formal parameters and actual arguments.
5251         (not_all_callers_have_enough_arguments_p) Renamed to
5252         some_callers_have_mismatched_arguments_p.
5253
5254 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
5255
5256         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
5257
5258 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
5259
5260         PR target/60034
5261         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
5262         section anchor.
5263
5264 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
5265
5266         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
5267         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
5268         Split out
5269         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
5270         Use FMAMODE_NOVF512 mode iterator.
5271         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
5272         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
5273         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
5274         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
5275         Split out
5276         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
5277         Use VF_128_256 mode iterator.
5278         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
5279         Ditto.
5280
5281 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5282
5283         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
5284         static chain if needed.
5285
5286 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
5287
5288         PR target/60697
5289         * lra-constraints.c (index_part_to_reg): New.
5290         (process_address): Use it.
5291
5292 2014-03-27  Jeff Law  <law@redhat.com>
5293             Jakub Jelinek  <jakub@redhat.com>
5294
5295         PR target/60648
5296         * expr.c (do_tablejump): Use simplify_gen_binary rather than
5297         gen_rtx_{PLUS,MULT} to build up the address expression.
5298
5299         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
5300         creating non-canonical RTL.
5301
5302 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5303
5304         PR ipa/60243
5305         * ipa-inline.c (want_inline_small_function_p): Short circuit large
5306         functions; reorganize to make cheap checks first.
5307         (inline_small_functions): Do not estimate growth when dumping;
5308         it is expensive.
5309         * ipa-inline.h (inline_summary): Add min_size.
5310         (growth_likely_positive): New function.
5311         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
5312         (set_cond_stmt_execution_predicate): Cleanup.
5313         (estimate_edge_size_and_time): Compute min_size.
5314         (estimate_calls_size_and_time): Likewise.
5315         (estimate_node_size_and_time): Likewise.
5316         (inline_update_overall_summary): Update min_size.
5317         (do_estimate_edge_time): Likewise.
5318         (do_estimate_edge_size): Update.
5319         (do_estimate_edge_hints): Update.
5320         (growth_likely_positive): New function.
5321
5322 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
5323
5324         PR target/60693
5325         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
5326         also if addr has VOIDmode.
5327
5328 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5329
5330         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
5331         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
5332         Declare extern.
5333         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
5334         instructions as well as AdvancedSIMD loads.
5335
5336 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5337
5338         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
5339         Use crypto_aese type.
5340         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
5341         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
5342         crypto_aese, crypto_aesmc.  Move to types.md.
5343         * config/arm/types.md (crypto_aes): Split into crypto_aese,
5344         crypto_aesmc.
5345         * config/arm/iterators.md (crypto_type): Likewise.
5346
5347 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5348
5349         * cgraph.c: Include expr.h and tree-dfa.h.
5350         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
5351         remove LHS.
5352
5353 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
5354
5355         PR target/60675
5356         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
5357         regs from checking multi-reg pseudos.
5358
5359 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5360
5361         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
5362
5363 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5364
5365         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
5366         if it would clobber the stack pointer, even temporarily.
5367
5368 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
5369
5370         * mode-switching.c: Make small adjustments to the top comment.
5371
5372 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
5373
5374         * config/rs6000/constraints.md (wD constraint): New constraint to
5375         match the constant integer to get the top DImode/DFmode out of a
5376         vector in a VSX register.
5377
5378         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
5379         match the constant integer to get the top DImode/DFmode out of a
5380         vector in a VSX register.
5381
5382         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
5383         for ISA 2.07.
5384
5385         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5386         vbpermq builtins.
5387
5388         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
5389         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
5390
5391         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
5392         Optimize vec_extract of 64-bit values, where the value being
5393         extracted is in the top word, where we can use scalar
5394         instructions.  Add direct move and store support.  Combine the big
5395         endian/little endian vector select load support into a single insn.
5396         (vsx_extract_<mode>_internal1): Likewise.
5397         (vsx_extract_<mode>_internal2): Likewise.
5398         (vsx_extract_<mode>_load): Likewise.
5399         (vsx_extract_<mode>_store): Likewise.
5400         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
5401         combined into vsx_extract_<mode>_load.
5402         (vsx_extract_<mode>_one_le): Likewise.
5403
5404         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
5405         define the top 64-bit vector element.
5406
5407         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
5408         constraint.
5409
5410         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5411         Document vec_vbpermq builtin.
5412
5413         PR target/60672
5414         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
5415         enable use of xxsldwi and xxpermdi builtin functions.
5416         (vec_xxpermdi): Likewise.
5417
5418         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5419         Document use of vec_xxsldwi and vec_xxpermdi builtins.
5420
5421 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
5422
5423         PR rtl-optimization/60650
5424         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
5425         first_p.  Use it.
5426         (find_spills_for): New.
5427         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
5428         Spill all pseudos on the second iteration.
5429
5430 2014-03-27  Marek Polacek  <polacek@redhat.com>
5431
5432         PR c/50347
5433         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
5434         types.
5435
5436 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5437
5438         * config/s390/s390.c (s390_can_use_return_insn): Check for
5439         call-saved FPRs on 31 bit.
5440
5441 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
5442
5443         PR middle-end/60682
5444         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
5445         if they need regimplification, just drop them instead of
5446         calling gimple_regimplify_operands on them.
5447
5448 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5449
5450         PR target/60580
5451         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
5452         (aarch64_frame_pointer_required): Adjust logic.
5453         (aarch64_can_eliminate): Adjust logic.
5454         (aarch64_override_options_after_change): Adjust logic.
5455
5456 2014-03-27  Dehao Chen  <dehao@google.com>
5457
5458         * ipa-inline.c (early_inliner): Update node's inline info.
5459
5460 2014-03-26  Dehao Chen  <dehao@google.com>
5461
5462         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
5463         compiler inserted conditional jumps for NAN float check.
5464
5465 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5466
5467         * ubsan.h (ubsan_create_data): Change second argument's type
5468         to const location_t *.
5469         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
5470         _("<unknown>").
5471         (ubsan_create_data): Change second argument to const location_t *PLOC.
5472         Create Loc field whenever PLOC is non-NULL.
5473         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
5474         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
5475         callers.
5476
5477         PR other/59545
5478         * real.c (real_to_integer2): Change type of low to UHWI.
5479
5480 2014-03-26  Tobias Burnus  <burnus@net-b.de>
5481
5482         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
5483         (CILK_SELF_SPECS): New define.
5484         (driver_self_specs): Use it.
5485
5486 2014-03-26  Richard Biener  <rguenther@suse.de>
5487
5488         * tree-pretty-print.c (percent_K_format): Implement special
5489         case for LTO and its stripped down BLOCK tree.
5490
5491 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5492
5493         PR sanitizer/60636
5494         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
5495
5496         * tree-vrp.c (simplify_internal_call_using_ranges): If only
5497         one range is range_int_cst_p, but not both, at least optimize
5498         addition/subtraction of 0 and multiplication by 0 or 1.
5499         * gimple-fold.c (gimple_fold_call): Fold
5500         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
5501         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
5502         INTEGER_CSTs, try to fold at least x * 0 and y - y.
5503
5504 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
5505
5506         PR rtl-optimization/60452
5507         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
5508         <case REG>: Return 1 for invalid offsets from the frame pointer.
5509
5510 2014-03-26  Marek Polacek  <polacek@redhat.com>
5511
5512         PR c/37428
5513         * doc/extend.texi (C Extensions): Mention variable-length arrays in
5514         a structure/union.
5515
5516 2014-03-26  Marek Polacek  <polacek@redhat.com>
5517
5518         PR c/39525
5519         * doc/extend.texi (Designated Inits): Describe what happens to omitted
5520         field members.
5521
5522 2014-03-26  Marek Polacek  <polacek@redhat.com>
5523
5524         PR other/59545
5525         * ira-color.c (update_conflict_hard_regno_costs): Perform the
5526         multiplication in unsigned type.
5527
5528 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5529
5530         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
5531
5532 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5533
5534         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
5535
5536 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5537
5538         PR ipa/60315
5539         * cif-code.def (UNREACHABLE) New code.
5540         * ipa-inline.c (inline_small_functions): Skip edges to
5541         __builtlin_unreachable.
5542         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
5543         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
5544         predicate to __bulitin_unreachable.
5545         (set_cond_stmt_execution_predicate): Fix issue when
5546         invert_tree_comparison returns ERROR_MARK.
5547         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
5548         propagate to inline clones.
5549         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
5550         to unreachable.
5551         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
5552         * cgraphclones.c (cgraph_clone_node): If call destination is already
5553         ureachable, do not redirect it back.
5554         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
5555         unreachable.
5556
5557 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5558
5559         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
5560         Do not modify inline clones.
5561
5562 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5563
5564         * config/i386/i386.md (general_sext_operand): New mode attr.
5565         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
5566         don't generate (sign_extend (const_int)).
5567         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
5568         operands[2].  Use We constraint instead of <i> and
5569         <general_sext_operand> predicate instead of <general_operand>.
5570         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
5571         * config/i386/constraints.md (We): New constraint.
5572         * config/i386/predicates.md (x86_64_sext_operand,
5573         sext_operand): New predicates.
5574
5575 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5576
5577         PR ipa/60600
5578         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
5579         inconsistent devirtualizations to __builtin_unreachable.
5580
5581 2014-03-25  Marek Polacek  <polacek@redhat.com>
5582
5583         PR c/35449
5584         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
5585
5586 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
5587
5588         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
5589         order of elements for big-endian.
5590
5591 2014-03-25  Richard Biener  <rguenther@suse.de>
5592
5593         PR middle-end/60635
5594         * gimplify-me.c (gimple_regimplify_operands): Update the
5595         re-gimplifed stmt.
5596
5597 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5598
5599         PR ipa/59176
5600         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
5601         (lto_output_varpool_node): Likewise.
5602         (input_overwrite_node): Likewise.
5603         (input_varpool_node): Likewise.
5604
5605 2014-03-25  Richard Biener  <rguenther@suse.de>
5606
5607         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
5608         (run_gcc): Likewise.
5609
5610 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5611
5612         * combine.c (simplify_compare_const): Add MODE argument.
5613         Handle mode_width 0 as very large mode_width.
5614         (try_combine, simplify_comparison): Adjust callers.
5615
5616         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
5617         type to avoid signed integer overflow.
5618         * explow.c (plus_constant): Likewise.
5619
5620 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5621
5622         * doc/generic.texi: Correct typos.
5623
5624 2014-03-24  Tobias Burnus  <burnus@net-b.de>
5625
5626         * doc/invoke.texi (-flto): Expand section about
5627         using static libraries with LTO.
5628
5629 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5630
5631         PR rtl-optimization/60501
5632         * optabs.def (addptr3_optab): New optab.
5633         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
5634         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
5635         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
5636
5637         * lra.c (emit_add3_insn): Use the addptr pattern if available.
5638
5639         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
5640
5641 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
5642
5643         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
5644         _mm512_set1_pd.
5645
5646         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
5647         (_mm256_undefined_ps): Define.
5648         (_mm256_undefined_pd): Define.
5649         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
5650         (_mm_undefined_pd): Define.
5651         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
5652         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
5653         (_mm512_undefined_ps): Define.
5654         (_mm512_undefined_pd): Define.
5655         Use _mm*_undefined_*.
5656         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
5657
5658 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
5659
5660         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
5661         (lshr_simd): DI mode added.
5662         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
5663         (aarch64_ushr_simddi): Likewise.
5664         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
5665         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
5666         (vshrd_n_u64): Likewise.
5667
5668 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5669
5670         * Makefile.in (s-macro_list): Depend on cc1.
5671
5672 2014-03-23  Teresa Johnson  <tejohnson@google.com>
5673
5674         * ipa-utils.c (ipa_print_order): Use specified dump file.
5675
5676 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
5677
5678         PR rtl-optimization/60601
5679         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
5680
5681         * gcc.c (eval_spec_function): Initialize save_growing_value.
5682
5683 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
5684
5685         PR sanitizer/60613
5686         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
5687         code == MINUS_EXPR, never swap op0 with op1.
5688
5689         * toplev.c (init_local_tick): Avoid signed integer multiplication
5690         overflow.
5691         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
5692         shift by first operand's bitsize.
5693
5694 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
5695
5696         PR target/60610
5697         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
5698         redefine to 1 or 0.
5699         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
5700         TARGET_ISA_64BIT_P(x).
5701
5702 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5703
5704         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
5705         pattern for vector nor instead of subtract from splat(-1).
5706         (altivec_expand_vec_perm_const_le): Likewise.
5707
5708 2014-03-21  Richard Henderson  <rth@twiddle.net>
5709
5710         PR target/60598
5711         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
5712         related insns after epilogue_completed.
5713
5714 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5715
5716         PR ipa/59176
5717         * cgraph.h (symtab_node): New flag body_removed.
5718         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
5719         when removing bodies.
5720         * symtab.c (dump_symtab_base): Dump body_removed flag.
5721         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
5722         had their bodies removed.
5723
5724 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5725
5726         PR ipa/60419
5727         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
5728         in the border.
5729
5730 2014-03-21  Richard Biener  <rguenther@suse.de>
5731
5732         PR tree-optimization/60577
5733         * tree-core.h (struct tree_base): Document nothrow_flag use
5734         in VAR_DECL_NONALIASED.
5735         * tree.h (VAR_DECL_NONALIASED): New.
5736         (may_be_aliased): Adjust.
5737         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
5738
5739 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5740
5741         * expr.c (expand_expr_real_1): Remove outdated comment.
5742
5743 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
5744
5745         PR middle-end/60597
5746         * ira.c (adjust_cleared_regs): Call copy_rtx on
5747         *reg_equiv[REGNO (loc)].src_p before passing it to
5748         simplify_replace_fn_rtx.
5749
5750         PR target/60568
5751         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
5752         into CONST, put pic register as first operand of PLUS.  Use
5753         gen_const_mem for both 32-bit and 64-bit PIC got loads.
5754
5755 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5756
5757         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
5758
5759 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5760
5761         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
5762         around for store forwarding issue in the FPU on the UT699.
5763         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
5764         loads and operations if -mfix-ut699 is specified.
5765         (divtf3_hq): Tweak attribute.
5766         (sqrttf2_hq): Likewise.
5767
5768 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5769
5770         * calls.c (store_one_arg): Remove incorrect const qualification on the
5771         type of the temporary.
5772         * cfgexpand.c (expand_return): Likewise.
5773         * expr.c (expand_constructor): Likewise.
5774         (expand_expr_real_1): Likewise.
5775
5776 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5777
5778         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
5779         of parts.
5780
5781 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
5782
5783         PR target/60039
5784         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
5785
5786 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
5787
5788         * config/arm/aarch-common-protos.h
5789         (alu_cost_table): Fix spelling of "extend".
5790         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
5791
5792 2014-03-19  Richard Biener  <rguenther@suse.de>
5793
5794         PR middle-end/60553
5795         * tree-core.h (tree_type_common): Re-order pointer members
5796         to reduce recursion depth during GC walks.
5797
5798 2014-03-19  Marek Polacek  <polacek@redhat.com>
5799
5800         PR sanitizer/60569
5801         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
5802         before accessing it.
5803
5804 2014-03-19  Richard Biener  <rguenther@suse.de>
5805
5806         PR lto/59543
5807         * lto-streamer-in.c (input_function): In WPA stage do not drop
5808         debug stmts.
5809
5810 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
5811
5812         PR tree-optimization/60559
5813         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
5814         with build_zero_cst assignment.
5815
5816 2014-03-18  Kai Tietz  <ktietz@redhat.com>
5817
5818         PR rtl-optimization/56356
5819         * sdbout.c (sdbout_parms): Verify that parms'
5820         incoming argument is valid.
5821         (sdbout_reg_parms): Likewise.
5822
5823 2014-03-18  Richard Henderson  <rth@redhat.com>
5824
5825         PR target/60562
5826         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
5827         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
5828         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
5829
5830 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
5831
5832         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
5833         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
5834         Italicize plugin event names in description.  Explain that
5835         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
5836         Remind that no GCC functions should be called after PLUGIN_FINISH.
5837         Explain what pragmas with expansion are.
5838
5839 2014-03-18  Martin Liska  <mliska@suse.cz>
5840
5841         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
5842         gimple call statement is update.
5843         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
5844         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
5845
5846 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5847
5848         PR sanitizer/60557
5849         * ubsan.c (ubsan_instrument_unreachable): Call
5850         initialize_sanitizer_builtins.
5851         (ubsan_pass): Likewise.
5852
5853         PR sanitizer/60535
5854         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
5855         varpool_finalize_decl instead of rest_of_decl_compilation.
5856
5857 2014-03-18  Richard Biener  <rguenther@suse.de>
5858
5859         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
5860         by using bitmap_and_compl instead of bitmap_and_compl_into.
5861         (df_rd_transfer_function): Likewise.
5862
5863 2014-03-18  Richard Biener  <rguenther@suse.de>
5864
5865         * doc/lto.texi (fresolution): Fix typo.
5866
5867 2014-03-18  Richard Biener  <rguenther@suse.de>
5868
5869         * doc/invoke.texi (flto): Update for changes in 4.9.
5870
5871 2014-03-18  Richard Biener  <rguenther@suse.de>
5872
5873         * doc/loop.texi: Remove section on the removed lambda framework.
5874         Update loop docs with recent changes in preserving loop structure.
5875
5876 2014-03-18  Richard Biener  <rguenther@suse.de>
5877
5878         * doc/lto.texi (-fresolution): Document.
5879
5880 2014-03-18  Richard Biener  <rguenther@suse.de>
5881
5882         * doc/contrib.texi: Adjust my name.
5883
5884 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5885
5886         PR ipa/58721
5887         * internal-fn.c: Include diagnostic-core.h.
5888         (expand_BUILTIN_EXPECT): New function.
5889         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
5890         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
5891         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
5892         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
5893         IFN_BUILTIN_EXPECT.
5894         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
5895         Revert 3 argument __builtin_expect code.
5896         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
5897         * gimple-fold.c (gimple_fold_call): Likewise.
5898         * tree.h (fold_builtin_expect): New prototype.
5899         * builtins.c (build_builtin_expect_predicate): Add predictor
5900         argument, if non-NULL, create 3 argument __builtin_expect.
5901         (fold_builtin_expect): No longer static.  Add ARG2 argument,
5902         pass it through to build_builtin_expect_predicate.
5903         (fold_builtin_2): Adjust caller.
5904         (fold_builtin_3): Handle BUILT_IN_EXPECT.
5905         * internal-fn.def (BUILTIN_EXPECT): New.
5906
5907 2014-03-18  Tobias Burnus  <burnus@net-b.de>
5908
5909         PR ipa/58721
5910         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
5911         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
5912         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
5913
5914 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
5915
5916         PR ipa/58721
5917         * predict.c (combine_predictions_for_bb): Fix up formatting.
5918         (expr_expected_value_1, expr_expected_value): Add predictor argument,
5919         fill what it points to if non-NULL.
5920         (tree_predict_by_opcode): Adjust caller, use the predictor.
5921         * predict.def (PRED_COMPARE_AND_SWAP): Add.
5922
5923 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
5924
5925         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
5926         proper constant for the store mode.
5927
5928 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
5929
5930         * symtab.c (change_decl_assembler_name): Fix transparent alias
5931         chain construction.
5932
5933 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
5934
5935         * config/aarch64/aarch64.c: Correct the comments about the
5936         aarch64 stack layout.
5937
5938 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
5939
5940         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
5941         check for GF_OMP_FOR_KIND_FOR.
5942
5943 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
5944
5945         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
5946         ymm and zmm register names.
5947
5948 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
5949
5950         PR target/60516
5951         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
5952         note creation for the 2010-08-31 changes.
5953
5954 2014-03-17  Marek Polacek  <polacek@redhat.com>
5955
5956         PR middle-end/60534
5957         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
5958         as -fno-tree-loop-vectorize.
5959         (expand_omp_simd): Likewise.
5960
5961 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
5962
5963         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
5964         (eligible_for_call_delay): New prototype.
5965         * config/sparc/sparc.c (tls_call_delay): Rename into...
5966         (eligible_for_call_delay): ...this.  Return false if the instruction
5967         cannot be put in the delay slot of a branch.
5968         (eligible_for_restore_insn): Simplify.
5969         (eligible_for_return_delay): Return false if the instruction cannot be
5970         put in the delay slot of a branch and simplify.
5971         (eligible_for_sibcall_delay): Return false if the instruction cannot be
5972         put in the delay slot of a branch.
5973         * config/sparc/sparc.md (fix_ut699): New attribute.
5974         (tls_call_delay): Delete.
5975         (in_call_delay): Reimplement.
5976         (eligible_for_sibcall_delay): Rename into...
5977         (in_sibcall_delay): ...this.
5978         (eligible_for_return_delay): Rename into...
5979         (in_return_delay): ...this.
5980         (in_branch_delay): Reimplement.
5981         (in_uncond_branch_delay): Delete.
5982         (in_annul_branch_delay): Delete.
5983
5984 2014-03-14  Richard Henderson  <rth@redhat.com>
5985
5986         PR target/60525
5987         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
5988         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
5989         (*floathi<X87MODEF>2_i387_with_temp): Remove.
5990         (floathi splitters): Remove.
5991         (float<SWI48x>xf2): New pattern.
5992         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
5993         code that tried to handle DImode for 32-bit, but which was excluded
5994         by the pattern's condition.  Drop allocation of stack temporary.
5995         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
5996         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
5997         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
5998         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
5999         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
6000         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
6001         (*float<SWI48><MODEF>2_sse_interunit): Remove.
6002         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
6003         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
6004         (*float<SWI48x><X87MODEF>2_i387): Remove.
6005         (all float _with_temp splitters): Remove.
6006         (*float<SWI48x><MODEF>2_i387): New pattern.
6007         (*float<SWI48><MODEF>2_sse): New pattern.
6008         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
6009         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
6010
6011 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
6012             Marek Polacek  <polacek@redhat.com>
6013
6014         PR middle-end/60484
6015         * common.opt (dump_base_name_prefixed): New Variable.
6016         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
6017         if x_dump_base_name_prefixed is already set, set it at the end.
6018
6019 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
6020
6021         PR rtl-optimization/60508
6022         * lra-constraints.c (get_reload_reg): Add new parameter
6023         in_subreg_p.
6024         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
6025         Pass the new parameter values.
6026
6027 2014-03-14  Richard Biener  <rguenther@suse.de>
6028
6029         * common.opt: Revert unintented changes from r205065.
6030         * opts.c: Likewise.
6031
6032 2014-03-14  Richard Biener  <rguenther@suse.de>
6033
6034         PR middle-end/60518
6035         * cfghooks.c (split_block): Properly adjust all loops the
6036         block was a latch of.
6037
6038 2014-03-14  Martin Jambor  <mjambor@suse.cz>
6039
6040         PR lto/60461
6041         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
6042         and simplify it.
6043
6044 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
6045
6046         PR target/59396
6047         * config/avr/avr.c (avr_set_current_function): Pass function name
6048         through default_strip_name_encoding before sanity checking instead
6049         of skipping the first char of the assembler name.
6050
6051 2014-03-13  Richard Henderson  <rth@redhat.com>
6052
6053         PR debug/60438
6054         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
6055         (ix86_force_to_memory, ix86_free_from_memory): Remove.
6056         * config/i386/i386-protos.h: Likewise.
6057         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
6058         in the expander instead of a splitter.
6059         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
6060         any possibility of requiring a memory.
6061         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
6062         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
6063         (fp branch splitters): Update for ix86_split_fp_branch.
6064         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
6065         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
6066         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
6067         (*fop_<MODEF>_2_i387): Remove f/r alternative.
6068         (*fop_<MODEF>_3_i387): Likewise.
6069         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
6070         (splitters for the fop_* register patterns): Remove.
6071         (fscalexf4_i387): Rename from *fscalexf4_i387.
6072         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
6073
6074 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
6075
6076         PR tree-optimization/59779
6077         * tree-dfa.c (get_ref_base_and_extent): Use double_int
6078         type for bitsize and maxsize instead of HOST_WIDE_INT.
6079
6080 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
6081
6082         PR rtl-optimization/57320
6083         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
6084         the CFG after thread_prologue_and_epilogue_insns.
6085
6086 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
6087
6088         PR rtl-optimization/57189
6089         * lra-constraints.c (process_alt_operands): Disfavor spilling
6090         vector pseudos.
6091
6092 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
6093
6094         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
6095
6096 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
6097
6098         PR tree-optimization/59025
6099         PR middle-end/60418
6100         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
6101         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
6102
6103 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
6104
6105         PR target/60486
6106         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
6107         calls of avr_out_plus_1.
6108
6109 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
6110
6111         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
6112         BB's single pred and update the father loop's latch info later.
6113
6114 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6115
6116         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
6117         (VEC_M): Likewise.
6118         (VEC_N): Likewise.
6119         (VEC_R): Likewise.
6120         (VEC_base): Likewise.
6121         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
6122         registers, we need to swap double words in little endian mode.
6123
6124         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
6125         to be a container mode for 128-bit integer operations added in ISA
6126         2.07.  Unlike TImode and PTImode, the preferred register set is
6127         the Altivec/VMX registers for the 128-bit operations.
6128
6129         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
6130         declarations.
6131         (rs6000_split_128bit_ok_p): Likewise.
6132
6133         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
6134         macros for creating ISA 2.07 normal and overloaded builtin
6135         functions with 3 arguments.
6136         (BU_P8V_OVERLOAD_3): Likewise.
6137         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
6138         for use as overloaded functions.
6139         (VPERM_1TI_UNS): Likewise.
6140         (VSEL_1TI): Likewise.
6141         (VSEL_1TI_UNS): Likewise.
6142         (ST_INTERNAL_1ti): Likewise.
6143         (LD_INTERNAL_1ti): Likewise.
6144         (XXSEL_1TI): Likewise.
6145         (XXSEL_1TI_UNS): Likewise.
6146         (VPERM_1TI): Likewise.
6147         (VPERM_1TI_UNS): Likewise.
6148         (XXPERMDI_1TI): Likewise.
6149         (SET_1TI): Likewise.
6150         (LXVD2X_V1TI): Likewise.
6151         (STXVD2X_V1TI): Likewise.
6152         (VEC_INIT_V1TI): Likewise.
6153         (VEC_SET_V1TI): Likewise.
6154         (VEC_EXT_V1TI): Likewise.
6155         (EQV_V1TI): Likewise.
6156         (NAND_V1TI): Likewise.
6157         (ORC_V1TI): Likewise.
6158         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
6159         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
6160         overloaded builtin.
6161         (VADDUQM): Likewise.
6162         (VSUBCUQ): Likewise.
6163         (VADDEUQM): Likewise.
6164         (VADDECUQ): Likewise.
6165         (VSUBEUQM): Likewise.
6166         (VSUBECUQ): Likewise.
6167
6168         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
6169         __int128_t and __uint128_t types.
6170         (__uint128_type): Likewise.
6171         (altivec_categorize_keyword): Add support for vector __int128_t,
6172         vector __uint128_t, vector __int128, and vector unsigned __int128
6173         as a container type for TImode operations that need to be done in
6174         VSX/Altivec registers.
6175         (rs6000_macro_to_expand): Likewise.
6176         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
6177         to support 128-bit integer instructions vaddcuq, vadduqm,
6178         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
6179         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
6180
6181         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
6182         for V1TImode, and set up preferences to use VSX/Altivec registers.
6183         Setup VSX reload handlers.
6184         (rs6000_debug_reg_global): Likewise.
6185         (rs6000_init_hard_regno_mode_ok): Likewise.
6186         (rs6000_preferred_simd_mode): Likewise.
6187         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
6188         (easy_altivec_constant): Likewise.
6189         (output_vec_const_move): Likewise.
6190         (rs6000_expand_vector_set): Convert V1TImode set and extract to
6191         simple move.
6192         (rs6000_expand_vector_extract): Likewise.
6193         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
6194         addressing.
6195         (rs6000_const_vec): Add support for V1TImode.
6196         (rs6000_emit_le_vsx_load): Swap double words when loading or
6197         storing TImode/V1TImode.
6198         (rs6000_emit_le_vsx_store): Likewise.
6199         (rs6000_emit_le_vsx_move): Likewise.
6200         (rs6000_emit_move): Add support for V1TImode.
6201         (altivec_expand_ld_builtin): Likewise.
6202         (altivec_expand_st_builtin): Likewise.
6203         (altivec_expand_vec_init_builtin): Likewise.
6204         (altivec_expand_builtin): Likewise.
6205         (rs6000_init_builtins): Add support for V1TImode type.  Add
6206         support for ISA 2.07 128-bit integer builtins.  Define type names
6207         for the VSX/Altivec vector types.
6208         (altivec_init_builtins): Add support for overloaded vector
6209         functions with V1TImode type.
6210         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
6211         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
6212         external function.
6213         (rs6000_split_128bit_ok_p): Likewise.
6214         (rs6000_handle_altivec_attribute): Create V1TImode from vector
6215         __int128_t and vector __uint128_t.
6216
6217         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
6218         and mode attributes.
6219         (VSX_M): Likewise.
6220         (VSX_M2): Likewise.
6221         (VSm): Likewise.
6222         (VSs): Likewise.
6223         (VSr): Likewise.
6224         (VSv): Likewise.
6225         (VS_scalar): Likewise.
6226         (VS_double): Likewise.
6227         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
6228
6229         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
6230         we support the ISA 2.07 128-bit integer arithmetic instructions.
6231         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
6232         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
6233         and TImode types for use with the builtin functions.
6234         (V1TI_type_node): Likewise.
6235         (unsigned_V1TI_type_node): Likewise.
6236         (intTI_type_internal_node): Likewise.
6237         (uintTI_type_internal_node): Likewise.
6238
6239         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
6240         128-bit builtin functions.
6241         (UNSPEC_VADDEUQM): Likewise.
6242         (UNSPEC_VADDECUQ): Likewise.
6243         (UNSPEC_VSUBCUQ): Likewise.
6244         (UNSPEC_VSUBEUQM): Likewise.
6245         (UNSPEC_VSUBECUQ): Likewise.
6246         (VM): Add V1TImode to vector mode iterators.
6247         (VM2): Likewise.
6248         (VI_unit): Likewise.
6249         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
6250         (altivec_vaddcuq): Likewise.
6251         (altivec_vsubuqm): Likewise.
6252         (altivec_vsubcuq): Likewise.
6253         (altivec_vaddeuqm): Likewise.
6254         (altivec_vaddecuq): Likewise.
6255         (altivec_vsubeuqm): Likewise.
6256         (altivec_vsubecuq): Likewise.
6257
6258         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
6259         mode iterators.
6260         (BOOL_128): Likewise.
6261         (BOOL_REGS_OUTPUT): Likewise.
6262         (BOOL_REGS_OP1): Likewise.
6263         (BOOL_REGS_OP2): Likewise.
6264         (BOOL_REGS_UNARY): Likewise.
6265         (BOOL_REGS_AND_CR0): Likewise.
6266
6267         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
6268         128-bit integer builtin support.
6269         (vec_vadduqm): Likewise.
6270         (vec_vaddecuq): Likewise.
6271         (vec_vaddeuqm): Likewise.
6272         (vec_vsubecuq): Likewise.
6273         (vec_vsubeuqm): Likewise.
6274         (vec_vsubcuq): Likewise.
6275         (vec_vsubuqm): Likewise.
6276
6277         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6278         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
6279         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
6280         128-bit integer add/subtract to ISA 2.07.
6281
6282 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
6283
6284         * config/arc/arc.c (arc_predicate_delay_insns):
6285         Fix third argument passed to conditionalize_nonjump.
6286
6287 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
6288
6289         * config/aarch64/aarch64-builtins.c
6290         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
6291         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
6292         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
6293         instead of __builtin_lfloor.
6294         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
6295
6296 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6297
6298         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
6299         (tree_ssa_ifcombine_bb_1): New function.
6300         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
6301         is an empty forwarder block to then_bb or vice versa and then_bb
6302         and else_bb are effectively swapped.
6303
6304 2014-03-12  Christian Bruel  <christian.bruel@st.com>
6305
6306         PR target/60264
6307         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
6308         REG_CFA_DEF_CFA note.
6309         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
6310         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
6311
6312 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
6313
6314         PR tree-optimization/60454
6315         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
6316
6317 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6318
6319         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
6320         Do not define target_cpu_default2 to generic.
6321         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
6322         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
6323         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
6324
6325 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6326             Marc Glisse  <marc.glisse@inria.fr>
6327
6328         PR tree-optimization/60502
6329         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
6330         instead of build_low_bits_mask.
6331
6332 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6333
6334         PR middle-end/60482
6335         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
6336         if there are multiple uses, but op doesn't live on E edge.
6337         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
6338         clobber stmts before __builtin_unreachable.
6339
6340 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
6341
6342         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
6343         hard_frame_pointer_rtx.
6344         * cse.c (cse_insn): Remove volatile check.
6345         * cselib.c (cselib_process_insn): Likewise.
6346         * dse.c (scan_insn): Likewise.
6347
6348 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6349
6350         * config/arc/arc.c (conditionalize_nonjump): New function,
6351         broken out of ...
6352         (arc_ifcvt): ... this.
6353         (arc_predicate_delay_insns): Use it.
6354
6355 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6356
6357         * config/arc/predicates.md (extend_operand): During/after reload,
6358         allow const_int_operand.
6359         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
6360         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
6361         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
6362         to "i".
6363         (umulsi3_highpart_i): Likewise.
6364
6365 2014-03-11  Richard Biener  <rguenther@suse.de>
6366
6367         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
6368         Add asserts to guard possible wrong-code bugs.
6369
6370 2014-03-11  Richard Biener  <rguenther@suse.de>
6371
6372         PR tree-optimization/60429
6373         PR tree-optimization/60485
6374         * tree-ssa-structalias.c (set_union_with_increment): Properly
6375         take into account all fields that overlap the shifted vars.
6376         (do_sd_constraint): Likewise.
6377         (do_ds_constraint): Likewise.
6378         (get_constraint_for_ptr_offset): Likewise.
6379
6380 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
6381
6382         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
6383         (nios2_compute_frame_layout):
6384         Add calculation of cfun->machine->fp_save_offset.
6385         (nios2_expand_prologue): Correct setting of frame pointer register
6386         in prologue.
6387         (nios2_expand_epilogue): Update recovery of stack pointer from
6388         frame pointer accordingly.
6389         (nios2_initial_elimination_offset): Update calculation of offset
6390         for eliminating to HARD_FRAME_POINTER_REGNUM.
6391
6392 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
6393
6394         PR ipa/60457
6395         * ipa.c (symtab_remove_unreachable_nodes): Don't call
6396         cgraph_get_create_node on VAR_DECLs.
6397
6398 2014-03-10  Richard Biener  <rguenther@suse.de>
6399
6400         PR middle-end/60474
6401         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
6402
6403 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
6404
6405         * config/vms/vms.opt (vms_float_format): New variable.
6406
6407 2014-03-08  Tobias Burnus  <burnus@net-b.de>
6408
6409         * doc/invoke.texi (-fcilkplus): Update implementation status.
6410
6411 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
6412             Richard Biener  <rguenther@suse.de>
6413
6414         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
6415         consistently accross all TUs.
6416         (run_gcc): Enable -fshort-double automatically at link at link-time
6417         and disallow override.
6418
6419 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
6420
6421         PR target/58271
6422         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
6423         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
6424         if they can't be used.
6425
6426 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6427
6428         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
6429         for Solaris 11/x86 ld.
6430         * configure: Regenerate.
6431
6432 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6433
6434         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
6435         (LIB_TLS_SPEC): Save as ld_tls_libs.
6436         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
6437         (HAVE_AS_IX86_TLSLDM): New test.
6438         * configure, config.in: Regenerate.
6439         * config/i386/i386.c (legitimize_tls_address): Fall back to
6440         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
6441         cannot support TLS_MODEL_LOCAL_DYNAMIC.
6442         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
6443         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
6444
6445 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
6446
6447         * common.opt (fira-loop-pressure): Mark as optimization.
6448
6449 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
6450
6451         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
6452         an OpenMP mappable type.
6453
6454 2014-03-06  Matthias Klose  <doko@ubuntu.com>
6455
6456         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
6457         MULTILIB_OSDIRNAMES is not defined.
6458
6459 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
6460             Meador Inge  <meadori@codesourcery.com>
6461
6462         PR target/58595
6463         * config/arm/arm.c (arm_tls_symbol_p): Remove.
6464         (arm_legitimize_address): Call legitimize_tls_address for any
6465         arm_tls_referenced_p expression, handle constant addend.  Call it
6466         before testing for !TARGET_ARM.
6467         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
6468
6469 2014-03-06  Richard Biener  <rguenther@suse.de>
6470
6471         PR middle-end/60445
6472         PR lto/60424
6473         PR lto/60427
6474         Revert
6475         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6476
6477         * tree-streamer.c (record_common_node): Assert we don't record
6478         nodes with type double.
6479         (preload_common_node): Skip type double, complex double and double
6480         pointer since it is now frontend dependent due to fshort-double option.
6481
6482 2014-03-06  Richard Biener  <rguenther@suse.de>
6483
6484         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
6485         or -fno-lto is specified and the linker has full plugin support.
6486         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
6487         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
6488         * lto-wrapper.c (merge_and_complain): Merge compile-time
6489         optimization levels.
6490         (run_gcc): And pass it through to the link options.
6491
6492 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
6493
6494         PR debug/60381
6495         Revert:
6496         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6497         PR debug/59992
6498         * cselib.c (remove_useless_values): Skip to avoid quadratic
6499         behavior if the condition moved from...
6500         (cselib_process_insn): ... here holds.
6501
6502 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6503
6504         PR plugins/59335
6505         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
6506         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
6507
6508         PR plugins/59335
6509         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
6510         (TM_H): Add x86-tune.def.
6511
6512 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6513
6514         * config/aarch64/aarch64.c (generic_tunings):
6515         Use cortexa57_extra_costs.
6516
6517 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6518
6519         PR lto/60404
6520         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
6521         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
6522         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
6523         cost for in_lto_p.
6524
6525 2014-03-04  Heiher  <r@hev.cc>
6526
6527         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
6528         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
6529
6530 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
6531
6532         * config/i386/predicates.md (const2356_operand): Change to ...
6533         (const2367_operand): ... this.
6534         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
6535         const2367_operand.
6536         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6537         (*avx512pf_scatterpf<mode>sf): Ditto.
6538         (avx512pf_scatterpf<mode>df): Ditto.
6539         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6540         (*avx512pf_scatterpf<mode>df): Ditto.
6541         * config/i386/i386.c (ix86_expand_builtin): Update
6542         incorrect hint operand error message.
6543
6544 2014-03-04  Richard Biener  <rguenther@suse.de>
6545
6546         * lto-section-in.c (lto_get_section_data): Fix const cast.
6547
6548 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6549
6550         * tree-streamer.c (record_common_node): Assert we don't record
6551         nodes with type double.
6552         (preload_common_node): Skip type double, complex double and double
6553         pointer since it is now frontend dependent due to fshort-double option.
6554
6555 2014-03-04  Richard Biener  <rguenther@suse.de>
6556
6557         PR lto/60405
6558         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
6559         (lto_input_toplevel_asms): Likewise.
6560         * lto-section-in.c (lto_get_section_data): Instead do it here
6561         for every section.
6562
6563 2014-03-04  Richard Biener  <rguenther@suse.de>
6564
6565         PR tree-optimization/60382
6566         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
6567         dead PHIs a reduction.
6568
6569 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
6570
6571         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
6572         hint value.
6573         (_mm_prefetch): Move out of GCC target("sse") pragma.
6574         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
6575         GCC target("prfchw") pragma.
6576         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
6577         for locality <= 2.
6578         * config/i386/i386.c (ix86_option_override_internal): Enable
6579         -mprfchw with -mprefetchwt1.
6580
6581 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6582
6583         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
6584         Mark as varying.
6585
6586 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6587
6588         * opts.h (CL_PCH_IGNORE): Define.
6589         * targhooks.c (option_affects_pch_p):
6590         Return false for options that have CL_PCH_IGNORE set.
6591         * opt-functions.awk: Process PchIgnore.
6592         * doc/options.texi: Document PchIgnore.
6593
6594         * config/arc/arc.opt (misize): Add PchIgnore property.
6595
6596 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6597
6598         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
6599         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
6600         constraint on constants to permit them being loaded into
6601         GENERAL_REGS or BASE_REGS.
6602
6603 2014-03-03  Nick Clifton  <nickc@redhat.com>
6604
6605         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
6606         anti-cacnonical alternatives.
6607         (negandhi3_real): New pattern.
6608         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
6609
6610 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
6611
6612         * config/avr/avr-mcus.def: Remove atxmega16x1.
6613         * config/avr/avr-tables.opt: Regenerate.
6614         * config/avr/t-multilib: Regenerate.
6615         * doc/avr-mmcu.texi: Regenerate.
6616
6617 2014-03-03  Tobias Grosser  <tobias@grosser.es>
6618             Mircea Namolaru  <mircea.namolaru@inria.fr>
6619
6620         PR tree-optimization/58028
6621         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
6622         scalar dimensions.
6623
6624 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6625
6626         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
6627         not handled by recognizers.
6628
6629 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
6630
6631         PR middle-end/60175
6632         * function.c (expand_function_end): Don't emit
6633         clobber_return_register sequence if clobber_after is a BARRIER.
6634         * cfgexpand.c (construct_exit_block): Append instructions before
6635         return_label to prev_bb.
6636
6637 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6638
6639         * config/rs6000/constraints.md: Document reserved use of "wc".
6640
6641 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6642
6643         PR ipa/60150
6644         * ipa.c (function_and_variable_visibility): When dissolving comdat
6645         group, also set all symbols to local.
6646
6647 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6648
6649         PR ipa/60306
6650
6651         Revert:
6652         2013-12-14   Jan Hubicka  <jh@suse.cz>
6653         PR middle-end/58477
6654         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
6655
6656 2014-03-02  Jon Beniston  <jon@beniston.com>
6657
6658         PR bootstrap/48230
6659         PR bootstrap/50927
6660         PR bootstrap/52466
6661         PR target/46898
6662         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
6663           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
6664         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
6665         (simple_return, *simple_return): New patterns
6666         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
6667         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
6668
6669 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
6670
6671         * dwarf2out.c (gen_subprogram_die): Tidy.
6672
6673 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
6674
6675         PR target/60071
6676         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
6677         (*mov_t_msb_neg_negc): ... this new insn.
6678
6679 2014-02-28  Jason Merrill  <jason@redhat.com>
6680
6681         PR c++/58678
6682         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
6683         function.
6684
6685 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
6686
6687         PR c++/60314
6688         * dwarf2out.c (decltype_auto_die): New static.
6689         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
6690         (gen_type_die_with_usage): Handle 'decltype(auto)'.
6691         (is_cxx_auto): Likewise.
6692
6693 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
6694
6695         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
6696         we are not using general regs only.
6697
6698 2014-02-28  Richard Biener  <rguenther@suse.de>
6699
6700         PR target/60280
6701         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
6702         previous fix and only allow to remove trivial pre-headers
6703         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
6704         (remove_forwarder_block): Properly update the latch of a loop.
6705
6706 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6707
6708         PR debug/59992
6709         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
6710         (cselib_preserved_hash_table): New.
6711         (preserve_constants_and_equivs): Move preserved vals to it.
6712         (cselib_find_slot): Look it up first.
6713         (cselib_init): Initialize it.
6714         (cselib_finish): Release it.
6715         (dump_cselib_table): Dump it.
6716
6717 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6718
6719         PR debug/59992
6720         * cselib.c (remove_useless_values): Skip to avoid quadratic
6721         behavior if the condition moved from...
6722         (cselib_process_insn): ... here holds.
6723
6724 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6725
6726         PR debug/57232
6727         * var-tracking.c (vt_initialize): Apply the same condition to
6728         preserve the CFA base value.
6729
6730 2014-02-28  Joey Ye  <joey.ye@arm.com>
6731
6732         PR target/PR60169
6733         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
6734         if reload in progress or completed.
6735
6736 2014-02-28  Tobias Burnus  <burnus@net-b.de>
6737
6738         PR middle-end/60147
6739         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
6740         NAMELIST_DECL.
6741
6742 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
6743
6744         * doc/tm.texi.in (Condition Code Status): Update documention for
6745         relative locations of cc0-setter and cc0-user.
6746
6747 2014-02-27  Jeff Law  <law@redhat.com>
6748
6749         PR rtl-optimization/52714
6750         * combine.c (try_combine): When splitting an unrecognized PARALLEL
6751         into two independent simple sets, if I3 is a jump, ensure the
6752         pattern we place into I3 is a (set (pc) ...).
6753
6754 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
6755             Jeff Law  <law@redhat.com>
6756
6757         PR rtl-optimization/49847
6758         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
6759         are in different blocks.
6760         * doc/tm.texi (Condition Code Status): Update documention for
6761         relative locations of cc0-setter and cc0-user.
6762
6763 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
6764
6765         PR target/59222
6766         * lra.c (lra_emit_add): Check SUBREG too.
6767
6768 2014-02-27  Andreas Schwab  <schwab@suse.de>
6769
6770         * config/m68k/m68k.c (m68k_option_override): Disable
6771         -flive-range-shrinkage for classic m68k.
6772         (m68k_override_options_after_change): Likewise.
6773
6774 2014-02-27  Marek Polacek  <polacek@redhat.com>
6775
6776         PR middle-end/59223
6777         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
6778         -Wmaybe-uninitialized.
6779
6780 2014-02-27  Alan Modra  <amodra@gmail.com>
6781
6782         PR target/57936
6783         * reload1.c (emit_input_reload_insns): When reload_override_in,
6784         set old to rl->in_reg when rl->in_reg is a subreg.
6785
6786 2014-02-26  Richard Biener  <rguenther@suse.de>
6787
6788         PR bootstrap/60343
6789         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
6790
6791 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6792
6793         * common/config/i386/predicates.md (const1256_operand): Remove.
6794         (const2356_operand): New.
6795         (const_1_to_2_operand): Remove.
6796         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
6797         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6798         (*avx512pf_gatherpf<mode>sf): Ditto.
6799         (avx512pf_gatherpf<mode>df): Ditto.
6800         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6801         (*avx512pf_gatherpf<mode>df): Ditto.
6802         (avx512pf_scatterpf<mode>sf): Ditto.
6803         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6804         (*avx512pf_scatterpf<mode>sf): Ditto.
6805         (avx512pf_scatterpf<mode>df): Ditto.
6806         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6807         (*avx512pf_scatterpf<mode>df): Ditto.
6808         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
6809
6810 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
6811
6812         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
6813         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
6814         (_mm512_mask_testn_epi64_mask): Move to ...
6815         * config/i386/avx512cdintrin.h: Here.
6816         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
6817         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
6818         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
6819         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
6820         TARGET_AVX512F from TARGET_AVX512CD.
6821
6822 2014-02-26  Richard Biener  <rguenther@suse.de>
6823
6824         PR ipa/60327
6825         * ipa.c (walk_polymorphic_call_targets): Properly guard
6826         call to inline_update_overall_summary.
6827
6828 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
6829
6830         PR target/60280
6831         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
6832         and latches only if requested.  Fix latch if it is removed.
6833         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
6834         LOOPS_HAVE_PREHEADERS.
6835
6836 2014-02-25  Andrew Pinski  <apinski@cavium.com>
6837
6838         * builtins.c (expand_builtin_thread_pointer): Create a new target
6839         when the target is NULL.
6840
6841 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
6842
6843         PR rtl-optimization/60317
6844         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6845         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6846         * lra-assigns.c: Include params.h.
6847         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
6848         other reload pseudos considerations.
6849
6850 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6851
6852         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
6853         to use canonical form for nor<mode>3.
6854
6855 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6856
6857         PR target/55426
6858         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
6859         conversions.
6860
6861 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6862
6863         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
6864         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
6865         (ix86_handle_option): Handle OPT_mprefetchwt1.
6866         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
6867         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
6868         PREFETCHWT1 CPUID.
6869         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6870         OPTION_MASK_ISA_PREFETCHWT1.
6871         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
6872         (PTA_PREFETCHWT1): New.
6873         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
6874         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
6875         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
6876         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
6877         (*prefetch_avx512pf_<mode>_: Change into ...
6878         (*prefetch_prefetchwt1_<mode>: This.
6879         * config/i386/i386.opt (mprefetchwt1): New.
6880         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
6881         (_mm_prefetch): Handle intent to write.
6882         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
6883
6884 2014-02-25  Richard Biener  <rguenther@suse.de>
6885
6886         PR middle-end/60291
6887         * emit-rtl.c (mem_attrs_htab): Remove.
6888         (mem_attrs_htab_hash): Likewise.
6889         (mem_attrs_htab_eq): Likewise.
6890         (set_mem_attrs): Always allocate new mem-attrs when something changed.
6891         (init_emit_once): Do not allocate mem_attrs_htab.
6892
6893 2014-02-25  Richard Biener  <rguenther@suse.de>
6894
6895         PR lto/60319
6896         * lto-opts.c (lto_write_options): Output non-explicit conservative
6897         -fwrapv, -fno-trapv and -fno-strict-overflow.
6898         * lto-wrapper.c (merge_and_complain): Handle merging those options.
6899         (run_gcc): And pass them through.
6900
6901 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6902
6903         * sel-sched.c (calculate_new_fences): New parameter ptime.
6904         Calculate it as a maximum over all fence cycles.
6905         (sel_sched_region_2): Adjust the call to calculate_new_fences.
6906         Print the final schedule timing when sched_verbose.
6907
6908 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6909
6910         PR rtl-optimization/60292
6911         * sel-sched.c (fill_vec_av_set): Do not reset target availability
6912         bit fot the fence instruction.
6913
6914 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
6915
6916         * calls.h: Fix typo in comment.
6917
6918 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
6919
6920         * config/pa/pa.c (pa_output_move_double): Don't valididate when
6921         adjusting offsetable addresses.
6922
6923 2014-02-24  Guozhi Wei  <carrot@google.com>
6924
6925         * sparseset.h (sparseset_pop): Fix the wrong index.
6926
6927 2014-02-24  Walter Lee  <walt@tilera.com>
6928
6929         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
6930         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
6931         triplet.
6932         * common/config/tilegx/tilegx-common.c
6933         (TARGET_DEFAULT_TARGET_FLAGS): Define.
6934         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
6935         (LINK_SPEC): Ditto.
6936         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
6937         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
6938         (tilegx_gimplify_va_arg_expr): Handle big endian.
6939         (tilegx_expand_unaligned_load): Ditto.
6940         (tilegx_expand_unaligned_store): Ditto.
6941         (TARGET_RETURN_IN_MSB): New.
6942         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
6943         (TARGET_ENDIAN_DEFAULT): New.
6944         (TARGET_BIG_ENDIAN): Handle big endian.
6945         (BYTES_BIG_ENDIAN): Ditto.
6946         (WORDS_BIG_ENDIAN): Ditto.
6947         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
6948         (ENDIAN_SPEC): New.
6949         (EXTRA_SPECS): New.
6950         * config/tilegx/tilegx.md (extv): Handle big endian.
6951         (extzv): Ditto.
6952         (insn_st<n>): Ditto.
6953         (insn_st<n>_add<bitsuffix>): Ditto.
6954         (insn_stnt<n>): Ditto.
6955         (insn_stnt<n>_add<bitsuffix>):Ditto.
6956         (vec_interleave_highv8qi): Handle big endian.
6957         (vec_interleave_highv8qi_be): New.
6958         (vec_interleave_highv8qi_le): New.
6959         (insn_v1int_h): Handle big endian.
6960         (vec_interleave_lowv8qi): Handle big endian.
6961         (vec_interleave_lowv8qi_be): New.
6962         (vec_interleave_lowv8qi_le): New.
6963         (insn_v1int_l): Handle big endian.
6964         (vec_interleave_highv4hi): Handle big endian.
6965         (vec_interleave_highv4hi_be): New.
6966         (vec_interleave_highv4hi_le): New.
6967         (insn_v2int_h): Handle big endian.
6968         (vec_interleave_lowv4hi): Handle big endian.
6969         (vec_interleave_lowv4hi_be): New.
6970         (vec_interleave_lowv4hi_le): New.
6971         (insn_v2int_l): Handle big endian.
6972         (vec_interleave_highv2si): Handle big endian.
6973         (vec_interleave_highv2si_be): New.
6974         (vec_interleave_highv2si_le): New.
6975         (insn_v4int_h): Handle big endian.
6976         (vec_interleave_lowv2si): Handle big endian.
6977         (vec_interleave_lowv2si_be): New.
6978         (vec_interleave_lowv2si_le): New.
6979         (insn_v4int_l): Handle big endian.
6980         * config/tilegx/tilegx.opt (mbig-endian): New option.
6981         (mlittle-endian): New option.
6982         * doc/install.texi: Document tilegxbe-linux.
6983         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
6984
6985 2014-02-24  Martin Jambor  <mjambor@suse.cz>
6986
6987         PR ipa/60266
6988         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
6989         there are no parameter descriptors.
6990
6991 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
6992
6993         PR rtl-optimization/60268
6994         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
6995         initialization to ...
6996         (sched_rgn_init): ... here.
6997         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
6998
6999 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7000
7001         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
7002         names.
7003
7004 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
7005
7006         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
7007         definition.
7008
7009 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7010
7011         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
7012         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
7013
7014 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7015
7016         * config/microblaze/predicates.md: Add cmp_op predicate.
7017         * config/microblaze/microblaze.md: Add branch_compare instruction
7018         which uses cmp_op predicate and emits cmp insn before branch.
7019         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
7020         to microblaze_expand_conditional_branch and consolidate logic.
7021         (microblaze_expand_conditional_branch): emit branch_compare
7022         insn instead of handling cmp op separate from branch insn.
7023
7024 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7025
7026         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
7027         to permit subregs.
7028
7029 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7030
7031         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
7032         define_insn with define_expand and new define_insn
7033         *altivec_lve<VI_char>x_internal.
7034         (altivec_stve<VI_char>x): Replace define_insn with define_expand
7035         and new define_insn *altivec_stve<VI_char>x_internal.
7036         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
7037         prototype.
7038         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
7039         lve*x built-ins.
7040         (altivec_expand_stvex_be): New function.
7041
7042 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
7043
7044         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
7045         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
7046         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
7047         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
7048
7049 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
7050
7051         PR target/60298
7052         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
7053         instead of emit_move_insn.
7054
7055 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7056
7057         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
7058         vspltw with vsldoi.
7059         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
7060         gen_altivec_vsumsws.
7061
7062 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7063
7064         * config/rs6000/altivec.md (altivec_lvxl): Rename as
7065         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
7066         (altivec_lvxl_<mode>): New define_expand incorporating
7067         -maltivec=be semantics where needed.
7068         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
7069         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
7070         semantics where needed.
7071         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
7072         (altivec_stvx_<mode>): New define_expand incorporating
7073         -maltivec=be semantics where needed.
7074         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
7075         VM2 iterator instead of V4SI.
7076         (altivec_stvxl_<mode>): New define_expand incorporating
7077         -maltivec=be semantics where needed.
7078         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
7079         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
7080         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
7081         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
7082         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
7083         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
7084         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
7085         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
7086         ALTIVEC_BUILTIN_STVXL.
7087         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
7088         (altivec_expand_stvx_be): Likewise.
7089         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
7090         (altivec_expand_lvx_be): Likewise.
7091         (altivec_expand_stvx_be): Likewise.
7092         (altivec_expand_builtin): Add cases for
7093         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
7094         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
7095         (altivec_init_builtins): Add definitions for
7096         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
7097         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
7098
7099 2014-02-21  Catherine Moore  <clm@codesourcery.com>
7100
7101         * doc/invoke.texi (mvirt, mno-virt): Document.
7102         * config/mips/mips.opt (mvirt): New option.
7103         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
7104
7105 2014-02-21  Richard Biener  <rguenther@suse.de>
7106
7107         PR tree-optimization/60276
7108         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
7109         (STMT_VINFO_MIN_NEG_DIST): New macro.
7110         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
7111         STMT_VINFO_MIN_NEG_DIST.
7112         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
7113         made for negative dependence distances still hold.
7114
7115 2014-02-21  Richard Biener  <rguenther@suse.de>
7116
7117         PR middle-end/60291
7118         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
7119         DECL_INITIAL for globals not in the current function context.
7120
7121 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
7122
7123         PR tree-optimization/56490
7124         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
7125         * tree-ssa-uninit.c: Include params.h.
7126         (compute_control_dep_chain): Add num_calls argument, return false
7127         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
7128         num_calls to recursive call.
7129         (find_predicates): Change dep_chain into normal array,
7130         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
7131         variable and adjust compute_control_dep_chain caller.
7132         (find_def_preds): Likewise.
7133
7134 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
7135
7136         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
7137         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
7138
7139 2014-02-21  Nick Clifton  <nickc@redhat.com>
7140
7141         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
7142         (pushhi1): Likewise.
7143         (popqi1): Add mode to pre_dec.
7144         (pophi1): Likewise.
7145
7146 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
7147
7148         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
7149         mode for mask of V8SFmode permutation.
7150
7151 2014-02-20  Richard Henderson  <rth@redhat.com>
7152
7153         PR c++/60272
7154         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
7155         a new pseudo for OLDVAL.
7156
7157 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
7158
7159         PR target/57896
7160         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
7161         gen_reg_rtx if d->testing_p.
7162         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
7163         if d->testing_p and we will certainly return true.
7164         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
7165         if d->testing_p.
7166
7167 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
7168
7169         * emit-rtl.c (gen_reg_rtx): Assert that
7170         crtl->emit.regno_pointer_align_length is non-zero.
7171
7172 2014-02-20  Richard Henderson  <rth@redhat.com>
7173
7174         PR c++/60272
7175         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
7176         on failure the store back into EXPECT.
7177
7178 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
7179             Sandra Loosemore  <sandra@codesourcery.com>
7180
7181         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
7182         * config/nios2/nios2.c (nios2_function_profiler): Add
7183         -fPIC (flag_pic == 2) support.
7184         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
7185         (nios2_large_offset_p): New function.
7186         (nios2_unspec_reloc_p): Move up position, update to use
7187         nios2_large_offset_p.
7188         (nios2_unspec_address): Remove function.
7189         (nios2_unspec_offset): New function.
7190         (nios2_large_got_address): New function.
7191         (nios2_got_address): Add large offset support.
7192         (nios2_legitimize_tls_address): Update usage of removed and new
7193         functions.
7194         (nios2_symbol_binds_local_p): New function.
7195         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
7196         (nios2_legitimize_address): Update to use nios2_large_offset_p.
7197         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
7198         (nios2_print_operand): Merge H/L processing, add hiadj/lo
7199         processing for (const (unspec ...)).
7200         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
7201
7202 2014-02-20  Richard Biener  <rguenther@suse.de>
7203
7204         * tree-cfg.c (replace_uses_by): Mark altered BBs before
7205         doing the substitution.
7206         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
7207
7208 2014-02-20  Martin Jambor  <mjambor@suse.cz>
7209
7210         PR ipa/55260
7211         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
7212         info when checking whether lattices are bottom.
7213
7214 2014-02-20  Richard Biener  <rguenther@suse.de>
7215
7216         PR middle-end/60221
7217         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
7218         regions at -O0.
7219
7220 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
7221
7222         PR ipa/58555
7223         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
7224         parameter specifying the scaling.
7225         (inline_call): Update.
7226         (want_inline_recursively): Guard division by zero.
7227         (recursive_inlining): Update.
7228         * ipa-inline.h (clone_inlined_nodes): Update.
7229
7230 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
7231
7232         PR target/60204
7233         * config/i386/i386.c (classify_argument): Pass structures of size
7234         64 bytes or less in register.
7235
7236 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
7237             Kirill Yukhin  <kirill.yukhin@intel.com>
7238
7239         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
7240         (_mm_rcp28_round_ss): Ditto.
7241         (_mm_rsqrt28_round_sd): Ditto.
7242         (_mm_rsqrt28_round_ss): Ditto.
7243         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
7244         (_mm_rcp14_round_ss): Ditto.
7245         (_mm_rsqrt14_round_sd): Ditto.
7246         (_mm_rsqrt14_round_ss): Ditto.
7247         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
7248         the first input operand, get rid of match_dup.
7249         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
7250         attribute to sse.
7251         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7252         Ditto.
7253         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
7254         operand as the first input operand, set type attribute.
7255         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7256         Set type attribute.
7257         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
7258         operand as the first input operand, set type attribute.
7259
7260 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7261
7262         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
7263         bit of zero.
7264
7265 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
7266
7267         PR target/60207
7268         * config/i386/i386.c (construct_container): Remove TFmode check
7269         for X86_64_INTEGER_CLASS.
7270
7271 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
7272
7273         PR target/59794
7274         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
7275         only when -Wpsabi is enabled.
7276
7277 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
7278
7279          PR target/59799
7280         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
7281         passing arrays in registers are the same as for structs, so remove the
7282         special case for them.
7283
7284 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
7285
7286         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
7287         destination type, extract only the valid bits if the source type is not
7288         integral and has a different mode.
7289
7290 2014-02-19  Richard Biener  <rguenther@suse.de>
7291
7292         PR ipa/60243
7293         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
7294         for all calls.
7295
7296 2014-02-19  Richard Biener  <rguenther@suse.de>
7297
7298         PR ipa/60243
7299         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
7300         (ipa_modify_call_arguments): Emit an argument load explicitely and
7301         preserve virtual SSA form there and for the replacement call.
7302         Do not update SSA form nor free dominance info.
7303
7304 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7305
7306         * ipa.c (function_and_variable_visibility): Also clear WEAK
7307         flag when disolving COMDAT_GROUP.
7308
7309 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7310
7311         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
7312         * ipa-prop.c (ipa_set_jf_known_type): Return early when
7313         not devirtualizing.
7314         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
7315         do more sanity checks.
7316         (detect_type_change): Return true when giving up early.
7317         (compute_complex_assign_jump_func): Fix type parameter of
7318         ipa_set_ancestor_jf.
7319         (compute_complex_ancestor_jump_func): Likewise.
7320         (update_jump_functions_after_inlining): Fix updating of
7321         ancestor function.
7322         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
7323
7324 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7325
7326         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
7327         inline clones when edge disappears.
7328
7329 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7330
7331         PR target/60203
7332         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
7333         Split 64-bit moves into 2 patterns.  Do not allow the use of
7334         direct move for TDmode in little endian, since the decimal value
7335         has little endian bytes within a word, but the 64-bit pieces are
7336         ordered in a big endian fashion, and normal subreg's of TDmode are
7337         not allowed.
7338         (mov<mode>_64bit_dm): Likewise.
7339         (movtd_64bit_nodm): Likewise.
7340
7341 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
7342
7343         PR tree-optimization/60174
7344         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
7345         statement of an SSA_NAME that occurs in an abnormal PHI node.
7346
7347 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
7348
7349         PR sanitizer/60142
7350         * final.c (SEEN_BB): Remove.
7351         (SEEN_NOTE, SEEN_EMITTED): Renumber.
7352         (final_scan_insn): Don't force_source_line on second
7353         NOTE_INSN_BASIC_BLOCK.
7354
7355 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
7356
7357         PR target/60205
7358         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
7359         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
7360         (type_natural_mode): Warn ABI change when %zmm register is not
7361         available for AVX512F vector value passing.
7362
7363 2014-02-18  Kai Tietz  <ktietz@redhat.com>
7364
7365         PR target/60193
7366         * config/i386/i386.c (ix86_expand_prologue): Use value in
7367         rax register as displacement when restoring %r10 or %rax.
7368         Fix wrong offset when restoring both registers.
7369
7370 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
7371
7372         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
7373         assertion with conditional return.
7374
7375 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
7376             Uros Bizjak  <ubizjak@gmail.com>
7377
7378         PR driver/60233
7379         * config/i386/driver-i386.c (host_detect_local_cpu): If
7380         YMM state is not saved by the OS, also clear has_f16c.  Move
7381         CPUID 0x80000001 handling before YMM state saving checking.
7382
7383 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
7384
7385         PR rtl-optimization/58960
7386         * haifa-sched.c (alloc_global_sched_pressure_data): New,
7387         factored out from ...
7388         (sched_init): ... here.
7389         (free_global_sched_pressure_data): New, factored out from ...
7390         (sched_finish): ... here.
7391         * sched-int.h (free_global_sched_pressure_data): Declare.
7392         * sched-rgn.c (nr_regions_initial): New static global.
7393         (haifa_find_rgns): Initialize it.
7394         (schedule_region): Disable sched-pressure for the newly
7395         generated regions.
7396
7397 2014-02-17  Richard Biener  <rguenther@suse.de>
7398
7399         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
7400         release SSA defs of pattern stmts.
7401
7402 2014-02-17  Richard Biener  <rguenther@suse.de>
7403
7404         * tree-inline.c (expand_call_inline): Release the virtual
7405         operand defined by the call we are about to inline.
7406
7407 2014-02-17  Richard Biener  <rguenther@suse.de>
7408
7409         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
7410
7411 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
7412             Ilya Tocar  <ilya.tocar@intel.com>
7413
7414         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
7415         arguments order in builtin.
7416         (_mm512_permutexvar_epi64): Ditto.
7417         (_mm512_mask_permutexvar_epi64): Ditto
7418         (_mm512_maskz_permutexvar_epi32): Ditto
7419         (_mm512_permutexvar_epi32): Ditto
7420         (_mm512_mask_permutexvar_epi32): Ditto
7421
7422 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7423
7424         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
7425         (p8_vmrgow): Likewise.
7426
7427 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7428
7429         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
7430         endian targets.
7431
7432 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
7433
7434         PR target/60203
7435         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
7436         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
7437         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
7438         using direct move instructions on ISA 2.07.  Also adjust
7439         instruction length for 64-bit.
7440         (mov<mode>_64bit, TFmode/TDmode): Likewise.
7441         (mov<mode>_32bit, TFmode/TDmode): Likewise.
7442
7443 2014-02-15  Alan Modra  <amodra@gmail.com>
7444
7445         PR target/58675
7446         PR target/57935
7447         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
7448         find_replacement on parts of insn rtl that might be reloaded.
7449
7450 2014-02-15  Richard Biener  <rguenther@suse.de>
7451
7452         PR tree-optimization/60183
7453         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
7454         (tree_ssa_phiprop): Calculate and free post-dominators.
7455
7456 2014-02-14  Jeff Law  <law@redhat.com>
7457
7458         PR rtl-optimization/60131
7459         * ree.c (get_extended_src_reg): New function.
7460         (combine_reaching_defs): Use it rather than assuming location of REG.
7461         (find_and_remove_re): Verify first operand of extension is
7462         a REG before adding the insns to the copy list.
7463
7464 2014-02-14  Roland McGrath  <mcgrathr@google.com>
7465
7466         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
7467         * configure: Regenerated.
7468         * config.in: Regenerated.
7469         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
7470         instead of ASM_SHORT.
7471
7472 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
7473             Richard Earnshaw  <rearnsha@arm.com>
7474
7475         PR rtl-optimization/59535
7476         * lra-constraints.c (process_alt_operands): Encourage alternative
7477         when unassigned pseudo class is superset of the alternative class.
7478         (inherit_reload_reg): Don't inherit when optimizing for code size.
7479         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
7480         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
7481         modes not less than 4 for Thumb1.
7482
7483 2014-02-14  Kyle McMartin  <kyle@redhat.com>
7484
7485         PR pch/60010
7486         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
7487
7488 2014-02-14  Richard Biener  <rguenther@suse.de>
7489
7490         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
7491         (get_frame_arg): Drop the assert with langhook types_compatible_p.
7492         Do not strip INDIRECT_REFs.
7493
7494 2014-02-14  Richard Biener  <rguenther@suse.de>
7495
7496         PR lto/60179
7497         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
7498         DECL_FUNCTION_SPECIFIC_TARGET.
7499         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
7500         * tree-streamer-out.c (pack_ts_target_option): Remove.
7501         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
7502         (write_ts_function_decl_tree_pointers): Do not stream
7503         DECL_FUNCTION_SPECIFIC_TARGET.
7504         * tree-streamer-in.c (unpack_ts_target_option): Remove.
7505         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
7506         (lto_input_ts_function_decl_tree_pointers): Do not stream
7507         DECL_FUNCTION_SPECIFIC_TARGET.
7508
7509 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
7510
7511         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
7512         (get_initial_def_for_induction, vectorizable_induction): Ignore
7513         debug stmts when looking for exit_phi.
7514         (vectorizable_live_operation): Fix up condition.
7515
7516 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7517
7518         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
7519         nreverse() because it changes the content of original tree list.
7520
7521 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7522
7523         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
7524         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
7525
7526 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7527
7528         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
7529         GNU coding standards.
7530
7531 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7532
7533         PR debug/60152
7534         * dwarf2out.c (gen_subprogram_die): Don't call
7535         add_calling_convention_attribute if subr_die is old_die.
7536
7537 2014-02-13  Sharad Singhai  <singhai@google.com>
7538
7539         * doc/optinfo.texi: Fix order of nodes.
7540
7541 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
7542
7543         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
7544         operands[2], not operands[3].
7545
7546 2014-02-13  Richard Biener  <rguenther@suse.de>
7547
7548         PR bootstrap/59878
7549         * doc/install.texi (ISL): Update recommended version to 0.12.2,
7550         mention the possibility of an in-tree build.
7551         (CLooG): Update recommended version to 0.18.1, mention the
7552         possibility of an in-tree build and clarify that the ISL
7553         bundled with CLooG does not work.
7554
7555 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7556
7557         PR target/43546
7558         * expr.c (compress_float_constant): If x is a hard register,
7559         extend into a pseudo and then move to x.
7560
7561 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7562
7563         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
7564         caused by bad second argument to warning_at() with -mhotpatch and
7565         nested functions (e.g. with gfortran).
7566
7567 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
7568
7569         * opts.c (option_name): Remove "enabled by default" rider.
7570
7571 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
7572
7573         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
7574
7575 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
7576             Uros Bizjak  <ubizjak@gmail.com>
7577
7578         PR target/60151
7579         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
7580         * configure: Regenerated.
7581
7582 2014-02-12  Richard Biener  <rguenther@suse.de>
7583
7584         * vec.c (vec_prefix::calculate_allocation): Move as
7585         inline variant to vec.h.
7586         (vec_prefix::calculate_allocation_1): New out-of-line version.
7587         * vec.h (vec_prefix::calculate_allocation_1): Declare.
7588         (vec_prefix::m_has_auto_buf): Rename to ...
7589         (vec_prefix::m_using_auto_storage): ... this.
7590         (vec_prefix::calculate_allocation): Inline the easy cases
7591         and dispatch to calculate_allocation_1 which doesn't need the
7592         prefix address.
7593         (va_heap::reserve): Use gcc_checking_assert.
7594         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
7595         m_using_auto_storage.
7596         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
7597         member and adjust.
7598         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
7599         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
7600         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
7601
7602 2014-02-12  Richard Biener  <rguenther@suse.de>
7603
7604         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
7605         when we found a dependence.
7606
7607 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
7608
7609         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
7610         common code...
7611         (maybe_fold_stmt): ... into this new function.
7612         * omp-low.c (lower_omp): Update comment.
7613
7614         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
7615         last use.
7616
7617         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
7618         dereference.
7619
7620 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
7621
7622         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
7623         identifiers in comments.
7624         (cortexa53_extra_costs): Likewise.
7625         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
7626         (cortexa7_extra_costs): Likewise.
7627         (cortexa12_extra_costs): Likewise.
7628         (cortexa15_extra_costs): Likewise.
7629         (v7m_extra_costs): Likewise.
7630
7631 2014-02-12  Richard Biener  <rguenther@suse.de>
7632
7633         PR middle-end/60092
7634         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
7635         of posix_memalign being successful.
7636         (lower_stmt): Restrict lowering of posix_memalign to when
7637         -ftree-bit-ccp is enabled.
7638
7639 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
7640
7641         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
7642         arg_loc.
7643         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
7644
7645 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
7646
7647         PR rtl-optimization/60116
7648         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
7649         other_insn once the combination has been validated.
7650
7651 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
7652
7653         PR lto/59468
7654         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
7655         and wrapper.
7656         * ipa-devirt.c: Include demangle.h
7657         (odr_violation_reported): New static variable.
7658         (add_type_duplicate): Update odr_violations.
7659         (maybe_record_node): Add completep parameter; update it.
7660         (record_target_from_binfo): Add COMPLETEP parameter;
7661         update it as needed.
7662         (possible_polymorphic_call_targets_1): Likewise.
7663         (struct polymorphic_call_target_d): Add nonconstruction_targets;
7664         rename FINAL to COMPLETE.
7665         (record_targets_from_bases): Sanity check we found the binfo;
7666         fix COMPLETEP updating.
7667         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
7668         parameter, fix computing of COMPLETEP.
7669         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
7670         at LTO time do demangling.
7671         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
7672         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
7673         parameter.
7674         (gimple_get_virt_method_for_binfo): Likewise.
7675         * gimple-fold.h (gimple_get_virt_method_for_binfo,
7676         gimple_get_virt_method_for_vtable): Update prototypes.
7677
7678 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
7679
7680         PR target/49008
7681         * genautomata.c (add_presence_absence): Fix typo with
7682         {final_}presence_list.
7683
7684 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7685
7686         PR target/60137
7687         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
7688         for VSX/Altivec vectors that land in GPR registers.
7689
7690 2014-02-11  Richard Henderson  <rth@redhat.com>
7691             Jakub Jelinek  <jakub@redhat.com>
7692
7693         PR debug/59776
7694         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
7695         around drhs if type conversion to lacc->type is not useless.
7696
7697 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7698
7699         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
7700         tuning struct.
7701         (cortex-a57.cortex-a53): Likewise.
7702         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
7703
7704 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7705
7706         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
7707         arm_restrict_it.
7708
7709 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
7710
7711         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
7712         add_options_for_arm_vfp3.
7713
7714 2014-02-11  Jeff Law  <law@redhat.com>
7715
7716         PR middle-end/54041
7717         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
7718         object with an undesirable mode.
7719
7720 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7721
7722         PR libgomp/60107
7723         * config/i386/sol2-9.h: New file.
7724         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
7725         *-*-solaris2.9*): Use it.
7726
7727 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7728
7729         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
7730         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
7731
7732 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7733
7734         * config/microblaze/microblaze.c: Extend mcpu version format
7735
7736 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
7737
7738         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
7739
7740 2014-02-10  Richard Henderson  <rth@redhat.com>
7741
7742         PR target/59927
7743         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
7744         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
7745         ms-abi vs -mno-accumulate-outgoing-args.
7746         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
7747         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
7748         respect to ms-abi.
7749
7750 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7751
7752         PR middle-end/60080
7753         * cfgexpand.c (expand_asm_operands): Attach source location to
7754         ASM_INPUT rtx objects.
7755         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
7756
7757 2014-02-10  Nick Clifton  <nickc@redhat.com>
7758
7759         * config/mn10300/mn10300.c (popcount): New function.
7760         (mn10300_expand_prologue): Include saved registers in stack usage
7761         count.
7762
7763 2014-02-10  Jeff Law  <law@redhat.com>
7764
7765         PR middle-end/52306
7766         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
7767         when changing the SET_DEST of a prior insn to avoid an input reload.
7768
7769 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7770
7771         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
7772         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
7773         -mcall-openbsd, or -mcall-linux.
7774         (CC1_ENDIAN_BIG_SPEC): Remove.
7775         (CC1_ENDIAN_LITTLE_SPEC): Remove.
7776         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7777         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
7778         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
7779         and %cc1_endian_default.
7780         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7781
7782 2014-02-10  Richard Biener  <rguenther@suse.de>
7783
7784         PR tree-optimization/60115
7785         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
7786         MEM_REF handling.  Properly verify that the accesses are not
7787         out of the objects bound.
7788
7789 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7790
7791         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
7792         coretex to cortex.
7793
7794 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
7795
7796         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
7797         proper constants and fix formatting.
7798         (possible_polymorphic_call_targets): Fix formatting.
7799
7800 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
7801             Ilya Tocar  <ilya.tocar@intel.com>
7802
7803         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
7804         (_mm512_loadu_epi32): Renamed into...
7805         (_mm512_loadu_si512): This.
7806         (_mm512_storeu_epi32): Renamed into...
7807         (_mm512_storeu_si512): This.
7808         (_mm512_maskz_ceil_ps): Removed.
7809         (_mm512_maskz_ceil_pd): Ditto.
7810         (_mm512_maskz_floor_ps): Ditto.
7811         (_mm512_maskz_floor_pd): Ditto.
7812         (_mm512_floor_round_ps): Ditto.
7813         (_mm512_floor_round_pd): Ditto.
7814         (_mm512_ceil_round_ps): Ditto.
7815         (_mm512_ceil_round_pd): Ditto.
7816         (_mm512_mask_floor_round_ps): Ditto.
7817         (_mm512_mask_floor_round_pd): Ditto.
7818         (_mm512_mask_ceil_round_ps): Ditto.
7819         (_mm512_mask_ceil_round_pd): Ditto.
7820         (_mm512_maskz_floor_round_ps): Ditto.
7821         (_mm512_maskz_floor_round_pd): Ditto.
7822         (_mm512_maskz_ceil_round_ps): Ditto.
7823         (_mm512_maskz_ceil_round_pd): Ditto.
7824         (_mm512_expand_pd): Ditto.
7825         (_mm512_expand_ps): Ditto.
7826         * config/i386/i386.c (ix86_builtins): Remove
7827         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
7828         (bdesc_args): Ditto.
7829         * config/i386/predicates.md (const1256_operand): New.
7830         (const_1_to_2_operand): Ditto.
7831         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
7832         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
7833         (*avx512pf_gatherpf<mode>sf): Ditto.
7834         (avx512pf_gatherpf<mode>df): Ditto.
7835         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7836         (*avx512pf_gatherpf<mode>df): Ditto.
7837         (avx512pf_scatterpf<mode>sf): Ditto.
7838         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7839         (*avx512pf_scatterpf<mode>sf): Ditto.
7840         (avx512pf_scatterpf<mode>df): Ditto.
7841         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7842         (*avx512pf_scatterpf<mode>df): Ditto.
7843         (avx512f_expand<mode>): Removed.
7844         (<shift_insn><mode>3<mask_name>): Change predicate type.
7845
7846 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7847
7848         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
7849         not at the end of datarefs vector use ordered_remove to avoid
7850         reordering datarefs vector.
7851
7852         PR c/59984
7853         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
7854         mark local addressable non-static vars as GOVD_PRIVATE
7855         instead of GOVD_LOCAL.
7856         * omp-low.c (lower_omp_for): Move gimple_bind_vars
7857         and BLOCK_VARS of gimple_bind_block to new_stmt rather
7858         than copying them.
7859
7860         PR middle-end/60092
7861         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
7862         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
7863         assume_aligned or alloc_align attributes.
7864         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
7865         arguments.  Handle also assume_aligned and alloc_align attributes.
7866         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
7867         calls to functions with assume_aligned or alloc_align attributes.
7868         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
7869
7870 2014-02-08  Terry Guo  <terry.guo@arm.com>
7871
7872         * doc/invoke.texi: Document ARM -march=armv7e-m.
7873
7874 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7875
7876         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
7877         flag on __cilkrts_rethrow builtin.
7878
7879         PR ipa/60026
7880         * ipa-cp.c (determine_versionability): Fail at -O0
7881         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
7882         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
7883
7884         Revert:
7885         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7886
7887         PR ipa/60026
7888         * tree-inline.c (copy_forbidden): Fail for
7889         __attribute__((optimize (0))) functions.
7890
7891 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7892
7893         * varpool.c: Include pointer-set.h.
7894         (varpool_remove_unreferenced_decls): Variables in other partitions
7895         will not be output; be however careful to not lose information
7896         about partitioning.
7897
7898 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7899
7900         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
7901         lookup in the vtable constructor.
7902
7903 2014-02-07  Jeff Law  <law@redhat.com>
7904
7905         PR target/40977
7906         * config/m68k/m68k.md (ashldi_extsi): Turn into a
7907         define_insn_and_split.
7908
7909         * ipa-inline.c (inline_small_functions): Fix typos.
7910
7911 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7912
7913         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
7914         (s390_can_use_return_insn): Declare.
7915         * config/s390/s390.h (EPILOGUE_USES): Define.
7916         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
7917         instructions.
7918         (s390_chunkify_start): Handle return JUMP_LABELs.
7919         (s390_early_mach): Emit a main_pool instruction on the entry edge.
7920         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
7921         (s390_can_use_return_insn): New functions.
7922         (s390_fix_long_loop_prediction): Handle conditional returns.
7923         (TARGET_SET_UP_BY_PROLOGUE): Define.
7924         * config/s390/s390.md (ANY_RETURN): New code iterator.
7925         (*creturn, *csimple_return, return, simple_return): New patterns.
7926
7927 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7928
7929         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
7930         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
7931         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
7932         REG_CFA_RESTORE list when deciding not to restore a register.
7933
7934 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7935
7936         * config/s390/s390.c: Include tree-pass.h and context.h.
7937         (s390_early_mach): New function, split out from...
7938         (s390_emit_prologue): ...here.
7939         (pass_data_s390_early_mach): New pass structure.
7940         (pass_s390_early_mach): New class.
7941         (s390_option_override): Create and register early_mach pass.
7942         Move to end of file.
7943
7944 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7945
7946         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
7947         to match for the exit block.
7948
7949 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7950
7951         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
7952         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
7953         Reject misaligned operands.
7954
7955 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7956
7957         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
7958
7959 2014-02-07  Richard Biener  <rguenther@suse.de>
7960
7961         PR middle-end/60092
7962         * gimple-low.c (lower_builtin_posix_memalign): New function.
7963         (lower_stmt): Call it to lower posix_memalign in a way
7964         to make alignment info accessible.
7965
7966 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
7967
7968         PR c++/60082
7969         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
7970         __builtin_setjmp_receiver.
7971
7972 2014-02-07  Richard Biener  <rguenther@suse.de>
7973
7974         PR middle-end/60092
7975         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
7976         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
7977         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
7978         Handle BUILT_IN_POSIX_MEMALIGN.
7979         (find_func_clobbers): Likewise.
7980         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7981         (call_may_clobber_ref_p_1): Likewise.
7982
7983 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7984
7985         PR ipa/59918
7986         * ipa-devirt.c (record_target_from_binfo): Remove overactive
7987         sanity check.
7988
7989 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7990
7991         PR ipa/59469
7992         * lto-cgraph.c (lto_output_node): Use
7993         symtab_get_symbol_partitioning_class.
7994         (lto_output_varpool_node): likewise.
7995         (symtab_get_symbol_partitioning_class): Move here from
7996         lto/lto-partition.c
7997         * cgraph.h (symbol_partitioning_class): Likewise.
7998         (symtab_get_symbol_partitioning_class): Declare.
7999
8000 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8001
8002         * ggc.h (ggc_internal_cleared_alloc): New macro.
8003         * vec.h (vec_safe_copy): Handle memory stats.
8004         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
8005         * target-globals.c (save_target_globals): Likewise.
8006
8007 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8008
8009         PR target/60077
8010         * expr.c (emit_move_resolve_push): Export; be bit more selective
8011         on when to clear alias set.
8012         * expr.h (emit_move_resolve_push): Declare.
8013         * function.h (struct function): Add tail_call_marked.
8014         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
8015         * config/i386/i386-protos.h (ix86_expand_push): Remove.
8016         * config/i386/i386.md (TImode move expander): De not call
8017         ix86_expand_push.
8018         (FP push expanders): Preserve memory attributes.
8019         * config/i386/sse.md (push<mode>1): Remove.
8020         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
8021         (ix86_expand_push): Remove.
8022         * config/i386/mmx.md (push<mode>1): Remove.
8023
8024 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8025
8026         PR rtl-optimization/60030
8027         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
8028         lopart with paradoxical subreg before shifting it up by hprec.
8029
8030 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8031
8032         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
8033         Remove extra newline at end of file.
8034         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
8035         (arm_issue_rate): Handle cortexa57.
8036         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
8037         (cortex-a57.cortex-a53): Likewise.
8038
8039 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8040
8041         PR target/59575
8042         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
8043         don't record in REG_FRAME_RELATED_EXPR registers not set in that
8044         bitmask.
8045         (arm_expand_prologue): Adjust all callers.
8046         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
8047         info, registers also at the lowest numbered registers side.  Use
8048         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
8049         XEXP.
8050
8051         PR debug/59992
8052         * var-tracking.c (adjust_mems): Before adding a SET to
8053         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
8054
8055 2014-02-06  Alan Modra  <amodra@gmail.com>
8056
8057         PR target/60032
8058         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
8059         change SDmode to DDmode when lra_in_progress.
8060
8061 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8062
8063         PR middle-end/59150
8064         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
8065         free_data_ref on the dr first, and before goto again also set dr
8066         to the next dr.  For simd_lane_access, free old datarefs[i] before
8067         overwriting it.  For get_vectype_for_scalar_type failure, don't
8068         free_data_ref if simd_lane_access.
8069
8070         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
8071
8072         PR target/60062
8073         * tree.h (opts_for_fn): New inline function.
8074         (opt_for_fn): Define.
8075         * config/i386/i386.c (ix86_function_regparm): Use
8076         opt_for_fn (decl, optimize) instead of optimize.
8077
8078 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8079
8080         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
8081         for SYMBOL_REF in large memory model.
8082
8083 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8084
8085         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
8086         and crypto support.
8087         (cortex-a57): Likewise.
8088         (cortex-a57.cortex-a53): Likewise.
8089
8090 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
8091             Kugan Vivekanandarajah  <kuganv@linaro.org>
8092
8093         * config/arm/arm.c (arm_vector_alignment_reachable): Check
8094         unaligned_access.
8095         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
8096
8097 2014-02-06  Richard Biener  <rguenther@suse.de>
8098
8099         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
8100         set_loop_copy and initialize_original_copy_tables.
8101
8102 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
8103
8104         * config/aarch64/aarch64-simd.md
8105         (aarch64_ashr_simddi): Change QI to SI.
8106
8107 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
8108             Jakub Jelinek  <jakub@redhat.com>
8109
8110         PR middle-end/60013
8111         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
8112         of the dataflow.
8113
8114 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8115
8116         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
8117         CODE_FOR_altivec_vpku[hw]um to
8118         CODE_FOR_altivec_vpku[hw]um_direct.
8119         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
8120         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
8121         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
8122         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
8123
8124 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8125
8126         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
8127         generation for -maltivec=be.
8128         (altivec_vsumsws): Simplify redundant test.
8129
8130 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8131
8132         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
8133         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
8134         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
8135         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
8136         gen_altivec_vpkuwum.
8137         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
8138         BYTES_BIG_ENDIAN.
8139         (altivec_vpks<VI_char>ss): Likewise.
8140         (altivec_vpks<VI_char>us): Likewise.
8141         (altivec_vpku<VI_char>us): Likewise.
8142         (altivec_vpku<VI_char>um): Likewise.
8143         (altivec_vpku<VI_char>um_direct): New (copy of
8144         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
8145         internal use).
8146         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
8147         target is little endian and -maltivec=be is not specified.
8148         (*altivec_vupkhs<VU_char>_direct): New (copy of
8149         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
8150         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
8151         target is little endian and -maltivec=be is not specified.
8152         (*altivec_vupkls<VU_char>_direct): New (copy of
8153         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
8154         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
8155         little endian and -maltivec=be is not specified.
8156         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
8157         little endian and -maltivec=be is not specified.
8158
8159 2014-02-05  Richard Henderson  <rth@redhat.com>
8160
8161         PR debug/52727
8162         * combine-stack-adj.c: Revert r206943.
8163         * sched-int.h (struct deps_desc): Add last_args_size.
8164         * sched-deps.c (init_deps): Initialize it.
8165         (sched_analyze_insn): Add OUTPUT dependencies between insns that
8166         contain REG_ARGS_SIZE notes.
8167
8168 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
8169
8170         * lto-cgraph.c (asm_nodes_output): Make global.
8171         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
8172         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
8173         (driver_handle_option): Handle OPT_fwpa.
8174
8175 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
8176
8177         PR ipa/59947
8178         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
8179         a comment typo and formatting issue.  If odr_hash hasn't been
8180         created, return vNULL and set *completep to false.
8181
8182         PR middle-end/57499
8183         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
8184         bb with no successors.
8185
8186 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
8187
8188         PR target/59718
8189         * doc/invoke.texi (-march): Clarify documentation for ARM.
8190         (-mtune): Likewise.
8191         (-mcpu): Likewise.
8192
8193 2014-02-05  Richard Biener  <rguenther@suse.de>
8194
8195         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
8196         when not vectorizing because of too many alias checks.
8197         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
8198         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
8199
8200 2014-02-05  Nick Clifton  <nickc@redhat.com>
8201
8202         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
8203         accept extended registers in any mode when compiling for the MN10300.
8204
8205 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
8206
8207         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
8208         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
8209         sanitization attributes.
8210         (can_inline_edge_p): Likewise.
8211         (sanitize_attrs_match_for_inline_p): New function.
8212
8213 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8214
8215         * ipa-prop.c (detect_type_change): Shor circuit testing of
8216         type changes on THIS pointer.
8217
8218 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
8219
8220         PR target/59777
8221         * config/pa/pa.c (legitimize_tls_address): Return original address
8222         if not passed a SYMBOL_REF rtx.
8223         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
8224         addresses.
8225         (pa_emit_move_sequence): Simplify TLS source operands.
8226         (pa_legitimate_constant_p): Reject all TLS constants.
8227         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
8228         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
8229
8230 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8231
8232         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
8233         groups when we know they are controlled by LTO.
8234         * varasm.c (default_binds_local_p_1): If object is in other partition,
8235         it will be resolved locally.
8236
8237 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8238
8239         * config/host-linux.c (linux_gt_pch_use_address): Don't
8240         use SSIZE_MAX because it is not always defined.
8241
8242 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
8243
8244         PR bootstrap/59913
8245         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
8246         threshold for pseudo splitting.
8247         (update_ebb_live_info): Process call argument hard registers and
8248         hard registers from insn definition too.
8249         (max_small_class_regs_num): New constant.
8250         (inherit_in_ebb): Update live hard regs through EBBs.  Update
8251         reloads_num only for small register classes.  Don't split for
8252         outputs of jumps.
8253
8254 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
8255
8256         PR ipa/60058
8257         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
8258         is non-null.
8259
8260 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8261
8262         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
8263         visibility is safe.
8264
8265 2014-02-04  Marek Polacek  <polacek@redhat.com>
8266
8267         * gdbinit.in (pel): Define.
8268
8269 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8270
8271         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
8272         behavior.
8273
8274 2014-02-04  Richard Biener  <rguenther@suse.de>
8275
8276         PR lto/59723
8277         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
8278         in function context local.
8279         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
8280         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
8281         similar to LTO_imported_decl_ref.
8282
8283 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
8284
8285         PR tree-optimization/60002
8286         * cgraphclones.c (build_function_decl_skip_args): Clear
8287         DECL_LANG_SPECIFIC.
8288
8289         PR tree-optimization/60023
8290         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
8291         false to gsi_replace.
8292         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
8293         has been in some EH region and vec_stmt could throw, add
8294         vec_stmt into the same EH region.
8295         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
8296         has no lhs, ignore it.
8297         * internal-fn.c (expand_MASK_LOAD): Likewise.
8298
8299         PR ipa/60026
8300         * tree-inline.c (copy_forbidden): Fail for
8301         __attribute__((optimize (0))) functions.
8302
8303         PR other/58712
8304         * omp-low.c (simd_clone_struct_copy): If from->inbranch
8305         is set, copy one less argument.
8306         (expand_simd_clones): Don't subtract clone_info->inbranch
8307         from simd_clone_struct_alloc argument.
8308
8309         PR rtl-optimization/57915
8310         * recog.c (simplify_while_replacing): If all unary/binary/relational
8311         operation arguments are constant, attempt to simplify those.
8312
8313         PR middle-end/59261
8314         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
8315         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
8316
8317 2014-02-04  Richard Biener  <rguenther@suse.de>
8318
8319         PR tree-optimization/60012
8320         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
8321         TBAA disambiguation to all DDRs.
8322
8323 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8324
8325         PR target/59788
8326         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
8327         (LINK_SPEC): Use it for -shared, -shared-libgcc.
8328
8329 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8330
8331         PR ipa/59882
8332         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
8333
8334 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8335
8336         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
8337         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
8338
8339 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8340
8341         PR ipa/59831
8342         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
8343         to figure out targets of polymorphic calls with known decl.
8344         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8345         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
8346         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
8347         (get_polymorphic_call_info): ... here.
8348         (get_polymorphic_call_info_from_invariant): New function.
8349
8350 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8351
8352         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
8353         lookup via vtable pointer; check for type consistency
8354         and turn inconsitent facts into UNREACHABLE.
8355         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8356         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
8357         type inconsistent querries; return UNREACHABLE instead.
8358
8359 2014-02-03  Richard Henderson  <rth@twiddle.net>
8360
8361         PR tree-opt/59924
8362         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
8363         already processed this node.
8364         (normalize_one_pred_1): Pass along mark_set.
8365         (normalize_one_pred): Create and destroy a pointer_set_t.
8366         (normalize_one_pred_chain): Likewise.
8367
8368 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
8369
8370         PR gcov-profile/58602
8371         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
8372
8373 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8374
8375         PR ipa/59831
8376         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
8377         -fno-devirtualize; try to devirtualize by the knowledge of
8378         virtual table pointer given by aggregate propagation.
8379         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8380         (ipa_print_node_jump_functions): Dump also offset that
8381         is relevant for polymorphic calls.
8382         (determine_known_aggregate_parts): Add arg_type parameter; use it
8383         instead of determining the type from pointer type.
8384         (ipa_compute_jump_functions_for_edge): Update call of
8385         determine_known_aggregate_parts.
8386         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
8387         (gimple_get_virt_method_for_binfo): ... here; simplify using
8388         vtable_pointer_value_to_vtable.
8389         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
8390         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
8391         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
8392         (vtable_pointer_value_to_vtable): Break out from ...; handle also
8393         POINTER_PLUS_EXPR.
8394         (vtable_pointer_value_to_binfo): ... here.
8395         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
8396
8397 2014-02-03  Teresa Johnson  <tejohnson@google.com>
8398
8399         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
8400         redef of outer loop index variable.
8401
8402 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
8403
8404         PR c++/53017
8405         PR c++/59211
8406         * doc/extend.texi (Function Attributes): Typo.
8407
8408 2014-02-03  Cong Hou  <congh@google.com>
8409
8410         PR tree-optimization/60000
8411         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
8412         if the vectorized statement is a store.  A store statement can only
8413         appear at the end of pattern statements.
8414
8415 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
8416
8417         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
8418         (ix86_option_override_internal): Default long double to 64-bit for
8419         32-bit Bionic and to 128-bit for 64-bit Bionic.
8420
8421         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
8422         TARGET_LONG_DOUBLE_128 is true.
8423         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
8424
8425         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
8426         (mlong-double-64): Negate -mlong-double-128.
8427         (mlong-double-128): New option.
8428
8429         * config/i386/i386-c.c (ix86_target_macros): Define
8430         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
8431
8432         * doc/invoke.texi: Document -mlong-double-128.
8433
8434 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
8435
8436         PR rtl-optimization/60024
8437         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
8438
8439 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
8440
8441         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
8442
8443 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8444
8445         PR rtl-optimization/57662
8446         * sel-sched.c (code_motion_path_driver): Do not mark already not
8447         existing blocks in the visiting bitmap.
8448
8449 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8450
8451         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
8452         on the insn being emitted.
8453
8454 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
8455             Will Deacon  <will.deacon@arm.com>
8456
8457         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
8458
8459 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8460
8461         * config/arm/arm-tables.opt: Regenerate.
8462
8463 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8464
8465         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
8466         for vector types other than V16QImode.
8467         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
8468         define_expand, and call altivec_expand_vec_perm_le when producing
8469         code with little endian element order.
8470         (*altivec_vperm_<mode>_internal): New insn having previous
8471         behavior of altivec_vperm_<mode>.
8472         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
8473         altivec_expand_vec_perm_le when producing code with little endian
8474         element order.
8475         (*altivec_vperm_<mode>_uns_internal): New insn having previous
8476         behavior of altivec_vperm_<mode>_uns.
8477
8478 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8479
8480         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
8481         (altivec_vsumsws): Add handling for -maltivec=be with a little
8482         endian target.
8483         (altivec_vsumsws_direct): New.
8484         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
8485         gen_altivec_vsumsws.
8486
8487 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8488
8489         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
8490         vtable_pointer_value_to_binfo): New functions.
8491         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
8492         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
8493
8494 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
8495
8496         * config/nios2/nios2.md (load_got_register): Initialize GOT
8497         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
8498         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
8499
8500 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8501
8502         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
8503         preserverd by passthrough, do not propagate the type.
8504
8505 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8506
8507         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
8508         (mips_atomic_assign_expand_fenv): New function.
8509         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
8510
8511 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8512
8513         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
8514         (__builtin_mips_set_fcsr): Likewise.
8515         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
8516         MIPS_USI_FTYPE_VOID.
8517         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
8518         (mips16_expand_set_fcsr): Likewise.
8519         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
8520         (mips16_set_fcsr_stub): Likewise.
8521         (mips16_get_fcsr_one_only_stub): New class.
8522         (mips16_set_fcsr_one_only_stub): Likewise.
8523         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
8524         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
8525         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
8526         (hard_float): New availability predicate.
8527         (mips_builtins): Add get_fcsr and set_fcsr.
8528         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
8529         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
8530         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
8531         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
8532         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
8533         patterns.
8534
8535 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8536
8537         * config/mips/mips.c (mips_one_only_stub): New class.
8538         (mips_need_mips16_rdhwr_p): Replace with...
8539         (mips16_rdhwr_stub): ...this new variable.
8540         (mips16_stub_call_address): New function.
8541         (mips16_rdhwr_one_only_stub): New class.
8542         (mips_expand_thread_pointer): Use mips16_stub_call_address.
8543         (mips_output_mips16_rdhwr): Delete.
8544         (mips_finish_stub): New function.
8545         (mips_code_end): Use it to handle rdhwr stubs.
8546
8547 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
8548
8549         PR target/60017
8550         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
8551         when calculating size of integer atomic types.
8552
8553 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
8554
8555         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
8556
8557 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
8558
8559         PR tree-optimization/60003
8560         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
8561         * profile.c (branch_prob): Use gimple_call_builtin_p
8562         to check for BUILT_IN_SETJMP_RECEIVER.
8563         * tree-inline.c (copy_bb): Call notice_special_calls.
8564
8565 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
8566
8567         PR bootstrap/59985
8568         * lra-constraints.c (process_alt_operands): Update reload_sum only
8569         on the first pass.
8570
8571 2014-01-31  Richard Henderson  <rth@redhat.com>
8572
8573         PR middle-end/60004
8574         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
8575         until after else_eh is processed.
8576
8577 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8578
8579         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
8580         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
8581         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
8582         in smmintrin.h, remove them.
8583         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
8584         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
8585         * config/i386/i386.md (ROUND_SAE): Fix value.
8586         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
8587         (const48_operand): New.
8588         * config/i386/subst.md (round), (round_expand): Use
8589         const_4_or_8_to_11_operand.
8590         (round_saeonly), (round_saeonly_expand): Use const48_operand.
8591
8592 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8593
8594         * config/i386/constraints.md (Yk): Swap meaning with k.
8595         * config/i386/i386.md (movhi_internal): Change Yk to k.
8596         (movqi_internal): Ditto.
8597         (*k<logic><mode>): Ditto.
8598         (*andhi_1): Ditto.
8599         (*andqi_1): Ditto.
8600         (kandn<mode>): Ditto.
8601         (*<code>hi_1): Ditto.
8602         (*<code>qi_1): Ditto.
8603         (kxnor<mode>): Ditto.
8604         (kortestzhi): Ditto.
8605         (kortestchi): Ditto.
8606         (kunpckhi): Ditto.
8607         (*one_cmplhi2_1): Ditto.
8608         (*one_cmplqi2_1): Ditto.
8609         * config/i386/sse.md (): Change k to Yk.
8610         (avx512f_load<mode>_mask): Ditto.
8611         (avx512f_blendm<mode>): Ditto.
8612         (avx512f_store<mode>_mask): Ditto.
8613         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
8614         (avx512f_storedqu<mode>_mask): Ditto.
8615         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
8616         Ditto.
8617         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
8618         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
8619         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
8620         (avx512f_maskcmp<mode>3): Ditto.
8621         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
8622         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
8623         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
8624         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
8625         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
8626         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
8627         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
8628         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
8629         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
8630         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
8631         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
8632         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
8633         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
8634         (vec_extract_lo_<mode>_maskm): Ditto.
8635         (vec_extract_hi_<mode>_maskm): Ditto.
8636         (avx512f_vternlog<mode>_mask): Ditto.
8637         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
8638         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
8639         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
8640         (avx512f_<code>v8div16qi2_mask): Ditto.
8641         (avx512f_<code>v8div16qi2_mask_store): Ditto.
8642         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
8643         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
8644         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
8645         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
8646         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
8647         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8648         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
8649         (*avx512pf_scatterpf<mode>df_mask): Ditto.
8650         (avx512cd_maskb_vec_dupv8di): Ditto.
8651         (avx512cd_maskw_vec_dupv16si): Ditto.
8652         (avx512f_vpermi2var<mode>3_maskz): Ditto.
8653         (avx512f_vpermi2var<mode>3_mask): Ditto.
8654         (avx512f_vpermi2var<mode>3_mask): Ditto.
8655         (avx512f_vpermt2var<mode>3_maskz): Ditto.
8656         (*avx512f_gathersi<mode>): Ditto.
8657         (*avx512f_gathersi<mode>_2): Ditto.
8658         (*avx512f_gatherdi<mode>): Ditto.
8659         (*avx512f_gatherdi<mode>_2): Ditto.
8660         (*avx512f_scattersi<mode>): Ditto.
8661         (*avx512f_scatterdi<mode>): Ditto.
8662         (avx512f_compress<mode>_mask): Ditto.
8663         (avx512f_compressstore<mode>_mask): Ditto.
8664         (avx512f_expand<mode>_mask): Ditto.
8665         * config/i386/subst.md (mask): Change k to Yk.
8666         (mask_scalar_merge): Ditto.
8667         (sd): Ditto.
8668
8669 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
8670
8671         * doc/extend.texi (Vector Extensions): Document ?: in C++.
8672
8673 2014-01-31  Richard Biener  <rguenther@suse.de>
8674
8675         PR middle-end/59990
8676         * builtins.c (fold_builtin_memory_op): Make sure to not
8677         use a floating-point mode or a boolean or enumeral type for
8678         the copy operation.
8679
8680 2014-01-30  DJ Delorie  <dj@redhat.com>
8681
8682         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
8683         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
8684         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
8685         whenever main() has an epilogue.
8686
8687 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8688
8689         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
8690         unused variable "field".
8691         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
8692         (vsx_mergeh_<mode>): Likewise.
8693         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
8694         (altivec_vmrghh): Likewise.
8695         (altivec_vmrghw): Likewise.
8696         (altivec_vmrglb): Likewise.
8697         (altivec_vmrglh): Likewise.
8698         (altivec_vmrglw): Likewise.
8699         (altivec_vspltb): Add missing uses.
8700         (altivec_vsplth): Likewise.
8701         (altivec_vspltw): Likewise.
8702         (altivec_vspltsf): Likewise.
8703
8704 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8705
8706         PR target/59923
8707         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
8708         frame related instructions.
8709
8710 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
8711
8712         PR rtl-optimization/59959
8713         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
8714         any reload of register whose subreg is invalid.
8715
8716 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8717
8718         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
8719         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
8720         Add missing return type - void.
8721
8722 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8723
8724         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
8725         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
8726         remove element index adjustment for endian (now handled in vsx.md
8727         and altivec.md).
8728         (altivec_expand_vec_perm_const): Use
8729         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
8730         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
8731         (vsx_xxspltw_<mode>): Adjust element index for little endian.
8732         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
8733         define_expand and a new define_insn *altivec_vspltb_internal;
8734         adjust for -maltivec=be on a little endian target.
8735         (altivec_vspltb_direct): New.
8736         (altivec_vsplth): Divide into a define_expand and a new
8737         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
8738         little endian target.
8739         (altivec_vsplth_direct): New.
8740         (altivec_vspltw): Divide into a define_expand and a new
8741         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
8742         little endian target.
8743         (altivec_vspltw_direct): New.
8744         (altivec_vspltsf): Divide into a define_expand and a new
8745         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
8746         a little endian target.
8747
8748 2014-01-30  Richard Biener  <rguenther@suse.de>
8749
8750         PR tree-optimization/59993
8751         * tree-ssa-forwprop.c (associate_pointerplus): Check we
8752         can propagate form the earlier stmt and avoid the transform
8753         when the intermediate result is needed.
8754
8755 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
8756
8757         * README.Portability: Fix typo.
8758
8759 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
8760
8761         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
8762         comparison_operator with ordered_comparison_operator.
8763
8764 2014-01-30  Nick Clifton  <nickc@redhat.com>
8765
8766         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
8767         Rename to mn10300_store_multiple_regs.
8768         * config/mn10300/mn10300.c: Likewise.
8769         * config/mn10300/mn10300.md (store_movm): Fix typo: call
8770         store_multiple_regs.
8771         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
8772         Call mn10300_store_multiple_regs.
8773
8774 2014-01-30  Nick Clifton  <nickc@redhat.com>
8775             DJ Delorie  <dj@redhat.com>
8776
8777         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
8778         %fp 2 to keep registers after it properly word-aligned.
8779         (rl78_alloc_physical_registers_umul): Handle the case where both
8780         input operands are the same.
8781
8782 2014-01-30  Richard Biener  <rguenther@suse.de>
8783
8784         PR tree-optimization/59903
8785         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
8786         check properly.
8787
8788 2014-01-30  Jason Merrill  <jason@redhat.com>
8789
8790         PR c++/59633
8791         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
8792
8793         PR c++/59645
8794         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
8795
8796 2014-01-30  Richard Biener  <rguenther@suse.de>
8797
8798         PR tree-optimization/59951
8799         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
8800
8801 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
8802
8803         PR target/59784
8804         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
8805         SFmode to DFmode case.
8806
8807 2014-01-29  DJ Delorie  <dj@redhat.com>
8808
8809         * config/msp430/msp430.opt (-minrt): New.
8810         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
8811         if -minrt given.
8812         (ENDFILE_SPEC): Likewise.
8813
8814 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
8815
8816         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
8817         (estimate_function_body_sizes): Use it.
8818
8819 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
8820
8821         PR c++/58561
8822         * dwarf2out.c (is_cxx_auto): New.
8823         (is_base_type): Use it.
8824         (gen_type_die_with_usage): Likewise.
8825
8826 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8827
8828         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
8829         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
8830         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
8831         -maltivec=be with LE targets.
8832         (vsx_mergeh_<mode>): Likewise.
8833         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
8834         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
8835         (altivec_vmrghb): Replace with define_expand and new
8836         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
8837         (altivec_vmrghb_direct): New define_insn.
8838         (altivec_vmrghh): Replace with define_expand and new
8839         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
8840         (altivec_vmrghh_direct): New define_insn.
8841         (altivec_vmrghw): Replace with define_expand and new
8842         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
8843         (altivec_vmrghw_direct): New define_insn.
8844         (*altivec_vmrghsf): Adjust for endianness.
8845         (altivec_vmrglb): Replace with define_expand and new
8846         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
8847         (altivec_vmrglb_direct): New define_insn.
8848         (altivec_vmrglh): Replace with define_expand and new
8849         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
8850         (altivec_vmrglh_direct): New define_insn.
8851         (altivec_vmrglw): Replace with define_expand and new
8852         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
8853         (altivec_vmrglw_direct): New define_insn.
8854         (*altivec_vmrglsf): Adjust for endianness.
8855         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8856         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8857         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8858         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8859         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8860         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8861         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8862         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8863
8864 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8865
8866         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
8867         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
8868         whitespace.
8869
8870 2014-01-29  Richard Biener  <rguenther@suse.de>
8871
8872         PR tree-optimization/58742
8873         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
8874         associate_pointerplus_align.
8875         (associate_pointerplus_diff): New function.
8876         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
8877         and associate_pointerplus_diff.
8878
8879 2014-01-29  Richard Biener  <rguenther@suse.de>
8880
8881         * lto-streamer.h (LTO_major_version): Bump to 3.
8882         (LTO_minor_version): Reset to 0.
8883
8884 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
8885
8886         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
8887         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
8888         (arm_file_start): Generate correct asm header for armv7ve.
8889         * config/arm/bpabi.h: Add multilib support for armv7ve.
8890         * config/arm/driver-arm.c: Change the architectures of cortex-a7
8891         and cortex-a15 to armv7ve.
8892         * config/arm/t-aprofile: Add multilib support for armv7ve.
8893         * doc/invoke.texi: Document -march=armv7ve.
8894
8895 2014-01-29  Richard Biener  <rguenther@suse.de>
8896
8897         PR tree-optimization/58742
8898         * tree-ssa-forwprop.c (associate_plusminus): Return true
8899         if we changed sth, defer EH cleanup to ...
8900         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
8901         (simplify_mult): New function.
8902
8903 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8904
8905         PR middle-end/59917
8906         PR tree-optimization/59920
8907         * tree.c (build_common_builtin_nodes): Remove
8908         __builtin_setjmp_dispatcher initialization.
8909         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
8910         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
8911         instead of gsi_after_labels + manually skipping debug stmts.
8912         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
8913         ignore bbs with IFN_ABNORMAL_DISPATCHER.
8914         * tree-inline.c (copy_edges_for_bb): Remove
8915         can_make_abnormal_goto argument, instead add abnormal_goto_dest
8916         argument.  Ignore computed_goto_p stmts.  Don't call
8917         make_abnormal_goto_edges.  If a call might need abnormal edges
8918         for non-local gotos, see if it already has an edge to
8919         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
8920         with true argument, don't do anything then, otherwise add
8921         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
8922         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
8923         caller.
8924         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
8925         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
8926         (lower_stmt): Don't set data->calls_builtin_setjmp.
8927         (lower_builtin_setjmp): Adjust comment.
8928         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
8929         * tree-cfg.c (found_computed_goto): Remove.
8930         (factor_computed_gotos): Remove.
8931         (make_goto_expr_edges): Return bool, true for computed gotos.
8932         Don't call make_abnormal_goto_edges.
8933         (build_gimple_cfg): Don't set found_computed_goto, don't call
8934         factor_computed_gotos.
8935         (computed_goto_p): No longer static.
8936         (make_blocks): Don't set found_computed_goto.
8937         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
8938         (make_edges): If make_goto_expr_edges returns true, push bb
8939         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
8940         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
8941         vector.  Record mapping between bbs and OpenMP regions if there
8942         are any, adjust make_gimple_omp_edges caller.  Call
8943         handle_abnormal_edges.
8944         (make_abnormal_goto_edges): Remove.
8945         * tree-cfg.h (make_abnormal_goto_edges): Remove.
8946         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
8947         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
8948         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
8949         * internal-fn.def (ABNORMAL_DISPATCHER): New.
8950         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
8951         filling *region also set *region_idx to (*region)->entry->index.
8952
8953         PR other/58712
8954         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
8955         For REGs set ORIGINAL_REGNO.
8956
8957 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
8958
8959         * doc/md.texi: Mention that a target shouldn't implement
8960         vec_widen_(s|u)mul_even/odd pair if it is less efficient
8961         than hi/lo pair.
8962
8963 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8964
8965         PR tree-optimization/59594
8966         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
8967         a copy of the datarefs vector rather than the vector itself.
8968
8969 2014-01-28  Jason Merrill  <jason@redhat.com>
8970
8971         PR c++/53756
8972         * dwarf2out.c (auto_die): New static.
8973         (gen_type_die_with_usage): Handle C++1y 'auto'.
8974         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
8975         on definition.
8976
8977 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
8978
8979         PR target/59672
8980         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
8981         (SPEC_X32): Likewise.
8982         (SPEC_64): Likewise.
8983         * config/i386/i386.c (ix86_option_override_internal): Turn off
8984         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
8985         for TARGET_16BIT.
8986         (x86_file_start): Output .code16gcc for TARGET_16BIT.
8987         * config/i386/i386.h (TARGET_16BIT): New macro.
8988         (TARGET_16BIT_P): Likewise.
8989         * config/i386/i386.opt: Add m16.
8990         * doc/invoke.texi: Document -m16.
8991
8992 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
8993
8994         PR preprocessor/59935
8995         * input.c (location_get_source_line): Bail out on when line number
8996         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
8997
8998 2014-01-28  Richard Biener  <rguenther@suse.de>
8999
9000         PR tree-optimization/58742
9001         * tree-ssa-forwprop.c (associate_plusminus): Handle
9002         pointer subtraction of the form (T)(P + A) - (T)P.
9003
9004 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9005
9006         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
9007         at const_int_cost.
9008
9009 2014-01-28  Richard Biener  <rguenther@suse.de>
9010
9011         Revert
9012         2014-01-28  Richard Biener  <rguenther@suse.de>
9013
9014         PR rtl-optimization/45364
9015         PR rtl-optimization/59890
9016         * var-tracking.c (local_get_addr_clear_given_value): Handle
9017         already cleared slot.
9018         (val_reset): Handle not allocated local_get_addr_cache.
9019         (vt_find_locations): Use post-order on the inverted CFG.
9020
9021 2014-01-28  Richard Biener  <rguenther@suse.de>
9022
9023         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
9024
9025 2014-01-28  Richard Biener  <rguenther@suse.de>
9026
9027         PR rtl-optimization/45364
9028         PR rtl-optimization/59890
9029         * var-tracking.c (local_get_addr_clear_given_value): Handle
9030         already cleared slot.
9031         (val_reset): Handle not allocated local_get_addr_cache.
9032         (vt_find_locations): Use post-order on the inverted CFG.
9033
9034 2014-01-28  Alan Modra  <amodra@gmail.com>
9035
9036         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
9037         * configure.ac <recursive call for build != host>: Define
9038         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
9039         and LD_FOR_BUILD too.
9040         * configure: Regenerate.
9041
9042 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
9043
9044         * config/i386/i386.c (get_builtin_code_for_version): Separate
9045         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
9046         Broadwell from Haswell.
9047
9048 2014-01-27  Steve Ellcey  <sellcey@mips.com>
9049
9050         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
9051         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
9052         * config/mips/mips.c (mips_option_override): Change setting
9053         of TARGET_DSP.
9054         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
9055         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
9056         Change from Mask to Var.
9057
9058 2014-01-27  Jeff Law  <law@redhat.com>
9059
9060         * ipa-inline.c (inline_small_functions): Fix typo.
9061
9062 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
9063
9064         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
9065         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
9066         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
9067         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
9068         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
9069         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
9070         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
9071         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
9072         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
9073         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
9074         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
9075         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
9076         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
9077         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
9078         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
9079         (_mm512_storeu_epi64): Ditto.
9080         (_mm512_cmpge_epi32_mask): Ditto.
9081         (_mm512_cmpge_epu32_mask): Ditto.
9082         (_mm512_cmpge_epi64_mask): Ditto.
9083         (_mm512_cmpge_epu64_mask): Ditto.
9084         (_mm512_cmple_epi32_mask): Ditto.
9085         (_mm512_cmple_epu32_mask): Ditto.
9086         (_mm512_cmple_epi64_mask): Ditto.
9087         (_mm512_cmple_epu64_mask): Ditto.
9088         (_mm512_cmplt_epi32_mask): Ditto.
9089         (_mm512_cmplt_epu32_mask): Ditto.
9090         (_mm512_cmplt_epi64_mask): Ditto.
9091         (_mm512_cmplt_epu64_mask): Ditto.
9092         (_mm512_cmpneq_epi32_mask): Ditto.
9093         (_mm512_cmpneq_epu32_mask): Ditto.
9094         (_mm512_cmpneq_epi64_mask): Ditto.
9095         (_mm512_cmpneq_epu64_mask): Ditto.
9096         (_mm512_expand_pd): Ditto.
9097         (_mm512_expand_ps): Ditto.
9098         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
9099         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
9100         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
9101         * config/i386/i386.c (ix86_builtins): Add
9102         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
9103         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
9104         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
9105         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
9106         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
9107         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
9108         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
9109         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
9110         IX86_BUILTIN_PMOVUSQW512_MEM.
9111         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
9112         __builtin_ia32_pmovsqd512mem_mask,
9113         __builtin_ia32_pmovqd512mem_mask,
9114         __builtin_ia32_pmovusqw512mem_mask,
9115         __builtin_ia32_pmovsqw512mem_mask,
9116         __builtin_ia32_pmovqw512mem_mask,
9117         __builtin_ia32_pmovusdw512mem_mask,
9118         __builtin_ia32_pmovsdw512mem_mask,
9119         __builtin_ia32_pmovdw512mem_mask,
9120         __builtin_ia32_pmovqb512mem_mask,
9121         __builtin_ia32_pmovusqb512mem_mask,
9122         __builtin_ia32_pmovsqb512mem_mask,
9123         __builtin_ia32_pmovusdb512mem_mask,
9124         __builtin_ia32_pmovsdb512mem_mask,
9125         __builtin_ia32_pmovdb512mem_mask.
9126         (bdesc_args): Add __builtin_ia32_expanddf512,
9127         __builtin_ia32_expandsf512.
9128         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
9129         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
9130         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
9131         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
9132         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
9133         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
9134         (avx512f_<code>v8div16qi2_mask_store): This.
9135         (avx512f_expand<mode>): New.
9136
9137 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
9138
9139         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
9140         New.
9141         (_mm512_mask_prefetch_i64gather_pd): Ditto.
9142         (_mm512_prefetch_i32scatter_pd): Ditto.
9143         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
9144         (_mm512_prefetch_i64scatter_pd): Ditto.
9145         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
9146         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
9147         (_mm512_mask_prefetch_i64gather_ps): Ditto.
9148         (_mm512_prefetch_i32scatter_ps): Ditto.
9149         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
9150         (_mm512_prefetch_i64scatter_ps): Ditto.
9151         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
9152         * config/i386/i386-builtin-types.def: Define
9153         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
9154         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
9155         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
9156         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
9157         IX86_BUILTIN_SCATTERPFQPD.
9158         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
9159         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
9160         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
9161         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
9162         __builtin_ia32_scatterpfqps.
9163         (ix86_expand_builtin): Expand new built-ins.
9164         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
9165         fix memory access data type.
9166         (*avx512pf_gatherpf<mode>_mask): Ditto.
9167         (*avx512pf_gatherpf<mode>): Ditto.
9168         (avx512pf_scatterpf<mode>): Ditto.
9169         (*avx512pf_scatterpf<mode>_mask): Ditto.
9170         (*avx512pf_scatterpf<mode>): Ditto.
9171         (GATHER_SCATTER_SF_MEM_MODE): New.
9172         (avx512pf_gatherpf<mode>df): Ditto.
9173         (*avx512pf_gatherpf<mode>df_mask): Ditto.
9174         (*avx512pf_scatterpf<mode>df): Ditto.
9175
9176 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
9177
9178         PR bootstrap/59934
9179         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
9180         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
9181         reached.
9182
9183 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
9184
9185         * common/config/arm/arm-common.c
9186         (arm_rewrite_mcpu): Handle multiple names.
9187         * config/arm/arm.h
9188         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
9189
9190 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
9191
9192         * gimple-builder.h (create_gimple_tmp): Delete.
9193
9194 2014-01-27  Christian Bruel  <christian.bruel@st.com>
9195
9196         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
9197         words comparisons.
9198
9199 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
9200
9201         * config/pa/pa.md (call): Generate indirect long calls to non-local
9202         functions when outputing 32-bit code.
9203         (call_value): Likewise except for special call to buggy powf function.
9204
9205         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
9206         portable runtime and PIC indirect calls.
9207         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
9208         and PIC call sequences.  Use ldo instead of blr to set return register
9209         in PIC call sequence.
9210
9211 2014-01-25  Walter Lee  <walt@tilera.com>
9212
9213         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
9214         avoid clobbering a live register.
9215
9216 2014-01-25  Walter Lee  <walt@tilera.com>
9217
9218         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
9219         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
9220         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
9221         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
9222
9223 2014-01-25  Walter Lee  <walt@tilera.com>
9224
9225         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
9226         arguments on even registers.
9227         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
9228         STACK_BOUNDARY.
9229         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
9230         (BIGGEST_ALIGNMENT): Ditto.
9231         (BIGGEST_FIELD_ALIGNMENT): Ditto.
9232
9233 2014-01-25  Walter Lee  <walt@tilera.com>
9234
9235         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
9236         insns before bundling.
9237         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
9238
9239 2014-01-25  Walter Lee  <walt@tilera.com>
9240
9241         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
9242         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
9243         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
9244
9245 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
9246
9247         * config/mips/constraints.md (kl): Delete.
9248         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
9249         define expands, using...
9250         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
9251         instructions for MIPS16.
9252         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
9253         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
9254
9255 2014-01-25  Walter Lee  <walt@tilera.com>
9256
9257         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
9258         (clzdi2): Ditto.
9259         (ffsdi2): Ditto.
9260
9261 2014-01-25  Walter Lee  <walt@tilera.com>
9262
9263         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
9264         (TARGET_EXPAND_TO_RTL_HOOK): Define.
9265
9266 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
9267
9268         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
9269         Handle XOR.
9270
9271 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
9272
9273         * print-rtl.c (in_call_function_usage): New var.
9274         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
9275         EXPR_LIST mode as mode and not as reg note name.
9276
9277         PR middle-end/59561
9278         * cfgloopmanip.c (copy_loop_info): If
9279         loop->warned_aggressive_loop_optimizations, make sure
9280         the flag is set in target loop too.
9281
9282 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
9283
9284         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
9285         flag_cilkplus.
9286         * builtins.def: Likewise.
9287         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
9288         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
9289         * ira.c (ira_setup_eliminable_regset): Likewise.
9290         * omp-low.c (gate_expand_omp): Likewise.
9291         (execute_lower_omp): Likewise.
9292         (diagnose_sb_0): Likewise.
9293         (gate_diagnose_omp_blocks): Likewise.
9294         (simd_clone_clauses_extract): Likewise.
9295         (gate): Likewise.
9296
9297 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9298
9299         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
9300         correction for little endian...
9301         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
9302         here.
9303
9304 2014-01-24  Jeff Law  <law@redhat.com>
9305
9306         PR tree-optimization/59919
9307         * tree-vrp.c (find_assert_locations_1): Do not register asserts
9308         for non-returning calls.
9309
9310 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
9311
9312         * common/config/aarch64/aarch64-common.c
9313         (aarch64_rewrite_mcpu): Handle multiple names.
9314         * config/aarch64/aarch64.h
9315         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
9316
9317 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
9318
9319         * input.c (add_file_to_cache_tab): Handle the case where fopen
9320         returns NULL.
9321
9322 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
9323
9324         PR target/59929
9325         * config/i386/i386.md (pushsf splitter): Get stack adjustment
9326         from push operand if code of push isn't PRE_DEC.
9327
9328 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
9329
9330         PR target/59909
9331         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
9332         -mquad-memory-atomic.  Update -mquad-memory documentation to say
9333         it is only used for non-atomic loads/stores.
9334
9335         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
9336         -mquad-memory or -mquad-memory-atomic switches.
9337
9338         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
9339         -mquad-memory-atomic to ISA 2.07 support.
9340
9341         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
9342         to separate support of normal quad word memory operations (ldq, stq)
9343         from the atomic quad word memory operations.
9344
9345         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9346         support to separate non-atomic quad word operations from atomic
9347         quad word operations.  Disable non-atomic quad word operations in
9348         little endian mode so that we don't have to swap words after the
9349         load and before the store.
9350         (quad_load_store_p): Add comment about atomic quad word support.
9351         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
9352         options printed with -mdebug=reg.
9353
9354         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
9355         -mquad-memory-atomic as the test for whether we have quad word
9356         atomic instructions.
9357         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
9358         or -mp8-vector are used, allow byte/half-word atomic operations.
9359
9360         * config/rs6000/sync.md (load_lockedti): Insure that the address
9361         is a proper indexed or indirect address for the lqarx instruction.
9362         On little endian systems, swap the hi/lo registers after the lqarx
9363         instruction.
9364         (load_lockedpti): Use indexed_or_indirect_operand predicate to
9365         insure the address is valid for the lqarx instruction.
9366         (store_conditionalti): Insure that the address is a proper indexed
9367         or indirect address for the stqcrx. instruction.  On little endian
9368         systems, swap the hi/lo registers before doing the stqcrx.
9369         instruction.
9370         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
9371         insure the address is valid for the stqcrx. instruction.
9372
9373         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9374         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
9375         type of quad memory support is available.
9376
9377 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
9378
9379         PR regression/59915
9380         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
9381         there is a danger of looping.
9382
9383 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
9384
9385         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9386         force flag_ira_loop_pressure if set via command line.
9387
9388 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9389
9390         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
9391         (ashr_simd): New builtin handling DI mode.
9392         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
9393         (aarch64_sshr_simddi): New match pattern.
9394         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
9395         (vshrd_n_s64): Likewise.
9396         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
9397
9398 2014-01-23  Nick Clifton  <nickc@redhat.com>
9399
9400         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
9401         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
9402         favour of mcu specific scripts.
9403         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
9404         430x multilibs.
9405
9406 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
9407             Alex Velenko  <Alex.Velenko@arm.com>
9408
9409         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
9410         (vaddv_s16): Likewise.
9411         (vaddv_s32): Likewise.
9412         (vaddv_u8): Likewise.
9413         (vaddv_u16): Likewise.
9414         (vaddv_u32): Likewise.
9415         (vaddvq_s8): Likewise.
9416         (vaddvq_s16): Likewise.
9417         (vaddvq_s32): Likewise.
9418         (vaddvq_s64): Likewise.
9419         (vaddvq_u8): Likewise.
9420         (vaddvq_u16): Likewise.
9421         (vaddvq_u32): Likewise.
9422         (vaddvq_u64): Likewise.
9423         (vaddv_f32): Likewise.
9424         (vaddvq_f32): Likewise.
9425         (vaddvq_f64): Likewise.
9426         (vmaxv_f32): Likewise.
9427         (vmaxv_s8): Likewise.
9428         (vmaxv_s16): Likewise.
9429         (vmaxv_s32): Likewise.
9430         (vmaxv_u8): Likewise.
9431         (vmaxv_u16): Likewise.
9432         (vmaxv_u32): Likewise.
9433         (vmaxvq_f32): Likewise.
9434         (vmaxvq_f64): Likewise.
9435         (vmaxvq_s8): Likewise.
9436         (vmaxvq_s16): Likewise.
9437         (vmaxvq_s32): Likewise.
9438         (vmaxvq_u8): Likewise.
9439         (vmaxvq_u16): Likewise.
9440         (vmaxvq_u32): Likewise.
9441         (vmaxnmv_f32): Likewise.
9442         (vmaxnmvq_f32): Likewise.
9443         (vmaxnmvq_f64): Likewise.
9444         (vminv_f32): Likewise.
9445         (vminv_s8): Likewise.
9446         (vminv_s16): Likewise.
9447         (vminv_s32): Likewise.
9448         (vminv_u8): Likewise.
9449         (vminv_u16): Likewise.
9450         (vminv_u32): Likewise.
9451         (vminvq_f32): Likewise.
9452         (vminvq_f64): Likewise.
9453         (vminvq_s8): Likewise.
9454         (vminvq_s16): Likewise.
9455         (vminvq_s32): Likewise.
9456         (vminvq_u8): Likewise.
9457         (vminvq_u16): Likewise.
9458         (vminvq_u32): Likewise.
9459         (vminnmv_f32): Likewise.
9460         (vminnmvq_f32): Likewise.
9461         (vminnmvq_f64): Likewise.
9462
9463 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
9464
9465         * config/aarch64/aarch64-simd.md
9466         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
9467         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
9468         (*aarch64_mul3_elt<mode>): Likewise.
9469         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
9470         (*aarch64_mul3_elt_to_64v2df): Likewise.
9471         (*aarch64_mla_elt<mode>): Likewise.
9472         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
9473         (*aarch64_mls_elt<mode>): Likewise.
9474         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
9475         (*aarch64_fma4_elt<mode>): Likewise.
9476         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
9477         (*aarch64_fma4_elt_to_64v2df): Likewise.
9478         (*aarch64_fnma4_elt<mode>): Likewise.
9479         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
9480         (*aarch64_fnma4_elt_to_64v2df): Likewise.
9481         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
9482         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
9483         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9484         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9485         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
9486         (aarch64_sqdmull_lane<mode>_internal): Likewise.
9487         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
9488
9489 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9490
9491         * config/aarch64/aarch64-simd.md
9492         (aarch64_be_checked_get_lane<mode>): New define_expand.
9493         * config/aarch64/aarch64-simd-builtins.def
9494         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
9495         New builtin definition.
9496         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
9497         Use new safe be builtin.
9498
9499 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9500
9501         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
9502         New define_insn.
9503         (aarch64_be_st1<mode>): Likewise.
9504         (aarch_ld1<VALL:mode>): Define_expand modified.
9505         (aarch_st1<VALL:mode>): Likewise.
9506         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
9507         (UNSPEC_ST1): Likewise.
9508
9509 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
9510
9511         * config/microblaze/microblaze.md: Add trap insn and attribute
9512
9513 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
9514
9515         PR preprocessor/58580
9516         * input.h (location_get_source_line): Take an additional line_size
9517         parameter.
9518         (void diagnostics_file_cache_fini): Declare new function.
9519         * input.c (struct fcache): New type.
9520         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
9521         New static constants.
9522         (diagnostic_file_cache_init, total_lines_num)
9523         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
9524         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
9525         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
9526         (get_next_line, read_next_line, goto_next_line, read_line_num):
9527         New static function definitions.
9528         (diagnostic_file_cache_fini): New function.
9529         (location_get_source_line): Take an additional output line_len
9530         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
9531         read_line_num.
9532         * diagnostic.c (diagnostic_finish): Call
9533         diagnostic_file_cache_fini.
9534         (adjust_line): Take an additional input parameter for the length
9535         of the line, rather than calculating it with strlen.
9536         (diagnostic_show_locus): Adjust the use of
9537         location_get_source_line and adjust_line with respect to their new
9538         signature.  While displaying a line now, do not stop at the first
9539         null byte.  Rather, display the zero byte as a space and keep
9540         going until we reach the size of the line.
9541         * Makefile.in: Add vec.o to OBJS-libcommon
9542
9543 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9544             Ilya Tocar     <ilya.tocar@intel.com>
9545
9546         * config/i386/avx512fintrin.h (_mm512_kmov): New.
9547         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
9548         (__builtin_ia32_kmov16): Ditto.
9549         * config/i386/i386.md (UNSPEC_KMOV): New.
9550         (kmovw): Ditto.
9551
9552 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9553
9554         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
9555         (_mm512_storeu_si512): Ditto.
9556
9557 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
9558
9559         PR target/52125
9560         * rtl.h (get_referenced_operands): Declare.
9561         * recog.c (get_referenced_operands): New function.
9562         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
9563         operands have been referenced when recording LO_SUM references.
9564
9565 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
9566
9567         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
9568
9569 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9570
9571         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
9572         Enable for generic and recent AMD targets.
9573
9574 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9575
9576         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
9577         ARG_SIZE note when adjustment was eliminated.
9578
9579 2014-01-22  Jeff Law  <law@redhat.com>
9580
9581         PR tree-optimization/59597
9582         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
9583         in file.  Accept new argument REGISTERING and use it to modify
9584         dump output appropriately.
9585         (register_jump_thread): Corresponding changes.
9586         (mark_threaded_blocks): Reinstate code to cancel unprofitable
9587         thread paths involving joiner blocks.  Add code to dump cancelled
9588         jump threading paths.
9589
9590 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
9591
9592         PR rtl-optimization/59477
9593         * lra-constraints.c (inherit_in_ebb): Process call for living hard
9594         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
9595
9596 2014-01-22  Tom Tromey  <tromey@redhat.com>
9597
9598         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
9599         PARAMS.
9600         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
9601
9602 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9603
9604         PR rtl-optimization/59896
9605         * lra-constraints.c (process_alt_operands): Check unused note for
9606         matched operands of insn with no output reloads.
9607
9608 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
9609
9610         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
9611         (mips_move_from_gpr_cost): Likewise.
9612
9613 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9614
9615         PR rtl-optimization/59858
9616         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
9617         ira_class_hard_regs_num.
9618         (process_alt_operands): Increase reject for dying matched operand.
9619
9620 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
9621
9622         PR target/59003
9623         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
9624         smaller than size, perform several stores or loads and stores
9625         at dst + count - size to store or copy all of size bytes, rather
9626         than just last modesize bytes.
9627
9628 2014-01-20  DJ Delorie  <dj@redhat.com>
9629
9630         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
9631         that CLOBBERs are REGs before propogating their values.
9632
9633 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
9634
9635         PR middle-end/59789
9636         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
9637         (cgraph_inline_failed_type): New function.
9638         * cgraph.h (DEFCIFCODE): Add type.
9639         (cgraph_inline_failed_type_t): New enum.
9640         (cgraph_inline_failed_type): New prototype.
9641         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
9642         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
9643         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
9644         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
9645         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
9646         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
9647         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
9648         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
9649         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
9650         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
9651         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
9652         OPTIMIZATION_MISMATCH.
9653         * tree-inline.c (expand_call_inline): Emit errors during
9654         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
9655
9656 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9657
9658         PR target/59685
9659         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
9660         mode attribute in insn output.
9661
9662 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
9663
9664         * output.h (output_constant): Delete.
9665         * varasm.c (output_constant): Make private.
9666
9667 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
9668
9669         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
9670
9671 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9672
9673         PR middle-end/59860
9674         * tree.h (fold_builtin_strcat): New prototype.
9675         * builtins.c (fold_builtin_strcat): No longer static.  Add len
9676         argument, if non-NULL, don't call c_strlen.  Optimize
9677         directly into __builtin_memcpy instead of __builtin_strcpy.
9678         (fold_builtin_2): Adjust fold_builtin_strcat caller.
9679         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
9680
9681 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9682
9683         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9684         for SImode_address_operand operands, having only a REG argument.
9685
9686 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9687
9688         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
9689         loader name using mbig-endian.
9690         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
9691
9692 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9693
9694         * doc/invoke.texi (-march): Clarify documentation for AArch64.
9695         (-mtune): Likewise.
9696         (-mcpu): Likewise.
9697
9698 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
9699
9700         * config/aarch64/aarch64-protos.h
9701         (aarch64_cannot_change_mode_class_ptr): Declare.
9702         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
9703         aarch64_cannot_change_mode_class_ptr): New.
9704         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
9705         backend hook aarch64_cannot_change_mode_class.
9706
9707 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9708
9709         * common/config/aarch64/aarch64-common.c
9710         (aarch64_handle_option): Don't handle any option order logic here.
9711         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
9712         selected_cpu, warn on architecture version mismatch.
9713         (aarch64_override_options): Fix parsing order for option strings.
9714
9715 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9716             Iain Sandoe  <iain@codesourcery.com>
9717
9718         PR bootstrap/59496
9719         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
9720         warning.  Amend comment to reflect current functionality.
9721
9722 2014-01-20  Richard Biener  <rguenther@suse.de>
9723
9724         PR middle-end/59860
9725         * builtins.c (fold_builtin_strcat): Remove case better handled
9726         by tree-ssa-strlen.c.
9727
9728 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
9729
9730         * config/aarch64/aarch64.opt
9731         (mcpu, march, mtune): Make case-insensitive.
9732
9733 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9734
9735         PR target/59880
9736         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9737         if operands[1] is a REG or ZERO_EXTEND of a REG.
9738
9739 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
9740
9741         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
9742
9743 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
9744
9745         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
9746         long non-pic millicode calls.
9747
9748 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9749
9750         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
9751
9752 2014-01-19  Kito Cheng  <kito@0xlab.org>
9753
9754         * builtins.c (expand_movstr): Check movstr expand done or fail.
9755
9756 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9757             H.J. Lu  <hongjiu.lu@intel.com>
9758
9759         PR target/59379
9760         * config/i386/i386.md (*lea<mode>): Zero-extend return register
9761         to DImode for zero-extended addresses.
9762
9763 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
9764
9765         PR rtl-optimization/57763
9766         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
9767         on the new indirect jump_insn and increment LABEL_NUSES (label).
9768
9769 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
9770
9771         PR bootstrap/59580
9772         PR bootstrap/59583
9773         * config.gcc (x86_archs): New variable.
9774         (x86_64_archs): Likewise.
9775         (x86_cpus): Likewise.
9776         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
9777         --with-arch/--with-cpu= options.
9778         Support --with-arch=/--with-cpu={nehalem,westmere,
9779         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
9780
9781 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9782
9783         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
9784         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
9785
9786 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9787
9788         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
9789
9790 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
9791
9792         PR target/58944
9793         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
9794         clear cpp_get_options (parse_in)->warn_unused_macros for
9795         ix86_target_macros_internal with cpp_define.
9796
9797 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
9798
9799         * jump.c (delete_related_insns): Keep (use (insn))s.
9800         * reorg.c (redundant_insn): Check for barriers too.
9801
9802 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9803
9804         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
9805
9806 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
9807
9808         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
9809         call to $$dyncall when TARGET_LONG_CALLS is true.
9810
9811 2014-01-17  Jeff Law  <law@redhat.com>
9812
9813         * ree.c (combine_set_extension): Temporarily disable test for
9814         changing number of hard registers.
9815
9816 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9817
9818         PR middle-end/58125
9819         * ipa-inline-analysis.c (inline_free_summary):
9820         Do not free summary of aliases.
9821
9822 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9823
9824         PR middle-end/59706
9825         * gimplify.c (gimplify_expr): Use create_tmp_var
9826         instead of create_tmp_var_raw.  If cond doesn't have
9827         integral type, don't add the IFN_ANNOTATE builtin at all.
9828
9829 2014-01-17  Martin Jambor  <mjambor@suse.cz>
9830
9831         PR ipa/59736
9832         * ipa-cp.c (prev_edge_clone): New variable.
9833         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
9834         Also resize prev_edge_clone vector.
9835         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
9836         (ipcp_edge_removal_hook): New function.
9837         (ipcp_driver): Register ipcp_edge_removal_hook.
9838
9839 2014-01-17  Andrew Pinski  <apinski@cavium.com>
9840             Steve Ellcey  <sellcey@mips.com>
9841
9842         PR target/59462
9843         * config/mips/mips.c (mips_print_operand): Check operand mode instead
9844         of operator mode.
9845
9846 2014-01-17  Jeff Law  <law@redhat.com>
9847
9848         PR middle-end/57904
9849         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
9850         so that pass_ccp runs first.
9851
9852 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9853
9854         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
9855         (ix86_adjust_cost): Use !TARGET_XXX.
9856         (do_reorder_for_imul): Likewise.
9857         (swap_top_of_ready_list): Likewise.
9858         (ix86_sched_reorder): Likewise.
9859
9860 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9861
9862         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9863         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
9864         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
9865         (intel_memset): New.  Duplicate slm_memset.
9866         (intel_cost): New.  Duplicate slm_cost.
9867         (m_INTEL): New macro.
9868         (processor_target_table): Add "intel".
9869         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
9870         with PROCESSOR_INTEL for "intel".
9871         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
9872         PROCESSOR_SILVERMONT.
9873         (ix86_issue_rate): Likewise.
9874         (ix86_adjust_cost): Likewise.
9875         (ia32_multipass_dfa_lookahead): Likewise.
9876         (swap_top_of_ready_list): Likewise.
9877         (ix86_sched_reorder): Likewise.
9878         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
9879         instead of TARGET_OPT_AGU.
9880         * config/i386/i386.h (TARGET_INTEL): New.
9881         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
9882         (processor_type): Add PROCESSOR_INTEL.
9883         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
9884         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
9885
9886 2014-01-17  Marek Polacek  <polacek@redhat.com>
9887
9888         PR c/58346
9889         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
9890         size is zero.
9891
9892 2014-01-17  Richard Biener  <rguenther@suse.de>
9893
9894         PR tree-optimization/46590
9895         * opts.c (default_options_table): Add entries for
9896         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
9897         all enabled at -O1 but not for -Og.
9898         * common.opt (fbranch-count-reg): Remove Init(1).
9899         (fmove-loop-invariants): Likewise.
9900         (ftree-pta): Likewise.
9901
9902 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9903
9904         * config/i386/i386.c (ix86_data_alignment): For compatibility with
9905         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
9906         decls to at least the GCC 4.8 used alignments.
9907
9908         PR fortran/59440
9909         * tree-nested.c (convert_nonlocal_reference_stmt,
9910         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
9911         of GIMPLE_BIND stmts, adjust associated decls.
9912
9913 2014-01-17  Richard Biener  <rguenther@suse.de>
9914
9915         PR tree-optimization/46590
9916         * vec.h (vec<>::bseach): New member function implementing
9917         binary search according to C89 bsearch.
9918         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
9919         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
9920         bitmap pointer again.  Make accesses_in_loop a flat array.
9921         (mem_ref_obstack): New global.
9922         (outermost_indep_loop): Adjust for mem_ref->stored changes.
9923         (mark_ref_stored): Likewise.
9924         (ref_indep_loop_p_2): Likewise.
9925         (set_ref_stored_in_loop): New helper function.
9926         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
9927         (memref_free): Adjust.
9928         (record_mem_ref_loc): Simplify.
9929         (gather_mem_refs_stmt): Adjust.
9930         (sort_locs_in_loop_postorder_cmp): New function.
9931         (analyze_memory_references): Sort accesses_in_loop after
9932         loop postorder number.
9933         (find_ref_loc_in_loop_cmp): New function.
9934         (for_all_locs_in_loop): Find relevant cluster of locs in
9935         accesses_in_loop and iterate without recursion.
9936         (execute_sm): Avoid uninit warning.
9937         (struct ref_always_accessed): Simplify.
9938         (ref_always_accessed::operator ()): Likewise.
9939         (ref_always_accessed_p): Likewise.
9940         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
9941         loop postorder numbers here.
9942         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
9943         numbers.
9944
9945 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9946
9947         PR c++/57945
9948         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
9949         on decls for which assemble_alias has been called.
9950
9951 2014-01-17  Nick Clifton  <nickc@redhat.com>
9952
9953         * config/msp430/msp430.opt: (mcpu): New option.
9954         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
9955         (msp430_option_override): Parse target_cpu.  If the MCU name
9956         matches a generic string, clear target_mcu.
9957         (msp430_attr): Allow numeric interrupt values up to 63.
9958         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
9959         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
9960         option.
9961         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
9962         Add mcpu matches.
9963         * config/msp430/msp430.md (popm): Use %J rather than %I.
9964         (addsi3): Use msp430_nonimmediate_operand for operand 2.
9965         (addhi_cy_i): Use immediate_operand for operand 2.
9966         * doc/invoke.texi: Document -mcpu option.
9967
9968 2014-01-17  Richard Biener  <rguenther@suse.de>
9969
9970         PR rtl-optimization/38518
9971         * df.h (df_analyze_loop): Declare.
9972         * df-core.c: Include cfgloop.h.
9973         (df_analyze_1): Split out main part of df_analyze.
9974         (df_analyze): Adjust.
9975         (loop_inverted_post_order_compute): New function.
9976         (loop_post_order_compute): Likewise.
9977         (df_analyze_loop): New function avoiding whole-function
9978         postorder computes.
9979         * loop-invariant.c (find_defs): Use df_analyze_loop.
9980         (find_invariants): Adjust.
9981         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
9982
9983 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
9984
9985         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
9986         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
9987
9988 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
9989
9990         * ipa-ref.c (ipa_remove_stmt_references): Fix references
9991         traversal when removing references.
9992
9993 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
9994
9995         PR ipa/59775
9996         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
9997
9998 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
9999
10000         PR middle-end/56791
10001         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
10002         pushing a reload for an autoinc when we had previously reloaded an
10003         inner part of the address.
10004
10005 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
10006
10007         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
10008         field.
10009         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
10010         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
10011         when not giving up or versioning for alias only because of
10012         loop->safelen.
10013         (vect_analyze_data_ref_dependences): Set to true.
10014         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
10015         is a GIMPLE_PHI.
10016         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
10017         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
10018         to the condition.
10019
10020         PR middle-end/58344
10021         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
10022
10023         PR target/59839
10024         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
10025         operand 0 predicate for gathers, use a new pseudo as subtarget.
10026
10027 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
10028
10029         PR middle-end/59609
10030         * lra-constraints.c (process_alt_operands): Add printing debug info.
10031         Check absence of input/output reloads for matched operands too.
10032
10033 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
10034
10035         PR rtl-optimization/59835
10036         * ira.c (ira_init_register_move_cost): Increase cost for
10037         impossible modes.
10038
10039 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
10040
10041         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
10042
10043 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
10044
10045         PR target/59780
10046         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
10047         non-register objects.  Use gen_(high/low)part more consistently.
10048         Fix assertions.
10049
10050 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
10051
10052         PR target/59844
10053         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
10054         endian support, remove tests for WORDS_BIG_ENDIAN.
10055         (p8_mfvsrd_3_<mode>): Likewise.
10056         (reload_gpr_from_vsx<mode>): Likewise.
10057         (reload_gpr_from_vsxsf): Likewise.
10058         (p8_mfvsrd_4_disf): Likewise.
10059
10060 2014-01-16  Richard Biener  <rguenther@suse.de>
10061
10062         PR rtl-optimization/46590
10063         * lcm.c (compute_antinout_edge): Use postorder iteration.
10064         (compute_laterin): Use inverted postorder iteration.
10065
10066 2014-01-16  Nick Clifton  <nickc@redhat.com>
10067
10068         PR middle-end/28865
10069         * varasm.c (output_constant): Return the number of bytes actually
10070         emitted.
10071         (output_constructor_array_range): Update the field size with the
10072         number of bytes emitted by output_constant.
10073         (output_constructor_regular_field): Likewise.  Also do not
10074         complain if the total number of bytes emitted is now greater
10075         than the expected fieldpos.
10076         * output.h (output_constant): Update prototype and descriptive comment.
10077
10078 2014-01-16  Marek Polacek  <polacek@redhat.com>
10079
10080         PR middle-end/59827
10081         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
10082         it is error_mark_node.
10083
10084 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
10085
10086         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
10087         VALID_AVX256_REG_OR_OI_MODE.
10088
10089 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
10090
10091         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
10092         current procedure should be profiled.
10093
10094 2014-01-15  Andrew Pinski  <apinski@cavium.com>
10095
10096         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
10097         of moving from/to the STACK_REG register class.
10098
10099 2014-01-15  Richard Henderson  <rth@redhat.com>
10100
10101         PR debug/54694
10102         * reginfo.c (global_regs_decl): Globalize.
10103         * rtl.h (global_regs_decl): Declare.
10104         * ira.c (do_reload): Diagnose frame_pointer_needed and it
10105         reserved via global_regs.
10106
10107 2014-01-15  Teresa Johnson  <tejohnson@google.com>
10108
10109         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
10110
10111 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
10112
10113         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
10114         and vmulosh rather than call gen_vec_widen_smult_*.
10115         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
10116         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
10117         (vec_widen_smult_even_v16qi): Likewise.
10118         (vec_widen_umult_even_v8hi): Likewise.
10119         (vec_widen_smult_even_v8hi): Likewise.
10120         (vec_widen_umult_odd_v16qi): Likewise.
10121         (vec_widen_smult_odd_v16qi): Likewise.
10122         (vec_widen_umult_odd_v8hi): Likewise.
10123         (vec_widen_smult_odd_v8hi): Likewise.
10124         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
10125         vmuloub rather than call gen_vec_widen_umult_*.
10126         (vec_widen_umult_lo_v16qi): Likewise.
10127         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
10128         vmulosb rather than call gen_vec_widen_smult_*.
10129         (vec_widen_smult_lo_v16qi): Likewise.
10130         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
10131         rather than call gen_vec_widen_umult_*.
10132         (vec_widen_umult_lo_v8hi): Likewise.
10133         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
10134         rather than call gen_vec_widen_smult_*.
10135         (vec_widen_smult_lo_v8hi): Likewise.
10136
10137 2014-01-15  Jeff Law  <law@redhat.com>
10138
10139         PR tree-optimization/59747
10140         * ree.c (find_and_remove_re): Properly handle case where a second
10141         eliminated extension requires widening a copy created for elimination
10142         of a prior extension.
10143         (combine_set_extension): Ensure that the number of hard regs needed
10144         for a destination register does not change when we widen it.
10145
10146 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
10147
10148         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
10149         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
10150         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
10151         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
10152         (avr-*-rtems*): Likewise.
10153         (bfin*-rtems*): Likewise.
10154         (moxie-*-rtems*): Likewise.
10155         (h8300-*-rtems*): Likewise.
10156         (i[34567]86-*-rtems*): Likewise.
10157         (lm32-*-rtems*): Likewise.
10158         (m32r-*-rtems*): Likewise.
10159         (m68k-*-rtems*): Likewise.
10160         (microblaze*-*-rtems*): Likewise.
10161         (mips*-*-rtems*): Likewise.
10162         (powerpc-*-rtems*): Likewise.
10163         (sh-*-rtems*): Likewise.
10164         (sparc-*-rtems*): Likewise.
10165         (sparc64-*-rtems*): Likewise.
10166         (v850-*-rtems*): Likewise.
10167         (m32c-*-rtems*): Likewise.
10168
10169 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
10170
10171         PR rtl-optimization/59511
10172         * ira.c (ira_init_register_move_cost): Use memory costs for some
10173         cases of register move cost calculations.
10174         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
10175         instead of BB frequency.
10176         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
10177         * lra-assigns.c (find_hard_regno_for): Ditto.
10178
10179 2014-01-15  Richard Biener  <rguenther@suse.de>
10180
10181         PR tree-optimization/59822
10182         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
10183         (vectorizable_load): Use it to hoist defs of uses of invariant
10184         loads out of the loop.
10185
10186 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
10187             Kugan Vivekanandarajah  <kuganv@linaro.org>
10188
10189         PR target/59695
10190         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
10191         truncation.
10192
10193 2014-01-15  Richard Biener  <rguenther@suse.de>
10194
10195         PR rtl-optimization/59802
10196         * lcm.c (compute_available): Use inverted postorder to seed
10197         the initial worklist.
10198
10199 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10200
10201         PR target/59803
10202         * config/s390/s390.c (s390_preferred_reload_class): Don't return
10203         ADDR_REGS for invalid symrefs in non-PIC code.
10204
10205 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
10206
10207         PR other/58712
10208         * builtins.c (determine_block_size): Initialize *probable_max_size
10209         even if len_rtx is CONST_INT.
10210
10211 2014-01-14  Andrew Pinski  <apinski@cavium.com>
10212
10213         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
10214         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
10215         (cortexa53_tunings): Likewise.
10216         (aarch64_sched_issue_rate): New function.
10217         (TARGET_SCHED_ISSUE_RATE): Define.
10218
10219 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
10220
10221         * ira-costs.c (find_costs_and_classes): Add missed
10222         ira_init_register_move_cost_if_necessary.
10223
10224 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
10225
10226         PR target/59787
10227         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
10228
10229 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
10230
10231         PR target/59794
10232         * config/i386/i386.c (type_natural_mode): Add a bool parameter
10233         to indicate if type is used for function return value.  Warn ABI
10234         change if the vector mode isn't available for function return value.
10235         (ix86_function_arg_advance): Pass false to type_natural_mode.
10236         (ix86_function_arg): Likewise.
10237         (ix86_gimplify_va_arg): Likewise.
10238         (function_arg_32): Don't warn ABI change.
10239         (ix86_function_value): Pass true to type_natural_mode.
10240         (ix86_return_in_memory): Likewise.
10241         (ix86_struct_value_rtx): Removed.
10242         (TARGET_STRUCT_VALUE_RTX): Likewise.
10243
10244 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10245
10246         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
10247         converting a conditional jump into a conditional return.
10248
10249 2014-01-14  Richard Biener  <rguenther@suse.de>
10250
10251         PR tree-optimization/58921
10252         PR tree-optimization/59006
10253         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
10254         hoisting invariant stmts.
10255         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
10256         invariant loads on the preheader edge if possible.
10257
10258 2014-01-14  Joey Ye  <joey.ye@arm.com>
10259
10260         * doc/plugin.texi (Building GCC plugins): Update to C++.
10261
10262 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
10263
10264         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
10265         (_mm_rcp28_round_ss): Ditto.
10266         (_mm_rsqrt28_round_sd): Ditto.
10267         (_mm_rsqrt28_round_ss): Ditto.
10268         (_mm_rcp28_sd): Ditto.
10269         (_mm_rcp28_ss): Ditto.
10270         (_mm_rsqrt28_sd): Ditto.
10271         (_mm_rsqrt28_ss): Ditto.
10272         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
10273         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
10274         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
10275         (IX86_BUILTIN_RCP28SD): Ditto.
10276         (IX86_BUILTIN_RCP28SS): Ditto.
10277         (IX86_BUILTIN_RSQRT28SD): Ditto.
10278         (IX86_BUILTIN_RSQRT28SS): Ditto.
10279         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
10280         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
10281         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
10282         (ix86_expand_special_args_builtin): Expand new FTYPE.
10283         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
10284         (srcp14<mode>): Make insn unary.
10285         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
10286         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
10287         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
10288         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
10289         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
10290         Fix rounding: make it SAE only.
10291         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
10292         Ditto.
10293         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
10294         Ditto.
10295         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
10296         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
10297         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
10298         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
10299         (round_saeonly_mask_scalar_operand4): Ditto.
10300         (round_saeonly_mask_scalar_op3): Ditto.
10301         (round_saeonly_mask_scalar_op4): Ditto.
10302
10303 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10304
10305         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10306         Implement -maltivec=be for vec_insert and vec_extract.
10307
10308 2014-01-10  DJ Delorie  <dj@redhat.com>
10309
10310         * config/msp430/msp430.md (call_internal): Don't allow memory
10311         references with SP as the base register.
10312         (call_value_internal): Likewise.
10313         * config/msp430/constraints.md (Yc): New.  For memory references
10314         that don't use SP as a base register.
10315
10316         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
10317         "an integer without a # prefix"
10318         * config/msp430/msp430.md (epilogue_helper): Use it.
10319
10320 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
10321
10322         PR target/59617
10323         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
10324         AVX512F gather builtins.
10325         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
10326         on gather decls with INTEGER_TYPE masktype.
10327         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
10328         directly into the builtin rather than hoisting it before loop.
10329
10330         PR tree-optimization/59387
10331         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
10332         (scev_const_prop): If folded_casts and type has undefined overflow,
10333         use force_gimple_operand instead of force_gimple_operand_gsi and
10334         for each added stmt if it is assign with
10335         arith_code_with_undefined_signed_overflow, call
10336         rewrite_to_defined_overflow.
10337         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
10338         gimple-fold.h instead.
10339         (arith_code_with_undefined_signed_overflow,
10340         rewrite_to_defined_overflow): Moved to ...
10341         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
10342         rewrite_to_defined_overflow): ... here.  No longer static.
10343         Include gimplify-me.h.
10344         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
10345         rewrite_to_defined_overflow): New prototypes.
10346
10347 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10348
10349         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
10350
10351 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
10352
10353         * builtins.c (get_object_alignment_2): Minor tweak.
10354         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
10355
10356 2014-01-13  Christian Bruel  <christian.bruel@st.com>
10357
10358         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
10359         optimized non constant lengths.
10360
10361 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
10362
10363         PR libgomp/59194
10364         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
10365         load as __atomic_load_N if possible.
10366
10367 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
10368
10369         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
10370         target parameter.
10371         (rs6000_expand_builtin): Adjust call.
10372
10373 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
10374
10375         PR target/58115
10376         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
10377         * config/rs6000/rs6000.c: Include target-globals.h.
10378         (rs6000_set_current_function): Instead of doing target_reinit
10379         unconditionally, use save_target_globals_default_opts and
10380         restore_target_globals.
10381
10382         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
10383         FPSCR.
10384         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
10385         (rs6000_expand_builtin): Handle mffs and mtfsf.
10386         (rs6000_init_builtins): Define mffs and mtfsf.
10387         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
10388         (rs6000_mffs): New pattern.
10389         (rs6000_mtfsf): New pattern.
10390
10391 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
10392
10393         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
10394         Start narrowing with START.  Apply candidate-use pair
10395         and check overall cost in narrowing.
10396         (iv_ca_prune): Pass new argument.
10397
10398 2014-01-10  Jeff Law  <law@redhat.com>
10399
10400         PR middle-end/59743
10401         * ree.c (combine_reaching_defs): Ensure the defining statement
10402         occurs before the extension when optimizing extensions with
10403         different source and destination hard registers.
10404
10405 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
10406
10407         PR ipa/58585
10408         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
10409         vtables into the type inheritance graph.
10410
10411 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10412
10413         PR rtl-optimization/59754
10414         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
10415         modes in the REGNO != REGNO case.
10416
10417 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10418
10419         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
10420
10421 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10422
10423         PR tree-optimization/59745
10424         * tree-predcom.c (tree_predictive_commoning_loop): Call
10425         free_affine_expand_cache if giving up because components is NULL.
10426
10427         * target-globals.c (save_target_globals): Allocate < 4KB structs using
10428         GC in payload of target_globals struct instead of allocating them on
10429         the heap and the larger structs separately using GC.
10430         * target-globals.h (struct target_globals): Make regs, hard_regs,
10431         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
10432         of GTY((skip)) and change type to void *.
10433         (reset_target_globals): Cast loads from those fields to corresponding
10434         types.
10435
10436 2014-01-10  Steve Ellcey  <sellcey@mips.com>
10437
10438         PR plugins/59335
10439         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
10440         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
10441         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
10442
10443 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
10444
10445         PR target/59744
10446         * aarch64-modes.def (CC_Zmode): New flags mode.
10447         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
10448         represents an equality.
10449         (aarch64_get_condition_code): Handle CC_Zmode.
10450         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
10451
10452 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10453
10454         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
10455         extraction in good case.
10456
10457 2014-01-10  Richard Biener  <rguenther@suse.de>
10458
10459         PR tree-optimization/59374
10460         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
10461         checking after SLP discovery.  Mark stmts not participating
10462         in any SLP instance properly.
10463
10464 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10465
10466         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
10467         when handling a SET rtx.
10468
10469 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10470
10471         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
10472         (cortex-a57): Likewise.
10473         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
10474
10475 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10476
10477         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
10478         non-iwmmxt builtins.
10479
10480 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
10481
10482         PR ipa/58252
10483         PR ipa/59226
10484         * ipa-devirt.c record_target_from_binfo): Take as argument
10485         stack of binfos and lookup matching one for virtual inheritance.
10486         (possible_polymorphic_call_targets_1): Update.
10487
10488 2014-01-10  Huacai Chen  <chenhc@lemote.com>
10489
10490         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
10491         kernel strings for Loongson-2E/2F/3A.
10492
10493 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10494
10495         PR middle-end/59670
10496         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
10497         is_gimple_call before calling gimple_call_internal_p.
10498
10499 2014-01-09  Steve Ellcey  <sellcey@mips.com>
10500
10501         * Makefile.in (TREE_FLOW_H): Remove.
10502         (TREE_SSA_H): Add file names from tree-flow.h.
10503         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
10504         * tree.h: Remove tree-flow.h reference.
10505         * hash-table.h: Remove tree-flow.h reference.
10506         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
10507         reference with tree-ssa-loop.h.
10508
10509 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10510
10511         * doc/invoke.texi: Add -maltivec={be,le} options, and document
10512         default element-order behavior for -maltivec.
10513         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
10514         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
10515         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
10516         when targeting big endian, at least for now.
10517         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
10518
10519 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10520
10521         PR middle-end/47735
10522         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
10523         var satisfies use_register_for_decl, just take into account type
10524         alignment, rather than decl alignment.
10525
10526         PR tree-optimization/59622
10527         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
10528         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
10529         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
10530         Don't devirtualize for inplace at all.  For targets.length () == 1,
10531         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
10532
10533 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10534
10535         * config/i386/i386.md (cpu): Remove the unused btver1.
10536
10537 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10538
10539         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
10540
10541 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10542
10543         PR target/58115
10544         * tree-core.h (struct target_globals): New forward declaration.
10545         (struct tree_target_option): Add globals field.
10546         * tree.h (TREE_TARGET_GLOBALS): Define.
10547         (prepare_target_option_nodes_for_pch): New prototype.
10548         * target-globals.h (struct target_globals): Define even if
10549         !SWITCHABLE_TARGET.
10550         * tree.c (prepare_target_option_node_for_pch,
10551         prepare_target_option_nodes_for_pch): New functions.
10552         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
10553         * config/i386/i386.c: Include target-globals.h.
10554         (ix86_set_current_function): Instead of doing target_reinit
10555         unconditionally, use save_target_globals_default_opts and
10556         restore_target_globals.
10557
10558 2014-01-09  Richard Biener  <rguenther@suse.de>
10559
10560         PR tree-optimization/59715
10561         * tree-cfg.h (split_critical_edges): Declare.
10562         * tree-cfg.c (split_critical_edges): Export.
10563         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
10564
10565 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
10566
10567         * cfgexpand.c (expand_stack_vars): Optionally disable
10568         asan stack protection.
10569         (expand_used_vars): Likewise.
10570         (partition_stack_vars): Likewise.
10571         * asan.c (asan_emit_stack_protection): Optionally disable
10572         after return stack usage.
10573         (instrument_derefs): Optionally disable memory access instrumentation.
10574         (instrument_builtin_call): Likewise.
10575         (instrument_strlen_call): Likewise.
10576         (asan_protect_global): Optionally disable global variables protection.
10577         * doc/invoke.texi: Added doc for new options.
10578         * params.def: Added new options.
10579         * params.h: Likewise.
10580
10581 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10582
10583         PR rtl-optimization/59724
10584         * ifcvt.c (cond_exec_process_if_block): Don't call
10585         flow_find_head_matching_sequence with 0 longest_match.
10586         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
10587         non-active insns if !stop_after.
10588         (try_head_merge_bb): Revert 2014-01-07 changes.
10589
10590 2014-01-08  Jeff Law  <law@redhat.com>
10591
10592         * ree.c (get_sub_rtx): New function, extracted from...
10593         (merge_def_and_ext): Here.
10594         (combine_reaching_defs): Use get_sub_rtx.
10595
10596 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
10597
10598         * cgraph.h (varpool_variable_node): Do not choke on null node.
10599
10600 2014-01-08  Catherine Moore  <clm@codesourcery.com>
10601
10602         * config/mips/mips.md (simple_return): Attempt to use JRC
10603         for microMIPS.
10604         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
10605
10606 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10607
10608         PR rtl-optimization/59137
10609         * reorg.c (steal_delay_list_from_target): Call update_block for
10610         elided insns.
10611         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
10612
10613 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10614
10615         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10616         two duplicate entries.
10617
10618 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10619
10620         Revert:
10621         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
10622
10623         * config/mips/mips.c (mips_truncated_op_cost): New function.
10624         (mips_rtx_costs): Adjust test for BADDU.
10625         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
10626
10627         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
10628
10629         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
10630         (*baddu_si): ...this new pattern.
10631
10632 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10633
10634         PR ipa/59722
10635         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
10636
10637 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10638
10639         PR middle-end/57748
10640         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
10641         inner_reference_p.
10642         (expand_expr, expand_normal): Adjust.
10643         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
10644         inner_reference_p. Use inner_reference_p to expand inner references.
10645         (store_expr): Adjust.
10646         * cfgexpand.c (expand_call_stmt): Adjust.
10647
10648 2014-01-08  Rong Xu  <xur@google.com>
10649
10650         * gcov-io.c (gcov_var): Move from gcov-io.h.
10651         (gcov_position): Ditto.
10652         (gcov_is_error): Ditto.
10653         (gcov_rewrite): Ditto.
10654         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
10655         only part to libgcc/libgcov.h.
10656
10657 2014-01-08  Marek Polacek  <polacek@redhat.com>
10658
10659         PR middle-end/59669
10660         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
10661
10662 2014-01-08  Marek Polacek  <polacek@redhat.com>
10663
10664         PR sanitizer/59667
10665         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
10666
10667 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10668
10669         PR rtl-optimization/59649
10670         * stor-layout.c (get_mode_bounds): For BImode return
10671         0 and STORE_FLAG_VALUE.
10672
10673 2014-01-08  Richard Biener  <rguenther@suse.de>
10674
10675         PR middle-end/59630
10676         * gimple.h (is_gimple_builtin_call): Remove.
10677         (gimple_builtin_call_types_compatible_p): New.
10678         (gimple_call_builtin_p): New overload.
10679         * gimple.c (is_gimple_builtin_call): Remove.
10680         (validate_call): Rename to ...
10681         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
10682         check return types.
10683         (validate_type): New static function.
10684         (gimple_call_builtin_p): New overload and adjust.
10685         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
10686         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
10687         (gimple_fold_stmt_to_constant_1): Likewise.
10688         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
10689
10690 2014-01-08  Richard Biener  <rguenther@suse.de>
10691
10692         PR middle-end/59471
10693         * gimplify.c (gimplify_expr): Gimplify register-register type
10694         VIEW_CONVERT_EXPRs to separate stmts.
10695
10696 2014-01-07  Jeff Law  <law@redhat.com>
10697
10698         PR middle-end/53623
10699         * ree.c (combine_set_extension): Handle case where source
10700         and destination registers in an extension insn are different.
10701         (combine_reaching_defs): Allow source and destination registers
10702         in extension to be different under limited circumstances.
10703         (add_removable_extension): Remove restriction that the
10704         source and destination registers in the extension are the same.
10705         (find_and_remove_re): Emit a copy from the extension's
10706         destination to its source after the defining insn if
10707         the source and destination registers are different.
10708
10709         PR middle-end/59285
10710         * ifcvt.c (merge_if_block): If we are merging a block with more than
10711         one successor with a block with no successors, remove any BARRIER
10712         after the second block.
10713
10714 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
10715
10716         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
10717
10718 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
10719
10720         PR target/59652
10721         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
10722         for 14-bit register offsets when INT14_OK_STRICT is false.
10723
10724 2014-01-07  Roland Stigge  <stigge@antcom.de>
10725             Michael Meissner  <meissner@linux.vnet.ibm.com>
10726
10727         PR 57386/target
10728         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
10729         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
10730
10731 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
10732
10733         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
10734         -mcpu.
10735
10736 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
10737
10738         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
10739         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
10740         rtx is const0_rtx or not.
10741
10742 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
10743
10744         PR target/58115
10745         * target-globals.c (save_target_globals): Remove this_fn_optab
10746         handling.
10747         * toplev.c: Include optabs.h.
10748         (target_reinit): Temporarily restore the global options if another
10749         set of options are in force.
10750
10751 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
10752
10753         PR rtl-optimization/58668
10754         * cfgcleanup.c (flow_find_cross_jump): Don't count
10755         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
10756         to determine what is counted.
10757         (flow_find_head_matching_sequence): Use active_insn_p to determine
10758         what is counted.
10759         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
10760         counting change.
10761         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
10762         determine what is counted.
10763
10764         PR tree-optimization/59643
10765         * tree-predcom.c (split_data_refs_to_components): If one dr is
10766         read and one write, determine_offset fails and the write isn't
10767         in the bad component, just put the read into the bad component.
10768
10769 2014-01-07  Mike Stump  <mikestump@comcast.net>
10770             Jakub Jelinek  <jakub@redhat.com>
10771
10772         PR pch/59436
10773         * tree-core.h (struct tree_optimization_option): Change optabs
10774         type from unsigned char * to void *.
10775         * optabs.c (init_tree_optimization_optabs): Adjust
10776         TREE_OPTIMIZATION_OPTABS initialization.
10777
10778 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
10779
10780         PR target/59644
10781         * config/i386/i386.h (struct machine_function): Add
10782         no_drap_save_restore field.
10783         * config/i386/i386.c (ix86_save_reg): Use
10784         !cfun->machine->no_drap_save_restore instead of
10785         crtl->stack_realign_needed.
10786         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
10787         this function clears frame_pointer_needed.  Set
10788         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
10789         and DRAP reg is needed.
10790
10791 2014-01-06  Marek Polacek  <polacek@redhat.com>
10792
10793         PR c/57773
10794         * doc/implement-c.texi: Mention that other integer types are
10795         permitted as bit-field types in strictly conforming mode.
10796
10797 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
10798
10799         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
10800         is newly allocated.
10801
10802 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
10803
10804         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
10805
10806 2014-01-06  Martin Jambor  <mjambor@suse.cz>
10807
10808         PR ipa/59008
10809         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
10810         to int.
10811         * ipa-prop.c (ipa_print_node_params): Fix indentation.
10812
10813 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
10814
10815         PR debug/59350
10816         PR debug/59510
10817         * var-tracking.c (add_stores): Preserve the value of the source even if
10818         we don't record the store.
10819
10820 2014-01-06  Terry Guo  <terry.guo@arm.com>
10821
10822         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
10823
10824 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
10825
10826         PR bootstrap/59541
10827         * config/darwin.c (darwin_function_section): Adjust return values to
10828         correspond to optimisation changes made in r206070.
10829
10830 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
10831
10832         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
10833         from prefetch_block tune setting.
10834         (nocona_cost): Correct size of prefetch block to 64.
10835
10836 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
10837
10838         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
10839         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
10840         used to save the static chain register in the computation of the offset
10841         from which the FP registers need to be restored.
10842
10843 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
10844
10845         PR tree-optimization/59519
10846         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
10847         ICE if get_current_def (current_new_name) is already non-NULL, as long
10848         as it is a phi result of some other phi in *new_exit_bb that has
10849         the same argument.
10850
10851         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
10852         or vmovdqu* for misaligned_operand.
10853         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
10854         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
10855         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
10856         aligned_mem for AVX512F masked aligned load and store builtins and for
10857         non-temporal moves.
10858
10859 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
10860
10861         PR tree-optimization/59651
10862         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
10863         Address range for negative step should be added by TYPE_SIZE_UNIT.
10864
10865 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
10866
10867         * config/m68k/m68k.c (handle_move_double): Handle pushes with
10868         overlapping registers also for registers other than the stack pointer.
10869
10870 2014-01-03  Marek Polacek  <polacek@redhat.com>
10871
10872         PR other/59661
10873         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
10874         __builtin_FILE.
10875
10876 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
10877
10878         PR target/59625
10879         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
10880         asm goto as jump.
10881
10882         * config/i386/i386.md (MODE_SIZE): New mode attribute.
10883         (push splitter): Use <P:MODE_SIZE> instead of
10884         GET_MODE_SIZE (<P:MODE>mode).
10885         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
10886         (mov -1, reg peephole2): Likewise.
10887         * config/i386/sse.md (*mov<mode>_internal,
10888         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
10889         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
10890         *<code><mode>3, *andnot<mode>3<mask_name>,
10891         <mask_codefor><code><mode>3<mask_name>): Likewise.
10892         * config/i386/subst.md (mask_mode512bit_condition,
10893         sd_mask_mode512bit_condition): Likewise.
10894
10895 2014-01-02  Xinliang David Li  <davidxl@google.com>
10896
10897         PR tree-optimization/59303
10898         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
10899         (dump_predicates): Better output format.
10900         (pred_equal_p): New function.
10901         (is_neq_relop_p): Ditto.
10902         (is_neq_zero_form_p): Ditto.
10903         (pred_expr_equal_p): Ditto.
10904         (pred_neg_p): Ditto.
10905         (simplify_pred): Ditto.
10906         (simplify_preds_2): Ditto.
10907         (simplify_preds_3): Ditto.
10908         (simplify_preds_4): Ditto.
10909         (simplify_preds): Ditto.
10910         (push_pred): Ditto.
10911         (push_to_worklist): Ditto.
10912         (get_pred_info_from_cmp): Ditto.
10913         (is_degenerated_phi): Ditto.
10914         (normalize_one_pred_1): Ditto.
10915         (normalize_one_pred): Ditto.
10916         (normalize_one_pred_chain): Ditto.
10917         (normalize_preds): Ditto.
10918         (normalize_cond_1): Remove function.
10919         (normalize_cond): Ditto.
10920         (is_gcond_subset_of): Ditto.
10921         (is_subset_of_any): Ditto.
10922         (is_or_set_subset_of): Ditto.
10923         (is_and_set_subset_of): Ditto.
10924         (is_norm_cond_subset_of): Ditto.
10925         (pred_chain_length_cmp): Ditto.
10926         (convert_control_dep_chain_into_preds): Type change.
10927         (find_predicates): Ditto.
10928         (find_def_preds): Ditto.
10929         (destroy_predicates_vecs): Ditto.
10930         (find_matching_predicates_in_rest_chains): Ditto.
10931         (use_pred_not_overlap_with_undef_path_pred): Ditto.
10932         (is_pred_expr_subset): Ditto.
10933         (is_pred_chain_subset_of): Ditto.
10934         (is_included_in): Ditto.
10935         (is_superset_of): Ditto.
10936
10937 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10938
10939         Update copyright years.
10940
10941 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10942
10943         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
10944         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
10945         config/arc/arc.md, config/arc/arc.opt,
10946         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
10947         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
10948         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
10949         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
10950         config/linux-protos.h, config/linux.c, config/winnt-c.c,
10951         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
10952         vtable-verify.c, vtable-verify.h: Use the standard form for the
10953         copyright notice.
10954
10955 2014-01-02  Tobias Burnus  <burnus@net-b.de>
10956
10957         * gcc.c (process_command): Update copyright notice dates.
10958         * gcov-dump.c: Ditto.
10959         * gcov.c: Ditto.
10960         * doc/cpp.texi: Bump @copying's copyright year.
10961         * doc/cppinternals.texi: Ditto.
10962         * doc/gcc.texi: Ditto.
10963         * doc/gccint.texi: Ditto.
10964         * doc/gcov.texi: Ditto.
10965         * doc/install.texi: Ditto.
10966         * doc/invoke.texi: Ditto.
10967
10968 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
10969
10970         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
10971
10972 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
10973
10974         * config/i386/sse.md (*mov<mode>_internal): Guard
10975         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
10976
10977         PR rtl-optimization/59647
10978         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
10979         new_rtx into UNSIGNED_FLOAT rtxes.
10980 \f
10981 Copyright (C) 2014 Free Software Foundation, Inc.
10982
10983 Copying and distribution of this file, with or without modification,
10984 are permitted in any medium without royalty provided the copyright
10985 notice and this notice are preserved.