gcc/ChangeLog

   1 2014-05-17  Marc Glisse  <marc.glisse@inria.fr>
   2
   3         PR tree-optimization/61140
   4         PR tree-optimization/61150
   5         PR tree-optimization/61197
   6         * tree-ssa-phiopt.c (value_replacement): Punt on multiple phis.
   7
   8 2014-05-17  Uros Bizjak  <ubizjak@gmail.com>
   9
  10         * doc/invoke.texi (free): Mention Alpha.  Also enabled at -Os.
  11
  12 2014-05-17  Richard Sandiford  <r.sandiford@uk.ibm.com>
  13
  14         * wide-int.cc: Only include longlong.h if W_TYPE_SIZE==32 or
  15         __SIZEOF_INT128__ is defined.
  16
  17 2014-05-17  Richard Sandiford  <rdsandiford@googlemail.com>
  18
  19         * config/rs6000/rs6000.c (rs6000_real_tls_symbol_ref_p): New function.
  20         (rs6000_delegitimize_address): Use it.
  21
  22 2014-05-17  Richard Sandiford  <rdsandiford@googlemail.com>
  23
  24         * emit-rtl.h (replace_equiv_address, replace_equiv_address_nv): Add an
  25         inplace argument.  Store the new address in the original MEM when true.
  26         * emit-rtl.c (change_address_1): Likewise.
  27         (adjust_address_1, adjust_automodify_address_1, offset_address):
  28         Update accordingly.
  29         * rtl.h (plus_constant): Add an inplace argument.
  30         * explow.c (plus_constant): Likewise.  Try to reuse the original PLUS
  31         when true.  Avoid generating (plus X (const_int 0)).
  32         * function.c (instantiate_virtual_regs_in_rtx): Adjust the PLUS
  33         in-place.  Pass true to plus_constant.
  34         (instantiate_virtual_regs_in_insn): Pass true to replace_equiv_address.
  35
  36 2014-05-16  Dehao Chen  <dehao@google.com>
  37
  38         * tree-cfg.c (gimple_merge_blocks): Updates bb count with max count.
  39
  40 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
  41
  42         PR target/54089
  43         * config/sh/predicates.md (negt_reg_shl31_operand): Match additional
  44         patterns.
  45         * config/sh/sh.md (*negt_msb): Merge SH2A and non-SH2A variants.
  46
  47 2014-05-16  Dehao Chen  <dehao@google.com>
  48
  49         * ira-int.h (REG_FREQ_FROM_EDGE_FREQ): Use
  50         optimize_function_for_size_p.
  51         * regs.h (REG_FREQ_FROM_BB): Likewise.
  52
  53 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
  54
  55         PR target/51244
  56         * config/sh/sh.c (sh_eval_treg_value): Handle t_reg_operand and
  57         negt_reg_operand cases.
  58         * config/sh/sh.md (*cset_zero): Likewise by using cbranch_treg_value
  59         predicate.
  60         * config/sh/predicates.md (cbranch_treg_value): Simplify.
  61
  62 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
  63
  64         * config/sh/sh.c (sh_option_override): Set branch cost to 2 for all
  65         target variants.
  66
  67 2014-05-16  David Malcolm  <dmalcolm@redhat.com>
  68
  69         Revert:
  70         2014-04-29  David Malcolm  <dmalcolm@redhat.com>
  71
  72         * tree-cfg.c (dump_function_to_file): Dump the return type of
  73         functions, in a line to itself before the function body, mimicking
  74         the layout of a C function.
  75
  76 2014-05-16  Dehao Chen  <dehao@google.com>
  77
  78         * cfghooks.c (make_forwarder_block): Use direct computation to
  79         get fall-through edge's count and frequency.
  80
  81 2014-05-16  Benno Schulenberg  <bensberg@justemail.net>
  82
  83         * config/arc/arc.c (arc_init): Fix typo in error message.
  84         * config/i386/i386.c (ix86_expand_builtin): Likewise.
  85         (split_stack_prologue_scratch_regno): Likewise.
  86         * fortran/check.c (gfc_check_fn_rc2008): Remove duplicate
  87         word from error message.
  88
  89 2014-05-16  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
  90
  91         * ira-costs.c: Fix typo in comment.
  92
  93 2014-05-16  David Wohlferd <dw@LimeGreenSocks.com>
  94
  95         * doc/extend.texi: (Visibility Pragmas) Fix misplaced @xref
  96
  97 2014-05-16  Jan Hubicka  <hubicka@ucw.cz>
  98
  99         * varpool.c (dump_varpool_node): Dump write-only flag.
 100         * lto-cgraph.c (lto_output_varpool_node, input_varpool_node): Stream
 101         write-only flag.
 102         * tree-cfg.c (execute_fixup_cfg): Remove statements setting
 103         write-only variables.
 104         * ipa.c (process_references): New function.
 105         (set_readonly_bit): New function.
 106         (set_writeonly_bit): New function.
 107         (clear_addressable_bit): New function.
 108         (ipa_discover_readonly_nonaddressable_var): Mark write only variables;
 109         fix handling of aliases.
 110         * cgraph.h (struct varpool_node): Add writeonly flag.
 111
 112 2014-05-16  Vladimir Makarov  <vmakarov@redhat.com>
 113
 114         PR rtl-optimization/60969
 115         * ira-costs.c (record_reg_classes): Allow only memory for pseudo.
 116         Calculate costs for this case.
 117
 118 2014-05-16  Eric Botcazou  <ebotcazou@adacore.com>
 119
 120         * fold-const (fold_unary_loc) <NON_LVALUE_EXPR>: New case.
 121         <CASE_CONVERT>: Pass arg0 instead of op0 to fold_convert_const.
 122
 123 2014-05-16  Richard Biener  <rguenther@suse.de>
 124
 125         PR tree-optimization/61194
 126         * tree-vect-patterns.c (adjust_bool_pattern): Also handle
 127         bool patterns ending in a COND_EXPR.
 128
 129 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 130
 131         * config/aarch64/aarch64.c (aarch64_rtx_mult_cost): Fix FNMUL case.
 132
 133 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 134
 135         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle the case
 136         where we were unable to cost an RTX.
 137
 138 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 139
 140         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost SYMBOL_REF,
 141         HIGH, LO_SUM.
 142
 143 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 144             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 145
 146         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost TRUNCATE.
 147
 148 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 149             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 150
 151         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost FMA,
 152         FLOAT_EXTEND, FLOAT_TRUNCATE, ABS, SMAX, and SMIN.
 153
 154 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 155             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 156
 157         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost comparison
 158         operators.
 159
 160 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 161             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 162
 163         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
 164         DIV/MOD.
 165
 166 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 167             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 168
 169         * config/aarch64/aarch64.c (aarch64_rtx_arith_op_extract_p): New.
 170         (aarch64_rtx_costs): Improve costs for SIGN/ZERO_EXTRACT.
 171
 172 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 173             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 174
 175         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
 176         rotates and shifts.
 177
 178 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 179             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 180
 181         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost
 182         ZERO_EXTEND and SIGN_EXTEND better.
 183
 184 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 185             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 186
 187         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve cost for
 188         logical operations.
 189
 190 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 191             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 192
 193         * config/aarch64/aarch64.c (aarch64_rtx_costs): Use address
 194         costs when costing loads and stores to memory.
 195
 196 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 197             Philip Tomsich  <philipp.tomsich@theobroma-systems.com>
 198
 199         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costing
 200         for SET RTX.
 201
 202 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 203
 204         * config/aarch64/aarch64.c (aarch64_rtx_costs): Set default costs.
 205
 206 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 207             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 208
 209         * config/aarch64/aarch64.c (aarch64_strip_shift_or_extend): Rename
 210         to...
 211         (aarch64_strip_extend): ...this, don't strip shifts, check RTX is
 212         well formed.
 213         (aarch64_rtx_mult_cost): New.
 214         (aarch64_rtx_costs): Use it, refactor as appropriate.
 215
 216 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 217             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
 218
 219         * config/aarch64/aarch64.c (aarch64_build_constant): Conditionally
 220         emit instructions, return number of instructions which would
 221         be emitted.
 222         (aarch64_add_constant): Update call to aarch64_build_constant.
 223         (aarch64_output_mi_thunk): Likewise.
 224         (aarch64_rtx_costs): Estimate cost of a CONST_INT, cost of
 225         a CONST_DOUBLE.
 226
 227 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 228
 229         * config/aarch64/aarch64.c (aarch64_rtx_costs_wrapper): New.
 230         (TARGET_RTX_COSTS): Call it.
 231
 232 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 233
 234         * config/aarch64/aarch64.c (cortexa57_addrcost_table): New.
 235         (cortexa57_vector_cost): Likewise.
 236         (cortexa57_tunings): Use them.
 237
 238 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
 239
 240         * config/aarch64/aarch64-protos.h (scale_addr_mode_cost): New.
 241         (cpu_addrcost_table): Use it.
 242         * config/aarch64/aarch64.c (generic_addrcost_table): Initialize it.
 243         (aarch64_address_cost): Rewrite using aarch64_classify_address,
 244         move it.
 245
 246 2014-05-16  Richard Biener  <rguenther@suse.de>
 247
 248         * tree-ssa-sccvn.c: Include tree-cfg.h and domwalk.h.
 249         (set_ssa_val_to): Handle unexpected sets to VN_TOP.
 250         (visit_phi): Ignore edges marked as not executable.
 251         (class cond_dom_walker): New.
 252         (cond_dom_walker::before_dom_children): Value-number
 253         control statements and mark successor edges as not
 254         executable if possible.
 255         (run_scc_vn): First walk all control statements in
 256         dominator order, marking edges as not executable.
 257         * tree-inline.c (copy_edges_for_bb): Be not confused
 258         about random edge flags.
 259
 260 2014-05-16  Richard Biener  <rguenther@suse.de>
 261
 262         * tree-ssa-sccvn.c (visit_use): Also constant-fold calls.
 263
 264 2014-05-15  Peter Bergner  <bergner@vnet.ibm.com>
 265
 266         PR target/61193
 267         * config/rs6000/htmxlintrin.h (_HTM_TBEGIN_STARTED): New define.
 268         (__TM_simple_begin): Use it.
 269         (__TM_begin): Likewise.
 270
 271 2014-05-15  Martin Jambor  <mjambor@suse.cz>
 272
 273         PR ipa/61085
 274         * ipa-prop.c (update_indirect_edges_after_inlining): Check
 275         type_preserved flag when the indirect edge is polymorphic.
 276
 277 2014-05-15  Martin Jambor  <mjambor@suse.cz>
 278
 279         PR tree-optimization/61090
 280         * tree-sra.c (sra_modify_expr): Pass the current gsi to
 281         build_ref_for_model.
 282
 283 2014-05-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 284
 285         * config/arm/arm.c (arm_option_override): Use the SCHED_PRESSURE_MODEL
 286         enum name for PARAM_SCHED_PRESSURE_ALGORITHM.
 287
 288 2014-05-15  Jakub Jelinek  <jakub@redhat.com>
 289
 290         PR tree-optimization/61158
 291         * fold-const.c (fold_binary_loc): If X is zero-extended and
 292         shiftc >= prec, make sure zerobits is all ones instead of
 293         invoking undefined behavior.
 294
 295 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 296
 297         * regcprop.h: New file.
 298         * regcprop.c (skip_debug_insn_p): New decl.
 299         (replace_oldest_value_reg): Check skip_debug_insn_p.
 300         (copyprop_hardreg_forward_bb_without_debug_insn): New function.
 301         * shrink-wrap.c: Include regcprop.h.
 302         (prepare_shrink_wrap): Call
 303         copyprop_hardreg_forward_bb_without_debug_insn.
 304
 305 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 306
 307         * shrink-wrap.h: Update comment.
 308         * shrink-wrap.c: Update comment.
 309         (next_block_for_reg): Rename to live_edge_for_reg.
 310         (live_edge_for_reg): Allow live_edge->dest has two predecessors.
 311         (move_insn_for_shrink_wrap): Split live_edge.
 312         (prepre_shrink_wrap): One more parameter for move_insn_for_shrink_wrap.
 313
 314 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
 315
 316         * config/sparc/sparc-protos.h (sparc_absnegfloat_split_legitimate):
 317         Delete.
 318         * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): Likewise.
 319         * config/sparc/sparc.md (fptype_ut699): New attribute.
 320         (in_branch_delay): Return false if -mfix-ut699 is specified and
 321         fptype_ut699 is set to single.
 322         (truncdfsf2): Add fptype_ut699 attribute.
 323         (fix_truncdfsi2): Likewise.
 324         (floatsisf2): Change fptype attribute.
 325         (fix_truncsfsi2): Likewise.
 326         (negtf2_notv9): Delete.
 327         (negtf2_v9): Likewise.
 328         (negtf2_hq): New instruction.
 329         (negtf2): New instruction and splitter.
 330         (negdf2_notv9): Rewrite.
 331         (abstf2_notv9): Delete.
 332         (abstf2_hq_v9): Likewise.
 333         (abstf2_v9): Likewise.
 334         (abstf2_hq): New instruction.
 335         (abstf2): New instruction and splitter.
 336         (absdf2_notv9): Rewrite.
 337
 338 2014-05-14  Cary Coutant  <ccoutant@google.com>
 339
 340         PR debug/61013
 341         * opts.c (common_handle_option): Don't special-case "-g".
 342         (set_debug_level): Default to at least level 2 with "-g".
 343
 344 2014-05-14  DJ Delorie  <dj@redhat.com>
 345
 346         * config/msp430/msp430.c (msp430_builtin): Add
 347         MSP430_BUILTIN_DELAY_CYCLES.
 348         (msp430_init_builtins): Register void __delay_cycles(long long).
 349         (msp430_builtin_decl): Add it.
 350         (cg_magic_constant): New.
 351         (msp430_expand_delay_cycles): New.
 352         (msp430_expand_builtin): Call it.
 353         (msp430_print_operand_raw): Change integer printing from "int" to
 354         HOST_WIDE_INT.
 355         * config/msp430/msp430.md (define_constants): Add delay_cycles tags.
 356         (delay_cycles_start): New.
 357         (delay_cycles_end): New.
 358         (delay_cycles_32): New.
 359         (delay_cycles_32x): New.
 360         (delay_cycles_16): New.
 361         (delay_cycles_16x): New.
 362         (delay_cycles_2): New.
 363         (delay_cycles_1): New.
 364         * doc/extend.texi: Document __delay_cycles().
 365
 366 2014-05-14  Sandra Loosemore  <sandra@codesourcery.com>
 367
 368         * config/nios2/nios2.md (nios2_cbranch): Fix paste-o in
 369         length attribute computation.
 370
 371 2014-05-14  Richard Sandiford  <rdsandiford@googlemail.com>
 372
 373         PR debug/61188
 374         * print-rtl.c (print_rtx): Suppress uids if flag_dump_unnumbered.
 375
 376 2014-05-14  Richard Sandiford  <r.sandiford@uk.ibm.com>
 377
 378         PR target/61084
 379         * config/sparc/sparc.md: Fix types of low and high in DI constant
 380         splitter.  Use gen_int_mode in some other splitters.
 381
 382 2014-05-14  Martin Jambor  <mjambor@suse.cz>
 383
 384         PR ipa/60897
 385         * ipa-prop.c (ipa_modify_formal_parameters): Reset DECL_LANG_SPECIFIC.
 386
 387 2014-05-14  James Norris  <jnorris@codesourcery.com>
 388
 389         * omp-low.c (expand_parallel_call): Remove shadow variable.
 390         (expand_omp_taskreg): Likewise.
 391
 392 2014-05-14  Ilya Tocar  <ilya.tocar@intel.com>
 393
 394         * common/config/i386/i386-common.c
 395         (OPTION_MASK_ISA_CLFLUSHOPT_SET): Define.
 396         (OPTION_MASK_ISA_XSAVES_SET): Ditto.
 397         (OPTION_MASK_ISA_XSAVEC_SET): Ditto.
 398         (OPTION_MASK_ISA_CLFLUSHOPT_UNSET): Ditto.
 399         (OPTION_MASK_ISA_XSAVES_UNSET): Ditto.
 400         (OPTION_MASK_ISA_XSAVEC_UNSET): Ditto.
 401         (ix86_handle_option): Handle OPT_mxsavec, OPT_mxsaves, OPT_mclflushopt.
 402         * config.gcc (i[34567]86-*-*): Add clflushoptintrin.h,
 403         xsavecintrin.h, xsavesintrin.h.
 404         (x86_64-*-*): Ditto.
 405         * config/i386/clflushoptintrin.h: New.
 406         * config/i386/xsavecintrin.h: Ditto.
 407         * config/i386/xsavesintrin.h: Ditto.
 408         * config/i386/cpuid.h (bit_CLFLUSHOPT): Define.
 409         (bit_XSAVES): Ditto.
 410         (bit_XSAVES): Ditto.
 411         * config/i386/driver-i386.c (host_detect_local_cpu): Handle
 412         -mclflushopt, -mxsavec, -mxsaves, -mno-xsaves, -mno-xsavec,
 413         -mno-clflushopt.
 414         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
 415         OPTION_MASK_ISA_CLFLUSHOPT, OPTION_MASK_ISA_XSAVEC,
 416         OPTION_MASK_ISA_XSAVES.
 417         * config/i386/i386.c (ix86_target_string): Handle -mclflushopt,
 418         -mxsavec, -mxsaves.
 419         (PTA_CLFLUSHOPT) Define.
 420         (PTA_XSAVEC): Ditto.
 421         (PTA_XSAVES): Ditto.
 422         (ix86_option_override_internal): Handle new options.
 423         (ix86_valid_target_attribute_inner_p): Ditto.
 424         (ix86_builtins): Add IX86_BUILTIN_XSAVEC, IX86_BUILTIN_XSAVEC64,
 425         IX86_BUILTIN_XSAVES, IX86_BUILTIN_XRSTORS, IX86_BUILTIN_XSAVES64,
 426         IX86_BUILTIN_XRSTORS64, IX86_BUILTIN_CLFLUSHOPT.
 427         (bdesc_special_args): Add __builtin_ia32_xsaves,
 428         __builtin_ia32_xrstors, __builtin_ia32_xsavec, __builtin_ia32_xsaves64,
 429         __builtin_ia32_xrstors64, __builtin_ia32_xsavec64.
 430         (ix86_init_mmx_sse_builtins): Add __builtin_ia32_clflushopt.
 431         (ix86_expand_builtin): Handle new builtins.
 432         * config/i386/i386.h (TARGET_CLFLUSHOPT) Define.
 433         (TARGET_CLFLUSHOPT_P): Ditto.
 434         (TARGET_XSAVEC): Ditto.
 435         (TARGET_XSAVEC_P): Ditto.
 436         (TARGET_XSAVES): Ditto.
 437         (TARGET_XSAVES_P): Ditto.
 438         * config/i386/i386.md (ANY_XSAVE): Add UNSPECV_XSAVEC, UNSPECV_XSAVES.
 439         (ANY_XSAVE64)" Add UNSPECV_XSAVEC64, UNSPECV_XSAVES64.
 440         (attr xsave): Add xsavec, xsavec64, xsaves, xsaves64.
 441         (ANY_XRSTOR): New.
 442         (ANY_XRSTOR64): Ditto.
 443         (xrstor): Ditto.
 444         (xrstor): Change into <xrstor>.
 445         (xrstor_rex64): Change into <xrstor>_rex64.
 446         (xrstor64): Change into <xrstor>64
 447         (clflushopt): New.
 448         * config/i386/i386.opt (mclflushopt): New.
 449         (mxsavec): Ditto.
 450         (mxsaves): Ditto.
 451         * config/i386/x86intrin.h: Add clflushoptintrin.h, xsavesintrin.h,
 452         xsavecintrin.h.
 453         * doc/invoke.texi: Document new options.
 454
 455 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
 456
 457         PR rtl-optimization/60866
 458         * sel-sched-ir (sel_init_new_insn): New parameter old_seqno.
 459         Default it to -1.  Pass it down to init_simplejump_data.
 460         (init_simplejump_data): New parameter old_seqno.  Pass it down
 461         to get_seqno_for_a_jump.
 462         (get_seqno_for_a_jump): New parameter old_seqno.  Use it for
 463         initializing new jump seqno as a last resort.  Add comment.
 464         (sel_redirect_edge_and_branch): Save old seqno of the conditional
 465         jump and pass it down to sel_init_new_insn.
 466         (sel_redirect_edge_and_branch_force): Likewise.
 467
 468 2014-05-14  Georg-Johann Lay  <avr@gjlay.de>
 469
 470         * config/avr/avr.h (REG_CLASS_CONTENTS): Use unsigned suffix for
 471         shifted values to avoid build warning.
 472
 473 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
 474
 475         * cfgcleanup.c (try_forward_edges): Use location_t for locations.
 476         * cfgrtl.c (rtl_merge_blocks): Fix comment.
 477         (cfg_layout_merge_blocks): Likewise.
 478         * except.c (emit_to_new_bb_before): Remove prev_bb local variable.
 479
 480 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
 481
 482         PR rtl-optimization/60901
 483         * config/i386/i386.c (ix86_dependencies_evaluation_hook): Check that
 484         bb predecessor belongs to the same scheduling region.  Adjust comment.
 485
 486 2014-05-13  Peter Bergner  <bergner@vnet.ibm.com>
 487
 488         * doc/sourcebuild.texi: (dfp_hw): Document.
 489         (p8vector_hw): Likewise.
 490         (powerpc_eabi_ok): Likewise.
 491         (powerpc_elfv2): Likewise.
 492         (powerpc_htm_ok): Likewise.
 493         (ppc_recip_hw): Likewise.
 494         (vsx_hw): Likewise.
 495
 496 2014-05-13  Cary Coutant  <ccoutant@google.com>
 497
 498         * opts.c (finish_options): Use -ggnu-pubnames with -gsplit-dwarf.
 499
 500 2014-05-13  David Malcolm  <dmalcolm@redhat.com>
 501
 502         * gengtype-parse.c (require3): Eliminate in favor of...
 503         (require4): New.
 504         (require_template_declaration): Update to support optional single *
 505         on a type.
 506
 507         * gengtype.c (get_ultimate_base_class): Add a non-const overload.
 508         (create_user_defined_type): Handle a single level of explicit
 509         pointerness within template arguments.
 510         (struct write_types_data): Add field "kind".
 511         (filter_type_name): Handle "*" character.
 512         (write_user_func_for_structure_ptr): Require a write_types_data
 513         rather than just a prefix string, so that we can look up the kind
 514         of the wtd and use it as an index into wrote_user_func_for_ptr,
 515         ensuring that such functions are written at most once.  Support
 516         subclasses by invoking the marking function of the ultimate base class.
 517         (write_user_func_for_structure_body): Require a write_types_data
 518         rather than just a prefix string, so that we can pass this to
 519         write_user_func_for_structure_ptr.
 520         (write_func_for_structure): Likewise.
 521         (ggc_wtd): Add initializer of new "kind" field.
 522         (pch_wtd): Likewise.
 523
 524         * gengtype.h (enum write_types_kinds): New.
 525         (struct type): Add field wrote_user_func_for_ptr to the "s"
 526         union member.
 527
 528 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
 529
 530         * fold-const.c (optimize_bit_field_compare): Use wi:: operations
 531         instead of const_binop.
 532         (fold_binary_loc): Likewise.
 533
 534 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
 535
 536         * tree-dfa.h (get_addr_base_and_unit_offset_1): Update array index
 537         calculation to match get_ref_base_and_extent.
 538
 539 2014-05-13  Catherine Moore  <clm@codesourcery.com>
 540             Sandra Loosemore  <sandra@codesourcery.com>
 541
 542         * configure.ac: Fix assembly for explicit JALR relocation check.
 543         * configure: Regenerate.
 544
 545 2014-05-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 546
 547         * config/arm/arm.c (neon_itype): Remove NEON_RESULTPAIR.
 548         (arm_init_neon_builtins): Remove handling of NEON_RESULTPAIR.
 549         Remove associated type declarations and initialisations.
 550         (arm_expand_neon_builtin): Likewise.
 551         (neon_emit_pair_result_insn): Delete.
 552         * config/arm/arm_neon_builtins (vtrn, vzip, vuzp): Delete.
 553         * config/arm/neon.md (neon_vtrn<mode>): Delete.
 554         (neon_vzip<mode>): Likewise.
 555         (neon_vuzp<mode>): Likewise.
 556
 557 2014-05-13  Richard Biener  <rguenther@suse.de>
 558
 559         PR ipa/60973
 560         * tree-inline.c (remap_gimple_stmt): Clear tail call flag,
 561         it needs revisiting whether the call still may be tail-called.
 562
 563 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 564
 565         * rtl.def (SYMBOL_REF): Remove middle "0" field.
 566         * rtl.h (block_symbol): Reduce number of fields to 2.
 567         (rtx_def): Add u2.symbol_ref_flags.
 568         (SYMBOL_REF_FLAGS): Use it.
 569         (SYMBOL_REF_DATA, SET_SYMBOL_REF_DECL, SYMBOL_REF_DECL)
 570         (SET_SYMBOL_REF_CONSTANT, SYMBOL_REF_CONSTANT): Lower index.
 571         * gengtype.c (adjust_field_rtx_def): Remove SYMBOL_REF_FLAGS handling.
 572         Lower index of SYMBOL_REF_DATA.
 573         * print-rtl.c (print_rtx): Lower index for SYMBOL_REF_DATA.
 574         Print SYMBOL_REF_FLAGS at the same time.
 575         * genattrtab.c (attr_rtx_1): Only initialize 1 "0" SYMBOL_REF field.
 576
 577 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 578
 579         * rtl.def (VAR_LOCATION): Remove "i" field.
 580         * rtl.h (rtx_def): Add u2.var_location_status.
 581         (PAT_VAR_LOCATION_STATUS): Use it.
 582         (gen_rtx_VAR_LOCATION): Declare.
 583         * gengenrtl.c (excluded_rtx): Add VAR_LOCATION.
 584         * emit-rtl.c (gen_rtx_VAR_LOCATION): New function.
 585         * var-tracking.c (emit_note_insn_var_location): Remove casts.
 586
 587 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 588
 589         * rtl.def (scratch): Fix outdated comment and remove "0" field.
 590         * gengtype.c (adjust_field_rtx_def): Update accordingly.
 591
 592 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 593
 594         * rtl.def (DEBUG_INSN, INSN, JUMP_INSN, CALL_INSN, JUMP_TABLE_DATA)
 595         (BARRIER, CODE_LABEL, NOTE): Remove first "i" field.
 596         * rtl.h (rtx_def): Add insn_uid to u2 field.
 597         (RTX_FLAG_CHECK8): Delete in favor of...
 598         (RTL_INSN_CHAIN_FLAG_CHECK): ...this new macro.
 599         (INSN_DELETED_P): Update accordingly.
 600         (INSN_UID): Use u2.insn_uid.
 601         (INSN_CHAIN_CODE_P): Define.
 602         (PREV_INSN, NEXT_INSN, BLOCK_FOR_INSN, PATTERN, INSN_LOCATION)
 603         (INSN_CODE, REG_NOTES, CALL_INSN_FUNCTION_USAGE, CODE_LABEL_NUMBER)
 604         (NOTE_DATA, NOTE_DELETED_LABEL_NAME, NOTE_BLOCK, NOTE_EH_HANDLER)
 605         (NOTE_BASIC_BLOCK, NOTE_VAR_LOCATION, NOTE_CFI, NOTE_LABEL_NUMBER)
 606         (NOTE_KIND, LABEL_NAME, LABEL_NUSES, JUMP_LABEL, LABEL_REFS): Lower
 607         indices accordingly.
 608         * print-rtl.c (print_rtx): Print INSN_UIDs before the main loop.
 609         Update indices for insn-chain rtxes.
 610         * gengtype.c (gen_rtx_next): Adjust test for insn-chain rtxes.
 611         (adjust_field_rtx_def): Lower '0' indices for all insn-chain rtxes.
 612         * emit-rtl.c (gen_label_rtx): Update gen_rtx_LABEL call.
 613         * caller-save.c (init_caller_save): Update gen_rtx_INSN calls.
 614         * combine.c (try_combine): Likewise.
 615         * ira.c (setup_prohibited_mode_move_regs): Likewise.
 616
 617 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 618
 619         * rtl.def (REG): Remove middle field.
 620         * rtl.h (rtx_def): Add orignal_regno to u2.
 621         (ORIGINAL_REGNO): Use it instead of field 1.
 622         (REG_ATTRS): Lower field index accordingly.
 623         * gengtype.c (adjust_field_rtx_def): Remove handling of
 624         ORIGINAL_REGNO.  Move REG_ATTRS index down.
 625         * print-rtl.c (print_rtx): Move ORIGINAL_REGNO handling to the
 626         code that prints the REGNO.
 627
 628 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 629
 630         * print-rtl.c (print_rtx): Guard whole '0' block with ifndef
 631         GENERATOR_FILE.
 632
 633 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 634
 635         * rtl.h (rtx_def): Mark u2 as GTY ((skip)).
 636
 637 2014-05-13  Bin Cheng  <bin.cheng@arm.com>
 638
 639         * tree-ssa-loop-ivopts.c (contain_complex_addr_expr): New.
 640         (alloc_iv): Lower base expressions containing ADDR_EXPR.
 641
 642 2014-05-13  Ian Bolton  <ian.bolton@arm.com>
 643
 644         * config/aarch64/aarch64-protos.h
 645         (aarch64_hard_regno_caller_save_mode): New prototype.
 646         * config/aarch64/aarch64.c (aarch64_hard_regno_caller_save_mode):
 647         New function.
 648         * config/aarch64/aarch64.h (HARD_REGNO_CALLER_SAVE_MODE): New macro.
 649
 650 2014-05-13  Christian Bruel  <christian.bruel@st.com>
 651
 652         * target.def (mode_switching): New hook vector.
 653         (mode_emit, mode_needed, mode_after, mode_entry): New hooks.
 654         (mode_exit, modepriority_to_mode): Likewise.
 655         * mode-switching.c (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Hookify.
 656         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 657         * target.h: Include tm.h and hard-reg-set.h.
 658         * doc/tm.texi.in (EMIT_MODE_SET, MODE_NEEDED, MODE_AFTER, MODE_ENTRY)
 659         (MODE_EXIT, MODE_PRIORITY_TO_MODE): Delete and hookify.
 660         * doc/tm.texi Regenerate.
 661         * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 662         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 663         * config/sh/sh.c (sh_emit_mode_set, sh_mode_priority): Hookify.
 664         (sh_mode_needed, sh_mode_after, sh_mode_entry, sh_mode_exit): Likewise.
 665         * config/i386/i386.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 666         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 667         * config/i386/i386-protos.h (ix86_mode_needed, ix86_mode_after)
 668         (ix86_mode_entrym, ix86_emit_mode_set): Remove external declaration.
 669         * config/i386/i386.c (ix86_mode_needed, ix86_mode_after,
 670         (ix86_mode_exit, ix86_mode_entry, ix86_mode_priority)
 671         (ix86_emit_mode_set): Hookify.
 672         * config/epiphany/epiphany.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY):
 673         Delete.
 674         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 675         * config/epiphany/epiphany-protos.h (epiphany_mode_needed)
 676         (emit_set_fp_mode, epiphany_mode_entry_exit, epiphany_mode_after)
 677         (epiphany_mode_priority_to_mode): Remove declaration.
 678         * config/epiphany/epiphany.c (emit_set_fp_mode): Hookify.
 679         (epiphany_mode_needed, epiphany_mode_priority_to_mode): Likewise.
 680         (epiphany_mode_entry, epiphany_mode_exit, epiphany_mode_after):
 681         Likewise.
 682         (epiphany_mode_priority_to_mode): Change priority type.  Hookify.
 683         (epiphany_mode_needed, epiphany_mode_entry_exit): Hookify.
 684         (epiphany_mode_after, epiphany_mode_entry, emit_set_fp_mode): Hookify.
 685
 686 2014-05-13  Jakub Jelinek  <jakub@redhat.com>
 687
 688         PR target/61060
 689         * config/i386/i386.c (ix86_expand_set_or_movmem): If count_exp
 690         is const0_rtx, return immediately.  Don't test count == 0 when
 691         it is always true.
 692
 693 2014-05-13  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 694
 695         * Makefile.in: add shrink-wrap.o.
 696         * config/i386/i386.c: include "shrink-wrap.h"
 697         * function.c: Likewise.
 698         (requires_stack_frame_p, next_block_for_reg,
 699         move_insn_for_shrink_wrap, prepare_shrink_wrap,
 700         dup_block_and_redirect): Move to shrink-wrap.c
 701         (thread_prologue_and_epilogue_insns): Extract three code segments
 702         as functions in shrink-wrap.c
 703         * function.h: Move #ifdef HAVE_simple_return ... #endif block to
 704         shrink-wrap.h
 705         * shrink-wrap.c: New file.
 706         * shrink-wrap.h: New file.
 707
 708 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
 709
 710         * doc/extend.texi: Reflect current numbers of pragmas.  Remove
 711         reference to Solaris.
 712
 713 2014-05-12  Mike Stump  <mikestump@comcast.net>
 714
 715         PR other/31778
 716         * genattrtab.c (filename): Add.
 717         (convert_set_attr_alternative): Improve error message.
 718         (check_defs): Restore read_md_filename for error messages.
 719         (gen_insn): Save filename.
 720
 721 2014-05-12  Dimitris Papavasiliou  <dpapavas@gmail.com>
 722
 723         * doc/invoke.texi: Document new switches -Wno-shadow-ivar,
 724         -fno-local-ivars and -fivar-visibility.
 725         * c-family/c.opt: Make -Wshadow also implicitly enable
 726         -Wshadow-ivar.
 727
 728 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
 729
 730         * doc/tm.texi: Remove reference to deleted macro.
 731         * doc/tm.texi.in: Likewise.
 732
 733 2014-05-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
 734
 735         PR target/60991
 736         * config/avr/avr.c (avr_out_store_psi): Use correct constant
 737         to restore Y.
 738
 739 2014-05-12  Georg-Johann Lay  <avr@gjlay.de>
 740
 741         * config/arm/arm.h (License): Add GCC Runtime Library Exception.
 742         * config/arm/aout.h (License): Same.
 743         * config/arm/bpabi.h (License): Same.
 744         * config/arm/elf.h (License): Same.
 745         * config/arm/linux-elf.h (License): Same.
 746         * config/arm/linux-gas.h (License): Same.
 747         * config/arm/netbsd-elf.h (License): Same.
 748         * config/arm/uclinux-eabi.h (License): Same.
 749         * config/arm/uclinux-elf.h (License): Same.
 750         * config/arm/vxworks.h (License): Same.
 751
 752 2014-05-11  Jakub Jelinek  <jakub@redhat.com>
 753
 754         * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
 755         * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
 756         number of operands to 3.
 757         (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
 758         * tree-nested.c (convert_nonlocal_omp_clauses,
 759         convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
 760         * gimplify.c (gimplify_scan_omp_clauses): Handle
 761         OMP_CLAUSE_LINEAR_STMT.
 762         * omp-low.c (lower_rec_input_clauses): Fix typo.
 763         (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
 764         cast between Fortran boolean_type_node and C _Bool if
 765         needed.
 766
 767 2014-05-11  Richard Sandiford  <rdsandiford@googlemail.com>
 768
 769         PR tree-optimization/61136
 770         * wide-int.h (multiple_of_p): Define a version that doesn't return
 771         the quotient.
 772         * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
 773         integer_zerop/const_binop pair.
 774         (multiple_of_p): Likewise, converting both operands to widest_int
 775         precision.
 776
 777 2014-05-09  Teresa Johnson  <tejohnson@google.com>
 778
 779         * cgraphunit.c (analyze_functions): Use correct dump file.
 780
 781 2014-05-09  Florian Weimer  <fweimer@redhat.com>
 782
 783         * cfgexpand.c (stack_protect_decl_p): New function, extracted from
 784         expand_used_vars.
 785         (stack_protect_return_slot_p): New function.
 786         (expand_used_vars): Call stack_protect_decl_p and
 787         stack_protect_return_slot_p for -fstack-protector-strong.
 788
 789 2014-05-09  Kenneth Zadeck  <zadeck@naturalbridge.com>
 790
 791         PR middle-end/61111
 792         * fold-const.c (fold_binary_loc): Changed width of mask.
 793
 794 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 795
 796         * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
 797         unsigned int initializers for regno_in, regno_out.
 798
 799 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 800
 801         PR target/61055
 802         * config/avr/avr.md (cc): Add new attribute set_vzn.
 803         (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
 804         Set cc insn attribute to set_vzn instead of set_zn for alternatives
 805         with INC, DEC or NEG.
 806         * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
 807         (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
 808         INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
 809
 810 2014-05-09  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 811
 812         Revert:
 813         2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 814
 815         * wide-int.cc (UTItype): Define.
 816         (UDWtype): Define for appropriate W_TYPE_SIZE.
 817
 818 2014-05-09  Richard Biener  <rguenther@suse.de>
 819
 820         * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
 821         * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
 822         (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
 823         (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
 824         ssa_propagate): Adjust.
 825
 826 2014-05-08  Jeff Law  <law@redhat.com>
 827
 828         PR tree-optimization/61009
 829         * tree-ssa-threadedge.c (thread_through_normal_block): Return a
 830         tri-state rather than a boolean.  When a block is too big to
 831         thread through, inform caller via negative return value.
 832         (thread_across_edge): If a block was too big for normal threading,
 833         then it's too big for a joiner too, so remove temporary equivalences
 834         and return immediately.
 835
 836 2014-05-08  Manuel López-Ibáñez  <manu@gcc.gnu.org>
 837             Matthias Klose  <doko@ubuntu.com>
 838
 839         PR driver/61106
 840         * optc-gen.awk: Fix option handling for -Wunused-parameter.
 841
 842 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 843
 844         PR target/59952
 845         * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
 846
 847 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 848
 849         PR target/61092
 850         * config/alpha/alpha.c: Include gimple-iterator.h.
 851         (alpha_gimple_fold_builtin): New function.  Move
 852         ALPHA_BUILTIN_UMULH folding from ...
 853         (alpha_fold_builtin): ... here.
 854         (TARGET_GIMPLE_FOLD_BUILTIN): New define.
 855
 856 2014-05-08  Wei Mi  <wmi@google.com>
 857
 858         PR target/58066
 859         * config/i386/i386.c (ix86_compute_frame_layout): Update
 860         preferred_stack_boundary for call, expanded from tls descriptor.
 861         * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
 862         to depend on SP register.
 863         (*tls_local_dynamic_base_32_gnu): Ditto.
 864         (*tls_local_dynamic_32_once): Ditto.
 865         (tls_global_dynamic_64_<mode>): Set
 866         ix86_tls_descriptor_calls_expanded_in_cfun.
 867         (tls_local_dynamic_base_64_<mode>): Ditto.
 868         (tls_global_dynamic_32): Set
 869         ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
 870         to depend on SP register.
 871         (tls_local_dynamic_base_32): Ditto.
 872
 873 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 874
 875         * config/arm/arm_neon.h: Update comment.
 876         * config/arm/neon-docgen.ml: Delete.
 877         * config/arm/neon-gen.ml: Delete.
 878         * doc/arm-neon-intrinsics.texi: Update comment.
 879
 880 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 881
 882         * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
 883         and v4sf versions.
 884         (vand, vorr, veor, vorn, vbic): Remove.
 885         * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
 886         iterator.
 887         (neon_vsub_unspec): Likewise.
 888         (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
 889
 890 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 891
 892         * config/arm/arm_neon.h (vadd_s8): GNU C implementation
 893         (vadd_s16): Likewise.
 894         (vadd_s32): Likewise.
 895         (vadd_f32): Likewise.
 896         (vadd_u8): Likewise.
 897         (vadd_u16): Likewise.
 898         (vadd_u32): Likewise.
 899         (vadd_s64): Likewise.
 900         (vadd_u64): Likewise.
 901         (vaddq_s8): Likewise.
 902         (vaddq_s16): Likewise.
 903         (vaddq_s32): Likewise.
 904         (vaddq_s64): Likewise.
 905         (vaddq_f32): Likewise.
 906         (vaddq_u8): Likewise.
 907         (vaddq_u16): Likewise.
 908         (vaddq_u32): Likewise.
 909         (vaddq_u64): Likewise.
 910         (vmul_s8): Likewise.
 911         (vmul_s16): Likewise.
 912         (vmul_s32): Likewise.
 913         (vmul_f32): Likewise.
 914         (vmul_u8): Likewise.
 915         (vmul_u16): Likewise.
 916         (vmul_u32): Likewise.
 917         (vmul_p8): Likewise.
 918         (vmulq_s8): Likewise.
 919         (vmulq_s16): Likewise.
 920         (vmulq_s32): Likewise.
 921         (vmulq_f32): Likewise.
 922         (vmulq_u8): Likewise.
 923         (vmulq_u16): Likewise.
 924         (vmulq_u32): Likewise.
 925         (vsub_s8): Likewise.
 926         (vsub_s16): Likewise.
 927         (vsub_s32): Likewise.
 928         (vsub_f32): Likewise.
 929         (vsub_u8): Likewise.
 930         (vsub_u16): Likewise.
 931         (vsub_u32): Likewise.
 932         (vsub_s64): Likewise.
 933         (vsub_u64): Likewise.
 934         (vsubq_s8): Likewise.
 935         (vsubq_s16): Likewise.
 936         (vsubq_s32): Likewise.
 937         (vsubq_s64): Likewise.
 938         (vsubq_f32): Likewise.
 939         (vsubq_u8): Likewise.
 940         (vsubq_u16): Likewise.
 941         (vsubq_u32): Likewise.
 942         (vsubq_u64): Likewise.
 943         (vand_s8): Likewise.
 944         (vand_s16): Likewise.
 945         (vand_s32): Likewise.
 946         (vand_u8): Likewise.
 947         (vand_u16): Likewise.
 948         (vand_u32): Likewise.
 949         (vand_s64): Likewise.
 950         (vand_u64): Likewise.
 951         (vandq_s8): Likewise.
 952         (vandq_s16): Likewise.
 953         (vandq_s32): Likewise.
 954         (vandq_s64): Likewise.
 955         (vandq_u8): Likewise.
 956         (vandq_u16): Likewise.
 957         (vandq_u32): Likewise.
 958         (vandq_u64): Likewise.
 959         (vorr_s8): Likewise.
 960         (vorr_s16): Likewise.
 961         (vorr_s32): Likewise.
 962         (vorr_u8): Likewise.
 963         (vorr_u16): Likewise.
 964         (vorr_u32): Likewise.
 965         (vorr_s64): Likewise.
 966         (vorr_u64): Likewise.
 967         (vorrq_s8): Likewise.
 968         (vorrq_s16): Likewise.
 969         (vorrq_s32): Likewise.
 970         (vorrq_s64): Likewise.
 971         (vorrq_u8): Likewise.
 972         (vorrq_u16): Likewise.
 973         (vorrq_u32): Likewise.
 974         (vorrq_u64): Likewise.
 975         (veor_s8): Likewise.
 976         (veor_s16): Likewise.
 977         (veor_s32): Likewise.
 978         (veor_u8): Likewise.
 979         (veor_u16): Likewise.
 980         (veor_u32): Likewise.
 981         (veor_s64): Likewise.
 982         (veor_u64): Likewise.
 983         (veorq_s8): Likewise.
 984         (veorq_s16): Likewise.
 985         (veorq_s32): Likewise.
 986         (veorq_s64): Likewise.
 987         (veorq_u8): Likewise.
 988         (veorq_u16): Likewise.
 989         (veorq_u32): Likewise.
 990         (veorq_u64): Likewise.
 991         (vbic_s8): Likewise.
 992         (vbic_s16): Likewise.
 993         (vbic_s32): Likewise.
 994         (vbic_u8): Likewise.
 995         (vbic_u16): Likewise.
 996         (vbic_u32): Likewise.
 997         (vbic_s64): Likewise.
 998         (vbic_u64): Likewise.
 999         (vbicq_s8): Likewise.
1000         (vbicq_s16): Likewise.
1001         (vbicq_s32): Likewise.
1002         (vbicq_s64): Likewise.
1003         (vbicq_u8): Likewise.
1004         (vbicq_u16): Likewise.
1005         (vbicq_u32): Likewise.
1006         (vbicq_u64): Likewise.
1007         (vorn_s8): Likewise.
1008         (vorn_s16): Likewise.
1009         (vorn_s32): Likewise.
1010         (vorn_u8): Likewise.
1011         (vorn_u16): Likewise.
1012         (vorn_u32): Likewise.
1013         (vorn_s64): Likewise.
1014         (vorn_u64): Likewise.
1015         (vornq_s8): Likewise.
1016         (vornq_s16): Likewise.
1017         (vornq_s32): Likewise.
1018         (vornq_s64): Likewise.
1019         (vornq_u8): Likewise.
1020         (vornq_u16): Likewise.
1021         (vornq_u32): Likewise.
1022         (vornq_u64): Likewise.
1023
1024 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1025
1026         * wide-int.cc (UTItype): Define.
1027         (UDWtype): Define for appropriate W_TYPE_SIZE.
1028
1029 2014-05-08  Marc Glisse  <marc.glisse@inria.fr>
1030
1031         PR tree-optimization/59100
1032         * tree-ssa-phiopt.c: Include tree-inline.h.
1033         (neutral_element_p, absorbing_element_p): New functions.
1034         (value_replacement): Handle conditional binary operations with a
1035         neutral or absorbing element.
1036
1037 2014-05-08  Richard Biener  <rguenther@suse.de>
1038
1039         * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
1040         folding the expression.
1041         (valueize_expr): Remove.
1042         (visit_reference_op_load): Do not valueize the result of
1043         vn_get_expr_for.
1044         (simplify_binary_expression): Likewise.
1045         (simplify_unary_expression): Likewise.
1046
1047 2014-05-08  Richard Biener  <rguenther@suse.de>
1048
1049         * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
1050         looking at TYPE_ARG_TYPES.
1051
1052 2014-05-08  Richard Biener  <rguenther@suse.de>
1053
1054         * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
1055         pointer propagation special-case.
1056
1057 2014-05-08  Bin Cheng  <bin.cheng@arm.com>
1058
1059         * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
1060         core part of address expressions.
1061
1062 2014-05-08  Alan Modra  <amodra@gmail.com>
1063
1064         PR target/60737
1065         * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
1066         loads and stores when -mno-strict-align at any alignment.
1067         (expand_block_clear): Similarly.  Also correct calculation of
1068         instruction count.
1069
1070 2014-05-07  Thomas Preud'homme  <thomas.preudhomme@arm.com>
1071
1072         PR middle-end/39246
1073         * tree-complex.c (expand_complex_move): Keep line info when expanding
1074         complex move.
1075         * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
1076         of complex expression. Use new argument to display correct location
1077         for values coming from phi statement.
1078         (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
1079         (warn_uninitialized_phi): Pass location of phi argument to
1080         warn_uninit.
1081         * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
1082         COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
1083
1084 2014-05-07  Segher Boessenkool  <segher@kernel.crashing.org>
1085
1086         * config/rs6000/predicates.md (indexed_address_mem): New.
1087         * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
1088         load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
1089         fpstore_ux, fpstore_u.
1090         (sign_extend, indexed, update): New.
1091         (cell_micro): Adjust.
1092         (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
1093         *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
1094         *movsi_internal1, *movsi_internal1_single, *movhi_internal,
1095         *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
1096         *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
1097         *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
1098         *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
1099         *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
1100         *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
1101         *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
1102         *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
1103         *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
1104         *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
1105         *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
1106         *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
1107
1108         * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
1109         * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
1110         *vsx_extract_<mode>_store): Adjust.
1111         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
1112         is_cracked_insn, insn_must_be_first_in_group,
1113         insn_must_be_last_in_group): Adjust.
1114
1115         * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
1116         Adjust.
1117         * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
1118         ppc440-fpstore): Adjust.
1119         * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
1120         ppc476-fpstore): Adjust.
1121         * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
1122         ppc601-fpstore): Adjust.
1123         * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
1124         Adjust.
1125         * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
1126         Adjust.
1127         * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
1128         ppc7450-fpstore): Adjust.
1129         * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
1130         * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
1131         * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
1132         Adjust.
1133         * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
1134         cell-fpload, cell-fpload-update, cell-store, cell-store-update,
1135         cell-fpstore, cell-fpstore-update): Adjust.
1136         * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
1137         ppce300c3_store, ppce300c3_fpstore): Adjust.
1138         * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
1139         e500mc_fpstore): Adjust.
1140         * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
1141         e500mc64_store, e500mc64_fpstore): Adjust.
1142         * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
1143         e5500_fpstore): Adjust.
1144         * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
1145         e6500_fpstore): Adjust.
1146         * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
1147         Adjust.
1148         * config/rs6000/power4.md (power4-load, power4-load-ext,
1149         power4-load-ext-update, power4-load-ext-update-indexed,
1150         power4-load-update-indexed, power4-load-update, power4-fpload,
1151         power4-fpload-update, power4-store, power4-store-update,
1152         power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
1153         Adjust.
1154         * config/rs6000/power5.md (power5-load, power5-load-ext,
1155         power5-load-ext-update, power5-load-ext-update-indexed,
1156         power5-load-update-indexed, power5-load-update, power5-fpload,
1157         power5-fpload-update, power5-store, power5-store-update,
1158         power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
1159         Adjust.
1160         * config/rs6000/power6.md (power6-load, power6-load-ext,
1161         power6-load-update, power6-load-update-indexed,
1162         power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
1163         power6-fpload-update, power6-store, power6-store-update,
1164         power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
1165         Adjust.
1166         * config/rs6000/power7.md (power7-load, power7-load-ext,
1167         power7-load-update, power7-load-update-indexed,
1168         power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
1169         power7-fpload-update, power7-store, power7-store-update,
1170         power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
1171         Adjust.
1172         * config/rs6000/power8.md (power8-load, power8-load-update,
1173         power8-load-ext, power8-load-ext-update, power8-fpload,
1174         power8-fpload-update, power8-store, power8-store-update-indexed,
1175         power8-fpstore, power8-fpstore-update): Adjust.
1176         * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
1177         Adjust.
1178         * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
1179         titan_lsu_store, titan_lsu_fpstore): Adjust.
1180         * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
1181
1182 2014-05-07  Oleg Endo  <olegendo@gcc.gnu.org>
1183
1184         PR target/60884
1185         * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
1186         unrolled byte insns.  Emit address increments after move insns.
1187
1188 2014-05-07  David Malcolm  <dmalcolm@redhat.com>
1189
1190         * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
1191         const_gimple, rather than a gimple.
1192         (gimple_call_builtin_p): Likewise, for the three variants.
1193
1194         * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
1195         (gimple_call_builtin_p): Likewise, for the three variants.
1196
1197 2014-05-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1198
1199         PR tree-optimization/61095
1200         * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
1201
1202 2014-05-07  Richard Biener  <rguenther@suse.de>
1203
1204         PR tree-optimization/61034
1205         * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
1206         (maybe_skip_until): Use translate to take into account
1207         lattices when trying to do disambiguations.
1208         (get_continuation_for_phi_1): Likewise.
1209         (get_continuation_for_phi): Adjust for added translate arguments.
1210         (walk_non_aliased_vuses): Likewise.
1211         * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
1212         (walk_non_aliased_vuses): Likewise.
1213         (call_may_clobber_ref_p_1): Declare.
1214         * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
1215         calls.  Stop early if we are only supposed to disambiguate.
1216         * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
1217
1218 2014-05-07  Joern Rennecke  <joern.rennecke@embecosm.com>
1219
1220         * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
1221         Emit an error when the function has arguments.
1222
1223 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
1224
1225         * cfgloop.h (unswitch_loops): Remove.
1226         * doc/passes.texi: Remove references to loop-unswitch.c
1227         * timevar.def (TV_LOOP_UNSWITCH): Remove.
1228
1229 2014-05-07  Evgeny Stupachenko  <evstupac@gmail.com>
1230
1231         * tree-vect-data-refs.c (vect_grouped_load_supported): New
1232         check for loads group of length 3.
1233         (vect_permute_load_chain): New permutations for loads group of
1234         length 3.
1235         * tree-vect-stmts.c (vect_model_load_cost): Change cost
1236         of vec_perm_shuffle for the new permutations.
1237
1238 2014-05-07  Alan Lawrence  <alan.lawrence@arm.com>
1239
1240         * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
1241         vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
1242         vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
1243         vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
1244         vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
1245         vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
1246         vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
1247         vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
1248
1249 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
1250
1251         * loop-unswitch.c: Delete.
1252
1253 2014-05-07  Richard Biener  <rguenther@suse.de>
1254
1255         * config.gcc: Always set need_64bit_hwint to yes.
1256
1257 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
1258
1259         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
1260         of using optimize_size.
1261
1262 2014-05-06  Mike Stump  <mikestump@comcast.net>
1263
1264         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
1265
1266 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
1267
1268         * config/i386/sse.md (*mov<mode>_internal)
1269         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
1270         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
1271         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
1272         (*<code><mode>3, *andnot<mode>3<mask_name>)
1273         (<mask_codefor><code><mode>3<mask_name>): Only consider
1274         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
1275
1276 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
1277
1278         Revert:
1279         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1280
1281         * lra-constraints.c (valid_address_p): Move earlier in file.
1282         Add a constraint argument to the address_info version.
1283         (satisfies_memory_constraint_p): New function.
1284         (satisfies_address_constraint_p): Likewise.
1285         (process_alt_operands, curr_insn_transform): Use them.
1286         (process_address): Pass the constraint to valid_address_p when
1287         checking address operands.
1288
1289 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
1290
1291         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
1292         to their respective blocks.  Fix inadvertent use of "node".
1293
1294 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
1295
1296         * emit-rtl.c (init_derived_machine_modes): New functionm, split
1297         out from...
1298         (init_emit_once): ...here.
1299         * rtl.h (init_derived_machine_modes): Declare.
1300         * toplev.c (do_compile): Call it even if no_backend.
1301
1302 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
1303             Mike Stump  <mikestump@comcast.net>
1304             Richard Sandiford  <rdsandiford@googlemail.com>
1305             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1306
1307         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
1308         (rtx_equal_for_memref_p): Update comment.
1309         (adjust_offset_for_component_ref): Use wide-int interfaces.
1310         * builtins.c (get_object_alignment_2): Likewise.
1311         (c_readstr): Likewise.
1312         (target_char_cast): Add comment.
1313         (determine_block_size): Use wide-int interfaces.
1314         (expand_builtin_signbit): Likewise.
1315         (fold_builtin_int_roundingfn): Likewise.
1316         (fold_builtin_bitop): Likewise.
1317         (fold_builtin_bswap): Likewise.
1318         (fold_builtin_logarithm): Use signop.
1319         (fold_builtin_pow): Likewise.
1320         (fold_builtin_memory_op): Use wide-int interfaces.
1321         (fold_builtin_object_size): Likewise.
1322         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
1323         nb_iterations_estimate.
1324         (record_niter_bound): Use wide-int interfaces.
1325         (get_estimated_loop_iterations_int): Likewise.
1326         (get_estimated_loop_iterations): Likewise.
1327         (get_max_loop_iterations): Likewise.
1328         * cfgloop.h: Include wide-int.h.
1329         (struct nb_iter_bound): Change bound to widest_int.
1330         (struct loop): Change nb_iterations_upper_bound and
1331         nb_iterations_estimate to widest_int.
1332         (record_niter_bound): Switch to use widest_int.
1333         (get_estimated_loop_iterations): Likewise.
1334         (get_max_loop_iterations): Likewise.
1335         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
1336         update for wide-int.
1337         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
1338         * combine.c (try_combine): Likewise.
1339         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
1340         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
1341         interfaces.
1342         (aarch64_float_const_representable_p): Likewise.
1343         * config/arc/arc.c: Include wide-int.h.
1344         (arc_can_use_doloop_p): Use wide-int interfaces.
1345         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1346         (vfp3_const_double_index): Likewise.
1347         * config/avr/avr.c (avr_out_round): Likewise.
1348         (avr_fold_builtin): Likewise.
1349         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1350         (bfin_can_use_doloop_p): Likewise.
1351         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1352         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
1353         * config/i386/i386.c: Include wide-int.h.
1354         (ix86_data_alignment): Use wide-int interfaces.
1355         (ix86_local_alignment): Likewise.
1356         (ix86_emit_swsqrtsf): Update real_from_integer.
1357         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
1358         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1359         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
1360         (zero_constant): Likewise.
1361         (input_operand): Likewise.
1362         (splat_input_operand): Likewise.
1363         (non_logical_cint_operand): Change const_double to const_wide_int.
1364         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
1365         (easy_altivec_constant): Remove comment.
1366         (paired_expand_vector_init): Use CONSTANT_P.
1367         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
1368         (rs6000_emit_move): Update checks.
1369         (rs6000_aggregate_candidate): Use wide-int interfaces.
1370         (rs6000_expand_ternop_builtin): Likewise.
1371         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
1372         (rs6000_assemble_integer): Likewise.
1373         (rs6000_hash_constant): Likewise.
1374         (output_toc): Likewise.
1375         (rs6000_rtx_costs): Likewise.
1376         (rs6000_emit_swrsqrt); Update call to real_from_integer.
1377         * config/rs6000/rs6000-c.c: Include wide-int.h.
1378         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
1379         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
1380         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
1381         Handle CONST_WIDE_INT.
1382         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
1383         Use tree_fits_uhwi_p.
1384         * config/sparc/sparc.c: Include wide-int.h.
1385         (sparc_fold_builtin): Use wide-int interfaces.
1386         * config/vax/vax.c: Include wide-int.h.
1387         (vax_float_literal): Use real_from_integer.
1388         * coretypes.h (struct hwivec_def): New.
1389         (hwivec): New.
1390         (const_hwivec): New.
1391         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
1392         (equiv_constant): Handle CONST_WIDE_INT.
1393         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
1394         (cselib_hash_rtx): Handle CONST_WIDE_INT.
1395         * dbxout.c (stabstr_U): Use wide-int interfaces.
1396         (dbxout_type): Update to use cst_fits_shwi_p.
1397         * defaults.h (LOG2_BITS_PER_UNIT): Define.
1398         (TARGET_SUPPORTS_WIDE_INT): Add default.
1399         * dfp.c: Include wide-int.h.
1400         (decimal_real_to_integer2): Use wide-int interfaces and rename to
1401         decimal_real_to_integer.
1402         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
1403         decimal_real_to_integer.
1404         * doc/generic.texi (Constant expressions): Update for wide_int.
1405         * doc/rtl.texi (const_double): Likewise.
1406         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
1407         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
1408         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
1409         (REAL_VALUE_FROM_INT): Remove.
1410         (TARGET_SUPPORTS_WIDE_INT): New.
1411         * doc/tm.texi: Regenerate.
1412         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
1413         * double-int.h: Include wide-int.h.
1414         (struct wi::int_traits): New.
1415         * dwarf2out.c (get_full_len): New.
1416         (dw_val_equal_p): Add case dw_val_class_wide_int.
1417         (size_of_loc_descr): Likewise.
1418         (output_loc_operands): Likewise.
1419         (insert_double): Remove.
1420         (insert_wide_int): New.
1421         (add_AT_wide): New.
1422         (print_die): Add case dw_val_class_wide_int.
1423         (attr_checksum): Likewise.
1424         (attr_checksum_ordered): Likewise.
1425         (same_dw_val_p): Likewise.
1426         (size_of_die): Likewise.
1427         (value_format): Likewise.
1428         (output_die): Likewise.
1429         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
1430         Use wide-int.
1431         (clz_loc_descriptor): Use wide-int interfaces.
1432         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
1433         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
1434         (round_up_to_align): Use wide-int interfaces.
1435         (field_byte_offset): Likewise.
1436         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
1437         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
1438         CONST_DOUBLE handling.  Use wide-int interfaces.
1439         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
1440         (gen_enumeration_type_die): Use add_AT_wide.
1441         (hash_loc_operands): Add case dw_val_class_wide_int.
1442         (compare_loc_operands): Likewise.
1443         * dwarf2out.h: Include wide-int.h.
1444         (wide_int_ptr): New.
1445         (enum dw_val_class): Add dw_val_class_wide_int.
1446         (struct dw_val_struct): Add val_wide.
1447         * emit-rtl.c (const_wide_int_htab): New.
1448         (const_wide_int_htab_hash): New.
1449         (const_wide_int_htab_eq): New.
1450         (lookup_const_wide_int): New.
1451         (const_double_htab_hash): Use wide-int interfaces.
1452         (const_double_htab_eq): Likewise.
1453         (rtx_to_double_int): Conditionally compile for wide-int.
1454         (immed_double_int_const): Rename to immed_wide_int_const and
1455         update for wide-int.
1456         (immed_double_const): Conditionally compile for wide-int.
1457         (init_emit_once): Use wide-int interfaces.
1458         * explow.c (plus_constant): Likewise.
1459         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
1460         (lshift_value): Use wide-int interfaces.
1461         (expand_mult): Likewise.
1462         (choose_multiplier): Likewise.
1463         (expand_smod_pow2): Likewise.
1464         (make_tree): Likewise.
1465         * expr.c (convert_modes): Consolidate handling of constants.
1466         Use wide-int interfaces.
1467         (emit_group_load_1): Add note.
1468         (store_expr): Update comment.
1469         (get_inner_reference): Use wide-int interfaces.
1470         (expand_constructor): Update comment.
1471         (expand_expr_real_2): Use wide-int interfaces.
1472         (expand_expr_real_1): Likewise.
1473         (reduce_to_bit_field_precision): Likewise.
1474         (const_vector_from_tree): Likewise.
1475         * final.c: Include wide-int-print.h.
1476         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
1477         * fixed-value.c: Include wide-int.h.
1478         (fixed_from_string): Use wide-int interfaces.
1479         (fixed_to_decimal): Likewise.
1480         (fixed_convert_from_real): Likewise.
1481         (real_convert_from_fixed): Likewise.
1482         * fold-const.h (mem_ref_offset): Return an offset_int.
1483         (div_if_zero_remainder): Remove code parameter.
1484         * fold-const.c (div_if_zero_remainder): Remove code parameter.
1485         Use wide-int interfaces.
1486         (may_negate_without_overflow_p): Use wide-int interfaces.
1487         (negate_expr_p): Likewise.
1488         (fold_negate_expr): Likewise.
1489         (int_const_binop_1): Likewise.
1490         (const_binop): Likewise.
1491         (fold_convert_const_int_from_int): Likewise.
1492         (fold_convert_const_int_from_real): Likewise.
1493         (fold_convert_const_int_from_fixed): Likewise.
1494         (fold_convert_const_fixed_from_int): Likewise.
1495         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
1496         (sign_bit_p): Use wide-int interfaces.
1497         (make_range_step): Likewise.
1498         (build_range_check): Likewise.  Pass an integer of the correct type
1499         instead of using integer_one_node.
1500         (range_predecessor): Pass an integer of the correct type instead
1501         of using integer_one_node.
1502         (range_successor): Likewise.
1503         (merge_ranges): Likewise.
1504         (unextend): Use wide-int interfaces.
1505         (extract_muldiv_1): Likewise.
1506         (fold_div_compare): Likewise.
1507         (fold_single_bit_test): Likewise.
1508         (fold_sign_changed_comparison): Likewise.
1509         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
1510         (fold_plusminus_mult_expr): Use wide-int interfaces.
1511         (native_encode_int): Likewise.
1512         (native_interpret_int): Likewise.
1513         (fold_unary_loc): Likewise.
1514         (pointer_may_wrap_p): Likewise.
1515         (size_low_cst): Likewise.
1516         (mask_with_tz): Likewise.
1517         (fold_binary_loc): Likewise.
1518         (fold_ternary_loc): Likewise.
1519         (multiple_of_p): Likewise.
1520         (tree_call_nonnegative_warnv_p): Update calls to
1521         tree_int_cst_min_precision and real_from_integer.
1522         (fold_negate_const): Use wide-int interfaces.
1523         (fold_abs_const): Likewise.
1524         (fold_relational_const): Use tree_int_cst_lt.
1525         (round_up_loc): Use wide-int interfaces.
1526         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
1527         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
1528         * gengtype.c: Remove include of double-int.h.
1529         (do_typedef): Use wide-int interfaces.
1530         (open_base_files): Add wide-int.h.
1531         (main): Add offset_int and widest_int typedefs.
1532         * gengtype-lex.l: Handle "^".
1533         (CXX_KEYWORD): Add "static".
1534         * gengtype-parse.c (require3): New.
1535         (require_template_declaration): Handle constant template arguments
1536         and nested templates.
1537         * gengtype-state.c: Don't include "double-int.h".
1538         * genpreds.c (write_one_predicate_function): Update comment.
1539         (write_tm_constrs_h): Add check for hval and lval use in
1540         CONST_WIDE_INT.
1541         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
1542         (add_to_sequence): Likewise.
1543         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
1544         and const_double_operand.
1545         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
1546         interfaces.
1547         * gimple-fold.c (get_base_constructor): Likewise.
1548         (fold_array_ctor_reference): Likewise.
1549         (fold_nonarray_ctor_reference): Likewise.
1550         (fold_const_aggregate_ref_1): Likewise.
1551         (gimple_val_nonnegative_real_p): Likewise.
1552         (gimple_fold_indirect_ref): Likewise.
1553         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
1554         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
1555         (struct slsr_cand_d): Change index to be widest_int.
1556         (struct incr_info_d): Change incr to be widest_int.
1557         (alloc_cand_and_find_basis): Use wide-int interfaces.
1558         (slsr_process_phi): Likewise.
1559         (backtrace_base_for_ref): Likewise.  Return a widest_int.
1560         (restructure_reference): Take a widest_int instead of a double_int.
1561         (slsr_process_ref): Use wide-int interfaces.
1562         (create_mul_ssa_cand): Likewise.
1563         (create_mul_imm_cand): Likewise.
1564         (create_add_ssa_cand): Likewise.
1565         (create_add_imm_cand): Take a widest_int instead of a double_int.
1566         (slsr_process_add): Use wide-int interfaces.
1567         (slsr_process_cast): Likewise.
1568         (slsr_process_copy): Likewise.
1569         (dump_candidate): Likewise.
1570         (dump_incr_vec): Likewise.
1571         (replace_ref): Likewise.
1572         (cand_increment): Likewise.  Return a widest_int.
1573         (cand_abs_increment): Likewise.
1574         (replace_mult_candidate): Take a widest_int instead of a double_int.
1575         (replace_unconditional_candidate): Use wide-int interfaces.
1576         (incr_vec_index): Take a widest_int instead of a double_int.
1577         (create_add_on_incoming_edge): Likewise.
1578         (create_phi_basis): Use wide-int interfaces.
1579         (replace_conditional_candidate): Likewise.
1580         (record_increment): Take a widest_int instead of a double_int.
1581         (record_phi_increments): Use wide-int interfaces.
1582         (phi_incr_cost): Take a widest_int instead of a double_int.
1583         (lowest_cost_path): Likewise.
1584         (total_savings): Likewise.
1585         (analyze_increments): Use wide-int interfaces.
1586         (ncd_with_phi): Take a widest_int instead of a double_int.
1587         (ncd_of_cand_and_phis): Likewise.
1588         (nearest_common_dominator_for_cands): Likewise.
1589         (insert_initializers): Use wide-int interfaces.
1590         (all_phi_incrs_profitable): Likewise.
1591         (replace_one_candidate): Likewise.
1592         (replace_profitable_candidates): Likewise.
1593         * godump.c: Include wide-int-print.h.
1594         (go_output_typedef): Use wide-int interfaces.
1595         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
1596         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1597         (build_loop_iteration_domains): Likewise.
1598         * hooks.h: Include wide-int.h rather than double-int.h.
1599         (hook_bool_dint_dint_uint_bool_true): Delete.
1600         (hook_bool_wint_wint_uint_bool_true): Declare.
1601         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
1602         (hook_bool_wint_wint_uint_bool_true): New.
1603         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
1604         interfaces.
1605         (ubsan_expand_si_overflow_mul_check): Likewise.
1606         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
1607         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1608         (get_ancestor_addr_info): Likewise.
1609         (ipa_modify_call_arguments): Likewise.
1610         * loop-doloop.c (doloop_modify): Likewise.
1611         (doloop_optimize): Likewise.
1612         * loop-iv.c (iv_number_of_iterations): Likewise.
1613         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
1614         (unroll_loop_constant_iterations): Likewise.
1615         (decide_unroll_runtime_iterations): Likewise.
1616         (unroll_loop_runtime_iterations): Likewise.
1617         (decide_peel_simple): Likewise.
1618         (decide_unroll_stupid): Likewise.
1619         * lto-streamer-in.c (streamer_read_wi): Add.
1620         (input_cfg): Use wide-int interfaces.
1621         (lto_input_tree_1): Likewise.
1622         * lto-streamer-out.c (streamer_write_wi): Add.
1623         (hash_tree): Use wide-int interfaces.
1624         (output_cfg): Likewise.
1625         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
1626         (GTFILES): Add wide-int.h and signop.h.
1627         (TAGS): Look for .cc files too.
1628         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
1629         * optabs.c (expand_subword_shift): Likewise.
1630         (expand_doubleword_shift): Likewise.
1631         (expand_absneg_bit): Likewise.
1632         (expand_copysign_absneg): Likewise.
1633         (expand_copysign_bit): Likewise.
1634         * postreload.c (reload_cse_simplify_set): Likewise.
1635         * predict.c (predict_iv_comparison): Likewise.
1636         * pretty-print.h: Include wide-int-print.h.
1637         (pp_wide_int) New.
1638         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
1639         * print-tree.c: Include wide-int-print.h.
1640         (print_node_brief): Use wide-int interfaces.
1641         (print_node): Likewise.
1642         * read-rtl.c (validate_const_wide_int): New.
1643         (read_rtx_code): Add CONST_WIDE_INT case.
1644         * real.c: Include wide-int.h.
1645         (real_to_integer2): Delete.
1646         (real_to_integer): New function, returning a wide_int.
1647         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1648         (ten_to_ptwo): Update call to real_from_integer.
1649         (real_digit): Likewise.
1650         * real.h: Include signop.h, wide-int.h and insn-modes.h.
1651         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
1652         (REAL_VALUE_TO_INT): Delete.
1653         (real_to_integer): Declare a wide-int form.
1654         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1655         * recog.c (const_int_operand): Improve comment.
1656         (const_scalar_int_operand): New.
1657         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
1658         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
1659         (split_double): Likewise.
1660         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
1661         (rtx_size): Likewise.
1662         (rtx_alloc_stat_v): New.
1663         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
1664         (cwi_output_hex): New.
1665         (iterative_hash_rtx): Handle CONST_WIDE_INT.
1666         (cwi_check_failed_bounds): New.
1667         * rtl.def (CONST_WIDE_INT): New.
1668         * rtl.h: Include <utility> and wide-int.h.
1669         (struct hwivec_def): New.
1670         (CWI_GET_NUM_ELEM): New.
1671         (CWI_PUT_NUM_ELEM): New.
1672         (struct rtx_def): Add num_elem and hwiv.
1673         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
1674         (CASE_CONST_UNIQUE): Likewise.
1675         (CASE_CONST_ANY): Likewise.
1676         (CONST_SCALAR_INT_P): Likewise.
1677         (CONST_WIDE_INT_P): New.
1678         (CWI_ELT): New.
1679         (HWIVEC_CHECK): New.
1680         (cwi_check_failed_bounds): New.
1681         (CWI_ELT): New.
1682         (HWIVEC_CHECK): New.
1683         (CONST_WIDE_INT_VEC) New.
1684         (CONST_WIDE_INT_NUNITS) New.
1685         (CONST_WIDE_INT_ELT) New.
1686         (rtx_mode_t): New type.
1687         (wi::int_traits <rtx_mode_t>): New.
1688         (wi::shwi): New.
1689         (wi::min_value): New.
1690         (wi::max_value): New.
1691         (rtx_alloc_v) New.
1692         (const_wide_int_alloc): New.
1693         (immed_wide_int_const): New.
1694         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
1695         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
1696         * signop.h: New file.
1697         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
1698         (simplify_const_unary_operation): Use wide-int interfaces.
1699         (simplify_binary_operation_1): Likewise.
1700         (simplify_const_binary_operation): Likewise.
1701         (simplify_const_relational_operation): Likewise.
1702         (simplify_immed_subreg): Likewise.
1703         * stmt.c (expand_case): Likewise.
1704         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
1705         signop rather than a bool.
1706         * stor-layout.c (layout_type): Use wide-int interfaces.
1707         (initialize_sizetypes): Update calls to
1708         set_min_and_max_values_for_integral_type.
1709         (set_min_and_max_values_for_integral_type): Take a signop rather
1710         than a bool.  Use wide-int interfaces.
1711         (fixup_signed_type): Update accordingly.  Remove
1712         HOST_BITS_PER_DOUBLE_INT limit.
1713         (fixup_unsigned_type): Likewise.
1714         * system.h (STATIC_CONSTANT_P): New.
1715         (STATIC_ASSERT): New.
1716         * target.def (can_use_doloop_p): Take widest_ints rather than
1717         double_ints.
1718         * target.h: Include wide-int.h rather than double-int.h.
1719         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
1720         than double_ints.
1721         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
1722         rather than INT_CST_LT_UNSIGNED.
1723         (can_use_doloop_if_innermost): Take widest_ints rather than
1724         double_ints.
1725         * tree-affine.c: Include wide-int-print.h.
1726         (double_int_ext_for_comb): Delete.
1727         (wide_int_ext_for_comb): New.
1728         (aff_combination_zero): Use wide-int interfaces.
1729         (aff_combination_const): Take a widest_int instead of a double_int.
1730         (aff_combination_elt): Use wide-int interfaces.
1731         (aff_combination_scale): Take a widest_int instead of a double_int.
1732         (aff_combination_add_elt): Likewise.
1733         (aff_combination_add_cst): Likewise.
1734         (aff_combination_add): Use wide-int interfaces.
1735         (aff_combination_convert): Likewise.
1736         (tree_to_aff_combination): Likewise.
1737         (add_elt_to_tree): Take a widest_int instead of a double_int.
1738         (aff_combination_to_tree): Use wide-int interfaces.
1739         (aff_combination_remove_elt): Likewise.
1740         (aff_combination_add_product): Take a widest_int instead of
1741         a double_int.
1742         (aff_combination_mult): Use wide-int interfaces.
1743         (aff_combination_expand): Likewise.
1744         (double_int_constant_multiple_p): Delete.
1745         (wide_int_constant_multiple_p): New.
1746         (aff_combination_constant_multiple_p): Take a widest_int pointer
1747         instead of a double_int pointer.
1748         (print_aff): Use wide-int interfaces.
1749         (get_inner_reference_aff): Take a widest_int pointer
1750         instead of a double_int pointer.
1751         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
1752         * tree-affine.h: Include wide-int.h.
1753         (struct aff_comb_elt): Change type of coef to widest_int.
1754         (struct affine_tree_combination): Change type of offset to widest_int.
1755         (double_int_ext_for_comb): Delete.
1756         (wide_int_ext_for_comb): New.
1757         (aff_combination_const): Use widest_int instead of double_int.
1758         (aff_combination_scale): Likewise.
1759         (aff_combination_add_elt): Likewise.
1760         (aff_combination_constant_multiple_p): Likewise.
1761         (get_inner_reference_aff): Likewise.
1762         (aff_comb_cannot_overlap_p): Likewise.
1763         (aff_combination_zero_p): Use wide-int interfaces.
1764         * tree.c: Include tree.h.
1765         (init_ttree): Use make_int_cst.
1766         (tree_code_size): Removed code for INTEGER_CST case.
1767         (tree_size): Add INTEGER_CST case.
1768         (make_node_stat): Update comment.
1769         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
1770         (build_int_cst_type): Use wide-int interfaces.
1771         (double_int_to_tree): Likewise.
1772         (double_int_fits_to_tree_p): Delete.
1773         (force_fit_type_double): Delete.
1774         (force_fit_type): New.
1775         (int_cst_hash_hash): Use wide-int interfaces.
1776         (int_cst_hash_eq): Likewise.
1777         (build_int_cst_wide): Delete.
1778         (wide_int_to_tree): New.
1779         (cache_integer_cst): Use wide-int interfaces.
1780         (build_low_bits_mask): Likewise.
1781         (cst_and_fits_in_hwi): Likewise.
1782         (real_value_from_int_cst): Likewise.
1783         (make_int_cst_stat): New.
1784         (integer_zerop): Use wide_int interfaces.
1785         (integer_onep): Likewise.
1786         (integer_all_onesp): Likewise.
1787         (integer_pow2p): Likewise.
1788         (integer_nonzerop): Likewise.
1789         (tree_log2): Likewise.
1790         (tree_floor_log2): Likewise.
1791         (tree_ctz): Likewise.
1792         (int_size_in_bytes): Likewise.
1793         (mem_ref_offset): Return an offset_int rather than a double_int.
1794         (build_type_attribute_qual_variant): Use wide_int interfaces.
1795         (type_hash_eq): Likewise
1796         (tree_int_cst_equal): Likewise.
1797         (tree_int_cst_lt): Delete.
1798         (tree_int_cst_compare): Likewise.
1799         (tree_fits_shwi_p): Use wide_int interfaces.
1800         (tree_fits_uhwi_p): Likewise.
1801         (tree_int_cst_sign_bit): Likewise.
1802         (tree_int_cst_sgn): Likewise.
1803         (tree_int_cst_min_precision): Take a signop rather than a bool.
1804         (simple_cst_equal): Use wide_int interfaces.
1805         (compare_tree_int): Likewise.
1806         (iterative_hash_expr): Likewise.
1807         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
1808         INT_CST_LT.
1809         (get_type_static_bounds): Use wide_int interfaces.
1810         (tree_int_cst_elt_check_failed): New.
1811         (build_common_tree_nodes): Reordered to set prec before filling in
1812         value.
1813         (int_cst_value): Check cst_and_fits_in_hwi.
1814         (widest_int_cst_value): Use wide_int interfaces.
1815         (upper_bound_in_type): Likewise.
1816         (lower_bound_in_type): Likewise.
1817         (num_ending_zeros): Likewise.
1818         (drop_tree_overflow): Likewise.
1819         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
1820         (gen_conditions_for_pow_cst_base): Likewise.
1821         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
1822         (group_case_labels_stmt): Use wide-int interfaces.
1823         (verify_gimple_assign_binary): Likewise.
1824         (print_loop): Likewise.
1825         * tree-chrec.c (tree_fold_binomial): Likewise.
1826         * tree-core.h (struct tree_base): Add int_length.
1827         (struct tree_int_cst): Change rep of value.
1828         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
1829         (dr_may_alias_p): Likewise.
1830         (max_stmt_executions_tree): Likewise.
1831         * tree.def (INTEGER_CST): Update comment.
1832         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
1833         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
1834         * tree-dump.c: Include wide-int.h and wide-int-print.h.
1835         (dequeue_and_dump): Use wide-int interfaces.
1836         * tree.h: Include wide-int.h.
1837         (NULL_TREE): Moved to earlier loc in file.
1838         (TREE_INT_CST_ELT_CHECK): New.
1839         (tree_int_cst_elt_check_failed): New.
1840         (TYPE_SIGN): New.
1841         (TREE_INT_CST): Delete.
1842         (TREE_INT_CST_LOW): Use wide-int interfaces.
1843         (TREE_INT_CST_HIGH): Delete.
1844         (TREE_INT_CST_NUNITS): New.
1845         (TREE_INT_CST_EXT_NUNITS): Likewise.
1846         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
1847         (TREE_INT_CST_ELT): Likewise.
1848         (INT_CST_LT): Delete.
1849         (tree_int_cst_elt_check): New (two forms).
1850         (type_code_size): Update comment.
1851         (make_int_cst_stat, make_int_cst): New.
1852         (tree_to_double_int): Delete.
1853         (double_int_fits_to_tree_p): Delete.
1854         (force_fit_type_double): Delete.
1855         (build_int_cstu): Replace with out-of-line function.
1856         (build_int_cst_wide): Delete.
1857         (tree_int_cst_lt): Define inline.
1858         (tree_int_cst_le): New.
1859         (tree_int_cst_compare): Define inline.
1860         (tree_int_cst_min_precision): Take a signop rather than a bool.
1861         (wi::int_traits <const_tree>): New.
1862         (wi::int_traits <tree>): New.
1863         (wi::extended_tree): New.
1864         (wi::int_traits <wi::extended_tree>): New.
1865         (wi::to_widest): New.
1866         (wi::to_offset): New.
1867         (wi::fits_to_tree_p): New.
1868         (wi::min_value): New.
1869         (wi::max_value): New.
1870         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
1871         (copy_tree_body_r): Likewise.
1872         * tree-object-size.c (compute_object_offset): Likewise.
1873         (addr_object_size): Likewise.
1874         * tree-predcom.c: Include wide-int-print.h.
1875         (struct dref_d): Change type of offset to widest_int.
1876         (dump_dref): Call wide-int printer.
1877         (aff_combination_dr_offset): Use wide-int interfaces.
1878         (determine_offset): Take a widest_int pointer rather than a
1879         double_int pointer.
1880         (split_data_refs_to_components): Use wide-int interfaces.
1881         (suitable_component_p): Likewise.
1882         (order_drefs): Likewise.
1883         (add_ref_to_chain): Likewise.
1884         (valid_initializer_p): Likewise.
1885         (determine_roots_comp): Likewise.
1886         * tree-pretty-print.c: Include wide-int-print.h.
1887         (dump_generic_node): Use wide-int interfaces.
1888         * tree-sra.c (sra_ipa_modify_expr): Likewise.
1889         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1890         (move_fixed_address_to_symbol): Likewise.
1891         (move_hint_to_base): Likewise.
1892         (move_pointer_to_base): Likewise.
1893         (move_variant_to_index): Likewise.
1894         (most_expensive_mult_to_index): Likewise.
1895         (addr_to_parts): Likewise.
1896         (copy_ref_info): Likewise.
1897         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1898         (indirect_refs_may_alias_p): Likewise.
1899         (stmt_kills_ref_p_1): Likewise.
1900         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
1901         * tree-ssa-ccp.c: Update comment at top of file.  Include
1902         wide-int-print.h.
1903         (struct prop_value_d): Change type of mask to widest_int.
1904         (extend_mask): New function.
1905         (dump_lattice_value): Use wide-int interfaces.
1906         (get_default_value): Likewise.
1907         (set_constant_value): Likewise.
1908         (set_value_varying): Likewise.
1909         (valid_lattice_transition): Likewise.
1910         (set_lattice_value): Likewise.
1911         (value_to_double_int): Delete.
1912         (value_to_wide_int): New.
1913         (get_value_from_alignment): Use wide-int interfaces.
1914         (get_value_for_expr): Likewise.
1915         (do_dbg_cnt): Likewise.
1916         (ccp_finalize): Likewise.
1917         (ccp_lattice_meet): Likewise.
1918         (bit_value_unop_1): Use widest_ints rather than double_ints.
1919         (bit_value_binop_1): Likewise.
1920         (bit_value_unop): Use wide-int interfaces.
1921         (bit_value_binop): Likewise.
1922         (bit_value_assume_aligned): Likewise.
1923         (evaluate_stmt): Likewise.
1924         (ccp_fold_stmt): Likewise.
1925         (visit_cond_stmt): Likewise.
1926         (ccp_visit_stmt): Likewise.
1927         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1928         (constant_pointer_difference): Likewise.
1929         (associate_pointerplus): Likewise.
1930         (combine_conversions): Likewise.
1931         * tree-ssa-loop.h: Include wide-int.h.
1932         (struct tree_niter_desc): Change type of max to widest_int.
1933         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
1934         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
1935         (remove_redundant_iv_tests): Likewise.
1936         (canonicalize_loop_induction_variables): Likewise.
1937         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
1938         (constant_multiple_of): Take a widest_int pointer instead of
1939         a double_int pointer.
1940         (get_computation_aff): Use wide-int interfaces.
1941         (ptr_difference_cost): Likewise.
1942         (difference_cost): Likewise.
1943         (get_loop_invariant_expr_id): Likewise.
1944         (get_computation_cost_at): Likewise.
1945         (iv_elimination_compare_lt): Likewise.
1946         (may_eliminate_iv): Likewise.
1947         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
1948         instead of double_int.
1949         (max_loop_iterations): Likewise.
1950         (max_stmt_executions): Likewise.
1951         (estimated_stmt_executions): Likewise.
1952         * tree-ssa-loop-niter.c: Include wide-int-print.h.
1953         (split_to_var_and_offset): Use wide-int interfaces.
1954         (determine_value_range): Likewise.
1955         (bound_difference_of_offsetted_base): Likewise.
1956         (bounds_add): Take a widest_int instead of a double_int.
1957         (number_of_iterations_ne_max): Use wide-int interfaces.
1958         (number_of_iterations_ne): Likewise.
1959         (number_of_iterations_lt_to_ne): Likewise.
1960         (assert_loop_rolls_lt): Likewise.
1961         (number_of_iterations_lt): Likewise.
1962         (number_of_iterations_le): Likewise.
1963         (number_of_iterations_cond): Likewise.
1964         (number_of_iterations_exit): Likewise.
1965         (finite_loop_p): Likewise.
1966         (derive_constant_upper_bound_assign): Likewise.
1967         (derive_constant_upper_bound): Return a widest_int.
1968         (derive_constant_upper_bound_ops): Likewise.
1969         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
1970         (record_estimate): Take a widest_int rather than a double_int.
1971         (record_nonwrapping_iv): Use wide-int interfaces.
1972         (double_int_cmp): Delete.
1973         (wide_int_cmp): New.
1974         (bound_index): Take a widest_int rather than a double_int.
1975         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
1976         (maybe_lower_iteration_bound): Likewise.
1977         (estimate_numbers_of_iterations_loop): Likewise.
1978         (estimated_loop_iterations): Take a widest_int pointer than than
1979         a double_int pointer.
1980         (estimated_loop_iterations_int): Use wide-int interfaces.
1981         (max_loop_iterations): Take a widest_int pointer than than
1982         a double_int pointer.
1983         (max_loop_iterations_int): Use wide-int interfaces.
1984         (max_stmt_executions): Take a widest_int pointer than than
1985         a double_int pointer.
1986         (estimated_stmt_executions): Likewise.
1987         (n_of_executions_at_most): Use wide-int interfaces.
1988         (scev_probably_wraps_p): Likewise.
1989         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
1990         to real_to_integer.
1991         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
1992         interfaces.
1993         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
1994         double_ints.  Adjust for trailing_wide_ints <3> representation.
1995         (set_nonzero_bits): Likewise.
1996         (get_range_info): Return wide_ints rather than double_ints.
1997         Adjust for trailing_wide_ints <3> representation.
1998         (get_nonzero_bits): Likewise.
1999         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
2000         representation.
2001         * tree-ssanames.h (struct range_info_def): Replace min, max and
2002         nonzero_bits with a trailing_wide_ints <3>.
2003         (set_range_info): Use wide_int_refs rather than double_ints.
2004         (set_nonzero_bits): Likewise.
2005         (get_range_info): Return wide_ints rather than double_ints.
2006         (get_nonzero_bits): Likewise.
2007         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
2008         * tree-ssa-pre.c (phi_translate_1): Likewise.
2009         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
2010         (acceptable_pow_call): Likewise.
2011         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
2012         interfaces.
2013         (vn_reference_fold_indirect): Likewise.
2014         (vn_reference_maybe_forwprop_address): Likewise.
2015         (valueize_refs_1): Likewise.
2016         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2017         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
2018         tree_int_cst_lt and tree_int_cst_le.
2019         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
2020         interfaces.
2021         (streamer_alloc_tree): Likewise.
2022         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
2023         (streamer_write_tree_header): Likewise.
2024         (streamer_write_integer_cst): Likewise.
2025         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
2026         (build_constructors): Likewise.
2027         (array_value_type): Likewise.
2028         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
2029         (vect_check_gather): Likewise.
2030         * tree-vect-generic.c (build_replicated_const): Likewise.
2031         (expand_vector_divmod): Likewise.
2032         * tree-vect-loop.c (vect_transform_loop): Likewise.
2033         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
2034         (vect_do_peeling_for_alignment): Likewise.
2035         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
2036         * tree-vrp.c: Include wide-int.h.
2037         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
2038         (extract_range_from_assert): Use wide-int interfaces.
2039         (vrp_int_const_binop): Likewise.
2040         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
2041         double_int pointers.
2042         (ranges_from_anti_range): Use wide-int interfaces.
2043         (quad_int_cmp): Delete.
2044         (quad_int_pair_sort): Likewise.
2045         (extract_range_from_binary_expr_1): Use wide-int interfaces.
2046         (extract_range_from_unary_expr_1): Likewise.
2047         (adjust_range_with_scev): Likewise.
2048         (masked_increment): Take and return wide_ints rather than double_ints.
2049         (register_edge_assert_for_2): Use wide-int interfaces.
2050         (check_array_ref): Likewise.
2051         (search_for_addr_array): Likewise.
2052         (maybe_set_nonzero_bits): Likewise.
2053         (union_ranges): Pass an integer of the correct type instead of
2054         using integer_one_node.
2055         (intersect_ranges): Likewise.
2056         (simplify_truth_ops_using_ranges): Likewise.
2057         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
2058         (range_fits_type_p): Likewise.
2059         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
2060         a bool.
2061         (simplify_conversion_using_ranges): Use wide-int interfaces.
2062         (simplify_float_conversion_using_ranges): Likewise.
2063         (vrp_finalize): Likewise.
2064         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
2065         (gimple_stringops_transform): Likewise.
2066         * varasm.c (decode_addr_const): Likewise.
2067         (const_hash_1): Likewise.
2068         (const_rtx_hash_1): Likewise
2069         (output_constant): Likewise.
2070         (array_size_for_constructor): Likewise.
2071         (output_constructor_regular_field): Likewise.
2072         (output_constructor_bitfield): Likewise.
2073         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
2074         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
2075         GENERATOR_FILEs.
2076         * gencheck.c: Define BITS_PER_UNIT.
2077         * wide-int.cc: New.
2078         * wide-int.h: New.
2079         * wide-int-print.cc: New.
2080         * wide-int-print.h: New.
2081
2082 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2083
2084         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
2085
2086 2014-05-06  Richard Biener  <rguenther@suse.de>
2087
2088         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
2089         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
2090         (TODO_verify_all): Adjust.
2091         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
2092         TODO_verify_stmts and TODO_verify_rtl_sharing.
2093         * bb-reorder.c: Likewise.
2094         * cfgexpand.c: Likewise.
2095         * cprop.c: Likewise.
2096         * cse.c: Likewise.
2097         * function.c: Likewise.
2098         * fwprop.c: Likewise.
2099         * gcse.c: Likewise.
2100         * gimple-ssa-isolate-paths.c: Likewise.
2101         * gimple-ssa-strength-reduction.c: Likewise.
2102         * ipa-split.c: Likewise.
2103         * loop-init.c: Likewise.
2104         * loop-unroll.c: Likewise.
2105         * lower-subreg.c: Likewise.
2106         * modulo-sched.c: Likewise.
2107         * postreload-gcse.c: Likewise.
2108         * predict.c: Likewise.
2109         * recog.c: Likewise.
2110         * sched-rgn.c: Likewise.
2111         * store-motion.c: Likewise.
2112         * tracer.c: Likewise.
2113         * trans-mem.c: Likewise.
2114         * tree-call-cdce.c: Likewise.
2115         * tree-cfg.c: Likewise.
2116         * tree-cfgcleanup.c: Likewise.
2117         * tree-complex.c: Likewise.
2118         * tree-eh.c: Likewise.
2119         * tree-emutls.c: Likewise.
2120         * tree-if-conv.c: Likewise.
2121         * tree-into-ssa.c: Likewise.
2122         * tree-loop-distribution.c: Likewise.
2123         * tree-object-size.c: Likewise.
2124         * tree-parloops.c: Likewise.
2125         * tree-pass.h: Likewise.
2126         * tree-sra.c: Likewise.
2127         * tree-ssa-ccp.c: Likewise.
2128         * tree-ssa-copy.c: Likewise.
2129         * tree-ssa-copyrename.c: Likewise.
2130         * tree-ssa-dce.c: Likewise.
2131         * tree-ssa-dom.c: Likewise.
2132         * tree-ssa-dse.c: Likewise.
2133         * tree-ssa-forwprop.c: Likewise.
2134         * tree-ssa-ifcombine.c: Likewise.
2135         * tree-ssa-loop-ch.c: Likewise.
2136         * tree-ssa-loop-ivcanon.c: Likewise.
2137         * tree-ssa-loop.c: Likewise.
2138         * tree-ssa-math-opts.c: Likewise.
2139         * tree-ssa-phiopt.c: Likewise.
2140         * tree-ssa-phiprop.c: Likewise.
2141         * tree-ssa-pre.c: Likewise.
2142         * tree-ssa-reassoc.c: Likewise.
2143         * tree-ssa-sink.c: Likewise.
2144         * tree-ssa-strlen.c: Likewise.
2145         * tree-ssa-tail-merge.c: Likewise.
2146         * tree-ssa-uncprop.c: Likewise.
2147         * tree-switch-conversion.c: Likewise.
2148         * tree-tailcall.c: Likewise.
2149         * tree-vect-generic.c: Likewise.
2150         * tree-vectorizer.c: Likewise.
2151         * tree-vrp.c: Likewise.
2152         * tsan.c: Likewise.
2153         * var-tracking.c: Likewise.
2154         * bt-load.c: Likewise.
2155         * cfgcleanup.c: Likewise.
2156         * combine-stack-adj.c: Likewise.
2157         * combine.c: Likewise.
2158         * compare-elim.c: Likewise.
2159         * config/epiphany/resolve-sw-modes.c: Likewise.
2160         * config/i386/i386.c: Likewise.
2161         * config/mips/mips.c: Likewise.
2162         * config/s390/s390.c: Likewise.
2163         * config/sh/sh_treg_combine.cc: Likewise.
2164         * config/sparc/sparc.c: Likewise.
2165         * dce.c: Likewise.
2166         * dse.c: Likewise.
2167         * final.c: Likewise.
2168         * ifcvt.c: Likewise.
2169         * mode-switching.c: Likewise.
2170         * passes.c: Likewise.
2171         * postreload.c: Likewise.
2172         * ree.c: Likewise.
2173         * reg-stack.c: Likewise.
2174         * regcprop.c: Likewise.
2175         * regrename.c: Likewise.
2176         * web.c: Likewise.
2177
2178 2014-05-06  Richard Biener  <rguenther@suse.de>
2179
2180         PR middle-end/61070
2181         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
2182         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
2183
2184 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
2185
2186         PR ipa/60965
2187         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
2188
2189 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
2190             Tom de Vries  <tom@codesourcery.com>
2191
2192         * target.def (call_fusage_contains_non_callee_clobbers): New
2193         DEFHOOKPOD.
2194         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
2195         Hooks to @menu.
2196         (@node Miscellaneous Register Hooks): New node.
2197         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
2198         * doc/tm.texi: Regenerate.
2199
2200 2014-05-05  Marek Polacek  <polacek@redhat.com>
2201
2202         PR driver/61065
2203         * opts.c (common_handle_option): Call error_at instead of warning_at.
2204
2205 2014-05-05  Richard Biener  <rguenther@suse.de>
2206
2207         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
2208         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
2209         under the TODO_verify_il umbrella.
2210
2211 2014-05-05  Richard Biener  <rguenther@suse.de>
2212
2213         * passes.c (execute_function_todo): Move TODO_verify_flow under
2214         the TODO_verify_ul umbrella.
2215
2216 2014-05-05  Richard Biener  <rguenther@suse.de>
2217
2218         PR middle-end/61010
2219         * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
2220         X & CST away from a CST that is the mask of a mode.
2221
2222 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2223
2224         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
2225         int argument to enum machine_mode.
2226         (picochip_class_max_nregs): Ditto.
2227         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
2228         (picochip_class_max_nregs): Ditto.
2229
2230 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2231
2232         * target.def: Add new target hook.
2233         * doc/tm.texi: Regenerate.
2234         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
2235         * targhooks.c (default_keep_leaf_when_profiled): New function.
2236
2237         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
2238         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
2239
2240 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
2241
2242         PR tree-optimization/60363
2243         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
2244         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
2245         (update_destination_phis): New parameter.
2246         (create_edge_and_update_destination_phis): Ditto.
2247         (ssa_fix_duplicate_block_edges): Pass new arguments.
2248         (thread_single_edge): Ditto.
2249
2250 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
2251
2252         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
2253         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
2254         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
2255         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
2256         Use RS6000_BTM_HARD_FLOAT.
2257         (BU_MISC_2): Likewise.
2258         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
2259         RS6000_BTM_HARD_FLOAT.
2260         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
2261         is explicitly used.
2262         (rs6000_invalid_builtin): Add hard floating builtin support.
2263         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
2264         hard float builtins.
2265         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
2266
2267 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2268
2269         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
2270         Add missing function* argument.
2271
2272 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
2273
2274         * lra-constraints.c (valid_address_p): Move earlier in file.
2275         Add a constraint argument to the address_info version.
2276         (satisfies_memory_constraint_p): New function.
2277         (satisfies_address_constraint_p): Likewise.
2278         (process_alt_operands, curr_insn_transform): Use them.
2279         (process_address): Pass the constraint to valid_address_p when
2280         checking address operands.
2281
2282 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
2283
2284         * config/mips/mips.c (mips_isa_rev): New variable.
2285         (mips_set_architecture): Set it.
2286         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
2287         from mips_isa_rev.
2288         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
2289         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
2290         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
2291         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
2292         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
2293         conditions in terms of mips_isa_rev.
2294         (mips_isa_rev): Declare.
2295
2296 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2297
2298         * config/sh/sh-mem.cc: Use tabs instead of spaces.
2299         (prob_unlikely, prob_likely): Make variables const.
2300
2301 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
2302
2303         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
2304
2305 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2306
2307         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
2308
2309 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2310
2311         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
2312         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
2313         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
2314         functions.
2315         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
2316         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
2317         sh_pass_in_reg_p.
2318         Replace usage of ROUND_REG with sh_round_reg.
2319         Use CEIL instead of ROUND_ADVANCE.
2320
2321 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
2322
2323         PR target/61026
2324         * config/sh/sh.c: Include stdlib headers before everything else.
2325
2326 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
2327
2328         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
2329         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
2330         (gimplify_adjust_omp_clauses): Simd region is never
2331         directly nested in combined parallel.  Instead, for linear
2332         with copyin/copyout, if in combined for simd loop, make decl
2333         firstprivate/lastprivate on OMP_FOR.
2334         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
2335         expand_omp_for_static_chunk): When setting endvar, also set
2336         fd->loop.v to the same value.
2337
2338 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
2339
2340         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
2341
2342 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
2343
2344         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
2345         expression.
2346
2347 2014-05-02  Marek Polacek  <polacek@redhat.com>
2348
2349         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
2350
2351 2014-05-02  Kito Cheng  <kito@0xlab.org>
2352
2353         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
2354         to a C expression marco.
2355         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
2356         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
2357         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
2358         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
2359         HONOR_REG_ALLOC_ORDER.
2360         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
2361
2362 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2363
2364         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
2365
2366 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2367
2368         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
2369
2370 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
2371
2372         * tree-if-conv.c (is_cond_scalar_reduction): New function.
2373         (convert_scalar_cond_reduction): Likewise.
2374         (predicate_scalar_phi): Add recognition and transformation
2375         of simple conditioanl reduction to be vectorizable.
2376
2377 2014-05-01  Marek Polacek  <polacek@redhat.com>
2378
2379         PR c/43245
2380         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
2381
2382 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
2383
2384         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
2385         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
2386         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
2387         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
2388         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
2389         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
2390         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
2391         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
2392
2393 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2394
2395         * config/arc/arc.opt (mlra): Move comment above option name
2396         to avoid mis-parsing as language options.
2397
2398 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2399
2400         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
2401         * config/sol2.h: ... here.
2402         * config/sol2-10.h: Remove.
2403
2404         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
2405         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
2406         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
2407         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
2408         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
2409         * config/sol2.h: ... here.
2410         (SECTION_NAME_FORMAT): Don't redefine.
2411         (STARTFILE_ARCH32_SPEC): Rename to ...
2412         (STARTFILE_ARCH_SPEC): ... this.
2413         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
2414         * config/sparc/sol2.h: ... here.
2415         (SECTION_NAME_FORMAT): Don't undef.
2416         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
2417         (SUBTARGET_EXTRA_SPECS): Remove.
2418         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
2419
2420         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
2421         (MD_STARTFILE_PREFIX): Remove.
2422         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
2423         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
2424         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
2425         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
2426         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
2427         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
2428         * config/i386/sol2.h: ... here.
2429         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
2430         * config/i386/sol2-bi.h: Remove.
2431         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
2432         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
2433
2434         * config/i386/t-sol2-64: Rename to ...
2435         * config/i386/t-sol2: ... this.
2436         * config/sparc/t-sol2-64: Rename to ...
2437         * config/sparc/t-sol2: ... this.
2438
2439         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
2440         sol2_tm_file_head, sol2_tm_file_tail.
2441         Include ${cpu_type}/sol2.h before sol2.h.
2442         Remove sol2-10.h.
2443         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
2444         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
2445         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
2446         Reflect i386/t-sol2-64 renaming.
2447         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
2448         Reflect sparc/t-sol2-64 renaming.
2449
2450 2014-04-30  Richard Biener  <rguenther@suse.de>
2451
2452         * passes.c (execute_function_todo): Move TODO_verify_stmts
2453         and TODO_verify_ssa under the TODO_verify_il umbrella.
2454         * tree-ssa.h (verify_ssa): Adjust prototype.
2455         * tree-ssa.c (verify_ssa): Add parameter to tell whether
2456         we should verify SSA operands.
2457         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
2458         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
2459         whether we should verify whether not throwing stmts have EH info.
2460         * graphite-scop-detection.c (create_sese_edges): Adjust.
2461         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
2462         * tree-eh.c (lower_try_finally_switch): Do not add the
2463         default case label twice.
2464
2465 2014-04-30  Marek Polacek  <polacek@redhat.com>
2466
2467         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
2468         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
2469         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
2470         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
2471
2472 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
2473
2474         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
2475         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
2476         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
2477         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
2478         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
2479         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
2480         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
2481         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
2482
2483 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
2484
2485         * tree-cfg.c (dump_function_to_file): Dump the return type of
2486         functions, in a line to itself before the function body, mimicking
2487         the layout of a C function.
2488
2489 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
2490
2491         PR tree-optimization/60971
2492         * tree-tailcall.c (process_assignment): Reject conversions which
2493         reduce precision.
2494
2495 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
2496
2497         * calls.c (initialize_argument_information): Always treat
2498         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
2499         (expand_call): Likewise.
2500         (emit_library_call_calue_1): Likewise.
2501         * expr.c (PUSH_ARGS_REVERSED): Do not define.
2502         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
2503         code accordingly.
2504
2505 2014-04-29  Nick Clifton  <nickc@redhat.com>
2506
2507         * config/msp430/msp430.md (umulsidi): Fix typo.
2508         (mulhisi3): Enable even inside interrupt handlers.
2509         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
2510         bigger return address pushed in large mode.
2511
2512 2014-04-29  Nick Clifton  <nickc@redhat.com>
2513
2514         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
2515         (arc_init_reg_tables): Use a machine_mode enum to iterate over
2516         available modes.
2517         * config/m32r/m32r.c (init_reg_tables): Likewise.
2518         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
2519         enum to hold the modes.
2520
2521 2014-04-29  Richard Biener  <rguenther@suse.de>
2522
2523         * dominance.c (free_dominance_info): Add overload with
2524         function parameter.
2525         (dom_info_state): Likewise.
2526         (dom_info_available_p): Likewise.
2527         * basic-block.h (free_dominance_info, dom_info_state,
2528         dom_info_available_p): Declare overloads.
2529         * passes.c (execute_function_todo): Verify that verifiers
2530         don't change dominator info state.  Drop dominator info
2531         for IPA pass invocations.
2532         * cgraph.c (release_function_body): Restore asserts that
2533         dominator information is released.
2534
2535 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
2536
2537         * doc/invoke.texi: Fix typo.
2538         * tree-vrp.c: Fix typos.
2539         * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
2540
2541 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2542
2543         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
2544
2545 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2546
2547         * config/aarch64/aarch64-builtins.c
2548         (aarch64_types_storestruct_lane_qualifiers): New.
2549         (TYPES_STORESTRUCT_LANE): Likewise.
2550         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
2551         (st3_lane): Likewise.
2552         (st4_lane): Likewise.
2553         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
2554         (vec_store_lanesci_lane<mode>): Likewise.
2555         (vec_store_lanesxi_lane<mode>): Likewise.
2556         (aarch64_st2_lane<VQ:mode>): Likewise.
2557         (aarch64_st3_lane<VQ:mode>): Likewise.
2558         (aarch64_st4_lane<VQ:mode>): Likewise.
2559         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
2560         * config/aarch64/arm_neon.h
2561         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
2562         use new macro arguments.
2563         (__ST3_LANE_FUNC): Likewise.
2564         (__ST4_LANE_FUNC): Likewise.
2565         * config/aarch64/iterators.md (V_TWO_ELEM): New.
2566         (V_THREE_ELEM): Likewise.
2567         (V_FOUR_ELEM): Likewise.
2568
2569 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
2570
2571         * doc/gimple.texi: Replace the description of the now-defunct
2572         union gimple_statement_d with a diagram showing the
2573         gimple_statement_base class hierarchy and its relationships to
2574         the GSS_ and GIMPLE_ enums.
2575
2576 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2577
2578         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
2579         * config/aarch64/aarch64.c
2580         (aarch64_cannot_change_mode_class): Weaken conditions.
2581         (aarch64_modes_tieable_p): New.
2582         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
2583
2584 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
2585
2586         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2587         (loadsync_<mode>): Change mode.
2588         (load_quadpti, store_quadpti): New.
2589         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2590         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2591
2592 2014-04-28  Martin Jambor  <mjambor@suse.cz>
2593
2594         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
2595         same alias type as the original statement.
2596         (subreplacement_assignment_data): New type.
2597         (handle_unscalarized_data_in_subtree): New type of parameter,
2598         generate new memory accesses with same alias type as the original
2599         statement.
2600         (load_assign_lhs_subreplacements): Likewise.
2601         (sra_modify_constructor_assign): Generate new memory accesses with
2602         same alias type as the original statement.
2603
2604 2014-04-28  Richard Biener  <rguenther@suse.de>
2605
2606         * tree-pass.h (TODO_verify_il): Define.
2607         (TODO_verify_all): Complete properly.
2608         * passes.c (execute_function_todo): Move existing loop-closed
2609         SSA verification under TODO_verify_il.
2610         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
2611         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
2612         Fix tree sharing issue.
2613
2614 2014-04-28  Richard Biener  <rguenther@suse.de>
2615
2616         PR middle-end/60092
2617         * builtins.def (DEF_C11_BUILTIN): Add.
2618         (BUILT_IN_ALIGNED_ALLOC): Likewise.
2619         * coretypes.h (enum function_class): Add function_c11_misc.
2620         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
2621         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
2622         (call_may_clobber_ref_p_1): Likewise.
2623         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2624         (mark_all_reaching_defs_necessary_1): Likewise.
2625         (propagate_necessity): Likewise.
2626         (eliminate_unnecessary_stmts): Likewise.
2627         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
2628
2629 2014-04-28  Richard Biener  <rguenther@suse.de>
2630
2631         * tree-vrp.c (vrp_var_may_overflow): Remove.
2632         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
2633         with overflow immediately bump to one before that value and
2634         let iteration figure out overflow status.
2635
2636 2014-04-28  Richard Biener  <rguenther@suse.de>
2637
2638         * configure.ac: Do valgrind header checks unconditionally.
2639         Add --enable-valgrind-annotations.
2640         * system.h: Guard valgrind header inclusion with
2641         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
2642         * alloc-pool.c (pool_alloc, pool_free): Use
2643         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
2644         to guard possibly dead code.
2645         * config.in: Regenerated.
2646         * configure: Likewise.
2647
2648 2014-04-28  Jeff Law  <law@redhat.com>
2649
2650         PR tree-optimization/60902
2651         * tree-ssa-threadedge.c
2652         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
2653         over real defs when invalidating outputs from statements that do not
2654         produce useful outputs for threading.
2655
2656 2014-04-28  Richard Biener  <rguenther@suse.de>
2657
2658         PR tree-optimization/60979
2659         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
2660         SCOPs that end in a block with a successor with abnormal
2661         predecessors.
2662
2663 2014-04-28  Richard Biener  <rguenther@suse.de>
2664
2665         * tree-pass.h (execute_pass_list): Adjust prototype.
2666         * passes.c (pass_manager::execute_early_local_passes): Adjust.
2667         (do_per_function): Change callback signature, push all actual
2668         work to the callbals.
2669         (do_per_function_toporder): Likewise.
2670         (execute_function_dump): Adjust.
2671         (execute_function_todo): Likewise.
2672         (clear_last_verified): Likewise.
2673         (verify_curr_properties): Likewise.
2674         (update_properties_after_pass): Likewise.
2675         (execute_pass_list_1): Split out from ...
2676         (execute_pass_list): ... here.  Adjust.
2677         (execute_ipa_pass_list): Likewise.
2678         * cgraphunit.c (cgraph_add_new_function): Adjust.
2679         (analyze_function): Likewise.
2680         (expand_function): Likewise.
2681         * cgraph.c (release_function_body): Free dominance info
2682         here instead of asserting it was magically freed elsewhere.
2683
2684 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
2685
2686         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
2687         * configure: Regenerate.
2688         * config/sparc/sparc.opt (muser-mode): New option.
2689         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
2690         for LEON3.
2691         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
2692         * doc/invoke.texi (SPARC options): Document -muser-mode.
2693
2694 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
2695
2696         * cselib.c (find_slot_memmode): Delete.
2697         (cselib_hasher): Change compare_type to a struct.
2698         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
2699         constants.
2700         (preserve_constants_and_equivs): Adjust for new compare_type.
2701         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
2702         (wrap_constant): Delete.
2703         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
2704
2705 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
2706
2707         * doc/install.texi (Building with profile feedback): Remove
2708         outdated sentence.
2709
2710 2014-04-26  Tom de Vries  <tom@codesourcery.com>
2711
2712         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
2713         array accesses.
2714
2715 2014-04-25  Cary Coutant  <ccoutant@google.com>
2716
2717         PR debug/60929
2718         * dwarf2out.c (should_move_die_to_comdat): A type definition
2719         can contain a subprogram definition, but don't move it to a
2720         comdat unit.
2721         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
2722         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
2723         from original DIE.
2724         (clone_tree_hash): Rename to...
2725         (clone_tree_partial): ...this; change callers.  Copy
2726         DW_TAG_subprogram DIEs as declarations.
2727         (copy_decls_walk): Don't copy children of a declaration into a
2728         type unit.
2729
2730 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
2731
2732         PR target/60969
2733         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
2734         alternative 12.
2735
2736 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
2737
2738         * config/arm/predicates.md (call_insn_operand): Add long_call check.
2739         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
2740         reg for long_call.
2741         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
2742         restriction.
2743
2744 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2745
2746         * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
2747
2748 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2749
2750         PR tree-optimization/60930
2751         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
2752         creating a multiply candidate by folding two constant
2753         multiplicands when the result overflows.
2754
2755 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
2756
2757         PR tree-optimization/60960
2758         * tree-vect-generic.c (expand_vector_operation): Only call
2759         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
2760
2761 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2762
2763         * expr.c (clobber_reg_mode): New function.
2764         * expr.h (clobber_reg): New function.
2765
2766 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2767
2768         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
2769         clobbers.
2770
2771 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
2772             Tom de Vries  <tom@codesourcery.com>
2773
2774         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
2775         handle.
2776         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
2777         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
2778         new argument to find_all_hard_reg_sets call.
2779
2780 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2781
2782         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
2783         Use HOST_WIDE_INT_C for mask literal.
2784         (aarch_rev16_shleft_mask_imm_p): Likewise.
2785
2786 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
2787
2788         PR target/60941
2789         * config/sparc/sparc.md (ashlsi3_extend): Delete.
2790
2791 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
2792
2793         PR preprocessor/56540
2794         * config/i386/i386-c.c (ix86_target_macros): Define
2795         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
2796
2797 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2798
2799         * configure.ac (tga_func): Remove.
2800         (LIB_TLS_SPEC): Remove.
2801         * configure: Regenerate.
2802         * config.in: Regenerate.
2803         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
2804
2805 2014-04-25  Richard Biener  <rguenther@suse.de>
2806
2807         PR ipa/60912
2808         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
2809         call stmt use/clobber sets during stmt walk instead of
2810         walking the possibly incomplete set of caller edges.
2811
2812 2014-04-25  Richard Biener  <rguenther@suse.de>
2813
2814         PR ipa/60911
2815         * passes.c (apply_ipa_transforms): Inline into only caller ...
2816         (execute_one_pass): ... here.  Properly bring in function
2817         bodies for nodes we want to apply IPA transforms to.
2818
2819 2014-04-24  Cong Hou  <congh@google.com>
2820
2821         PR tree-optimization/60896
2822         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
2823         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
2824         (vect_mark_pattern_stmts): Set the def type of all statements in
2825         PATTERN_DEF_SEQ as vect_internal_def.
2826
2827 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
2828
2829         * doc/extend.texi (PowerPC Built-in Functions): Document new
2830         powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
2831         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
2832
2833         * config/rs6000/predicates.md (const_0_to_3_operand): New
2834         predicate to match 0..3 integer constants.
2835
2836         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
2837         to support adding miscellaneous builtin functions.
2838         (BU_DFP_MISC_2): Likewise.
2839         (BU_P7_MISC_1): Likewise.
2840         (BU_P7_MISC_2): Likewise.
2841         (BU_P8V_MISC_3): Likewise.
2842         (BU_MISC_1): Likewise.
2843         (BU_MISC_2): Likewise.
2844         (DIVWE): Add extended divide builtin functions.
2845         (DIVWEO): Likewise.
2846         (DIVWEU): Likewise.
2847         (DIVWEUO): Likewise.
2848         (DIVDE): Likewise.
2849         (DIVDEO): Likewise.
2850         (DIVDEU): Likewise.
2851         (DIVDEUO): Likewise.
2852         (DXEX): Add decimal floating-point builtin functions.
2853         (DXEXQ): Likewise.
2854         (DDEDPD): Likewise.
2855         (DDEDPDQ): Likewise.
2856         (DENBCD): Likewise.
2857         (DENBCDQ): Likewise.
2858         (DIEX): Likewise.
2859         (DIEXQ): Likewise.
2860         (DSCLI): Likewise.
2861         (DSCLIQ): Likewise.
2862         (DSCRI): Likewise.
2863         (DSCRIQ): Likewise.
2864         (CDTBCD): Add new BCD builtin functions.
2865         (CBCDTD): Likewise.
2866         (ADDG6S): Likewise.
2867         (BCDADD): Likewise.
2868         (BCDADD_LT): Likewise.
2869         (BCDADD_EQ): Likewise.
2870         (BCDADD_GT): Likewise.
2871         (BCDADD_OV): Likewise.
2872         (BCDSUB): Likewise.
2873         (BCDSUB_LT): Likewise.
2874         (BCDSUB_EQ): Likewise.
2875         (BCDSUB_GT): Likewise.
2876         (BCDSUB_OV): Likewise.
2877         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
2878         (UNPACK_TD): Likewise.
2879         (PACK_TF): Likewise.
2880         (UNPACK_TF): Likewise.
2881         (UNPACK_TF_0): Likewise.
2882         (UNPACK_TF_1): Likewise.
2883         (PACK_V1TI): Likewise.
2884         (UNPACK_V1TI): Likewise.
2885
2886         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2887         support for decimal floating point builtin functions.
2888         (rs6000_expand_ternop_builtin): Add checks for the new builtin
2889         functions that take constant arguments.
2890         (rs6000_invalid_builtin): Add decimal floating point builtin support.
2891         (rs6000_init_builtins): Setup long double, _Decimal64, and
2892         _Decimal128 types for new builtin functions.
2893         (builtin_function_type): Set the unsigned flags appropriately for
2894         the new builtin functions.
2895         (rs6000_opt_masks): Add support for decimal floating point builtin
2896         functions.
2897
2898         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
2899         floating point builtin functions.
2900         (RS6000_BTM_COMMON): Likewise.
2901         (RS6000_BTI_long_double): Likewise.
2902         (RS6000_BTI_dfloat64): Likewise.
2903         (RS6000_BTI_dfloat128): Likewise.
2904         (long_double_type_internal_node): Likewise.
2905         (dfloat64_type_internal_node): Likewise.
2906         (dfloat128_type_internal_node): Likewise.
2907
2908         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
2909         2.07 bcd arithmetic instructions.
2910         (UNSPEC_BCDSUB): Likewise.
2911         (UNSPEC_BCD_OVERFLOW): Likewise.
2912         (UNSPEC_BCD_ADD_SUB): Likewise.
2913         (bcd_add_sub): Likewise.
2914         (BCD_TEST): Likewise.
2915         (bcd<bcd_add_sub>): Likewise.
2916         (bcd<bcd_add_sub>_test): Likewise.
2917         (bcd<bcd_add_sub>_test2): Likewise.
2918         (bcd<bcd_add_sub>_<code>): Likewise.
2919         (peephole2 for combined bcd ops): Likewise.
2920
2921         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
2922         decimal floating point builtin functions.
2923         (UNSPEC_DENBCD): Likewise.
2924         (UNSPEC_DXEX): Likewise.
2925         (UNSPEC_DIEX): Likewise.
2926         (UNSPEC_DSCLI): Likewise.
2927         (UNSPEC_DSCRI): Likewise.
2928         (D64_D128): Likewise.
2929         (dfp_suffix): Likewise.
2930         (dfp_ddedpd_<mode>): Likewise.
2931         (dfp_denbcd_<mode>): Likewise.
2932         (dfp_dxex_<mode>): Likewise.
2933         (dfp_diex_<mode>): Likewise.
2934         (dfp_dscli_<mode>): Likewise.
2935         (dfp_dscri_<mode>): Likewise.
2936
2937         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
2938         builtin functions.
2939         (UNSPEC_CDTBCD): Likewise.
2940         (UNSPEC_CBCDTD): Likewise.
2941         (UNSPEC_DIVE): Add support for new extended divide builtin functions.
2942         (UNSPEC_DIVEO): Likewise.
2943         (UNSPEC_DIVEU): Likewise.
2944         (UNSPEC_DIVEUO): Likewise.
2945         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
2946         pack/unpack 128-bit types.
2947         (UNSPEC_PACK_128BIT): Likewise.
2948         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
2949         (udiv<mode>3): Use idiv_ldiv mode attribute.
2950         (div<mode>3): Likewise.
2951         (addg6s): Add new BCD builtin functions.
2952         (cdtbcd): Likewise.
2953         (cbcdtd): Likewise.
2954         (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
2955         (div_extend): Likewise.
2956         (div<div_extend>_<mode>"): Likewise.
2957         (FP128_64): Add support for new builtin functions to pack/unpack
2958         128-bit types.
2959         (unpack<mode>): Likewise.
2960         (unpacktf_0): Likewise.
2961         (unpacktf_1): Likewise.
2962         (unpack<mode>_dm): Likewise.
2963         (unpack<mode>_nodm): Likewise.
2964         (pack<mode>): Likewise.
2965         (unpackv1ti): Likewise.
2966         (packv1ti): Likewise.
2967
2968 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
2969
2970         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
2971         is disabled.
2972
2973 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
2974
2975         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
2976         * gimplify.c (omp_is_private): Change last argument's type to int.
2977         Only diagnose lastprivate if the simd argument is 1, only diagnose
2978         linear if the simd argument is 2.
2979         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
2980         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
2981         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
2982         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
2983         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
2984         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2985         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
2986         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2987         * tree-nested.c (convert_nonlocal_omp_clauses,
2988         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
2989
2990 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
2991
2992         PR target/60822
2993         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
2994         operand 1.
2995
2996 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
2997
2998         * flag-types.h (enum ivar_visibility): Add.
2999
3000 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
3001
3002         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
3003         function * argument.
3004
3005 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
3006
3007         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
3008
3009 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
3010             Tom de Vries  <tom@codesourcery.com>
3011
3012         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
3013         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
3014         reg-note.
3015         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
3016         * emit-rtl.c (try_split): Same.
3017
3018 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
3019             Tom de Vries  <tom@codesourcery.com>
3020
3021         * common.opt (fuse-caller-save): New option.
3022
3023 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
3024
3025         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
3026         elements for big-endian.
3027
3028 2014-04-24  Richard Biener  <rguenther@suse.de>
3029
3030         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
3031         during TER and instead use the sepops interface for expanding
3032         non-GIMPLE_SINGLE_RHS.
3033
3034 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3035
3036         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
3037         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
3038
3039 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3040
3041         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
3042         assembler 64-bit option.
3043         * configure: Regenerate.
3044
3045 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3046
3047         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
3048         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
3049         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
3050         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
3051         (TARGET_CRYPTO): Take TARGET_SIMD into account.
3052
3053 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3054
3055         * config/aarch64/aarch64-builtins.c
3056         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
3057         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
3058         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
3059         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
3060         builtins.
3061         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
3062         (Vrevsuff): New mode attribute.
3063
3064 2014-04-24  Terry Guo  <terry.guo@arm.com>
3065
3066         * config/arm/arm.h (machine_function): Define variable
3067         after_arm_reorg here.
3068         * config/arm/arm.c (after_arm_reorg): Remove the definition.
3069         (arm_split_constant): Update the way to access variable
3070         after_arm_reorg.
3071         (arm_reorg): Ditto.
3072         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
3073
3074 2014-04-23  Tom de Vries  <tom@codesourcery.com>
3075
3076         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
3077
3078 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
3079
3080         * is-a.h: Update comments to reflect the following changes to the
3081         "pointerness" of the API, making the template parameter match the
3082         return type, allowing use of is-a.h with typedefs of pointers.
3083         (is_a_helper::cast): Return a T rather then a pointer to a T, so
3084         that the return type matches the parameter to the is_a_helper.
3085         (as_a): Likewise.
3086         (dyn_cast): Likewise.
3087
3088         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
3089         pointer from the is-a.h API.
3090
3091         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
3092         (is_a_helper <cgraph_node *>::test): ...this, matching change to
3093         is-a.h API.
3094         (is_a_helper <varpool_node>::test): Likewise, convert to...
3095         (is_a_helper <varpool_node *>::test): ...this.
3096
3097         (varpool_first_variable): Update for removal of implicit pointer
3098         from the is-a.h API.
3099         (varpool_next_variable): Likewise.
3100         (varpool_first_static_initializer): Likewise.
3101         (varpool_next_static_initializer): Likewise.
3102         (varpool_first_defined_variable): Likewise.
3103         (varpool_next_defined_variable): Likewise.
3104         (cgraph_first_defined_function): Likewise.
3105         (cgraph_next_defined_function): Likewise.
3106         (cgraph_first_function): Likewise.
3107         (cgraph_next_function): Likewise.
3108         (cgraph_first_function_with_gimple_body): Likewise.
3109         (cgraph_next_function_with_gimple_body): Likewise.
3110         (cgraph_alias_target): Likewise.
3111         (varpool_alias_target): Likewise.
3112         (cgraph_function_or_thunk_node): Likewise.
3113         (varpool_variable_node): Likewise.
3114         (symtab_real_symbol_p): Likewise.
3115         * cgraphunit.c (referred_to_p): Likewise.
3116         (analyze_functions): Likewise.
3117         (handle_alias_pairs): Likewise.
3118         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
3119         * gimple-ssa.h (gimple_vuse_op): Likewise.
3120         (gimple_vdef_op): Likewise.
3121         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
3122         * gimple.c (gimple_build_asm_1): Likewise.
3123         (gimple_build_try): Likewise.
3124         (gimple_build_resx): Likewise.
3125         (gimple_build_eh_dispatch): Likewise.
3126         (gimple_build_omp_for): Likewise.
3127         (gimple_omp_for_set_clauses): Likewise.
3128
3129         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
3130         (is_a_helper <gimple_statement_asm *>::test): ...this.
3131         (is_a_helper <gimple_statement_bind>::test): Convert to...
3132         (is_a_helper <gimple_statement_bind *>::test): ...this.
3133         (is_a_helper <gimple_statement_call>::test): Convert to...
3134         (is_a_helper <gimple_statement_call *>::test): ...this.
3135         (is_a_helper <gimple_statement_catch>::test): Convert to...
3136         (is_a_helper <gimple_statement_catch *>::test): ...this.
3137         (is_a_helper <gimple_statement_resx>::test): Convert to...
3138         (is_a_helper <gimple_statement_resx *>::test): ...this.
3139         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
3140         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
3141         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
3142         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
3143         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
3144         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
3145         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
3146         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
3147         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
3148         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
3149         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
3150         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
3151         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
3152         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
3153         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
3154         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
3155         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
3156         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
3157         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
3158         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
3159         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
3160         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
3161         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
3162         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
3163         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
3164         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
3165         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
3166         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
3167         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
3168         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
3169         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
3170         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
3171         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
3172         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
3173         (is_a_helper <gimple_statement_phi>::test): Convert to...
3174         (is_a_helper <gimple_statement_phi *>::test): ...this.
3175         (is_a_helper <gimple_statement_transaction>::test): Convert to...
3176         (is_a_helper <gimple_statement_transaction *>::test): ...this.
3177         (is_a_helper <gimple_statement_try>::test): Convert to...
3178         (is_a_helper <gimple_statement_try *>::test): ...this.
3179         (is_a_helper <gimple_statement_wce>::test): Convert to...
3180         (is_a_helper <gimple_statement_wce *>::test): ...this.
3181         (is_a_helper <const gimple_statement_asm>::test): Convert to...
3182         (is_a_helper <const gimple_statement_asm *>::test): ...this.
3183         (is_a_helper <const gimple_statement_bind>::test): Convert to...
3184         (is_a_helper <const gimple_statement_bind *>::test): ...this.
3185         (is_a_helper <const gimple_statement_call>::test): Convert to...
3186         (is_a_helper <const gimple_statement_call *>::test): ...this.
3187         (is_a_helper <const gimple_statement_catch>::test): Convert to...
3188         (is_a_helper <const gimple_statement_catch *>::test): ...this.
3189         (is_a_helper <const gimple_statement_resx>::test): Convert to...
3190         (is_a_helper <const gimple_statement_resx *>::test): ...this.
3191         (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
3192         (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
3193         (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
3194         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
3195         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
3196         Convert to...
3197         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
3198         ...this.
3199         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
3200         Convert to...
3201         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
3202         ...this.
3203         (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
3204         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
3205         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
3206         to...
3207         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
3208         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
3209         to...
3210         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
3211         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
3212         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
3213         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
3214         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
3215         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
3216         to...
3217         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
3218         (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
3219         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
3220         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
3221         to...
3222         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
3223         (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
3224         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
3225         (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
3226         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
3227         (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
3228         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
3229         (is_a_helper <const gimple_statement_phi>::test): Convert to...
3230         (is_a_helper <const gimple_statement_phi *>::test): ...this.
3231         (is_a_helper <const gimple_statement_transaction>::test): Convert to...
3232         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
3233         (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
3234         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
3235         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
3236         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
3237         (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
3238         to...
3239         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
3240         ...this.
3241         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
3242         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
3243
3244         (gimple_use_ops): Update for removal of implicit pointer from the
3245         is-a.h API.
3246         (gimple_set_use_ops): Likewise.
3247         (gimple_vuse): Likewise.
3248         (gimple_vdef): Likewise.
3249         (gimple_vuse_ptr): Likewise.
3250         (gimple_vdef_ptr): Likewise.
3251         (gimple_set_vuse): Likewise.
3252         (gimple_set_vdef): Likewise.
3253         (gimple_omp_return_set_lhs): Likewise.
3254         (gimple_omp_return_lhs): Likewise.
3255         (gimple_omp_return_lhs_ptr): Likewise.
3256         (gimple_call_fntype): Likewise.
3257         (gimple_call_set_fntype): Likewise.
3258         (gimple_call_set_internal_fn): Likewise.
3259         (gimple_call_use_set): Likewise.
3260         (gimple_call_clobber_set): Likewise.
3261         (gimple_bind_vars): Likewise.
3262         (gimple_bind_set_vars): Likewise.
3263         (gimple_bind_body_ptr): Likewise.
3264         (gimple_bind_set_body): Likewise.
3265         (gimple_bind_add_stmt): Likewise.
3266         (gimple_bind_block): Likewise.
3267         (gimple_bind_set_block): Likewise.
3268         (gimple_asm_ninputs): Likewise.
3269         (gimple_asm_noutputs): Likewise.
3270         (gimple_asm_nclobbers): Likewise.
3271         (gimple_asm_nlabels): Likewise.
3272         (gimple_asm_input_op): Likewise.
3273         (gimple_asm_input_op_ptr): Likewise.
3274         (gimple_asm_output_op): Likewise.
3275         (gimple_asm_output_op_ptr): Likewise.
3276         (gimple_asm_set_output_op): Likewise.
3277         (gimple_asm_clobber_op): Likewise.
3278         (gimple_asm_set_clobber_op): Likewise.
3279         (gimple_asm_label_op): Likewise.
3280         (gimple_asm_set_label_op): Likewise.
3281         (gimple_asm_string): Likewise.
3282         (gimple_catch_types): Likewise.
3283         (gimple_catch_types_ptr): Likewise.
3284         (gimple_catch_handler_ptr): Likewise.
3285         (gimple_catch_set_types): Likewise.
3286         (gimple_catch_set_handler): Likewise.
3287         (gimple_eh_filter_types): Likewise.
3288         (gimple_eh_filter_types_ptr): Likewise.
3289         (gimple_eh_filter_failure_ptr): Likewise.
3290         (gimple_eh_filter_set_types): Likewise.
3291         (gimple_eh_filter_set_failure): Likewise.
3292         (gimple_eh_must_not_throw_fndecl): Likewise.
3293         (gimple_eh_must_not_throw_set_fndecl): Likewise.
3294         (gimple_eh_else_n_body_ptr): Likewise.
3295         (gimple_eh_else_e_body_ptr): Likewise.
3296         (gimple_eh_else_set_n_body): Likewise.
3297         (gimple_eh_else_set_e_body): Likewise.
3298         (gimple_try_eval_ptr): Likewise.
3299         (gimple_try_cleanup_ptr): Likewise.
3300         (gimple_try_set_eval): Likewise.
3301         (gimple_try_set_cleanup): Likewise.
3302         (gimple_wce_cleanup_ptr): Likewise.
3303         (gimple_wce_set_cleanup): Likewise.
3304         (gimple_phi_capacity): Likewise.
3305         (gimple_phi_num_args): Likewise.
3306         (gimple_phi_result): Likewise.
3307         (gimple_phi_result_ptr): Likewise.
3308         (gimple_phi_set_result): Likewise.
3309         (gimple_phi_arg): Likewise.
3310         (gimple_phi_set_arg): Likewise.
3311         (gimple_resx_region): Likewise.
3312         (gimple_resx_set_region): Likewise.
3313         (gimple_eh_dispatch_region): Likewise.
3314         (gimple_eh_dispatch_set_region): Likewise.
3315         (gimple_omp_critical_name): Likewise.
3316         (gimple_omp_critical_name_ptr): Likewise.
3317         (gimple_omp_critical_set_name): Likewise.
3318         (gimple_omp_for_clauses): Likewise.
3319         (gimple_omp_for_clauses_ptr): Likewise.
3320         (gimple_omp_for_set_clauses): Likewise.
3321         (gimple_omp_for_collapse): Likewise.
3322         (gimple_omp_for_index): Likewise.
3323         (gimple_omp_for_index_ptr): Likewise.
3324         (gimple_omp_for_set_index): Likewise.
3325         (gimple_omp_for_initial): Likewise.
3326         (gimple_omp_for_initial_ptr): Likewise.
3327         (gimple_omp_for_set_initial): Likewise.
3328         (gimple_omp_for_final): Likewise.
3329         (gimple_omp_for_final_ptr): Likewise.
3330         (gimple_omp_for_set_final): Likewise.
3331         (gimple_omp_for_incr): Likewise.
3332         (gimple_omp_for_incr_ptr): Likewise.
3333         (gimple_omp_for_set_incr): Likewise.
3334         (gimple_omp_for_pre_body_ptr): Likewise.
3335         (gimple_omp_for_set_pre_body): Likewise.
3336         (gimple_omp_parallel_clauses): Likewise.
3337         (gimple_omp_parallel_clauses_ptr): Likewise.
3338         (gimple_omp_parallel_set_clauses): Likewise.
3339         (gimple_omp_parallel_child_fn): Likewise.
3340         (gimple_omp_parallel_child_fn_ptr): Likewise.
3341         (gimple_omp_parallel_set_child_fn): Likewise.
3342         (gimple_omp_parallel_data_arg): Likewise.
3343         (gimple_omp_parallel_data_arg_ptr): Likewise.
3344         (gimple_omp_parallel_set_data_arg): Likewise.
3345         (gimple_omp_task_clauses): Likewise.
3346         (gimple_omp_task_clauses_ptr): Likewise.
3347         (gimple_omp_task_set_clauses): Likewise.
3348         (gimple_omp_task_child_fn): Likewise.
3349         (gimple_omp_task_child_fn_ptr): Likewise.
3350         (gimple_omp_task_set_child_fn): Likewise.
3351         (gimple_omp_task_data_arg): Likewise.
3352         (gimple_omp_task_data_arg_ptr): Likewise.
3353         (gimple_omp_task_set_data_arg): Likewise.
3354         (gimple_omp_taskreg_clauses): Likewise.
3355         (gimple_omp_taskreg_clauses_ptr): Likewise.
3356         (gimple_omp_taskreg_set_clauses): Likewise.
3357         (gimple_omp_taskreg_child_fn): Likewise.
3358         (gimple_omp_taskreg_child_fn_ptr): Likewise.
3359         (gimple_omp_taskreg_set_child_fn): Likewise.
3360         (gimple_omp_taskreg_data_arg): Likewise.
3361         (gimple_omp_taskreg_data_arg_ptr): Likewise.
3362         (gimple_omp_taskreg_set_data_arg): Likewise.
3363         (gimple_omp_task_copy_fn): Likewise.
3364         (gimple_omp_task_copy_fn_ptr): Likewise.
3365         (gimple_omp_task_set_copy_fn): Likewise.
3366         (gimple_omp_task_arg_size): Likewise.
3367         (gimple_omp_task_arg_size_ptr): Likewise.
3368         (gimple_omp_task_set_arg_size): Likewise.
3369         (gimple_omp_task_arg_align): Likewise.
3370         (gimple_omp_task_arg_align_ptr): Likewise.
3371         (gimple_omp_task_set_arg_align): Likewise.
3372         (gimple_omp_single_clauses): Likewise.
3373         (gimple_omp_single_clauses_ptr): Likewise.
3374         (gimple_omp_single_set_clauses): Likewise.
3375         (gimple_omp_target_clauses): Likewise.
3376         (gimple_omp_target_clauses_ptr): Likewise.
3377         (gimple_omp_target_set_clauses): Likewise.
3378         (gimple_omp_target_child_fn): Likewise.
3379         (gimple_omp_target_child_fn_ptr): Likewise.
3380         (gimple_omp_target_set_child_fn): Likewise.
3381         (gimple_omp_target_data_arg): Likewise.
3382         (gimple_omp_target_data_arg_ptr): Likewise.
3383         (gimple_omp_target_set_data_arg): Likewise.
3384         (gimple_omp_teams_clauses): Likewise.
3385         (gimple_omp_teams_clauses_ptr): Likewise.
3386         (gimple_omp_teams_set_clauses): Likewise.
3387         (gimple_omp_sections_clauses): Likewise.
3388         (gimple_omp_sections_clauses_ptr): Likewise.
3389         (gimple_omp_sections_set_clauses): Likewise.
3390         (gimple_omp_sections_control): Likewise.
3391         (gimple_omp_sections_control_ptr): Likewise.
3392         (gimple_omp_sections_set_control): Likewise.
3393         (gimple_omp_for_set_cond): Likewise.
3394         (gimple_omp_for_cond): Likewise.
3395         (gimple_omp_atomic_store_set_val): Likewise.
3396         (gimple_omp_atomic_store_val): Likewise.
3397         (gimple_omp_atomic_store_val_ptr): Likewise.
3398         (gimple_omp_atomic_load_set_lhs): Likewise.
3399         (gimple_omp_atomic_load_lhs): Likewise.
3400         (gimple_omp_atomic_load_lhs_ptr): Likewise.
3401         (gimple_omp_atomic_load_set_rhs): Likewise.
3402         (gimple_omp_atomic_load_rhs): Likewise.
3403         (gimple_omp_atomic_load_rhs_ptr): Likewise.
3404         (gimple_omp_continue_control_def): Likewise.
3405         (gimple_omp_continue_control_def_ptr): Likewise.
3406         (gimple_omp_continue_set_control_def): Likewise.
3407         (gimple_omp_continue_control_use): Likewise.
3408         (gimple_omp_continue_control_use_ptr): Likewise.
3409         (gimple_omp_continue_set_control_use): Likewise.
3410         (gimple_transaction_body_ptr): Likewise.
3411         (gimple_transaction_label): Likewise.
3412         (gimple_transaction_label_ptr): Likewise.
3413         (gimple_transaction_set_body): Likewise.
3414         (gimple_transaction_set_label): Likewise.
3415
3416         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
3417         * ipa-inline-analysis.c (inline_write_summary): Likewise.
3418         * ipa-ref.c (ipa_record_reference): Likewise.
3419         * ipa-reference.c (analyze_function): Likewise.
3420         (ipa_reference_write_optimization_summary): Likewise.
3421         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
3422         (address_taken_from_non_vtable_p): Likewise.
3423         (comdat_can_be_unshared_p_1): Likewise.
3424         * lto-cgraph.c (lto_output_ref): Likewise.
3425         (add_references): Likewise.
3426         (compute_ltrans_boundary): Likewise.
3427         (output_symtab): Likewise.
3428         (input_ref): Likewise.
3429         (input_cgraph_1): Likewise.
3430         (output_cgraph_opt_summary): Likewise.
3431         * lto-streamer-out.c (lto_output): Likewise.
3432         (output_symbol_p): Likewise.
3433         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
3434         (lsei_start_function_in_partition): Likewise.
3435         (lsei_next_variable_in_partition): Likewise.
3436         (lsei_start_variable_in_partition): Likewise.
3437         * symtab.c (insert_to_assembler_name_hash): Likewise.
3438         (unlink_from_assembler_name_hash): Likewise.
3439         (symtab_unregister_node): Likewise.
3440         (symtab_remove_node): Likewise.
3441         (dump_symtab_node): Likewise.
3442         (verify_symtab_base): Likewise.
3443         (verify_symtab_node): Likewise.
3444         (symtab_make_decl_local): Likewise.
3445         (symtab_alias_ultimate_target): Likewise.
3446         (symtab_resolve_alias): Likewise.
3447         (symtab_get_symbol_partitioning_class): Likewise.
3448         * tree-phinodes.c (allocate_phi_node): Likewise.
3449         (reserve_phi_args_for_new_edge): Likewise.
3450         (remove_phi_args): Likewise.
3451         * varpool.c (varpool_node_for_asm): Likewise.
3452         (varpool_remove_unreferenced_decls): Likewise.
3453
3454 2014-04-23  Jeff Law  <law@redhat.com>
3455
3456         PR tree-optimization/60902
3457         * tree-ssa-threadedge.c
3458         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
3459         invalidate outputs from statements that do not produce useful
3460         outputs for threading.
3461
3462 2014-04-23  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3463
3464         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
3465         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
3466         machine descriptions for Stack Smashing Protector.
3467
3468 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
3469
3470         * aarch64.md (<optab>_rol<mode>3): New pattern.
3471         (<optab>_rolsi3_uxtw): Likewise.
3472         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
3473
3474 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
3475
3476         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
3477         (arm_cortex_a12_tune): Likewise.
3478
3479 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3480
3481         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
3482
3483 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3484
3485         * config/arm/arm.md (arm_rev16si2): New pattern.
3486         (arm_rev16si2_alt): Likewise.
3487         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
3488
3489 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3490
3491         * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
3492         (rev16<mode>2_alt): Likewise.
3493         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
3494         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
3495         (aarch_rev16_shleft_mask_imm_p): Likewise.
3496         (aarch_rev16_p_1): Likewise.
3497         (aarch_rev16_p): Likewise.
3498         * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
3499         (aarch_rev16_shright_mask_imm_p): Likewise.
3500         (aarch_rev16_shleft_mask_imm_p): Likewise.
3501
3502 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3503
3504         * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
3505         * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
3506         rev cost.
3507         (cortex_a53_extra_costs): Likewise.
3508         (cortex_a57_extra_costs): Likewise.
3509         * config/arm/arm.c (cortexa9_extra_costs): Likewise.
3510         (cortexa7_extra_costs): Likewise.
3511         (cortexa8_extra_costs): Likewise.
3512         (cortexa12_extra_costs): Likewise.
3513         (cortexa15_extra_costs): Likewise.
3514         (v7m_extra_costs): Likewise.
3515         (arm_new_rtx_costs): Handle BSWAP.
3516
3517 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3518
3519         * config/arm/arm.c (cortexa8_extra_costs): New table.
3520         (arm_cortex_a8_tune): New tuning struct.
3521         * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
3522
3523 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3524
3525         * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
3526
3527 2014-04-23  Richard Biener  <rguenther@suse.de>
3528
3529         * Makefile.in (OBJS): Remove loop-unswitch.o.
3530         * tree-pass.h (make_pass_rtl_unswitch): Remove.
3531         * passes.def (pass_rtl_unswitch): Likewise.
3532         * loop-init.c (gate_rtl_unswitch): Likewise.
3533         (rtl_unswitch): Likewise.
3534         (pass_data_rtl_unswitch): Likewise.
3535         (pass_rtl_unswitch): Likewise.
3536         (make_pass_rtl_unswitch): Likewise.
3537         * rtl.h (reversed_condition): Likewise.
3538         (compare_and_jump_seq): Likewise.
3539         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
3540         and make static.
3541         * loop-unroll.c (compare_and_jump_seq): Likewise.
3542
3543 2014-04-23  Richard Biener  <rguenther@suse.de>
3544
3545         PR tree-optimization/60903
3546         * tree-ssa-loop-im.c (analyze_memory_references): Remove
3547         commented code block.
3548         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
3549         loop flags to newly created BBs and edges.
3550
3551 2014-04-23  Nick Clifton  <nickc@redhat.com>
3552
3553         * config/msp430/msp430.c (msp430_handle_option): Move function
3554         to msp430-common.c
3555         (msp430_option_override): Simplify mcu and mcpu option handling.
3556         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
3557         support for -mhwmult command line option.
3558         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
3559         -mhwmult command line option.
3560         (msp430_hwmult_enabled): Delete.
3561         (msp43o_output_labelref): Add support for -mhwmult command line option.
3562         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
3563         (umulsidi3): Likewise.
3564         * config/msp430/msp430.opt (mmcu): Add Report attribute.
3565         (mcpu, mlarge, msmall): Likewise.
3566         (mhwmult): New option.
3567         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
3568         prototype.
3569         (msp430_is_f5_mcu): Remove prototype.
3570         (msp430_use_f5_series_hwmult): Add prototype.
3571         * config/msp430/msp430-opts.h: New file.
3572         * common/config/msp430: New directory.
3573         * common/config/msp430/msp430-common.c: New file.
3574         * config.gcc (msp430): Remove target_has_targetm_common.
3575         * doc/invoke.texi: Document -mhwmult command line option.
3576
3577 2014-04-23  Nick Clifton  <nickc@redhat.com>
3578
3579         * config/i386/cygwin.h (ENDFILE_SPEC): Include
3580         default-manifest.o if it can be found in the search path.
3581         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
3582
3583 2014-04-23  Terry Guo  <terry.guo@arm.com>
3584
3585         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
3586
3587 2014-04-23  Richard Biener  <rguenther@suse.de>
3588
3589         PR middle-end/60895
3590         * tree-inline.c (declare_return_variable): Use mark_addressable.
3591
3592 2014-04-23  Richard Biener  <rguenther@suse.de>
3593
3594         PR middle-end/60891
3595         * loop-init.c (loop_optimizer_init): Make sure to apply
3596         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
3597
3598 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3599
3600         PR sanitizer/60275
3601         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
3602         New options.
3603         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
3604         if flag_sanitize_undefined_trap_on_error.
3605         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
3606         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
3607         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
3608         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3609         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
3610         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
3611         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
3612         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
3613         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
3614         * ubsan.c (ubsan_instrument_unreachable): Return
3615         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
3616         (ubsan_expand_null_ifn): Emit __builtin_trap ()
3617         if flag_sanitize_undefined_trap_on_error and
3618         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
3619         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
3620         instrument_bool_enum_load): Emit __builtin_trap () if
3621         flag_sanitize_undefined_trap_on_error and
3622         __builtin_handle_*_abort () if !flag_sanitize_recover.
3623         * doc/invoke.texi (-fsanitize-recover,
3624         -fsanitize-undefined-trap-on-error): Document.
3625
3626 2014-04-22  Christian Bruel  <christian.bruel@st.com>
3627
3628         * config/sh/sh.md (mov<mode>): Replace movQIHI.
3629          Force immediates to SImode.
3630
3631 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
3632
3633         * config/nios2/nios2.md (UNSPEC_ROUND): New.
3634         (lroundsfsi2): New.
3635         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
3636         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
3637         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
3638         (nios2_fpu_insn): Add entry for round.
3639         (N2FPU_NO_ERRNO_P): Define.
3640         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
3641         flag_errno_math.
3642         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
3643
3644 2014-04-22  Richard Henderson  <rth@redhat.com>
3645
3646         * config/aarch64/aarch64 (addti3, subti3): New expanders.
3647         (add<GPI>3_compare0): Remove leading * from name.
3648         (add<GPI>3_carryin): Likewise.
3649         (sub<GPI>3_compare0): Likewise.
3650         (sub<GPI>3_carryin): Likewise.
3651         (<su_optab>mulditi3): New expander.
3652         (multi3): New expander.
3653         (madd<GPI>): Remove leading * from name.
3654
3655 2014-04-22  Martin Jambor  <mjambor@suse.cz>
3656
3657         * cgraphclones.c (cgraph_function_versioning): Copy
3658         ipa_transforms_to_apply instead of asserting it is empty.
3659
3660 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
3661
3662         PR target/60868
3663         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
3664         on count_exp to get mode.
3665
3666 2014-04-22  Andrew Pinski  <apinski@cavium.com>
3667
3668         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3669         Handle TLS for ILP32.
3670         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
3671         (tlsie_small_<mode>): this and handle PTR.
3672         (tlsie_small_sidi): New pattern.
3673         (tlsle_small): Change to an expand to handle ILP32.
3674         (tlsle_small_<mode>): New pattern.
3675         (tlsdesc_small): Rename to ...
3676         (tlsdesc_small_<mode>): this and handle PTR.
3677
3678 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3679
3680         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
3681
3682 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3683
3684         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3685         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
3686         (aarch64_types_signed_poly_qualifiers): Likewise.
3687         (aarch64_types_unsigned_signed_qualifiers): Likewise.
3688         (aarch64_types_poly_signed_qualifiers): Likewise.
3689         (TYPES_REINTERP_SS): Type macro added.
3690         (TYPES_REINTERP_SU): Likewise.
3691         (TYPES_REINTERP_SP): Likewise.
3692         (TYPES_REINTERP_US): Likewise.
3693         (TYPES_REINTERP_PS): Likewise.
3694         (aarch64_fold_builtin): New expression folding added.
3695         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
3696         Declarations removed.
3697         (REINTERP_SS): Declarations added.
3698         (REINTERP_US): Likewise.
3699         (REINTERP_PS): Likewise.
3700         (REINTERP_SU): Likewise.
3701         (REINTERP_SP): Likewise.
3702         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
3703         (vreinterpretq_p8_f64): Likewise.
3704         (vreinterpret_p16_f64): Likewise.
3705         (vreinterpretq_p16_f64): Likewise.
3706         (vreinterpret_f32_f64): Likewise.
3707         (vreinterpretq_f32_f64): Likewise.
3708         (vreinterpret_f64_f32): Likewise.
3709         (vreinterpret_f64_p8): Likewise.
3710         (vreinterpret_f64_p16): Likewise.
3711         (vreinterpret_f64_s8): Likewise.
3712         (vreinterpret_f64_s16): Likewise.
3713         (vreinterpret_f64_s32): Likewise.
3714         (vreinterpret_f64_s64): Likewise.
3715         (vreinterpret_f64_u8): Likewise.
3716         (vreinterpret_f64_u16): Likewise.
3717         (vreinterpret_f64_u32): Likewise.
3718         (vreinterpret_f64_u64): Likewise.
3719         (vreinterpretq_f64_f32): Likewise.
3720         (vreinterpretq_f64_p8): Likewise.
3721         (vreinterpretq_f64_p16): Likewise.
3722         (vreinterpretq_f64_s8): Likewise.
3723         (vreinterpretq_f64_s16): Likewise.
3724         (vreinterpretq_f64_s32): Likewise.
3725         (vreinterpretq_f64_s64): Likewise.
3726         (vreinterpretq_f64_u8): Likewise.
3727         (vreinterpretq_f64_u16): Likewise.
3728         (vreinterpretq_f64_u32): Likewise.
3729         (vreinterpretq_f64_u64): Likewise.
3730         (vreinterpret_s64_f64): Likewise.
3731         (vreinterpretq_s64_f64): Likewise.
3732         (vreinterpret_u64_f64): Likewise.
3733         (vreinterpretq_u64_f64): Likewise.
3734         (vreinterpret_s8_f64): Likewise.
3735         (vreinterpretq_s8_f64): Likewise.
3736         (vreinterpret_s16_f64): Likewise.
3737         (vreinterpretq_s16_f64): Likewise.
3738         (vreinterpret_s32_f64): Likewise.
3739         (vreinterpretq_s32_f64): Likewise.
3740         (vreinterpret_u8_f64): Likewise.
3741         (vreinterpretq_u8_f64): Likewise.
3742         (vreinterpret_u16_f64): Likewise.
3743         (vreinterpretq_u16_f64): Likewise.
3744         (vreinterpret_u32_f64): Likewise.
3745         (vreinterpretq_u32_f64): Likewise.
3746
3747 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3748
3749         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3750         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
3751         (vreinterpret_p8_s8): Likewise.
3752         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
3753         (vreinterpret_p8_s16): Likewise.
3754         (vreinterpret_p8_s32): Likewise.
3755         (vreinterpret_p8_s64): Likewise.
3756         (vreinterpret_p8_f32): Likewise.
3757         (vreinterpret_p8_u8): Likewise.
3758         (vreinterpret_p8_u16): Likewise.
3759         (vreinterpret_p8_u32): Likewise.
3760         (vreinterpret_p8_u64): Likewise.
3761         (vreinterpret_p8_p16): Likewise.
3762         (vreinterpretq_p8_s8): Likewise.
3763         (vreinterpretq_p8_s16): Likewise.
3764         (vreinterpretq_p8_s32): Likewise.
3765         (vreinterpretq_p8_s64): Likewise.
3766         (vreinterpretq_p8_f32): Likewise.
3767         (vreinterpretq_p8_u8): Likewise.
3768         (vreinterpretq_p8_u16): Likewise.
3769         (vreinterpretq_p8_u32): Likewise.
3770         (vreinterpretq_p8_u64): Likewise.
3771         (vreinterpretq_p8_p16): Likewise.
3772         (vreinterpret_p16_s8): Likewise.
3773         (vreinterpret_p16_s16): Likewise.
3774         (vreinterpret_p16_s32): Likewise.
3775         (vreinterpret_p16_s64): Likewise.
3776         (vreinterpret_p16_f32): Likewise.
3777         (vreinterpret_p16_u8): Likewise.
3778         (vreinterpret_p16_u16): Likewise.
3779         (vreinterpret_p16_u32): Likewise.
3780         (vreinterpret_p16_u64): Likewise.
3781         (vreinterpret_p16_p8): Likewise.
3782         (vreinterpretq_p16_s8): Likewise.
3783         (vreinterpretq_p16_s16): Likewise.
3784         (vreinterpretq_p16_s32): Likewise.
3785         (vreinterpretq_p16_s64): Likewise.
3786         (vreinterpretq_p16_f32): Likewise.
3787         (vreinterpretq_p16_u8): Likewise.
3788         (vreinterpretq_p16_u16): Likewise.
3789         (vreinterpretq_p16_u32): Likewise.
3790         (vreinterpretq_p16_u64): Likewise.
3791         (vreinterpretq_p16_p8): Likewise.
3792         (vreinterpret_f32_s8): Likewise.
3793         (vreinterpret_f32_s16): Likewise.
3794         (vreinterpret_f32_s32): Likewise.
3795         (vreinterpret_f32_s64): Likewise.
3796         (vreinterpret_f32_u8): Likewise.
3797         (vreinterpret_f32_u16): Likewise.
3798         (vreinterpret_f32_u32): Likewise.
3799         (vreinterpret_f32_u64): Likewise.
3800         (vreinterpret_f32_p8): Likewise.
3801         (vreinterpret_f32_p16): Likewise.
3802         (vreinterpretq_f32_s8): Likewise.
3803         (vreinterpretq_f32_s16): Likewise.
3804         (vreinterpretq_f32_s32): Likewise.
3805         (vreinterpretq_f32_s64): Likewise.
3806         (vreinterpretq_f32_u8): Likewise.
3807         (vreinterpretq_f32_u16): Likewise.
3808         (vreinterpretq_f32_u32): Likewise.
3809         (vreinterpretq_f32_u64): Likewise.
3810         (vreinterpretq_f32_p8): Likewise.
3811         (vreinterpretq_f32_p16): Likewise.
3812         (vreinterpret_s64_s8): Likewise.
3813         (vreinterpret_s64_s16): Likewise.
3814         (vreinterpret_s64_s32): Likewise.
3815         (vreinterpret_s64_f32): Likewise.
3816         (vreinterpret_s64_u8): Likewise.
3817         (vreinterpret_s64_u16): Likewise.
3818         (vreinterpret_s64_u32): Likewise.
3819         (vreinterpret_s64_u64): Likewise.
3820         (vreinterpret_s64_p8): Likewise.
3821         (vreinterpret_s64_p16): Likewise.
3822         (vreinterpretq_s64_s8): Likewise.
3823         (vreinterpretq_s64_s16): Likewise.
3824         (vreinterpretq_s64_s32): Likewise.
3825         (vreinterpretq_s64_f32): Likewise.
3826         (vreinterpretq_s64_u8): Likewise.
3827         (vreinterpretq_s64_u16): Likewise.
3828         (vreinterpretq_s64_u32): Likewise.
3829         (vreinterpretq_s64_u64): Likewise.
3830         (vreinterpretq_s64_p8): Likewise.
3831         (vreinterpretq_s64_p16): Likewise.
3832         (vreinterpret_u64_s8): Likewise.
3833         (vreinterpret_u64_s16): Likewise.
3834         (vreinterpret_u64_s32): Likewise.
3835         (vreinterpret_u64_s64): Likewise.
3836         (vreinterpret_u64_f32): Likewise.
3837         (vreinterpret_u64_u8): Likewise.
3838         (vreinterpret_u64_u16): Likewise.
3839         (vreinterpret_u64_u32): Likewise.
3840         (vreinterpret_u64_p8): Likewise.
3841         (vreinterpret_u64_p16): Likewise.
3842         (vreinterpretq_u64_s8): Likewise.
3843         (vreinterpretq_u64_s16): Likewise.
3844         (vreinterpretq_u64_s32): Likewise.
3845         (vreinterpretq_u64_s64): Likewise.
3846         (vreinterpretq_u64_f32): Likewise.
3847         (vreinterpretq_u64_u8): Likewise.
3848         (vreinterpretq_u64_u16): Likewise.
3849         (vreinterpretq_u64_u32): Likewise.
3850         (vreinterpretq_u64_p8): Likewise.
3851         (vreinterpretq_u64_p16): Likewise.
3852         (vreinterpret_s8_s16): Likewise.
3853         (vreinterpret_s8_s32): Likewise.
3854         (vreinterpret_s8_s64): Likewise.
3855         (vreinterpret_s8_f32): Likewise.
3856         (vreinterpret_s8_u8): Likewise.
3857         (vreinterpret_s8_u16): Likewise.
3858         (vreinterpret_s8_u32): Likewise.
3859         (vreinterpret_s8_u64): Likewise.
3860         (vreinterpret_s8_p8): Likewise.
3861         (vreinterpret_s8_p16): Likewise.
3862         (vreinterpretq_s8_s16): Likewise.
3863         (vreinterpretq_s8_s32): Likewise.
3864         (vreinterpretq_s8_s64): Likewise.
3865         (vreinterpretq_s8_f32): Likewise.
3866         (vreinterpretq_s8_u8): Likewise.
3867         (vreinterpretq_s8_u16): Likewise.
3868         (vreinterpretq_s8_u32): Likewise.
3869         (vreinterpretq_s8_u64): Likewise.
3870         (vreinterpretq_s8_p8): Likewise.
3871         (vreinterpretq_s8_p16): Likewise.
3872         (vreinterpret_s16_s8): Likewise.
3873         (vreinterpret_s16_s32): Likewise.
3874         (vreinterpret_s16_s64): Likewise.
3875         (vreinterpret_s16_f32): Likewise.
3876         (vreinterpret_s16_u8): Likewise.
3877         (vreinterpret_s16_u16): Likewise.
3878         (vreinterpret_s16_u32): Likewise.
3879         (vreinterpret_s16_u64): Likewise.
3880         (vreinterpret_s16_p8): Likewise.
3881         (vreinterpret_s16_p16): Likewise.
3882         (vreinterpretq_s16_s8): Likewise.
3883         (vreinterpretq_s16_s32): Likewise.
3884         (vreinterpretq_s16_s64): Likewise.
3885         (vreinterpretq_s16_f32): Likewise.
3886         (vreinterpretq_s16_u8): Likewise.
3887         (vreinterpretq_s16_u16): Likewise.
3888         (vreinterpretq_s16_u32): Likewise.
3889         (vreinterpretq_s16_u64): Likewise.
3890         (vreinterpretq_s16_p8): Likewise.
3891         (vreinterpretq_s16_p16): Likewise.
3892         (vreinterpret_s32_s8): Likewise.
3893         (vreinterpret_s32_s16): Likewise.
3894         (vreinterpret_s32_s64): Likewise.
3895         (vreinterpret_s32_f32): Likewise.
3896         (vreinterpret_s32_u8): Likewise.
3897         (vreinterpret_s32_u16): Likewise.
3898         (vreinterpret_s32_u32): Likewise.
3899         (vreinterpret_s32_u64): Likewise.
3900         (vreinterpret_s32_p8): Likewise.
3901         (vreinterpret_s32_p16): Likewise.
3902         (vreinterpretq_s32_s8): Likewise.
3903         (vreinterpretq_s32_s16): Likewise.
3904         (vreinterpretq_s32_s64): Likewise.
3905         (vreinterpretq_s32_f32): Likewise.
3906         (vreinterpretq_s32_u8): Likewise.
3907         (vreinterpretq_s32_u16): Likewise.
3908         (vreinterpretq_s32_u32): Likewise.
3909         (vreinterpretq_s32_u64): Likewise.
3910         (vreinterpretq_s32_p8): Likewise.
3911         (vreinterpretq_s32_p16): Likewise.
3912         (vreinterpret_u8_s8): Likewise.
3913         (vreinterpret_u8_s16): Likewise.
3914         (vreinterpret_u8_s32): Likewise.
3915         (vreinterpret_u8_s64): Likewise.
3916         (vreinterpret_u8_f32): Likewise.
3917         (vreinterpret_u8_u16): Likewise.
3918         (vreinterpret_u8_u32): Likewise.
3919         (vreinterpret_u8_u64): Likewise.
3920         (vreinterpret_u8_p8): Likewise.
3921         (vreinterpret_u8_p16): Likewise.
3922         (vreinterpretq_u8_s8): Likewise.
3923         (vreinterpretq_u8_s16): Likewise.
3924         (vreinterpretq_u8_s32): Likewise.
3925         (vreinterpretq_u8_s64): Likewise.
3926         (vreinterpretq_u8_f32): Likewise.
3927         (vreinterpretq_u8_u16): Likewise.
3928         (vreinterpretq_u8_u32): Likewise.
3929         (vreinterpretq_u8_u64): Likewise.
3930         (vreinterpretq_u8_p8): Likewise.
3931         (vreinterpretq_u8_p16): Likewise.
3932         (vreinterpret_u16_s8): Likewise.
3933         (vreinterpret_u16_s16): Likewise.
3934         (vreinterpret_u16_s32): Likewise.
3935         (vreinterpret_u16_s64): Likewise.
3936         (vreinterpret_u16_f32): Likewise.
3937         (vreinterpret_u16_u8): Likewise.
3938         (vreinterpret_u16_u32): Likewise.
3939         (vreinterpret_u16_u64): Likewise.
3940         (vreinterpret_u16_p8): Likewise.
3941         (vreinterpret_u16_p16): Likewise.
3942         (vreinterpretq_u16_s8): Likewise.
3943         (vreinterpretq_u16_s16): Likewise.
3944         (vreinterpretq_u16_s32): Likewise.
3945         (vreinterpretq_u16_s64): Likewise.
3946         (vreinterpretq_u16_f32): Likewise.
3947         (vreinterpretq_u16_u8): Likewise.
3948         (vreinterpretq_u16_u32): Likewise.
3949         (vreinterpretq_u16_u64): Likewise.
3950         (vreinterpretq_u16_p8): Likewise.
3951         (vreinterpretq_u16_p16): Likewise.
3952         (vreinterpret_u32_s8): Likewise.
3953         (vreinterpret_u32_s16): Likewise.
3954         (vreinterpret_u32_s32): Likewise.
3955         (vreinterpret_u32_s64): Likewise.
3956         (vreinterpret_u32_f32): Likewise.
3957         (vreinterpret_u32_u8): Likewise.
3958         (vreinterpret_u32_u16): Likewise.
3959         (vreinterpret_u32_u64): Likewise.
3960         (vreinterpret_u32_p8): Likewise.
3961         (vreinterpret_u32_p16): Likewise.
3962         (vreinterpretq_u32_s8): Likewise.
3963         (vreinterpretq_u32_s16): Likewise.
3964         (vreinterpretq_u32_s32): Likewise.
3965         (vreinterpretq_u32_s64): Likewise.
3966         (vreinterpretq_u32_f32): Likewise.
3967         (vreinterpretq_u32_u8): Likewise.
3968         (vreinterpretq_u32_u16): Likewise.
3969         (vreinterpretq_u32_u64): Likewise.
3970         (vreinterpretq_u32_p8): Likewise.
3971         (vreinterpretq_u32_p16): Likewise.
3972
3973 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3974
3975         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
3976         Pattern extended.
3977         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
3978         (sqabs): Likewise.
3979         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
3980         (vqnegd_s64): Likewise.
3981         (vqabs_s64): Likewise.
3982         (vqabsd_s64): Likewise.
3983
3984 2014-04-22  Richard Henderson  <rth@redhat.com>
3985
3986         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
3987         computation to the top of the loop.
3988
3989 2014-04-22  Renlin  <renlin.li@arm.com>
3990             Jiong Wang  <jiong.wang@arm.com>
3991
3992         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
3993         * config/aarch64/aarch64.c (aarch64_layout_frame)
3994         (aarch64_initial_elimination_offset): Likewise.
3995
3996 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3997
3998         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
3999         Fix indentation.
4000
4001 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
4002
4003         * machmode.h (bitwise_mode_for_mode): Declare.
4004         * stor-layout.h (bitwise_type_for_mode): Likewise.
4005         * stor-layout.c (bitwise_mode_for_mode): New function.
4006         (bitwise_type_for_mode): Likewise.
4007         * builtins.c (fold_builtin_memory_op): Use it instead of
4008         int_mode_for_mode and build_nonstandard_integer_type.
4009
4010 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4011
4012         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
4013         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
4014         (*-*-solaris2*): Simplify.
4015         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
4016         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
4017         *-*-solaris2.9* handling.
4018
4019         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
4020         as bug.
4021         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
4022         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
4023         handling, simplify.
4024         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
4025         * configure: Regenerate.
4026
4027         * config/i386/sol2-9.h: Remove.
4028
4029         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
4030         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
4031         Remove Solaris 9 references.
4032
4033 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
4034
4035         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
4036         (floatuns<GPI:mode><GPF:mode>2): Remove.
4037         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
4038         and floatuns conversions.
4039         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
4040         and floatuns conversions.
4041         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
4042         (w1,w2): New mode attributes for inequal width conversions.
4043
4044 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
4045
4046         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
4047         the output asm format.
4048
4049 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
4050
4051         * config/aarch64/aarch64-simd.md
4052         (aarch64_cm<optab>di): Always split.
4053         (*aarch64_cm<optab>di): New.
4054         (aarch64_cmtstdi): Always split.
4055         (*aarch64_cmtstdi): New.
4056
4057 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
4058
4059         PR tree-optimization/60823
4060         * omp-low.c (ipa_simd_modify_function_body): Go through
4061         all SSA_NAMEs and for those refering to vector arguments
4062         which are going to be replaced adjust SSA_NAME_VAR and,
4063         if it is a default definition, change it into a non-default
4064         definition assigned at the beginning of function from new_decl.
4065         (ipa_simd_modify_stmt_ops): Rewritten.
4066         * tree-dfa.c (set_ssa_default_def): When removing default def,
4067         check for NULL loc instead of NULL *loc.
4068
4069 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4070
4071         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
4072         restrictions on core registers for DImode values in Thumb2.
4073
4074 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
4075
4076         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
4077         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
4078
4079 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
4080
4081         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
4082         (*iordi_notzesidi_di): Likewise.
4083         (*iordi_notsesidi_di): Likewise.
4084
4085 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
4086
4087         * config/arm/arm-protos.h (tune_params): New struct members.
4088         * config/arm/arm.c: Initialise tune_params per processor.
4089         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
4090         for speed, based on new tune_params.
4091
4092 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
4093
4094         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
4095         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
4096         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
4097         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
4098         * config/aarch64/arm_neon.h (vrnd_f64): Added.
4099         (vrnda_f64): Likewise.
4100         (vrndi_f64): Likewise.
4101         (vrndm_f64): Likewise.
4102         (vrndn_f64): Likewise.
4103         (vrndp_f64): Likewise.
4104         (vrndx_f64): Likewise.
4105
4106 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4107
4108         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
4109         GET_MODE_SIZE argument is enum machine_mode.
4110
4111 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
4112
4113         PR target/60910
4114         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
4115         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
4116
4117 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
4118
4119         PR middle-end/60281
4120         * asan.c (asan_emit_stack_protection): Force the base to align to
4121         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
4122         appropriate bits if STRICT_ALIGNMENT.
4123         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
4124         when asan is on.
4125         (expand_used_vars): Leave a space in the stack frame for alignment
4126         if STRICT_ALIGNMENT.
4127
4128 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
4129
4130         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
4131         than a gimple.
4132         (gimple_store_p): Likewise.
4133         (gimple_assign_load_p): Likewise.
4134         (gimple_assign_cast_p): Likewise.
4135         (gimple_clobber_p): Likewise.
4136
4137         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
4138         rather than a gimple.
4139         (gimple_assign_cast_p): Likewise.
4140
4141 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
4142
4143         PR target/60735
4144         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
4145         If mode is DDmode and TARGET_E500_DOUBLE allow move.
4146
4147         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
4148         more debug information for E500 if -mdebug=reg.
4149
4150 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
4151
4152         PR target/60909
4153         * config/i386/i386.c (ix86_expand_builtin)
4154         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
4155         register for target RTX.
4156         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
4157
4158 2014-04-18  Cong Hou  <congh@google.com>
4159
4160         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
4161         the widen-mult pattern by handling two operands with different sizes,
4162         and operands whose size is smaller than half of the result type.
4163
4164 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
4165
4166         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
4167         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
4168         (do_estimate_edge_time): Compute it.
4169         * ipa-inline.c (want_inline_small_function_p): Bypass
4170         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
4171
4172 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
4173
4174         * ipa-inline.c (spec_rem): New static variable.
4175         (dump_overall_stats): New function.
4176         (dump_inline_stats): New function.
4177
4178 2014-04-18  Richard Henderson  <rth@redhat.com>
4179
4180         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
4181         to GET_MODE_SIZE, not a reg_class_t.
4182
4183 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4184
4185         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
4186         (vsx_xxmrglw_<mode>): Likewise.
4187
4188 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
4189
4190         PR target/60876
4191         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
4192         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
4193         (rs6000_init_hard_regno_mode_ok): Likewise.
4194
4195 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
4196
4197         * ipa-inline.c (inline_small_functions): Account only non-cold
4198         functions.
4199         * doc/invoke.texi (inline-unit-growth): Update documentation.
4200
4201 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
4202
4203         * config/rs6000/rs6000.md (addti3, subti3): New.
4204
4205 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
4206
4207         PR target/60863
4208         * config/i386/i386.c (ix86_expand_clear): Remove outdated
4209         comment.  Check optimize_insn_for_size_p instead of
4210         optimize_insn_for_speed_p.
4211
4212 2014-04-17  Martin Jambor  <mjambor@suse.cz>
4213
4214         * gimple-iterator.c (gsi_start_edge): New function.
4215         * gimple-iterator.h (gsi_start_edge): Declare.
4216         * tree-sra.c (single_non_eh_succ): New function.
4217         (disqualify_ops_if_throwing_stmt): Renamed to
4218         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
4219         having one non-EH successor BB.
4220         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
4221         generate loads into replacements.
4222         (sra_modify_assign): Likewise and and also use the simple path for
4223         such statements.
4224         (sra_modify_function_body): Commit statements on edges.
4225
4226 2014-04-17  Richard Biener  <rguenther@suse.de>
4227
4228         PR middle-end/60849
4229         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
4230         comparison results and add clarifying comment.
4231
4232 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
4233
4234         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
4235         (blank_mode): Initialize it.
4236         (emit_mode_size_inline, emit_mode_nunits_inline,
4237         emit_mode_inner_inline): New functions.
4238         (emit_insn_modes_h): Call them and surround their output with
4239         #if GCC_VERSION >= 4001 ... #endif.
4240         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
4241         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
4242         mode_* arrays if the argument is __builtin_constant_p.
4243         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
4244         is enum machine_mode.
4245
4246 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4247
4248         * passes.c (opt_pass::execute): Adjust.
4249         (pass_manager::execute_pass_mode_switching): Likewise.
4250         (early_local_passes::execute): Likewise.
4251         (execute_one_pass): Pass cfun to the pass's execute method.
4252         * tree-pass.h (opt_pass::execute): Add function * argument.
4253         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
4254         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
4255         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
4256         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4257         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
4258         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
4259         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
4260         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
4261         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
4262         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
4263         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
4264         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
4265         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
4266         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
4267         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
4268         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
4269         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
4270         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
4271         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
4272         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4273         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4274         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4275         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4276         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4277         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4278         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4279         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
4280         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
4281         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
4282         Adjust.
4283
4284 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4285
4286         * passes.c (opt_pass::gate): Take function * argument.
4287         (gate_all_early_local_passes): Merge into
4288         (early_local_passes::gate): this.
4289         (gate_all_early_optimizations): Merge into
4290         (all_early_optimizations::gate): this.
4291         (gate_all_optimizations): Mege into
4292         (all_optimizations::gate): this.
4293         (gate_all_optimizations_g): Merge into
4294         (all_optimizations_g::gate): this.
4295         (gate_rest_of_compilation): Mege into
4296         (rest_of_compilation::gate): this.
4297         (gate_postreload): Merge into
4298         (postreload::gate): this.
4299         (dump_one_pass): Pass cfun to the pass's gate method.
4300         (execute_ipa_summary_passes): Likewise.
4301         (execute_one_pass): Likewise.
4302         (ipa_write_summaries_2): Likewise.
4303         (ipa_write_optimization_summaries_1): Likewise.
4304         (ipa_read_summaries_1): Likewise.
4305         (ipa_read_optimization_summaries_1): Likewise.
4306         (execute_ipa_stmt_fixups): Likewise.
4307         * tree-pass.h (opt_pass::gate): Add function * argument.
4308         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
4309         combine-stack-adj.c, combine.c, compare-elim.c,
4310         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4311         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
4312         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
4313         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
4314         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
4315         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
4316         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
4317         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
4318         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
4319         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
4320         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
4321         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
4322         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
4323         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
4324         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4325         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4326         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4327         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4328         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4329         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4330         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4331         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4332         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
4333         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
4334         var-tracking.c, vtable-verify.c, web.c: Adjust.
4335
4336 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4337
4338         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
4339         * configure: Regenerate.
4340
4341 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4342
4343         * passes.c (dump_one_pass): don't check pass->has_gate.
4344         (execute_ipa_summary_passes): Likewise.
4345         (execute_one_pass): Likewise.
4346         (ipa_write_summaries_2): Likewise.
4347         (ipa_write_optimization_summaries_1): Likewise.
4348         (ipa_read_optimization_summaries_1): Likewise.
4349         (execute_ipa_stmt_fixups): Likewise.
4350         * tree-pass.h (pass_data::has_gate): Remove.
4351         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
4352         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
4353         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
4354         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
4355         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
4356         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
4357         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
4358         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
4359         gimple-low.c, gimple-ssa-isolate-paths.c,
4360         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
4361         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
4362         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
4363         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
4364         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
4365         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
4366         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
4367         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
4368         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
4369         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
4370         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
4371         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4372         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4373         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
4374         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
4375         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
4376         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4377         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4378         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
4379         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
4380         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
4381         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
4382         Adjust.
4383
4384 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4385
4386         * pass_manager.h (pass_manager::register_dump_files_1): Remove
4387         declaration.
4388         * passes.c (pass_manager::register_dump_files_1): Merge into
4389         (pass_manager::register_dump_files): this, and remove its handling of
4390         properties since the pass always has the properties anyway.
4391         (pass_manager::pass_manager): Adjust.
4392
4393 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
4394
4395         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
4396         * passes.c (pass_manager::register_dump_files_1): Remove dead code
4397         dealing with properties.
4398         (pass_manager::register_dump_files): Adjust.
4399
4400 2014-03-20  Mark Wielaard  <mjw@redhat.com>
4401
4402         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
4403         then represent the bound as normal constant value.
4404
4405 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
4406
4407         PR target/60847
4408         Forward port from 4.8 branch
4409         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
4410
4411         * config/i386/bmiintrin.h (_blsi_u32): New.
4412         (_blsi_u64): Ditto.
4413         (_blsr_u32): Ditto.
4414         (_blsr_u64): Ditto.
4415         (_blsmsk_u32): Ditto.
4416         (_blsmsk_u64): Ditto.
4417         (_tzcnt_u32): Ditto.
4418         (_tzcnt_u64): Ditto.
4419
4420 2014-04-17  Kito Cheng  <kito@0xlab.org>
4421
4422         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
4423
4424 2014-04-17  Richard Biener  <rguenther@suse.de>
4425
4426         PR middle-end/60849
4427         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
4428         boolean results for comparisons.
4429
4430 2014-04-17  Richard Biener  <rguenther@suse.de>
4431
4432         PR tree-optimization/60836
4433         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
4434         initial PHI args to be gimple values.
4435
4436 2014-04-17   Richard Biener  <rguenther@suse.de>
4437
4438         PR tree-optimization/60841
4439         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
4440         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
4441         of stmts to SLP build.
4442         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4443         (vect_analyze_slp): Likewise.
4444         (vect_analyze_slp_instance): Likewise.
4445         (vect_build_slp_tree): Limit overall SLP tree growth.
4446         * tree-vectorizer.h (vect_analyze_data_refs,
4447         vect_analyze_slp): Adjust prototypes.
4448
4449 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4450
4451         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
4452         Silvermont.
4453
4454 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4455
4456         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
4457         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
4458         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
4459         for TARGET_SLOW_PSHUFB
4460
4461 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
4462
4463         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
4464         * config/i386/i386.c (intel_cost): Ditto.
4465
4466 2014-04-17  Joey Ye  <joey.ye@arm.com>
4467
4468         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
4469
4470 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4471
4472         * opts.c (common_handle_option): Disable -fipa-reference coorectly
4473         with -fuse-profile.
4474
4475 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4476
4477         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
4478         (type_all_derivations_known_p): New predicate.
4479         (type_all_ctors_visible_p): New predicate.
4480         (type_possibly_instantiated_p): New predicate.
4481         (get_odr_type): Compute all_derivations_known.
4482         (dump_odr_type): Dump the flag.
4483         (maybe_record_type): Cleanup.
4484         (record_target_from_binfo): Add bases_to_consider array;
4485         record bases for types w/o instances and skip CXX destructor.
4486         (possible_polymorphic_call_targets_1): Add bases_to_consider
4487         and consider_construction parameters; check if type may have instance.
4488         (get_polymorphic_call_info): Set maybe_in_construction to true
4489         when we know nothing.
4490         (record_targets_from_bases): Skip CXX destructors; they are
4491         never called for types in construction.
4492         (possible_polymorphic_call_targets): Do not record target when
4493         type may not have instance.
4494
4495 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
4496
4497         PR ipa/60854
4498         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
4499         external aliases alive, too.
4500
4501 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
4502
4503         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
4504         definition.
4505
4506 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
4507
4508         * final.c (compute_alignments): Do not apply loop alignment to a block
4509         falling through to the exit.
4510
4511 2014-04-16  Catherine Moore  <clm@codesourcery.com>
4512
4513         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
4514         Adjust constraints for microMIPS store patterns.
4515
4516 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
4517
4518         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
4519
4520 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
4521
4522         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
4523         (append_use): Run at -O0.
4524         (append_vdef): Likewise.
4525         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
4526         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
4527
4528 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
4529
4530         PR tree-optimization/60844
4531         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
4532         (propagate_op_to_single_use, remove_visited_stmt_chain,
4533         linearize_expr, repropagate_negates, reassociate_bb): Use it
4534         instead of gsi_remove.
4535
4536 2014-04-16  Martin Jambor  <mjambor@suse.cz>
4537
4538         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
4539         ipa_transforms_to_apply.
4540         (cgraph_function_versioning): Assert that old_node has empty
4541         ipa_transforms_to_apply.
4542         * trans-mem.c (ipa_tm_create_version): Likewise.
4543         * tree-inline.c (tree_function_versioning): Do not duplicate
4544         ipa_transforms_to_apply.
4545
4546 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4547
4548         PR target/60817
4549         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
4550         x86_64-*-* cases.
4551         Pass necessary as flags on 64-bit Solaris/x86.
4552         Use lowercase relocs for x86_64-*-*.
4553         * configure: Regenerate.
4554
4555 2014-04-15  Jan Hubicka  <jh@suse.cz>
4556
4557         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
4558         (maybe_record_node, likely_target_p): Use it.
4559
4560 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4561
4562         PR target/60839
4563         Revert following patch
4564
4565         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4566
4567         PR target/60735
4568         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4569         software floating point or no floating point registers, do not
4570         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4571         in GPRs that occurs after we tested for GPRs that would never be
4572         true.
4573
4574         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4575         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4576         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4577         specifically allow DDmode, since that does not use the SPE SIMD
4578         instructions.
4579
4580 2014-03-21  Mark Wielaard  <mjw@redhat.com>
4581
4582         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
4583         as unsigned or int depending on type and value used.
4584
4585 2014-04-15  Richard Biener  <rguenther@suse.de>
4586
4587         PR rtl-optimization/56965
4588         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
4589         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
4590         ... here.
4591         * alias.c (true_dependence_1): Do not call
4592         nonoverlapping_component_refs_p.
4593         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
4594         nonoverlapping_component_refs_p.
4595         (indirect_refs_may_alias_p): Likewise.
4596
4597 2014-04-15  Teresa Johnson  <tejohnson@google.com>
4598
4599         * cfg.c (dump_bb_info): Fix flags check.
4600         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
4601
4602 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4603
4604         PR rtl-optimization/60663
4605         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
4606         avoid 0 cost.
4607
4608 2014-04-15  Richard Biener  <rguenther@suse.de>
4609
4610         * lto-streamer.h (LTO_major_version): Bump to 4.
4611
4612 2014-04-15  Richard Biener  <rguenther@suse.de>
4613
4614         * common.opt (lto_partition_model): New enum.
4615         (flto-partition=): Merge separate options with a single with argument,
4616         add -flto-partition=one support.
4617         * flag-types.h (enum lto_partition_model): Declare.
4618         * opts.c (finish_options): Remove duplicate -flto-partition=
4619         option check.
4620         * lto-wrapper.c (run_gcc): Adjust.
4621
4622 2014-04-15  Richard Biener  <rguenther@suse.de>
4623
4624         * alias.c (ncr_compar): New function.
4625         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
4626
4627 2014-04-15  Richard Biener  <rguenther@suse.de>
4628
4629         * alias.c (record_component_aliases): Do not walk BINFOs.
4630
4631 2014-04-15  Richard Biener  <rguenther@suse.de>
4632
4633         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4634         Add struct function argument and adjust.
4635         (find_func_aliases_for_call): Likewise.
4636         (find_func_aliases): Likewise.
4637         (find_func_clobbers): Likewise.
4638         (intra_create_variable_infos): Likewise.
4639         (compute_points_to_sets): Likewise.
4640         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
4641
4642 2014-04-15  Richard Biener  <rguenther@suse.de>
4643
4644         * tree.c (iterative_hash_expr): Use enum tree_code_class
4645         to store TREE_CODE_CLASS.
4646         (tree_block): Likewise.
4647         (tree_set_block): Likewise.
4648         * tree.h (fold_build_pointer_plus_loc): Use
4649         convert_to_ptrofftype_loc.
4650
4651 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
4652
4653         PR plugins/59335
4654         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
4655         added in 4.9.
4656
4657 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
4658
4659         * cfgloop.h (struct loop): Move force_vectorize down.
4660         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
4661         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
4662         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
4663         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
4664         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
4665         * tree-core.h (enum annot_expr_kind): Add new kind values.
4666         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
4667         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
4668         kinds.
4669         * tree.def (ANNOTATE_EXPR): Tweak comment.
4670
4671 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4672
4673         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
4674         cxa_pure_virtual).
4675
4676 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
4677
4678         * tree.h (TYPE_IDENTIFIER): Declare.
4679         * tree.c (subrange_type_for_debug_p): Use it.
4680         * godump.c (go_format_type): Likewise.
4681         * dwarf2out.c (is_cxx_auto, modified_type_die,
4682         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
4683         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
4684
4685 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
4686
4687         PR lto/60820
4688         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
4689
4690 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
4691
4692         * config/i386/i386.c (examine_argument): Return bool.  Return true if
4693         parameter should be passed in memory.
4694         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
4695         (construct_container): Update calls to examine_argument.
4696         (function_arg_advance_64): Ditto.
4697         (return_in_memory_32): Merge with ix86_return_in_memory.
4698         (return_in_memory_64): Ditto.
4699         (return_in_memory_ms_64): Ditto.
4700
4701 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4702
4703         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
4704         * coverage.c (coverage_compute_profile_id): Handle externally visible
4705         symbols.
4706
4707 2014-04-14  Martin Jambor  <mjambor@suse.cz>
4708
4709         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
4710         DECL_DISREGARD_INLINE_LIMITS functions.
4711
4712 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4713
4714         PR target/60827
4715         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
4716
4717 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4718
4719         PR target/60827
4720         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
4721         optimize_insn_for_speed_p instead of
4722         optimize_function_for_speed_p.
4723
4724 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
4725
4726         * doc/invoke.texi (free): Document AArch64.
4727
4728 2014-04-14  Richard Biener  <rguenther@suse.de>
4729
4730         PR tree-optimization/60042
4731         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
4732         (insert_into_preds_of_block): Do not prevent PHI insertion
4733         for REFERENCE exprs here ...
4734         (eliminate_dom_walker::before_dom_children): ... but prevent
4735         their use here under similar conditions when applied to the
4736         IL after PRE optimizations.
4737
4738 2014-04-14  Richard Biener  <rguenther@suse.de>
4739
4740         * passes.def: Move early points-to after early SRA.
4741
4742 2014-04-14  Richard Biener  <rguenther@suse.de>
4743
4744         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
4745         check for which sign-changes we allow when forwarding
4746         a converted value into a switch.
4747
4748 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4749
4750         * stor-layout.c (place_field): Finalize non-constant offset for the
4751         field, if any.
4752
4753 2014-04-14  Richard Biener  <rguenther@suse.de>
4754
4755         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
4756         as argument.
4757         (expand_switch_using_bit_tests_p): Likewise.
4758         (process_switch): Compute and pass on speed_p based on the
4759         switch stmt.
4760         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
4761         optimize_bb_for_speed_p.
4762
4763 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4764
4765         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
4766         * function.h (struct function): Rename has_force_vect_loops into
4767         has_force_vectorize_loops.
4768         * lto-streamer-in.c (input_cfg): Adjust for renaming.
4769         (input_struct_function_base): Likewise.
4770         * lto-streamer-out.c (output_cfg): Likewise.
4771         (output_struct_function_base): Likewise.
4772         * omp-low.c (expand_omp_simd): Likewise.
4773         * tree-cfg.c (move_sese_region_to_fn): Likewise.
4774         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4775         (version_loop_for_if_conversion): Likewise.
4776         (tree_if_conversion): Likewise.
4777         (main_tree_if_conversion): Likewise.
4778         (gate_tree_if_conversion): Likewise.
4779         * tree-inline.c (copy_loops): Likewise.
4780         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
4781         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
4782         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
4783         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
4784         * tree-vectorizer.c (vectorize_loops): Likewise.
4785         * tree-vectorizer.h (unlimited_cost_model): Likewise.
4786
4787 2014-04-14  Richard Biener  <rguenther@suse.de>
4788
4789         PR lto/60720
4790         * lto-streamer-out.c (wrap_refs): New function.
4791         (lto_output): Wrap symbol references in global initializes in
4792         type-preserving MEM_REFs.
4793
4794 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4795
4796         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
4797
4798 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4799
4800         * config/sh/sh.md (setmemqi): New expand pattern.
4801         * config/sh/sh.h (CLEAR_RATIO): Define.
4802         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
4803         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
4804
4805 2014-04-14  Richard Biener  <rguenther@suse.de>
4806
4807         PR middle-end/55022
4808         * fold-const.c (negate_expr_p): Don't negate directional rounding
4809         division.
4810         (fold_negate_expr): Likewise.
4811
4812 2014-04-14  Richard Biener  <rguenther@suse.de>
4813
4814         PR tree-optimization/59817
4815         PR tree-optimization/60453
4816         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
4817         recursion to catch all CHRECs in the scalar evolution and restrict
4818         the predicate for the remains appropriately.
4819
4820 2014-04-12  Catherine Moore  <clm@codesourcery.com>
4821
4822         * config/mips/constraints.md: Add new register constraint "kb".
4823         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
4824         (*movhi_internal): Likewise.
4825         (*movqi_internal): Likewise.
4826         * config/mips/mips.h (M16_STORE_REGS): New register class.
4827         (REG_CLASS_NAMES): Add M16_STORE_REGS.
4828         (REG_CLASS_CONTENTS): Likewise.
4829         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
4830
4831 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4832
4833         PR c/60194
4834         * doc/invoke.texi (-Wformat-signedness): Document it.
4835         (Wformat=2): Mention that this enables -Wformat-signedness.
4836
4837 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4838
4839         * common/config/epiphany/epiphany-common.c
4840         (epiphany_option_optimization_table): Enable section anchors by
4841         default at -O1 or higher.
4842         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
4843         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
4844         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
4845         carries no extra cost.
4846         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
4847         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
4848         * config/epiphany/predicates.md (memclob_operand): New predicate.
4849         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
4850         Use memclob_operand predicate and X constraint for operand 3.
4851
4852 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4853
4854         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
4855         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
4856         its operands.
4857
4858 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4859
4860         PR rtl-optimization/60651
4861         * mode-switching.c (optimize_mode_switching): Make sure to emit
4862         sets of a lower numbered entity before sets of a higher numbered
4863         entity to a mode of the same or lower priority.
4864         When creating a seginfo for a basic block that starts with a code
4865         label, move the insertion point past the code label.
4866         (new_seginfo): Document and enforce requirement that
4867         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
4868         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
4869         * doc/tm.texi: Regenerate.
4870
4871 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4872
4873         PR target/60811
4874         * config/arc/arc.c (arc_save_restore): Fix assert typo.
4875
4876 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
4877
4878         * BASE-VER: Set to 4.10.0.
4879
4880 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4881
4882         PR other/59055
4883         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
4884         * doc/gcc.texi (Service): Update description in the @menu
4885         * doc/invoke.texi (Option Summary): Remove misplaced and
4886         duplicated @menu.
4887
4888 2014-04-11  Steve Ellcey  <sellcey@mips.com>
4889             Jakub Jelinek  <jakub@redhat.com>
4890
4891         PR middle-end/60556
4892         * expr.c (convert_move): Use emit_store_flag_force instead of
4893         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
4894         argument to it.
4895
4896 2014-04-11  Richard Biener  <rguenther@suse.de>
4897
4898         PR middle-end/60797
4899         * varasm.c (assemble_alias): Avoid endless error reporting
4900         recursion by setting TREE_ASM_WRITTEN.
4901
4902 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4903
4904         * config/s390/s390.md: Add a splitter for NOT rtx.
4905
4906 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
4907
4908         PR rtl-optimization/60663
4909         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
4910
4911 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
4912             Jakub Jelinek  <jakub@redhat.com>
4913
4914         PR lto/60567
4915         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
4916         flag from decl_node to node.
4917
4918 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4919
4920         PR debug/60655
4921         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
4922         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
4923         ameliorating the cases where it can be.
4924
4925 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
4926
4927         Revert
4928         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4929
4930         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4931         (loadsync_<mode>): Change mode.
4932         (load_quadpti, store_quadpti): New.
4933         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4934         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4935         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
4936
4937 2014-04-09  Cong Hou  <congh@google.com>
4938
4939         PR testsuite/60773
4940         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
4941         documentation.
4942
4943 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4944
4945         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
4946         instead of vnor to exploit possible fusion opportunity in the
4947         future.
4948         (altivec_expand_vec_perm_const_le): Likewise.
4949
4950 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4951
4952         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4953         (loadsync_<mode>): Change mode.
4954         (load_quadpti, store_quadpti): New.
4955         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4956         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4957
4958 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
4959
4960         PR target/60763
4961         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
4962         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
4963         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
4964
4965 2014-04-08  Richard Biener  <rguenther@suse.de>
4966
4967         PR middle-end/60706
4968         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
4969         a 64bit widest int print double-int similar to on HWI64 hosts.
4970
4971 2014-04-08  Richard Biener  <rguenther@suse.de>
4972
4973         PR tree-optimization/60785
4974         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
4975         default defs properly.
4976
4977 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
4978
4979         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
4980         (Weffc++): Likewise.
4981
4982 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
4983
4984         * ipa-devirt.c (maybe_record_node): When node is not recorded,
4985         set completep to false rather than true.
4986
4987 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
4988
4989         PR target/60504
4990         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
4991         ARM_TARGET2_DWARF_FORMAT.
4992
4993 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
4994
4995         PR target/60609
4996         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
4997         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
4998         ADDR_DIFF_VEC.
4999
5000 2014-04-07  Richard Biener  <rguenther@suse.de>
5001
5002         PR tree-optimization/60766
5003         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
5004         (may_eliminate_iv): Convert cand_value_at result to desired type.
5005
5006 2014-04-07  Jason Merrill  <jason@redhat.com>
5007
5008         PR c++/60731
5009         * common.opt (-fno-gnu-unique): Add.
5010         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
5011
5012 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5013
5014         * haifa-sched.c: Fix outdated function reference and minor
5015         grammar errors in introductory comment.
5016
5017 2014-04-07  Richard Biener  <rguenther@suse.de>
5018
5019         PR middle-end/60750
5020         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
5021         for noreturn calls.
5022         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
5023
5024 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
5025
5026         PR debug/55794
5027         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
5028         size accounting for thunks.
5029         (pa_asm_output_mi_thunk): Use final_start_function() and
5030         final_end_function() to output function start and end directives.
5031
5032 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
5033
5034         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
5035         device specific ISA/ feature information. Remove short_sp and
5036         errata_skip ds.  Add avr_device_specific_features enum to have device
5037         specific info.
5038         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
5039         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
5040         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
5041         updated device specific info.
5042         * config/avr/avr-mcus.def: Merge device specific details to
5043         dev_attribute field.
5044         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
5045         errata_skip.
5046         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
5047         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
5048         assembler if RMW isa supported by current device.
5049         * config/avr/genmultilib.awk: Update as device info structure changed.
5050         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
5051
5052 2014-04-04  Cong Hou  <congh@google.com>
5053
5054         PR tree-optimization/60656
5055         * tree-vect-stmts.c (supportable_widening_operation):
5056         Fix a bug that elements in a vector with vect_used_by_reduction
5057         property are incorrectly reordered when the operation on it is not
5058         consistant with the one in reduction operation.
5059
5060 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
5061
5062         PR rtl-optimization/60155
5063         * gcse.c (record_set_data): New function.
5064         (single_set_gcse): New function.
5065         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
5066         (hoist_code): Likewise.
5067         (get_pressure_class_and_nregs): Likewise.
5068
5069 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
5070
5071         * explow.c (probe_stack_range): Emit a final optimization blockage.
5072
5073 2014-04-04  Anthony Green  <green@moxielogic.com>
5074
5075         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
5076         typos.
5077
5078 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
5079
5080         PR ipa/59626
5081         * lto-cgraph.c (input_overwrite_node): Check that partitioning
5082         flags are set only during streaming.
5083         * ipa.c (process_references, walk_polymorphic_call_targets,
5084         symtab_remove_unreachable_nodes): Drop bodies of always inline
5085         after early inlining.
5086         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
5087
5088 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
5089         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5090
5091         PR debug/60655
5092         * dwarf2out.c (const_ok_for_output_1): Reject expressions
5093         containing a NOT.
5094
5095 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5096
5097         PR bootstrap/60743
5098         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
5099         duration.
5100         (cortex_a53_fdivd): Likewise.
5101
5102 2014-04-04  Martin Jambor  <mjambor@suse.cz>
5103
5104         PR ipa/60640
5105         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
5106         Adjust all callers.
5107         * cgraph.c (clone_of_p): Also return true if thunks match.
5108         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
5109         cgraph_function_or_thunk_node and an obsolete comment.
5110         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
5111         file.
5112         (build_function_decl_skip_args): Likewise.
5113         (set_new_clone_decl_and_node_flags): New function.
5114         (duplicate_thunk_for_node): Likewise.
5115         (redirect_edge_duplicating_thunks): Likewise.
5116         (cgraph_clone_node): New parameter args_to_skip, pass it to
5117         redirect_edge_duplicating_thunks which is called instead of
5118         cgraph_redirect_edge_callee.
5119         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
5120         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
5121
5122 2014-04-04  Jeff Law  <law@redhat.com>
5123
5124         PR target/60657
5125         * config/arm/predicates.md (const_int_I_operand): New predicate.
5126         (const_int_M_operand): Similarly.
5127         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
5128         const_int_operand.
5129         (insv_t2, extv_reg, extzv_t2): Likewise.
5130         (load_multiple_with_writeback): Similarly for const_int_I_operand.
5131         (pop_multiple_with_writeback_and_return): Likewise.
5132         (vfp_pop_multiple_with_writeback): Likewise
5133
5134 2014-04-04  Richard Biener  <rguenther@suse.de>
5135
5136         PR ipa/60746
5137         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
5138         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
5139         non-GIMPLE_LABELs.
5140         * gimplify.h (gimple_add_tmp_var_fn): Declare.
5141         * gimplify.c (gimple_add_tmp_var_fn): New function.
5142         * gimple-expr.h (create_tmp_reg_fn): Declare.
5143         * gimple-expr.c (create_tmp_reg_fn): New function.
5144         * gimple-low.c (record_vars_into): Don't change cfun.
5145         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
5146         code generation without cfun.
5147
5148 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
5149
5150         PR bootstrap/60719
5151         * Makefile.in (install-driver): Fix shell scripting.
5152
5153 2014-04-03  Cong Hou  <congh@google.com>
5154
5155         PR tree-optimization/60505
5156         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
5157         threshold of number of iterations below which no vectorization
5158         will be done.
5159         * tree-vect-loop.c (new_loop_vec_info):
5160         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
5161         * tree-vect-loop.c (vect_analyze_loop_operations):
5162         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
5163         * tree-vect-loop.c (vect_transform_loop):
5164         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
5165         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
5166         of iterations of the loop and see if we should build the epilogue.
5167
5168 2014-04-03  Richard Biener  <rguenther@suse.de>
5169
5170         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
5171         (streamer_tree_cache_create): Adjust.
5172         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
5173         to allow optional nodes array.
5174         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
5175         (streamer_tree_cache_append): Likewise.
5176         (streamer_tree_cache_create): Create nodes array optionally
5177         as specified by parameter.
5178         * lto-streamer-out.c (create_output_block): Avoid maintaining
5179         the node array in the writer cache.
5180         (DFS_write_tree): Remove assertion.
5181         (produce_asm_for_decls): Free the out decl state hash table early.
5182         * lto-streamer-in.c (lto_data_in_create): Adjust for
5183         streamer_tree_cache_create prototype change.
5184
5185 2014-04-03  Richard Biener  <rguenther@suse.de>
5186
5187         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
5188         set TREE_CHAIN to NULL_TREE.
5189
5190 2014-04-03  Richard Biener  <rguenther@suse.de>
5191
5192         PR tree-optimization/60740
5193         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
5194         over all GIMPLE_COND operands.
5195
5196 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
5197
5198         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
5199         (Weffc++): Remove Scott's numbering, merge lists and reference
5200         Wnon-virtual-dtor.
5201
5202 2014-04-03  Nick Clifton  <nickc@redhat.com>
5203
5204         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
5205         properly.
5206
5207 2014-04-03  Martin Jambor  <mjambor@suse.cz>
5208
5209         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
5210         mention gcc_unreachable before failing.
5211         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
5212         removed symbols.
5213
5214 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
5215
5216         PR ipa/60659
5217         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
5218         inconsistent code and instead mark the context inconsistent.
5219         (possible_polymorphic_call_targets): For inconsistent contexts
5220         return empty complete list.
5221
5222 2014-04-02  Anthony Green  <green@moxielogic.com>
5223
5224         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
5225         (extendqisi2, extendhisi2): Define.
5226         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
5227         (WCHAR_TYPE): Change to unsigned int.
5228
5229 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5230
5231         PR tree-optimization/60733
5232         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
5233         insertion point for PHI candidates to be the end of the feeding
5234         block for the PHI argument.
5235
5236 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
5237
5238         PR rtl-optimization/60650
5239         * lra-constraints.c (process_alt_operands): Decrease reject for
5240         earlyclobber matching.
5241
5242 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5243
5244         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
5245
5246 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5247
5248         * config/spu/spu.c (pad_bb): Do not crash when the last
5249         insn is CODE_FOR_blockage.
5250
5251 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5252
5253         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
5254         lies outside the target mode.
5255
5256 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5257
5258         PR target/60735
5259         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
5260         software floating point or no floating point registers, do not
5261         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
5262         in GPRs that occurs after we tested for GPRs that would never be
5263         true.
5264
5265         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
5266         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
5267         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
5268         specifically allow DDmode, since that does not use the SPE SIMD
5269         instructions.
5270
5271 2014-04-02  Richard Biener  <rguenther@suse.de>
5272
5273         PR middle-end/60729
5274         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
5275         MODE_INTs.  Properly use negv_optab.
5276         (expand_abs): Likewise.
5277
5278 2014-04-02  Richard Biener  <rguenther@suse.de>
5279
5280         PR bootstrap/60719
5281         * Makefile.in (install-driver): Guard extra installs with special
5282         names properly.
5283
5284 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
5285
5286         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5287         Document vec_vgbbd.
5288
5289 2014-04-01  Richard Henderson  <rth@redhat.com>
5290
5291         PR target/60704
5292         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
5293         alternative enabled before register allocation.
5294
5295 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
5296
5297         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
5298         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
5299         typo.
5300         (nios2_large_got_address): Remove unneeded 'sym' parameter.
5301         (nios2_got_address): Update nios2_large_got_address call site.
5302         (nios2_delegitimize_address): New function.
5303         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
5304         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
5305         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
5306
5307 2014-04-01  Martin Husemann  <martin@duskware.de>
5308
5309         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
5310         for -mabi=32.
5311
5312 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
5313
5314         PR rtl-optimization/60604
5315         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
5316         check from register_operand.
5317         (register_operand): Redefine in terms of general_operand.
5318         (nonmemory_operand): Use register_operand for the non-constant cases.
5319
5320 2014-04-01  Richard Biener  <rguenther@suse.de>
5321
5322         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
5323
5324 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5325
5326         * doc/invoke.texi (mapp-regs): Clarify.
5327
5328 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
5329
5330         * config/i386/avx512fintrin.h (__v32hi): Define type.
5331         (__v64qi): Likewise.
5332         (_mm512_set1_epi8): Define.
5333         (_mm512_set1_epi16): Define.
5334         (_mm512_set4_epi32): Define.
5335         (_mm512_set4_epi64): Define.
5336         (_mm512_set4_pd): Define.
5337         (_mm512_set4_ps): Define.
5338         (_mm512_setr4_epi64): Define.
5339         (_mm512_setr4_epi32): Define.
5340         (_mm512_setr4_pd): Define.
5341         (_mm512_setr4_ps): Define.
5342         (_mm512_setzero_epi32): Define.
5343
5344 2014-03-31  Martin Jambor  <mjambor@suse.cz>
5345
5346         PR middle-end/60647
5347         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
5348         callsite_arguments_match_p.  Updated all callers.  Also check types of
5349         corresponding formal parameters and actual arguments.
5350         (not_all_callers_have_enough_arguments_p) Renamed to
5351         some_callers_have_mismatched_arguments_p.
5352
5353 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
5354
5355         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
5356
5357 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
5358
5359         PR target/60034
5360         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
5361         section anchor.
5362
5363 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
5364
5365         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
5366         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
5367         Split out
5368         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
5369         Use FMAMODE_NOVF512 mode iterator.
5370         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
5371         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
5372         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
5373         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
5374         Split out
5375         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
5376         Use VF_128_256 mode iterator.
5377         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
5378         Ditto.
5379
5380 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5381
5382         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
5383         static chain if needed.
5384
5385 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
5386
5387         PR target/60697
5388         * lra-constraints.c (index_part_to_reg): New.
5389         (process_address): Use it.
5390
5391 2014-03-27  Jeff Law  <law@redhat.com>
5392             Jakub Jelinek  <jakub@redhat.com>
5393
5394         PR target/60648
5395         * expr.c (do_tablejump): Use simplify_gen_binary rather than
5396         gen_rtx_{PLUS,MULT} to build up the address expression.
5397
5398         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
5399         creating non-canonical RTL.
5400
5401 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5402
5403         PR ipa/60243
5404         * ipa-inline.c (want_inline_small_function_p): Short circuit large
5405         functions; reorganize to make cheap checks first.
5406         (inline_small_functions): Do not estimate growth when dumping;
5407         it is expensive.
5408         * ipa-inline.h (inline_summary): Add min_size.
5409         (growth_likely_positive): New function.
5410         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
5411         (set_cond_stmt_execution_predicate): Cleanup.
5412         (estimate_edge_size_and_time): Compute min_size.
5413         (estimate_calls_size_and_time): Likewise.
5414         (estimate_node_size_and_time): Likewise.
5415         (inline_update_overall_summary): Update min_size.
5416         (do_estimate_edge_time): Likewise.
5417         (do_estimate_edge_size): Update.
5418         (do_estimate_edge_hints): Update.
5419         (growth_likely_positive): New function.
5420
5421 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
5422
5423         PR target/60693
5424         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
5425         also if addr has VOIDmode.
5426
5427 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5428
5429         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
5430         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
5431         Declare extern.
5432         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
5433         instructions as well as AdvancedSIMD loads.
5434
5435 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5436
5437         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
5438         Use crypto_aese type.
5439         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
5440         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
5441         crypto_aese, crypto_aesmc.  Move to types.md.
5442         * config/arm/types.md (crypto_aes): Split into crypto_aese,
5443         crypto_aesmc.
5444         * config/arm/iterators.md (crypto_type): Likewise.
5445
5446 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
5447
5448         * cgraph.c: Include expr.h and tree-dfa.h.
5449         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
5450         remove LHS.
5451
5452 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
5453
5454         PR target/60675
5455         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
5456         regs from checking multi-reg pseudos.
5457
5458 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5459
5460         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
5461
5462 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5463
5464         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
5465         if it would clobber the stack pointer, even temporarily.
5466
5467 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
5468
5469         * mode-switching.c: Make small adjustments to the top comment.
5470
5471 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
5472
5473         * config/rs6000/constraints.md (wD constraint): New constraint to
5474         match the constant integer to get the top DImode/DFmode out of a
5475         vector in a VSX register.
5476
5477         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
5478         match the constant integer to get the top DImode/DFmode out of a
5479         vector in a VSX register.
5480
5481         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
5482         for ISA 2.07.
5483
5484         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5485         vbpermq builtins.
5486
5487         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
5488         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
5489
5490         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
5491         Optimize vec_extract of 64-bit values, where the value being
5492         extracted is in the top word, where we can use scalar
5493         instructions.  Add direct move and store support.  Combine the big
5494         endian/little endian vector select load support into a single insn.
5495         (vsx_extract_<mode>_internal1): Likewise.
5496         (vsx_extract_<mode>_internal2): Likewise.
5497         (vsx_extract_<mode>_load): Likewise.
5498         (vsx_extract_<mode>_store): Likewise.
5499         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
5500         combined into vsx_extract_<mode>_load.
5501         (vsx_extract_<mode>_one_le): Likewise.
5502
5503         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
5504         define the top 64-bit vector element.
5505
5506         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
5507         constraint.
5508
5509         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5510         Document vec_vbpermq builtin.
5511
5512         PR target/60672
5513         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
5514         enable use of xxsldwi and xxpermdi builtin functions.
5515         (vec_xxpermdi): Likewise.
5516
5517         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5518         Document use of vec_xxsldwi and vec_xxpermdi builtins.
5519
5520 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
5521
5522         PR rtl-optimization/60650
5523         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
5524         first_p.  Use it.
5525         (find_spills_for): New.
5526         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
5527         Spill all pseudos on the second iteration.
5528
5529 2014-03-27  Marek Polacek  <polacek@redhat.com>
5530
5531         PR c/50347
5532         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
5533         types.
5534
5535 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5536
5537         * config/s390/s390.c (s390_can_use_return_insn): Check for
5538         call-saved FPRs on 31 bit.
5539
5540 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
5541
5542         PR middle-end/60682
5543         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
5544         if they need regimplification, just drop them instead of
5545         calling gimple_regimplify_operands on them.
5546
5547 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5548
5549         PR target/60580
5550         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
5551         (aarch64_frame_pointer_required): Adjust logic.
5552         (aarch64_can_eliminate): Adjust logic.
5553         (aarch64_override_options_after_change): Adjust logic.
5554
5555 2014-03-27  Dehao Chen  <dehao@google.com>
5556
5557         * ipa-inline.c (early_inliner): Update node's inline info.
5558
5559 2014-03-26  Dehao Chen  <dehao@google.com>
5560
5561         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
5562         compiler inserted conditional jumps for NAN float check.
5563
5564 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5565
5566         * ubsan.h (ubsan_create_data): Change second argument's type
5567         to const location_t *.
5568         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
5569         _("<unknown>").
5570         (ubsan_create_data): Change second argument to const location_t *PLOC.
5571         Create Loc field whenever PLOC is non-NULL.
5572         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
5573         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
5574         callers.
5575
5576         PR other/59545
5577         * real.c (real_to_integer2): Change type of low to UHWI.
5578
5579 2014-03-26  Tobias Burnus  <burnus@net-b.de>
5580
5581         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
5582         (CILK_SELF_SPECS): New define.
5583         (driver_self_specs): Use it.
5584
5585 2014-03-26  Richard Biener  <rguenther@suse.de>
5586
5587         * tree-pretty-print.c (percent_K_format): Implement special
5588         case for LTO and its stripped down BLOCK tree.
5589
5590 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5591
5592         PR sanitizer/60636
5593         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
5594
5595         * tree-vrp.c (simplify_internal_call_using_ranges): If only
5596         one range is range_int_cst_p, but not both, at least optimize
5597         addition/subtraction of 0 and multiplication by 0 or 1.
5598         * gimple-fold.c (gimple_fold_call): Fold
5599         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
5600         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
5601         INTEGER_CSTs, try to fold at least x * 0 and y - y.
5602
5603 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
5604
5605         PR rtl-optimization/60452
5606         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
5607         <case REG>: Return 1 for invalid offsets from the frame pointer.
5608
5609 2014-03-26  Marek Polacek  <polacek@redhat.com>
5610
5611         PR c/37428
5612         * doc/extend.texi (C Extensions): Mention variable-length arrays in
5613         a structure/union.
5614
5615 2014-03-26  Marek Polacek  <polacek@redhat.com>
5616
5617         PR c/39525
5618         * doc/extend.texi (Designated Inits): Describe what happens to omitted
5619         field members.
5620
5621 2014-03-26  Marek Polacek  <polacek@redhat.com>
5622
5623         PR other/59545
5624         * ira-color.c (update_conflict_hard_regno_costs): Perform the
5625         multiplication in unsigned type.
5626
5627 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5628
5629         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
5630
5631 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5632
5633         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
5634
5635 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5636
5637         PR ipa/60315
5638         * cif-code.def (UNREACHABLE) New code.
5639         * ipa-inline.c (inline_small_functions): Skip edges to
5640         __builtlin_unreachable.
5641         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
5642         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
5643         predicate to __bulitin_unreachable.
5644         (set_cond_stmt_execution_predicate): Fix issue when
5645         invert_tree_comparison returns ERROR_MARK.
5646         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
5647         propagate to inline clones.
5648         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
5649         to unreachable.
5650         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
5651         * cgraphclones.c (cgraph_clone_node): If call destination is already
5652         ureachable, do not redirect it back.
5653         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
5654         unreachable.
5655
5656 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5657
5658         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
5659         Do not modify inline clones.
5660
5661 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5662
5663         * config/i386/i386.md (general_sext_operand): New mode attr.
5664         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
5665         don't generate (sign_extend (const_int)).
5666         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
5667         operands[2].  Use We constraint instead of <i> and
5668         <general_sext_operand> predicate instead of <general_operand>.
5669         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
5670         * config/i386/constraints.md (We): New constraint.
5671         * config/i386/predicates.md (x86_64_sext_operand,
5672         sext_operand): New predicates.
5673
5674 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5675
5676         PR ipa/60600
5677         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
5678         inconsistent devirtualizations to __builtin_unreachable.
5679
5680 2014-03-25  Marek Polacek  <polacek@redhat.com>
5681
5682         PR c/35449
5683         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
5684
5685 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
5686
5687         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
5688         order of elements for big-endian.
5689
5690 2014-03-25  Richard Biener  <rguenther@suse.de>
5691
5692         PR middle-end/60635
5693         * gimplify-me.c (gimple_regimplify_operands): Update the
5694         re-gimplifed stmt.
5695
5696 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5697
5698         PR ipa/59176
5699         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
5700         (lto_output_varpool_node): Likewise.
5701         (input_overwrite_node): Likewise.
5702         (input_varpool_node): Likewise.
5703
5704 2014-03-25  Richard Biener  <rguenther@suse.de>
5705
5706         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
5707         (run_gcc): Likewise.
5708
5709 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5710
5711         * combine.c (simplify_compare_const): Add MODE argument.
5712         Handle mode_width 0 as very large mode_width.
5713         (try_combine, simplify_comparison): Adjust callers.
5714
5715         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
5716         type to avoid signed integer overflow.
5717         * explow.c (plus_constant): Likewise.
5718
5719 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5720
5721         * doc/generic.texi: Correct typos.
5722
5723 2014-03-24  Tobias Burnus  <burnus@net-b.de>
5724
5725         * doc/invoke.texi (-flto): Expand section about
5726         using static libraries with LTO.
5727
5728 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5729
5730         PR rtl-optimization/60501
5731         * optabs.def (addptr3_optab): New optab.
5732         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
5733         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
5734         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
5735
5736         * lra.c (emit_add3_insn): Use the addptr pattern if available.
5737
5738         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
5739
5740 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
5741
5742         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
5743         _mm512_set1_pd.
5744
5745         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
5746         (_mm256_undefined_ps): Define.
5747         (_mm256_undefined_pd): Define.
5748         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
5749         (_mm_undefined_pd): Define.
5750         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
5751         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
5752         (_mm512_undefined_ps): Define.
5753         (_mm512_undefined_pd): Define.
5754         Use _mm*_undefined_*.
5755         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
5756
5757 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
5758
5759         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
5760         (lshr_simd): DI mode added.
5761         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
5762         (aarch64_ushr_simddi): Likewise.
5763         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
5764         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
5765         (vshrd_n_u64): Likewise.
5766
5767 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5768
5769         * Makefile.in (s-macro_list): Depend on cc1.
5770
5771 2014-03-23  Teresa Johnson  <tejohnson@google.com>
5772
5773         * ipa-utils.c (ipa_print_order): Use specified dump file.
5774
5775 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
5776
5777         PR rtl-optimization/60601
5778         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
5779
5780         * gcc.c (eval_spec_function): Initialize save_growing_value.
5781
5782 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
5783
5784         PR sanitizer/60613
5785         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
5786         code == MINUS_EXPR, never swap op0 with op1.
5787
5788         * toplev.c (init_local_tick): Avoid signed integer multiplication
5789         overflow.
5790         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
5791         shift by first operand's bitsize.
5792
5793 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
5794
5795         PR target/60610
5796         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
5797         redefine to 1 or 0.
5798         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
5799         TARGET_ISA_64BIT_P(x).
5800
5801 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5802
5803         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
5804         pattern for vector nor instead of subtract from splat(-1).
5805         (altivec_expand_vec_perm_const_le): Likewise.
5806
5807 2014-03-21  Richard Henderson  <rth@twiddle.net>
5808
5809         PR target/60598
5810         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
5811         related insns after epilogue_completed.
5812
5813 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5814
5815         PR ipa/59176
5816         * cgraph.h (symtab_node): New flag body_removed.
5817         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
5818         when removing bodies.
5819         * symtab.c (dump_symtab_base): Dump body_removed flag.
5820         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
5821         had their bodies removed.
5822
5823 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5824
5825         PR ipa/60419
5826         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
5827         in the border.
5828
5829 2014-03-21  Richard Biener  <rguenther@suse.de>
5830
5831         PR tree-optimization/60577
5832         * tree-core.h (struct tree_base): Document nothrow_flag use
5833         in VAR_DECL_NONALIASED.
5834         * tree.h (VAR_DECL_NONALIASED): New.
5835         (may_be_aliased): Adjust.
5836         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
5837
5838 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5839
5840         * expr.c (expand_expr_real_1): Remove outdated comment.
5841
5842 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
5843
5844         PR middle-end/60597
5845         * ira.c (adjust_cleared_regs): Call copy_rtx on
5846         *reg_equiv[REGNO (loc)].src_p before passing it to
5847         simplify_replace_fn_rtx.
5848
5849         PR target/60568
5850         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
5851         into CONST, put pic register as first operand of PLUS.  Use
5852         gen_const_mem for both 32-bit and 64-bit PIC got loads.
5853
5854 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5855
5856         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
5857
5858 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5859
5860         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
5861         around for store forwarding issue in the FPU on the UT699.
5862         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
5863         loads and operations if -mfix-ut699 is specified.
5864         (divtf3_hq): Tweak attribute.
5865         (sqrttf2_hq): Likewise.
5866
5867 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5868
5869         * calls.c (store_one_arg): Remove incorrect const qualification on the
5870         type of the temporary.
5871         * cfgexpand.c (expand_return): Likewise.
5872         * expr.c (expand_constructor): Likewise.
5873         (expand_expr_real_1): Likewise.
5874
5875 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5876
5877         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
5878         of parts.
5879
5880 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
5881
5882         PR target/60039
5883         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
5884
5885 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
5886
5887         * config/arm/aarch-common-protos.h
5888         (alu_cost_table): Fix spelling of "extend".
5889         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
5890
5891 2014-03-19  Richard Biener  <rguenther@suse.de>
5892
5893         PR middle-end/60553
5894         * tree-core.h (tree_type_common): Re-order pointer members
5895         to reduce recursion depth during GC walks.
5896
5897 2014-03-19  Marek Polacek  <polacek@redhat.com>
5898
5899         PR sanitizer/60569
5900         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
5901         before accessing it.
5902
5903 2014-03-19  Richard Biener  <rguenther@suse.de>
5904
5905         PR lto/59543
5906         * lto-streamer-in.c (input_function): In WPA stage do not drop
5907         debug stmts.
5908
5909 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
5910
5911         PR tree-optimization/60559
5912         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
5913         with build_zero_cst assignment.
5914
5915 2014-03-18  Kai Tietz  <ktietz@redhat.com>
5916
5917         PR rtl-optimization/56356
5918         * sdbout.c (sdbout_parms): Verify that parms'
5919         incoming argument is valid.
5920         (sdbout_reg_parms): Likewise.
5921
5922 2014-03-18  Richard Henderson  <rth@redhat.com>
5923
5924         PR target/60562
5925         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
5926         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
5927         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
5928
5929 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
5930
5931         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
5932         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
5933         Italicize plugin event names in description.  Explain that
5934         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
5935         Remind that no GCC functions should be called after PLUGIN_FINISH.
5936         Explain what pragmas with expansion are.
5937
5938 2014-03-18  Martin Liska  <mliska@suse.cz>
5939
5940         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
5941         gimple call statement is update.
5942         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
5943         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
5944
5945 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5946
5947         PR sanitizer/60557
5948         * ubsan.c (ubsan_instrument_unreachable): Call
5949         initialize_sanitizer_builtins.
5950         (ubsan_pass): Likewise.
5951
5952         PR sanitizer/60535
5953         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
5954         varpool_finalize_decl instead of rest_of_decl_compilation.
5955
5956 2014-03-18  Richard Biener  <rguenther@suse.de>
5957
5958         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
5959         by using bitmap_and_compl instead of bitmap_and_compl_into.
5960         (df_rd_transfer_function): Likewise.
5961
5962 2014-03-18  Richard Biener  <rguenther@suse.de>
5963
5964         * doc/lto.texi (fresolution): Fix typo.
5965
5966 2014-03-18  Richard Biener  <rguenther@suse.de>
5967
5968         * doc/invoke.texi (flto): Update for changes in 4.9.
5969
5970 2014-03-18  Richard Biener  <rguenther@suse.de>
5971
5972         * doc/loop.texi: Remove section on the removed lambda framework.
5973         Update loop docs with recent changes in preserving loop structure.
5974
5975 2014-03-18  Richard Biener  <rguenther@suse.de>
5976
5977         * doc/lto.texi (-fresolution): Document.
5978
5979 2014-03-18  Richard Biener  <rguenther@suse.de>
5980
5981         * doc/contrib.texi: Adjust my name.
5982
5983 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5984
5985         PR ipa/58721
5986         * internal-fn.c: Include diagnostic-core.h.
5987         (expand_BUILTIN_EXPECT): New function.
5988         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
5989         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
5990         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
5991         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
5992         IFN_BUILTIN_EXPECT.
5993         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
5994         Revert 3 argument __builtin_expect code.
5995         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
5996         * gimple-fold.c (gimple_fold_call): Likewise.
5997         * tree.h (fold_builtin_expect): New prototype.
5998         * builtins.c (build_builtin_expect_predicate): Add predictor
5999         argument, if non-NULL, create 3 argument __builtin_expect.
6000         (fold_builtin_expect): No longer static.  Add ARG2 argument,
6001         pass it through to build_builtin_expect_predicate.
6002         (fold_builtin_2): Adjust caller.
6003         (fold_builtin_3): Handle BUILT_IN_EXPECT.
6004         * internal-fn.def (BUILTIN_EXPECT): New.
6005
6006 2014-03-18  Tobias Burnus  <burnus@net-b.de>
6007
6008         PR ipa/58721
6009         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
6010         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
6011         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
6012
6013 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
6014
6015         PR ipa/58721
6016         * predict.c (combine_predictions_for_bb): Fix up formatting.
6017         (expr_expected_value_1, expr_expected_value): Add predictor argument,
6018         fill what it points to if non-NULL.
6019         (tree_predict_by_opcode): Adjust caller, use the predictor.
6020         * predict.def (PRED_COMPARE_AND_SWAP): Add.
6021
6022 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
6023
6024         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
6025         proper constant for the store mode.
6026
6027 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
6028
6029         * symtab.c (change_decl_assembler_name): Fix transparent alias
6030         chain construction.
6031
6032 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
6033
6034         * config/aarch64/aarch64.c: Correct the comments about the
6035         aarch64 stack layout.
6036
6037 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
6038
6039         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
6040         check for GF_OMP_FOR_KIND_FOR.
6041
6042 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
6043
6044         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
6045         ymm and zmm register names.
6046
6047 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
6048
6049         PR target/60516
6050         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
6051         note creation for the 2010-08-31 changes.
6052
6053 2014-03-17  Marek Polacek  <polacek@redhat.com>
6054
6055         PR middle-end/60534
6056         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
6057         as -fno-tree-loop-vectorize.
6058         (expand_omp_simd): Likewise.
6059
6060 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
6061
6062         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
6063         (eligible_for_call_delay): New prototype.
6064         * config/sparc/sparc.c (tls_call_delay): Rename into...
6065         (eligible_for_call_delay): ...this.  Return false if the instruction
6066         cannot be put in the delay slot of a branch.
6067         (eligible_for_restore_insn): Simplify.
6068         (eligible_for_return_delay): Return false if the instruction cannot be
6069         put in the delay slot of a branch and simplify.
6070         (eligible_for_sibcall_delay): Return false if the instruction cannot be
6071         put in the delay slot of a branch.
6072         * config/sparc/sparc.md (fix_ut699): New attribute.
6073         (tls_call_delay): Delete.
6074         (in_call_delay): Reimplement.
6075         (eligible_for_sibcall_delay): Rename into...
6076         (in_sibcall_delay): ...this.
6077         (eligible_for_return_delay): Rename into...
6078         (in_return_delay): ...this.
6079         (in_branch_delay): Reimplement.
6080         (in_uncond_branch_delay): Delete.
6081         (in_annul_branch_delay): Delete.
6082
6083 2014-03-14  Richard Henderson  <rth@redhat.com>
6084
6085         PR target/60525
6086         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
6087         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
6088         (*floathi<X87MODEF>2_i387_with_temp): Remove.
6089         (floathi splitters): Remove.
6090         (float<SWI48x>xf2): New pattern.
6091         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
6092         code that tried to handle DImode for 32-bit, but which was excluded
6093         by the pattern's condition.  Drop allocation of stack temporary.
6094         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
6095         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
6096         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
6097         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
6098         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
6099         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
6100         (*float<SWI48><MODEF>2_sse_interunit): Remove.
6101         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
6102         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
6103         (*float<SWI48x><X87MODEF>2_i387): Remove.
6104         (all float _with_temp splitters): Remove.
6105         (*float<SWI48x><MODEF>2_i387): New pattern.
6106         (*float<SWI48><MODEF>2_sse): New pattern.
6107         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
6108         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
6109
6110 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
6111             Marek Polacek  <polacek@redhat.com>
6112
6113         PR middle-end/60484
6114         * common.opt (dump_base_name_prefixed): New Variable.
6115         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
6116         if x_dump_base_name_prefixed is already set, set it at the end.
6117
6118 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
6119
6120         PR rtl-optimization/60508
6121         * lra-constraints.c (get_reload_reg): Add new parameter
6122         in_subreg_p.
6123         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
6124         Pass the new parameter values.
6125
6126 2014-03-14  Richard Biener  <rguenther@suse.de>
6127
6128         * common.opt: Revert unintented changes from r205065.
6129         * opts.c: Likewise.
6130
6131 2014-03-14  Richard Biener  <rguenther@suse.de>
6132
6133         PR middle-end/60518
6134         * cfghooks.c (split_block): Properly adjust all loops the
6135         block was a latch of.
6136
6137 2014-03-14  Martin Jambor  <mjambor@suse.cz>
6138
6139         PR lto/60461
6140         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
6141         and simplify it.
6142
6143 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
6144
6145         PR target/59396
6146         * config/avr/avr.c (avr_set_current_function): Pass function name
6147         through default_strip_name_encoding before sanity checking instead
6148         of skipping the first char of the assembler name.
6149
6150 2014-03-13  Richard Henderson  <rth@redhat.com>
6151
6152         PR debug/60438
6153         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
6154         (ix86_force_to_memory, ix86_free_from_memory): Remove.
6155         * config/i386/i386-protos.h: Likewise.
6156         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
6157         in the expander instead of a splitter.
6158         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
6159         any possibility of requiring a memory.
6160         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
6161         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
6162         (fp branch splitters): Update for ix86_split_fp_branch.
6163         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
6164         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
6165         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
6166         (*fop_<MODEF>_2_i387): Remove f/r alternative.
6167         (*fop_<MODEF>_3_i387): Likewise.
6168         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
6169         (splitters for the fop_* register patterns): Remove.
6170         (fscalexf4_i387): Rename from *fscalexf4_i387.
6171         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
6172
6173 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
6174
6175         PR tree-optimization/59779
6176         * tree-dfa.c (get_ref_base_and_extent): Use double_int
6177         type for bitsize and maxsize instead of HOST_WIDE_INT.
6178
6179 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
6180
6181         PR rtl-optimization/57320
6182         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
6183         the CFG after thread_prologue_and_epilogue_insns.
6184
6185 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
6186
6187         PR rtl-optimization/57189
6188         * lra-constraints.c (process_alt_operands): Disfavor spilling
6189         vector pseudos.
6190
6191 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
6192
6193         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
6194
6195 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
6196
6197         PR tree-optimization/59025
6198         PR middle-end/60418
6199         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
6200         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
6201
6202 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
6203
6204         PR target/60486
6205         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
6206         calls of avr_out_plus_1.
6207
6208 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
6209
6210         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
6211         BB's single pred and update the father loop's latch info later.
6212
6213 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6214
6215         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
6216         (VEC_M): Likewise.
6217         (VEC_N): Likewise.
6218         (VEC_R): Likewise.
6219         (VEC_base): Likewise.
6220         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
6221         registers, we need to swap double words in little endian mode.
6222
6223         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
6224         to be a container mode for 128-bit integer operations added in ISA
6225         2.07.  Unlike TImode and PTImode, the preferred register set is
6226         the Altivec/VMX registers for the 128-bit operations.
6227
6228         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
6229         declarations.
6230         (rs6000_split_128bit_ok_p): Likewise.
6231
6232         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
6233         macros for creating ISA 2.07 normal and overloaded builtin
6234         functions with 3 arguments.
6235         (BU_P8V_OVERLOAD_3): Likewise.
6236         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
6237         for use as overloaded functions.
6238         (VPERM_1TI_UNS): Likewise.
6239         (VSEL_1TI): Likewise.
6240         (VSEL_1TI_UNS): Likewise.
6241         (ST_INTERNAL_1ti): Likewise.
6242         (LD_INTERNAL_1ti): Likewise.
6243         (XXSEL_1TI): Likewise.
6244         (XXSEL_1TI_UNS): Likewise.
6245         (VPERM_1TI): Likewise.
6246         (VPERM_1TI_UNS): Likewise.
6247         (XXPERMDI_1TI): Likewise.
6248         (SET_1TI): Likewise.
6249         (LXVD2X_V1TI): Likewise.
6250         (STXVD2X_V1TI): Likewise.
6251         (VEC_INIT_V1TI): Likewise.
6252         (VEC_SET_V1TI): Likewise.
6253         (VEC_EXT_V1TI): Likewise.
6254         (EQV_V1TI): Likewise.
6255         (NAND_V1TI): Likewise.
6256         (ORC_V1TI): Likewise.
6257         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
6258         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
6259         overloaded builtin.
6260         (VADDUQM): Likewise.
6261         (VSUBCUQ): Likewise.
6262         (VADDEUQM): Likewise.
6263         (VADDECUQ): Likewise.
6264         (VSUBEUQM): Likewise.
6265         (VSUBECUQ): Likewise.
6266
6267         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
6268         __int128_t and __uint128_t types.
6269         (__uint128_type): Likewise.
6270         (altivec_categorize_keyword): Add support for vector __int128_t,
6271         vector __uint128_t, vector __int128, and vector unsigned __int128
6272         as a container type for TImode operations that need to be done in
6273         VSX/Altivec registers.
6274         (rs6000_macro_to_expand): Likewise.
6275         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
6276         to support 128-bit integer instructions vaddcuq, vadduqm,
6277         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
6278         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
6279
6280         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
6281         for V1TImode, and set up preferences to use VSX/Altivec registers.
6282         Setup VSX reload handlers.
6283         (rs6000_debug_reg_global): Likewise.
6284         (rs6000_init_hard_regno_mode_ok): Likewise.
6285         (rs6000_preferred_simd_mode): Likewise.
6286         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
6287         (easy_altivec_constant): Likewise.
6288         (output_vec_const_move): Likewise.
6289         (rs6000_expand_vector_set): Convert V1TImode set and extract to
6290         simple move.
6291         (rs6000_expand_vector_extract): Likewise.
6292         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
6293         addressing.
6294         (rs6000_const_vec): Add support for V1TImode.
6295         (rs6000_emit_le_vsx_load): Swap double words when loading or
6296         storing TImode/V1TImode.
6297         (rs6000_emit_le_vsx_store): Likewise.
6298         (rs6000_emit_le_vsx_move): Likewise.
6299         (rs6000_emit_move): Add support for V1TImode.
6300         (altivec_expand_ld_builtin): Likewise.
6301         (altivec_expand_st_builtin): Likewise.
6302         (altivec_expand_vec_init_builtin): Likewise.
6303         (altivec_expand_builtin): Likewise.
6304         (rs6000_init_builtins): Add support for V1TImode type.  Add
6305         support for ISA 2.07 128-bit integer builtins.  Define type names
6306         for the VSX/Altivec vector types.
6307         (altivec_init_builtins): Add support for overloaded vector
6308         functions with V1TImode type.
6309         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
6310         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
6311         external function.
6312         (rs6000_split_128bit_ok_p): Likewise.
6313         (rs6000_handle_altivec_attribute): Create V1TImode from vector
6314         __int128_t and vector __uint128_t.
6315
6316         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
6317         and mode attributes.
6318         (VSX_M): Likewise.
6319         (VSX_M2): Likewise.
6320         (VSm): Likewise.
6321         (VSs): Likewise.
6322         (VSr): Likewise.
6323         (VSv): Likewise.
6324         (VS_scalar): Likewise.
6325         (VS_double): Likewise.
6326         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
6327
6328         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
6329         we support the ISA 2.07 128-bit integer arithmetic instructions.
6330         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
6331         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
6332         and TImode types for use with the builtin functions.
6333         (V1TI_type_node): Likewise.
6334         (unsigned_V1TI_type_node): Likewise.
6335         (intTI_type_internal_node): Likewise.
6336         (uintTI_type_internal_node): Likewise.
6337
6338         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
6339         128-bit builtin functions.
6340         (UNSPEC_VADDEUQM): Likewise.
6341         (UNSPEC_VADDECUQ): Likewise.
6342         (UNSPEC_VSUBCUQ): Likewise.
6343         (UNSPEC_VSUBEUQM): Likewise.
6344         (UNSPEC_VSUBECUQ): Likewise.
6345         (VM): Add V1TImode to vector mode iterators.
6346         (VM2): Likewise.
6347         (VI_unit): Likewise.
6348         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
6349         (altivec_vaddcuq): Likewise.
6350         (altivec_vsubuqm): Likewise.
6351         (altivec_vsubcuq): Likewise.
6352         (altivec_vaddeuqm): Likewise.
6353         (altivec_vaddecuq): Likewise.
6354         (altivec_vsubeuqm): Likewise.
6355         (altivec_vsubecuq): Likewise.
6356
6357         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
6358         mode iterators.
6359         (BOOL_128): Likewise.
6360         (BOOL_REGS_OUTPUT): Likewise.
6361         (BOOL_REGS_OP1): Likewise.
6362         (BOOL_REGS_OP2): Likewise.
6363         (BOOL_REGS_UNARY): Likewise.
6364         (BOOL_REGS_AND_CR0): Likewise.
6365
6366         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
6367         128-bit integer builtin support.
6368         (vec_vadduqm): Likewise.
6369         (vec_vaddecuq): Likewise.
6370         (vec_vaddeuqm): Likewise.
6371         (vec_vsubecuq): Likewise.
6372         (vec_vsubeuqm): Likewise.
6373         (vec_vsubcuq): Likewise.
6374         (vec_vsubuqm): Likewise.
6375
6376         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6377         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
6378         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
6379         128-bit integer add/subtract to ISA 2.07.
6380
6381 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
6382
6383         * config/arc/arc.c (arc_predicate_delay_insns):
6384         Fix third argument passed to conditionalize_nonjump.
6385
6386 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
6387
6388         * config/aarch64/aarch64-builtins.c
6389         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
6390         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
6391         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
6392         instead of __builtin_lfloor.
6393         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
6394
6395 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6396
6397         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
6398         (tree_ssa_ifcombine_bb_1): New function.
6399         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
6400         is an empty forwarder block to then_bb or vice versa and then_bb
6401         and else_bb are effectively swapped.
6402
6403 2014-03-12  Christian Bruel  <christian.bruel@st.com>
6404
6405         PR target/60264
6406         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
6407         REG_CFA_DEF_CFA note.
6408         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
6409         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
6410
6411 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
6412
6413         PR tree-optimization/60454
6414         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
6415
6416 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6417
6418         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
6419         Do not define target_cpu_default2 to generic.
6420         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
6421         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
6422         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
6423
6424 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6425             Marc Glisse  <marc.glisse@inria.fr>
6426
6427         PR tree-optimization/60502
6428         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
6429         instead of build_low_bits_mask.
6430
6431 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
6432
6433         PR middle-end/60482
6434         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
6435         if there are multiple uses, but op doesn't live on E edge.
6436         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
6437         clobber stmts before __builtin_unreachable.
6438
6439 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
6440
6441         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
6442         hard_frame_pointer_rtx.
6443         * cse.c (cse_insn): Remove volatile check.
6444         * cselib.c (cselib_process_insn): Likewise.
6445         * dse.c (scan_insn): Likewise.
6446
6447 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6448
6449         * config/arc/arc.c (conditionalize_nonjump): New function,
6450         broken out of ...
6451         (arc_ifcvt): ... this.
6452         (arc_predicate_delay_insns): Use it.
6453
6454 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6455
6456         * config/arc/predicates.md (extend_operand): During/after reload,
6457         allow const_int_operand.
6458         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
6459         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
6460         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
6461         to "i".
6462         (umulsi3_highpart_i): Likewise.
6463
6464 2014-03-11  Richard Biener  <rguenther@suse.de>
6465
6466         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
6467         Add asserts to guard possible wrong-code bugs.
6468
6469 2014-03-11  Richard Biener  <rguenther@suse.de>
6470
6471         PR tree-optimization/60429
6472         PR tree-optimization/60485
6473         * tree-ssa-structalias.c (set_union_with_increment): Properly
6474         take into account all fields that overlap the shifted vars.
6475         (do_sd_constraint): Likewise.
6476         (do_ds_constraint): Likewise.
6477         (get_constraint_for_ptr_offset): Likewise.
6478
6479 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
6480
6481         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
6482         (nios2_compute_frame_layout):
6483         Add calculation of cfun->machine->fp_save_offset.
6484         (nios2_expand_prologue): Correct setting of frame pointer register
6485         in prologue.
6486         (nios2_expand_epilogue): Update recovery of stack pointer from
6487         frame pointer accordingly.
6488         (nios2_initial_elimination_offset): Update calculation of offset
6489         for eliminating to HARD_FRAME_POINTER_REGNUM.
6490
6491 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
6492
6493         PR ipa/60457
6494         * ipa.c (symtab_remove_unreachable_nodes): Don't call
6495         cgraph_get_create_node on VAR_DECLs.
6496
6497 2014-03-10  Richard Biener  <rguenther@suse.de>
6498
6499         PR middle-end/60474
6500         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
6501
6502 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
6503
6504         * config/vms/vms.opt (vms_float_format): New variable.
6505
6506 2014-03-08  Tobias Burnus  <burnus@net-b.de>
6507
6508         * doc/invoke.texi (-fcilkplus): Update implementation status.
6509
6510 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
6511             Richard Biener  <rguenther@suse.de>
6512
6513         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
6514         consistently accross all TUs.
6515         (run_gcc): Enable -fshort-double automatically at link at link-time
6516         and disallow override.
6517
6518 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
6519
6520         PR target/58271
6521         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
6522         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
6523         if they can't be used.
6524
6525 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6526
6527         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
6528         for Solaris 11/x86 ld.
6529         * configure: Regenerate.
6530
6531 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6532
6533         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
6534         (LIB_TLS_SPEC): Save as ld_tls_libs.
6535         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
6536         (HAVE_AS_IX86_TLSLDM): New test.
6537         * configure, config.in: Regenerate.
6538         * config/i386/i386.c (legitimize_tls_address): Fall back to
6539         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
6540         cannot support TLS_MODEL_LOCAL_DYNAMIC.
6541         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
6542         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
6543
6544 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
6545
6546         * common.opt (fira-loop-pressure): Mark as optimization.
6547
6548 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
6549
6550         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
6551         an OpenMP mappable type.
6552
6553 2014-03-06  Matthias Klose  <doko@ubuntu.com>
6554
6555         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
6556         MULTILIB_OSDIRNAMES is not defined.
6557
6558 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
6559             Meador Inge  <meadori@codesourcery.com>
6560
6561         PR target/58595
6562         * config/arm/arm.c (arm_tls_symbol_p): Remove.
6563         (arm_legitimize_address): Call legitimize_tls_address for any
6564         arm_tls_referenced_p expression, handle constant addend.  Call it
6565         before testing for !TARGET_ARM.
6566         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
6567
6568 2014-03-06  Richard Biener  <rguenther@suse.de>
6569
6570         PR middle-end/60445
6571         PR lto/60424
6572         PR lto/60427
6573         Revert
6574         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6575
6576         * tree-streamer.c (record_common_node): Assert we don't record
6577         nodes with type double.
6578         (preload_common_node): Skip type double, complex double and double
6579         pointer since it is now frontend dependent due to fshort-double option.
6580
6581 2014-03-06  Richard Biener  <rguenther@suse.de>
6582
6583         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
6584         or -fno-lto is specified and the linker has full plugin support.
6585         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
6586         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
6587         * lto-wrapper.c (merge_and_complain): Merge compile-time
6588         optimization levels.
6589         (run_gcc): And pass it through to the link options.
6590
6591 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
6592
6593         PR debug/60381
6594         Revert:
6595         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6596         PR debug/59992
6597         * cselib.c (remove_useless_values): Skip to avoid quadratic
6598         behavior if the condition moved from...
6599         (cselib_process_insn): ... here holds.
6600
6601 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6602
6603         PR plugins/59335
6604         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
6605         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
6606
6607         PR plugins/59335
6608         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
6609         (TM_H): Add x86-tune.def.
6610
6611 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6612
6613         * config/aarch64/aarch64.c (generic_tunings):
6614         Use cortexa57_extra_costs.
6615
6616 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6617
6618         PR lto/60404
6619         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
6620         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
6621         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
6622         cost for in_lto_p.
6623
6624 2014-03-04  Heiher  <r@hev.cc>
6625
6626         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
6627         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
6628
6629 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
6630
6631         * config/i386/predicates.md (const2356_operand): Change to ...
6632         (const2367_operand): ... this.
6633         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
6634         const2367_operand.
6635         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6636         (*avx512pf_scatterpf<mode>sf): Ditto.
6637         (avx512pf_scatterpf<mode>df): Ditto.
6638         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6639         (*avx512pf_scatterpf<mode>df): Ditto.
6640         * config/i386/i386.c (ix86_expand_builtin): Update
6641         incorrect hint operand error message.
6642
6643 2014-03-04  Richard Biener  <rguenther@suse.de>
6644
6645         * lto-section-in.c (lto_get_section_data): Fix const cast.
6646
6647 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6648
6649         * tree-streamer.c (record_common_node): Assert we don't record
6650         nodes with type double.
6651         (preload_common_node): Skip type double, complex double and double
6652         pointer since it is now frontend dependent due to fshort-double option.
6653
6654 2014-03-04  Richard Biener  <rguenther@suse.de>
6655
6656         PR lto/60405
6657         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
6658         (lto_input_toplevel_asms): Likewise.
6659         * lto-section-in.c (lto_get_section_data): Instead do it here
6660         for every section.
6661
6662 2014-03-04  Richard Biener  <rguenther@suse.de>
6663
6664         PR tree-optimization/60382
6665         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
6666         dead PHIs a reduction.
6667
6668 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
6669
6670         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
6671         hint value.
6672         (_mm_prefetch): Move out of GCC target("sse") pragma.
6673         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
6674         GCC target("prfchw") pragma.
6675         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
6676         for locality <= 2.
6677         * config/i386/i386.c (ix86_option_override_internal): Enable
6678         -mprfchw with -mprefetchwt1.
6679
6680 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6681
6682         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
6683         Mark as varying.
6684
6685 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6686
6687         * opts.h (CL_PCH_IGNORE): Define.
6688         * targhooks.c (option_affects_pch_p):
6689         Return false for options that have CL_PCH_IGNORE set.
6690         * opt-functions.awk: Process PchIgnore.
6691         * doc/options.texi: Document PchIgnore.
6692
6693         * config/arc/arc.opt (misize): Add PchIgnore property.
6694
6695 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6696
6697         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
6698         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
6699         constraint on constants to permit them being loaded into
6700         GENERAL_REGS or BASE_REGS.
6701
6702 2014-03-03  Nick Clifton  <nickc@redhat.com>
6703
6704         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
6705         anti-cacnonical alternatives.
6706         (negandhi3_real): New pattern.
6707         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
6708
6709 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
6710
6711         * config/avr/avr-mcus.def: Remove atxmega16x1.
6712         * config/avr/avr-tables.opt: Regenerate.
6713         * config/avr/t-multilib: Regenerate.
6714         * doc/avr-mmcu.texi: Regenerate.
6715
6716 2014-03-03  Tobias Grosser  <tobias@grosser.es>
6717             Mircea Namolaru  <mircea.namolaru@inria.fr>
6718
6719         PR tree-optimization/58028
6720         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
6721         scalar dimensions.
6722
6723 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6724
6725         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
6726         not handled by recognizers.
6727
6728 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
6729
6730         PR middle-end/60175
6731         * function.c (expand_function_end): Don't emit
6732         clobber_return_register sequence if clobber_after is a BARRIER.
6733         * cfgexpand.c (construct_exit_block): Append instructions before
6734         return_label to prev_bb.
6735
6736 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6737
6738         * config/rs6000/constraints.md: Document reserved use of "wc".
6739
6740 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6741
6742         PR ipa/60150
6743         * ipa.c (function_and_variable_visibility): When dissolving comdat
6744         group, also set all symbols to local.
6745
6746 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6747
6748         PR ipa/60306
6749
6750         Revert:
6751         2013-12-14   Jan Hubicka  <jh@suse.cz>
6752         PR middle-end/58477
6753         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
6754
6755 2014-03-02  Jon Beniston  <jon@beniston.com>
6756
6757         PR bootstrap/48230
6758         PR bootstrap/50927
6759         PR bootstrap/52466
6760         PR target/46898
6761         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
6762           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
6763         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
6764         (simple_return, *simple_return): New patterns
6765         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
6766         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
6767
6768 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
6769
6770         * dwarf2out.c (gen_subprogram_die): Tidy.
6771
6772 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
6773
6774         PR target/60071
6775         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
6776         (*mov_t_msb_neg_negc): ... this new insn.
6777
6778 2014-02-28  Jason Merrill  <jason@redhat.com>
6779
6780         PR c++/58678
6781         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
6782         function.
6783
6784 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
6785
6786         PR c++/60314
6787         * dwarf2out.c (decltype_auto_die): New static.
6788         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
6789         (gen_type_die_with_usage): Handle 'decltype(auto)'.
6790         (is_cxx_auto): Likewise.
6791
6792 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
6793
6794         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
6795         we are not using general regs only.
6796
6797 2014-02-28  Richard Biener  <rguenther@suse.de>
6798
6799         PR target/60280
6800         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
6801         previous fix and only allow to remove trivial pre-headers
6802         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
6803         (remove_forwarder_block): Properly update the latch of a loop.
6804
6805 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6806
6807         PR debug/59992
6808         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
6809         (cselib_preserved_hash_table): New.
6810         (preserve_constants_and_equivs): Move preserved vals to it.
6811         (cselib_find_slot): Look it up first.
6812         (cselib_init): Initialize it.
6813         (cselib_finish): Release it.
6814         (dump_cselib_table): Dump it.
6815
6816 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6817
6818         PR debug/59992
6819         * cselib.c (remove_useless_values): Skip to avoid quadratic
6820         behavior if the condition moved from...
6821         (cselib_process_insn): ... here holds.
6822
6823 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6824
6825         PR debug/57232
6826         * var-tracking.c (vt_initialize): Apply the same condition to
6827         preserve the CFA base value.
6828
6829 2014-02-28  Joey Ye  <joey.ye@arm.com>
6830
6831         PR target/PR60169
6832         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
6833         if reload in progress or completed.
6834
6835 2014-02-28  Tobias Burnus  <burnus@net-b.de>
6836
6837         PR middle-end/60147
6838         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
6839         NAMELIST_DECL.
6840
6841 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
6842
6843         * doc/tm.texi.in (Condition Code Status): Update documention for
6844         relative locations of cc0-setter and cc0-user.
6845
6846 2014-02-27  Jeff Law  <law@redhat.com>
6847
6848         PR rtl-optimization/52714
6849         * combine.c (try_combine): When splitting an unrecognized PARALLEL
6850         into two independent simple sets, if I3 is a jump, ensure the
6851         pattern we place into I3 is a (set (pc) ...).
6852
6853 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
6854             Jeff Law  <law@redhat.com>
6855
6856         PR rtl-optimization/49847
6857         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
6858         are in different blocks.
6859         * doc/tm.texi (Condition Code Status): Update documention for
6860         relative locations of cc0-setter and cc0-user.
6861
6862 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
6863
6864         PR target/59222
6865         * lra.c (lra_emit_add): Check SUBREG too.
6866
6867 2014-02-27  Andreas Schwab  <schwab@suse.de>
6868
6869         * config/m68k/m68k.c (m68k_option_override): Disable
6870         -flive-range-shrinkage for classic m68k.
6871         (m68k_override_options_after_change): Likewise.
6872
6873 2014-02-27  Marek Polacek  <polacek@redhat.com>
6874
6875         PR middle-end/59223
6876         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
6877         -Wmaybe-uninitialized.
6878
6879 2014-02-27  Alan Modra  <amodra@gmail.com>
6880
6881         PR target/57936
6882         * reload1.c (emit_input_reload_insns): When reload_override_in,
6883         set old to rl->in_reg when rl->in_reg is a subreg.
6884
6885 2014-02-26  Richard Biener  <rguenther@suse.de>
6886
6887         PR bootstrap/60343
6888         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
6889
6890 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6891
6892         * common/config/i386/predicates.md (const1256_operand): Remove.
6893         (const2356_operand): New.
6894         (const_1_to_2_operand): Remove.
6895         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
6896         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6897         (*avx512pf_gatherpf<mode>sf): Ditto.
6898         (avx512pf_gatherpf<mode>df): Ditto.
6899         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6900         (*avx512pf_gatherpf<mode>df): Ditto.
6901         (avx512pf_scatterpf<mode>sf): Ditto.
6902         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6903         (*avx512pf_scatterpf<mode>sf): Ditto.
6904         (avx512pf_scatterpf<mode>df): Ditto.
6905         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6906         (*avx512pf_scatterpf<mode>df): Ditto.
6907         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
6908
6909 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
6910
6911         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
6912         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
6913         (_mm512_mask_testn_epi64_mask): Move to ...
6914         * config/i386/avx512cdintrin.h: Here.
6915         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
6916         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
6917         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
6918         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
6919         TARGET_AVX512F from TARGET_AVX512CD.
6920
6921 2014-02-26  Richard Biener  <rguenther@suse.de>
6922
6923         PR ipa/60327
6924         * ipa.c (walk_polymorphic_call_targets): Properly guard
6925         call to inline_update_overall_summary.
6926
6927 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
6928
6929         PR target/60280
6930         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
6931         and latches only if requested.  Fix latch if it is removed.
6932         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
6933         LOOPS_HAVE_PREHEADERS.
6934
6935 2014-02-25  Andrew Pinski  <apinski@cavium.com>
6936
6937         * builtins.c (expand_builtin_thread_pointer): Create a new target
6938         when the target is NULL.
6939
6940 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
6941
6942         PR rtl-optimization/60317
6943         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6944         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6945         * lra-assigns.c: Include params.h.
6946         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
6947         other reload pseudos considerations.
6948
6949 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6950
6951         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
6952         to use canonical form for nor<mode>3.
6953
6954 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6955
6956         PR target/55426
6957         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
6958         conversions.
6959
6960 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6961
6962         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
6963         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
6964         (ix86_handle_option): Handle OPT_mprefetchwt1.
6965         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
6966         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
6967         PREFETCHWT1 CPUID.
6968         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6969         OPTION_MASK_ISA_PREFETCHWT1.
6970         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
6971         (PTA_PREFETCHWT1): New.
6972         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
6973         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
6974         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
6975         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
6976         (*prefetch_avx512pf_<mode>_: Change into ...
6977         (*prefetch_prefetchwt1_<mode>: This.
6978         * config/i386/i386.opt (mprefetchwt1): New.
6979         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
6980         (_mm_prefetch): Handle intent to write.
6981         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
6982
6983 2014-02-25  Richard Biener  <rguenther@suse.de>
6984
6985         PR middle-end/60291
6986         * emit-rtl.c (mem_attrs_htab): Remove.
6987         (mem_attrs_htab_hash): Likewise.
6988         (mem_attrs_htab_eq): Likewise.
6989         (set_mem_attrs): Always allocate new mem-attrs when something changed.
6990         (init_emit_once): Do not allocate mem_attrs_htab.
6991
6992 2014-02-25  Richard Biener  <rguenther@suse.de>
6993
6994         PR lto/60319
6995         * lto-opts.c (lto_write_options): Output non-explicit conservative
6996         -fwrapv, -fno-trapv and -fno-strict-overflow.
6997         * lto-wrapper.c (merge_and_complain): Handle merging those options.
6998         (run_gcc): And pass them through.
6999
7000 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
7001
7002         * sel-sched.c (calculate_new_fences): New parameter ptime.
7003         Calculate it as a maximum over all fence cycles.
7004         (sel_sched_region_2): Adjust the call to calculate_new_fences.
7005         Print the final schedule timing when sched_verbose.
7006
7007 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
7008
7009         PR rtl-optimization/60292
7010         * sel-sched.c (fill_vec_av_set): Do not reset target availability
7011         bit fot the fence instruction.
7012
7013 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
7014
7015         * calls.h: Fix typo in comment.
7016
7017 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
7018
7019         * config/pa/pa.c (pa_output_move_double): Don't valididate when
7020         adjusting offsetable addresses.
7021
7022 2014-02-24  Guozhi Wei  <carrot@google.com>
7023
7024         * sparseset.h (sparseset_pop): Fix the wrong index.
7025
7026 2014-02-24  Walter Lee  <walt@tilera.com>
7027
7028         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
7029         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
7030         triplet.
7031         * common/config/tilegx/tilegx-common.c
7032         (TARGET_DEFAULT_TARGET_FLAGS): Define.
7033         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
7034         (LINK_SPEC): Ditto.
7035         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
7036         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
7037         (tilegx_gimplify_va_arg_expr): Handle big endian.
7038         (tilegx_expand_unaligned_load): Ditto.
7039         (tilegx_expand_unaligned_store): Ditto.
7040         (TARGET_RETURN_IN_MSB): New.
7041         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
7042         (TARGET_ENDIAN_DEFAULT): New.
7043         (TARGET_BIG_ENDIAN): Handle big endian.
7044         (BYTES_BIG_ENDIAN): Ditto.
7045         (WORDS_BIG_ENDIAN): Ditto.
7046         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
7047         (ENDIAN_SPEC): New.
7048         (EXTRA_SPECS): New.
7049         * config/tilegx/tilegx.md (extv): Handle big endian.
7050         (extzv): Ditto.
7051         (insn_st<n>): Ditto.
7052         (insn_st<n>_add<bitsuffix>): Ditto.
7053         (insn_stnt<n>): Ditto.
7054         (insn_stnt<n>_add<bitsuffix>):Ditto.
7055         (vec_interleave_highv8qi): Handle big endian.
7056         (vec_interleave_highv8qi_be): New.
7057         (vec_interleave_highv8qi_le): New.
7058         (insn_v1int_h): Handle big endian.
7059         (vec_interleave_lowv8qi): Handle big endian.
7060         (vec_interleave_lowv8qi_be): New.
7061         (vec_interleave_lowv8qi_le): New.
7062         (insn_v1int_l): Handle big endian.
7063         (vec_interleave_highv4hi): Handle big endian.
7064         (vec_interleave_highv4hi_be): New.
7065         (vec_interleave_highv4hi_le): New.
7066         (insn_v2int_h): Handle big endian.
7067         (vec_interleave_lowv4hi): Handle big endian.
7068         (vec_interleave_lowv4hi_be): New.
7069         (vec_interleave_lowv4hi_le): New.
7070         (insn_v2int_l): Handle big endian.
7071         (vec_interleave_highv2si): Handle big endian.
7072         (vec_interleave_highv2si_be): New.
7073         (vec_interleave_highv2si_le): New.
7074         (insn_v4int_h): Handle big endian.
7075         (vec_interleave_lowv2si): Handle big endian.
7076         (vec_interleave_lowv2si_be): New.
7077         (vec_interleave_lowv2si_le): New.
7078         (insn_v4int_l): Handle big endian.
7079         * config/tilegx/tilegx.opt (mbig-endian): New option.
7080         (mlittle-endian): New option.
7081         * doc/install.texi: Document tilegxbe-linux.
7082         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
7083
7084 2014-02-24  Martin Jambor  <mjambor@suse.cz>
7085
7086         PR ipa/60266
7087         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
7088         there are no parameter descriptors.
7089
7090 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
7091
7092         PR rtl-optimization/60268
7093         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
7094         initialization to ...
7095         (sched_rgn_init): ... here.
7096         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
7097
7098 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7099
7100         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
7101         names.
7102
7103 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
7104
7105         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
7106         definition.
7107
7108 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7109
7110         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
7111         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
7112
7113 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
7114
7115         * config/microblaze/predicates.md: Add cmp_op predicate.
7116         * config/microblaze/microblaze.md: Add branch_compare instruction
7117         which uses cmp_op predicate and emits cmp insn before branch.
7118         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
7119         to microblaze_expand_conditional_branch and consolidate logic.
7120         (microblaze_expand_conditional_branch): emit branch_compare
7121         insn instead of handling cmp op separate from branch insn.
7122
7123 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7124
7125         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
7126         to permit subregs.
7127
7128 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7129
7130         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
7131         define_insn with define_expand and new define_insn
7132         *altivec_lve<VI_char>x_internal.
7133         (altivec_stve<VI_char>x): Replace define_insn with define_expand
7134         and new define_insn *altivec_stve<VI_char>x_internal.
7135         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
7136         prototype.
7137         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
7138         lve*x built-ins.
7139         (altivec_expand_stvex_be): New function.
7140
7141 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
7142
7143         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
7144         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
7145         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
7146         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
7147
7148 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
7149
7150         PR target/60298
7151         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
7152         instead of emit_move_insn.
7153
7154 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7155
7156         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
7157         vspltw with vsldoi.
7158         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
7159         gen_altivec_vsumsws.
7160
7161 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7162
7163         * config/rs6000/altivec.md (altivec_lvxl): Rename as
7164         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
7165         (altivec_lvxl_<mode>): New define_expand incorporating
7166         -maltivec=be semantics where needed.
7167         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
7168         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
7169         semantics where needed.
7170         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
7171         (altivec_stvx_<mode>): New define_expand incorporating
7172         -maltivec=be semantics where needed.
7173         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
7174         VM2 iterator instead of V4SI.
7175         (altivec_stvxl_<mode>): New define_expand incorporating
7176         -maltivec=be semantics where needed.
7177         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
7178         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
7179         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
7180         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
7181         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
7182         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
7183         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
7184         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
7185         ALTIVEC_BUILTIN_STVXL.
7186         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
7187         (altivec_expand_stvx_be): Likewise.
7188         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
7189         (altivec_expand_lvx_be): Likewise.
7190         (altivec_expand_stvx_be): Likewise.
7191         (altivec_expand_builtin): Add cases for
7192         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
7193         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
7194         (altivec_init_builtins): Add definitions for
7195         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
7196         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
7197
7198 2014-02-21  Catherine Moore  <clm@codesourcery.com>
7199
7200         * doc/invoke.texi (mvirt, mno-virt): Document.
7201         * config/mips/mips.opt (mvirt): New option.
7202         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
7203
7204 2014-02-21  Richard Biener  <rguenther@suse.de>
7205
7206         PR tree-optimization/60276
7207         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
7208         (STMT_VINFO_MIN_NEG_DIST): New macro.
7209         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
7210         STMT_VINFO_MIN_NEG_DIST.
7211         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
7212         made for negative dependence distances still hold.
7213
7214 2014-02-21  Richard Biener  <rguenther@suse.de>
7215
7216         PR middle-end/60291
7217         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
7218         DECL_INITIAL for globals not in the current function context.
7219
7220 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
7221
7222         PR tree-optimization/56490
7223         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
7224         * tree-ssa-uninit.c: Include params.h.
7225         (compute_control_dep_chain): Add num_calls argument, return false
7226         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
7227         num_calls to recursive call.
7228         (find_predicates): Change dep_chain into normal array,
7229         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
7230         variable and adjust compute_control_dep_chain caller.
7231         (find_def_preds): Likewise.
7232
7233 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
7234
7235         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
7236         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
7237
7238 2014-02-21  Nick Clifton  <nickc@redhat.com>
7239
7240         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
7241         (pushhi1): Likewise.
7242         (popqi1): Add mode to pre_dec.
7243         (pophi1): Likewise.
7244
7245 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
7246
7247         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
7248         mode for mask of V8SFmode permutation.
7249
7250 2014-02-20  Richard Henderson  <rth@redhat.com>
7251
7252         PR c++/60272
7253         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
7254         a new pseudo for OLDVAL.
7255
7256 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
7257
7258         PR target/57896
7259         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
7260         gen_reg_rtx if d->testing_p.
7261         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
7262         if d->testing_p and we will certainly return true.
7263         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
7264         if d->testing_p.
7265
7266 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
7267
7268         * emit-rtl.c (gen_reg_rtx): Assert that
7269         crtl->emit.regno_pointer_align_length is non-zero.
7270
7271 2014-02-20  Richard Henderson  <rth@redhat.com>
7272
7273         PR c++/60272
7274         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
7275         on failure the store back into EXPECT.
7276
7277 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
7278             Sandra Loosemore  <sandra@codesourcery.com>
7279
7280         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
7281         * config/nios2/nios2.c (nios2_function_profiler): Add
7282         -fPIC (flag_pic == 2) support.
7283         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
7284         (nios2_large_offset_p): New function.
7285         (nios2_unspec_reloc_p): Move up position, update to use
7286         nios2_large_offset_p.
7287         (nios2_unspec_address): Remove function.
7288         (nios2_unspec_offset): New function.
7289         (nios2_large_got_address): New function.
7290         (nios2_got_address): Add large offset support.
7291         (nios2_legitimize_tls_address): Update usage of removed and new
7292         functions.
7293         (nios2_symbol_binds_local_p): New function.
7294         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
7295         (nios2_legitimize_address): Update to use nios2_large_offset_p.
7296         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
7297         (nios2_print_operand): Merge H/L processing, add hiadj/lo
7298         processing for (const (unspec ...)).
7299         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
7300
7301 2014-02-20  Richard Biener  <rguenther@suse.de>
7302
7303         * tree-cfg.c (replace_uses_by): Mark altered BBs before
7304         doing the substitution.
7305         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
7306
7307 2014-02-20  Martin Jambor  <mjambor@suse.cz>
7308
7309         PR ipa/55260
7310         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
7311         info when checking whether lattices are bottom.
7312
7313 2014-02-20  Richard Biener  <rguenther@suse.de>
7314
7315         PR middle-end/60221
7316         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
7317         regions at -O0.
7318
7319 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
7320
7321         PR ipa/58555
7322         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
7323         parameter specifying the scaling.
7324         (inline_call): Update.
7325         (want_inline_recursively): Guard division by zero.
7326         (recursive_inlining): Update.
7327         * ipa-inline.h (clone_inlined_nodes): Update.
7328
7329 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
7330
7331         PR target/60204
7332         * config/i386/i386.c (classify_argument): Pass structures of size
7333         64 bytes or less in register.
7334
7335 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
7336             Kirill Yukhin  <kirill.yukhin@intel.com>
7337
7338         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
7339         (_mm_rcp28_round_ss): Ditto.
7340         (_mm_rsqrt28_round_sd): Ditto.
7341         (_mm_rsqrt28_round_ss): Ditto.
7342         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
7343         (_mm_rcp14_round_ss): Ditto.
7344         (_mm_rsqrt14_round_sd): Ditto.
7345         (_mm_rsqrt14_round_ss): Ditto.
7346         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
7347         the first input operand, get rid of match_dup.
7348         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
7349         attribute to sse.
7350         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7351         Ditto.
7352         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
7353         operand as the first input operand, set type attribute.
7354         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7355         Set type attribute.
7356         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
7357         operand as the first input operand, set type attribute.
7358
7359 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7360
7361         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
7362         bit of zero.
7363
7364 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
7365
7366         PR target/60207
7367         * config/i386/i386.c (construct_container): Remove TFmode check
7368         for X86_64_INTEGER_CLASS.
7369
7370 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
7371
7372         PR target/59794
7373         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
7374         only when -Wpsabi is enabled.
7375
7376 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
7377
7378          PR target/59799
7379         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
7380         passing arrays in registers are the same as for structs, so remove the
7381         special case for them.
7382
7383 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
7384
7385         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
7386         destination type, extract only the valid bits if the source type is not
7387         integral and has a different mode.
7388
7389 2014-02-19  Richard Biener  <rguenther@suse.de>
7390
7391         PR ipa/60243
7392         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
7393         for all calls.
7394
7395 2014-02-19  Richard Biener  <rguenther@suse.de>
7396
7397         PR ipa/60243
7398         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
7399         (ipa_modify_call_arguments): Emit an argument load explicitely and
7400         preserve virtual SSA form there and for the replacement call.
7401         Do not update SSA form nor free dominance info.
7402
7403 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7404
7405         * ipa.c (function_and_variable_visibility): Also clear WEAK
7406         flag when disolving COMDAT_GROUP.
7407
7408 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7409
7410         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
7411         * ipa-prop.c (ipa_set_jf_known_type): Return early when
7412         not devirtualizing.
7413         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
7414         do more sanity checks.
7415         (detect_type_change): Return true when giving up early.
7416         (compute_complex_assign_jump_func): Fix type parameter of
7417         ipa_set_ancestor_jf.
7418         (compute_complex_ancestor_jump_func): Likewise.
7419         (update_jump_functions_after_inlining): Fix updating of
7420         ancestor function.
7421         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
7422
7423 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
7424
7425         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
7426         inline clones when edge disappears.
7427
7428 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7429
7430         PR target/60203
7431         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
7432         Split 64-bit moves into 2 patterns.  Do not allow the use of
7433         direct move for TDmode in little endian, since the decimal value
7434         has little endian bytes within a word, but the 64-bit pieces are
7435         ordered in a big endian fashion, and normal subreg's of TDmode are
7436         not allowed.
7437         (mov<mode>_64bit_dm): Likewise.
7438         (movtd_64bit_nodm): Likewise.
7439
7440 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
7441
7442         PR tree-optimization/60174
7443         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
7444         statement of an SSA_NAME that occurs in an abnormal PHI node.
7445
7446 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
7447
7448         PR sanitizer/60142
7449         * final.c (SEEN_BB): Remove.
7450         (SEEN_NOTE, SEEN_EMITTED): Renumber.
7451         (final_scan_insn): Don't force_source_line on second
7452         NOTE_INSN_BASIC_BLOCK.
7453
7454 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
7455
7456         PR target/60205
7457         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
7458         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
7459         (type_natural_mode): Warn ABI change when %zmm register is not
7460         available for AVX512F vector value passing.
7461
7462 2014-02-18  Kai Tietz  <ktietz@redhat.com>
7463
7464         PR target/60193
7465         * config/i386/i386.c (ix86_expand_prologue): Use value in
7466         rax register as displacement when restoring %r10 or %rax.
7467         Fix wrong offset when restoring both registers.
7468
7469 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
7470
7471         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
7472         assertion with conditional return.
7473
7474 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
7475             Uros Bizjak  <ubizjak@gmail.com>
7476
7477         PR driver/60233
7478         * config/i386/driver-i386.c (host_detect_local_cpu): If
7479         YMM state is not saved by the OS, also clear has_f16c.  Move
7480         CPUID 0x80000001 handling before YMM state saving checking.
7481
7482 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
7483
7484         PR rtl-optimization/58960
7485         * haifa-sched.c (alloc_global_sched_pressure_data): New,
7486         factored out from ...
7487         (sched_init): ... here.
7488         (free_global_sched_pressure_data): New, factored out from ...
7489         (sched_finish): ... here.
7490         * sched-int.h (free_global_sched_pressure_data): Declare.
7491         * sched-rgn.c (nr_regions_initial): New static global.
7492         (haifa_find_rgns): Initialize it.
7493         (schedule_region): Disable sched-pressure for the newly
7494         generated regions.
7495
7496 2014-02-17  Richard Biener  <rguenther@suse.de>
7497
7498         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
7499         release SSA defs of pattern stmts.
7500
7501 2014-02-17  Richard Biener  <rguenther@suse.de>
7502
7503         * tree-inline.c (expand_call_inline): Release the virtual
7504         operand defined by the call we are about to inline.
7505
7506 2014-02-17  Richard Biener  <rguenther@suse.de>
7507
7508         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
7509
7510 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
7511             Ilya Tocar  <ilya.tocar@intel.com>
7512
7513         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
7514         arguments order in builtin.
7515         (_mm512_permutexvar_epi64): Ditto.
7516         (_mm512_mask_permutexvar_epi64): Ditto
7517         (_mm512_maskz_permutexvar_epi32): Ditto
7518         (_mm512_permutexvar_epi32): Ditto
7519         (_mm512_mask_permutexvar_epi32): Ditto
7520
7521 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7522
7523         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
7524         (p8_vmrgow): Likewise.
7525
7526 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7527
7528         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
7529         endian targets.
7530
7531 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
7532
7533         PR target/60203
7534         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
7535         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
7536         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
7537         using direct move instructions on ISA 2.07.  Also adjust
7538         instruction length for 64-bit.
7539         (mov<mode>_64bit, TFmode/TDmode): Likewise.
7540         (mov<mode>_32bit, TFmode/TDmode): Likewise.
7541
7542 2014-02-15  Alan Modra  <amodra@gmail.com>
7543
7544         PR target/58675
7545         PR target/57935
7546         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
7547         find_replacement on parts of insn rtl that might be reloaded.
7548
7549 2014-02-15  Richard Biener  <rguenther@suse.de>
7550
7551         PR tree-optimization/60183
7552         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
7553         (tree_ssa_phiprop): Calculate and free post-dominators.
7554
7555 2014-02-14  Jeff Law  <law@redhat.com>
7556
7557         PR rtl-optimization/60131
7558         * ree.c (get_extended_src_reg): New function.
7559         (combine_reaching_defs): Use it rather than assuming location of REG.
7560         (find_and_remove_re): Verify first operand of extension is
7561         a REG before adding the insns to the copy list.
7562
7563 2014-02-14  Roland McGrath  <mcgrathr@google.com>
7564
7565         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
7566         * configure: Regenerated.
7567         * config.in: Regenerated.
7568         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
7569         instead of ASM_SHORT.
7570
7571 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
7572             Richard Earnshaw  <rearnsha@arm.com>
7573
7574         PR rtl-optimization/59535
7575         * lra-constraints.c (process_alt_operands): Encourage alternative
7576         when unassigned pseudo class is superset of the alternative class.
7577         (inherit_reload_reg): Don't inherit when optimizing for code size.
7578         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
7579         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
7580         modes not less than 4 for Thumb1.
7581
7582 2014-02-14  Kyle McMartin  <kyle@redhat.com>
7583
7584         PR pch/60010
7585         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
7586
7587 2014-02-14  Richard Biener  <rguenther@suse.de>
7588
7589         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
7590         (get_frame_arg): Drop the assert with langhook types_compatible_p.
7591         Do not strip INDIRECT_REFs.
7592
7593 2014-02-14  Richard Biener  <rguenther@suse.de>
7594
7595         PR lto/60179
7596         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
7597         DECL_FUNCTION_SPECIFIC_TARGET.
7598         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
7599         * tree-streamer-out.c (pack_ts_target_option): Remove.
7600         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
7601         (write_ts_function_decl_tree_pointers): Do not stream
7602         DECL_FUNCTION_SPECIFIC_TARGET.
7603         * tree-streamer-in.c (unpack_ts_target_option): Remove.
7604         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
7605         (lto_input_ts_function_decl_tree_pointers): Do not stream
7606         DECL_FUNCTION_SPECIFIC_TARGET.
7607
7608 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
7609
7610         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
7611         (get_initial_def_for_induction, vectorizable_induction): Ignore
7612         debug stmts when looking for exit_phi.
7613         (vectorizable_live_operation): Fix up condition.
7614
7615 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7616
7617         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
7618         nreverse() because it changes the content of original tree list.
7619
7620 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7621
7622         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
7623         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
7624
7625 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7626
7627         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
7628         GNU coding standards.
7629
7630 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7631
7632         PR debug/60152
7633         * dwarf2out.c (gen_subprogram_die): Don't call
7634         add_calling_convention_attribute if subr_die is old_die.
7635
7636 2014-02-13  Sharad Singhai  <singhai@google.com>
7637
7638         * doc/optinfo.texi: Fix order of nodes.
7639
7640 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
7641
7642         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
7643         operands[2], not operands[3].
7644
7645 2014-02-13  Richard Biener  <rguenther@suse.de>
7646
7647         PR bootstrap/59878
7648         * doc/install.texi (ISL): Update recommended version to 0.12.2,
7649         mention the possibility of an in-tree build.
7650         (CLooG): Update recommended version to 0.18.1, mention the
7651         possibility of an in-tree build and clarify that the ISL
7652         bundled with CLooG does not work.
7653
7654 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7655
7656         PR target/43546
7657         * expr.c (compress_float_constant): If x is a hard register,
7658         extend into a pseudo and then move to x.
7659
7660 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7661
7662         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
7663         caused by bad second argument to warning_at() with -mhotpatch and
7664         nested functions (e.g. with gfortran).
7665
7666 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
7667
7668         * opts.c (option_name): Remove "enabled by default" rider.
7669
7670 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
7671
7672         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
7673
7674 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
7675             Uros Bizjak  <ubizjak@gmail.com>
7676
7677         PR target/60151
7678         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
7679         * configure: Regenerated.
7680
7681 2014-02-12  Richard Biener  <rguenther@suse.de>
7682
7683         * vec.c (vec_prefix::calculate_allocation): Move as
7684         inline variant to vec.h.
7685         (vec_prefix::calculate_allocation_1): New out-of-line version.
7686         * vec.h (vec_prefix::calculate_allocation_1): Declare.
7687         (vec_prefix::m_has_auto_buf): Rename to ...
7688         (vec_prefix::m_using_auto_storage): ... this.
7689         (vec_prefix::calculate_allocation): Inline the easy cases
7690         and dispatch to calculate_allocation_1 which doesn't need the
7691         prefix address.
7692         (va_heap::reserve): Use gcc_checking_assert.
7693         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
7694         m_using_auto_storage.
7695         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
7696         member and adjust.
7697         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
7698         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
7699         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
7700
7701 2014-02-12  Richard Biener  <rguenther@suse.de>
7702
7703         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
7704         when we found a dependence.
7705
7706 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
7707
7708         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
7709         common code...
7710         (maybe_fold_stmt): ... into this new function.
7711         * omp-low.c (lower_omp): Update comment.
7712
7713         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
7714         last use.
7715
7716         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
7717         dereference.
7718
7719 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
7720
7721         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
7722         identifiers in comments.
7723         (cortexa53_extra_costs): Likewise.
7724         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
7725         (cortexa7_extra_costs): Likewise.
7726         (cortexa12_extra_costs): Likewise.
7727         (cortexa15_extra_costs): Likewise.
7728         (v7m_extra_costs): Likewise.
7729
7730 2014-02-12  Richard Biener  <rguenther@suse.de>
7731
7732         PR middle-end/60092
7733         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
7734         of posix_memalign being successful.
7735         (lower_stmt): Restrict lowering of posix_memalign to when
7736         -ftree-bit-ccp is enabled.
7737
7738 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
7739
7740         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
7741         arg_loc.
7742         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
7743
7744 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
7745
7746         PR rtl-optimization/60116
7747         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
7748         other_insn once the combination has been validated.
7749
7750 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
7751
7752         PR lto/59468
7753         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
7754         and wrapper.
7755         * ipa-devirt.c: Include demangle.h
7756         (odr_violation_reported): New static variable.
7757         (add_type_duplicate): Update odr_violations.
7758         (maybe_record_node): Add completep parameter; update it.
7759         (record_target_from_binfo): Add COMPLETEP parameter;
7760         update it as needed.
7761         (possible_polymorphic_call_targets_1): Likewise.
7762         (struct polymorphic_call_target_d): Add nonconstruction_targets;
7763         rename FINAL to COMPLETE.
7764         (record_targets_from_bases): Sanity check we found the binfo;
7765         fix COMPLETEP updating.
7766         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
7767         parameter, fix computing of COMPLETEP.
7768         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
7769         at LTO time do demangling.
7770         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
7771         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
7772         parameter.
7773         (gimple_get_virt_method_for_binfo): Likewise.
7774         * gimple-fold.h (gimple_get_virt_method_for_binfo,
7775         gimple_get_virt_method_for_vtable): Update prototypes.
7776
7777 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
7778
7779         PR target/49008
7780         * genautomata.c (add_presence_absence): Fix typo with
7781         {final_}presence_list.
7782
7783 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7784
7785         PR target/60137
7786         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
7787         for VSX/Altivec vectors that land in GPR registers.
7788
7789 2014-02-11  Richard Henderson  <rth@redhat.com>
7790             Jakub Jelinek  <jakub@redhat.com>
7791
7792         PR debug/59776
7793         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
7794         around drhs if type conversion to lacc->type is not useless.
7795
7796 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7797
7798         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
7799         tuning struct.
7800         (cortex-a57.cortex-a53): Likewise.
7801         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
7802
7803 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7804
7805         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
7806         arm_restrict_it.
7807
7808 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
7809
7810         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
7811         add_options_for_arm_vfp3.
7812
7813 2014-02-11  Jeff Law  <law@redhat.com>
7814
7815         PR middle-end/54041
7816         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
7817         object with an undesirable mode.
7818
7819 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7820
7821         PR libgomp/60107
7822         * config/i386/sol2-9.h: New file.
7823         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
7824         *-*-solaris2.9*): Use it.
7825
7826 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7827
7828         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
7829         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
7830
7831 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7832
7833         * config/microblaze/microblaze.c: Extend mcpu version format
7834
7835 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
7836
7837         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
7838
7839 2014-02-10  Richard Henderson  <rth@redhat.com>
7840
7841         PR target/59927
7842         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
7843         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
7844         ms-abi vs -mno-accumulate-outgoing-args.
7845         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
7846         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
7847         respect to ms-abi.
7848
7849 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7850
7851         PR middle-end/60080
7852         * cfgexpand.c (expand_asm_operands): Attach source location to
7853         ASM_INPUT rtx objects.
7854         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
7855
7856 2014-02-10  Nick Clifton  <nickc@redhat.com>
7857
7858         * config/mn10300/mn10300.c (popcount): New function.
7859         (mn10300_expand_prologue): Include saved registers in stack usage
7860         count.
7861
7862 2014-02-10  Jeff Law  <law@redhat.com>
7863
7864         PR middle-end/52306
7865         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
7866         when changing the SET_DEST of a prior insn to avoid an input reload.
7867
7868 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7869
7870         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
7871         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
7872         -mcall-openbsd, or -mcall-linux.
7873         (CC1_ENDIAN_BIG_SPEC): Remove.
7874         (CC1_ENDIAN_LITTLE_SPEC): Remove.
7875         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7876         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
7877         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
7878         and %cc1_endian_default.
7879         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7880
7881 2014-02-10  Richard Biener  <rguenther@suse.de>
7882
7883         PR tree-optimization/60115
7884         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
7885         MEM_REF handling.  Properly verify that the accesses are not
7886         out of the objects bound.
7887
7888 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7889
7890         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
7891         coretex to cortex.
7892
7893 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
7894
7895         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
7896         proper constants and fix formatting.
7897         (possible_polymorphic_call_targets): Fix formatting.
7898
7899 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
7900             Ilya Tocar  <ilya.tocar@intel.com>
7901
7902         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
7903         (_mm512_loadu_epi32): Renamed into...
7904         (_mm512_loadu_si512): This.
7905         (_mm512_storeu_epi32): Renamed into...
7906         (_mm512_storeu_si512): This.
7907         (_mm512_maskz_ceil_ps): Removed.
7908         (_mm512_maskz_ceil_pd): Ditto.
7909         (_mm512_maskz_floor_ps): Ditto.
7910         (_mm512_maskz_floor_pd): Ditto.
7911         (_mm512_floor_round_ps): Ditto.
7912         (_mm512_floor_round_pd): Ditto.
7913         (_mm512_ceil_round_ps): Ditto.
7914         (_mm512_ceil_round_pd): Ditto.
7915         (_mm512_mask_floor_round_ps): Ditto.
7916         (_mm512_mask_floor_round_pd): Ditto.
7917         (_mm512_mask_ceil_round_ps): Ditto.
7918         (_mm512_mask_ceil_round_pd): Ditto.
7919         (_mm512_maskz_floor_round_ps): Ditto.
7920         (_mm512_maskz_floor_round_pd): Ditto.
7921         (_mm512_maskz_ceil_round_ps): Ditto.
7922         (_mm512_maskz_ceil_round_pd): Ditto.
7923         (_mm512_expand_pd): Ditto.
7924         (_mm512_expand_ps): Ditto.
7925         * config/i386/i386.c (ix86_builtins): Remove
7926         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
7927         (bdesc_args): Ditto.
7928         * config/i386/predicates.md (const1256_operand): New.
7929         (const_1_to_2_operand): Ditto.
7930         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
7931         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
7932         (*avx512pf_gatherpf<mode>sf): Ditto.
7933         (avx512pf_gatherpf<mode>df): Ditto.
7934         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7935         (*avx512pf_gatherpf<mode>df): Ditto.
7936         (avx512pf_scatterpf<mode>sf): Ditto.
7937         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7938         (*avx512pf_scatterpf<mode>sf): Ditto.
7939         (avx512pf_scatterpf<mode>df): Ditto.
7940         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7941         (*avx512pf_scatterpf<mode>df): Ditto.
7942         (avx512f_expand<mode>): Removed.
7943         (<shift_insn><mode>3<mask_name>): Change predicate type.
7944
7945 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7946
7947         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
7948         not at the end of datarefs vector use ordered_remove to avoid
7949         reordering datarefs vector.
7950
7951         PR c/59984
7952         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
7953         mark local addressable non-static vars as GOVD_PRIVATE
7954         instead of GOVD_LOCAL.
7955         * omp-low.c (lower_omp_for): Move gimple_bind_vars
7956         and BLOCK_VARS of gimple_bind_block to new_stmt rather
7957         than copying them.
7958
7959         PR middle-end/60092
7960         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
7961         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
7962         assume_aligned or alloc_align attributes.
7963         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
7964         arguments.  Handle also assume_aligned and alloc_align attributes.
7965         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
7966         calls to functions with assume_aligned or alloc_align attributes.
7967         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
7968
7969 2014-02-08  Terry Guo  <terry.guo@arm.com>
7970
7971         * doc/invoke.texi: Document ARM -march=armv7e-m.
7972
7973 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7974
7975         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
7976         flag on __cilkrts_rethrow builtin.
7977
7978         PR ipa/60026
7979         * ipa-cp.c (determine_versionability): Fail at -O0
7980         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
7981         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
7982
7983         Revert:
7984         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7985
7986         PR ipa/60026
7987         * tree-inline.c (copy_forbidden): Fail for
7988         __attribute__((optimize (0))) functions.
7989
7990 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7991
7992         * varpool.c: Include pointer-set.h.
7993         (varpool_remove_unreferenced_decls): Variables in other partitions
7994         will not be output; be however careful to not lose information
7995         about partitioning.
7996
7997 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7998
7999         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
8000         lookup in the vtable constructor.
8001
8002 2014-02-07  Jeff Law  <law@redhat.com>
8003
8004         PR target/40977
8005         * config/m68k/m68k.md (ashldi_extsi): Turn into a
8006         define_insn_and_split.
8007
8008         * ipa-inline.c (inline_small_functions): Fix typos.
8009
8010 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8011
8012         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
8013         (s390_can_use_return_insn): Declare.
8014         * config/s390/s390.h (EPILOGUE_USES): Define.
8015         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
8016         instructions.
8017         (s390_chunkify_start): Handle return JUMP_LABELs.
8018         (s390_early_mach): Emit a main_pool instruction on the entry edge.
8019         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
8020         (s390_can_use_return_insn): New functions.
8021         (s390_fix_long_loop_prediction): Handle conditional returns.
8022         (TARGET_SET_UP_BY_PROLOGUE): Define.
8023         * config/s390/s390.md (ANY_RETURN): New code iterator.
8024         (*creturn, *csimple_return, return, simple_return): New patterns.
8025
8026 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8027
8028         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
8029         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
8030         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
8031         REG_CFA_RESTORE list when deciding not to restore a register.
8032
8033 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8034
8035         * config/s390/s390.c: Include tree-pass.h and context.h.
8036         (s390_early_mach): New function, split out from...
8037         (s390_emit_prologue): ...here.
8038         (pass_data_s390_early_mach): New pass structure.
8039         (pass_s390_early_mach): New class.
8040         (s390_option_override): Create and register early_mach pass.
8041         Move to end of file.
8042
8043 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8044
8045         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
8046         to match for the exit block.
8047
8048 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8049
8050         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
8051         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
8052         Reject misaligned operands.
8053
8054 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8055
8056         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
8057
8058 2014-02-07  Richard Biener  <rguenther@suse.de>
8059
8060         PR middle-end/60092
8061         * gimple-low.c (lower_builtin_posix_memalign): New function.
8062         (lower_stmt): Call it to lower posix_memalign in a way
8063         to make alignment info accessible.
8064
8065 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
8066
8067         PR c++/60082
8068         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
8069         __builtin_setjmp_receiver.
8070
8071 2014-02-07  Richard Biener  <rguenther@suse.de>
8072
8073         PR middle-end/60092
8074         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
8075         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
8076         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
8077         Handle BUILT_IN_POSIX_MEMALIGN.
8078         (find_func_clobbers): Likewise.
8079         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8080         (call_may_clobber_ref_p_1): Likewise.
8081
8082 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8083
8084         PR ipa/59918
8085         * ipa-devirt.c (record_target_from_binfo): Remove overactive
8086         sanity check.
8087
8088 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8089
8090         PR ipa/59469
8091         * lto-cgraph.c (lto_output_node): Use
8092         symtab_get_symbol_partitioning_class.
8093         (lto_output_varpool_node): likewise.
8094         (symtab_get_symbol_partitioning_class): Move here from
8095         lto/lto-partition.c
8096         * cgraph.h (symbol_partitioning_class): Likewise.
8097         (symtab_get_symbol_partitioning_class): Declare.
8098
8099 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8100
8101         * ggc.h (ggc_internal_cleared_alloc): New macro.
8102         * vec.h (vec_safe_copy): Handle memory stats.
8103         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
8104         * target-globals.c (save_target_globals): Likewise.
8105
8106 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
8107
8108         PR target/60077
8109         * expr.c (emit_move_resolve_push): Export; be bit more selective
8110         on when to clear alias set.
8111         * expr.h (emit_move_resolve_push): Declare.
8112         * function.h (struct function): Add tail_call_marked.
8113         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
8114         * config/i386/i386-protos.h (ix86_expand_push): Remove.
8115         * config/i386/i386.md (TImode move expander): De not call
8116         ix86_expand_push.
8117         (FP push expanders): Preserve memory attributes.
8118         * config/i386/sse.md (push<mode>1): Remove.
8119         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
8120         (ix86_expand_push): Remove.
8121         * config/i386/mmx.md (push<mode>1): Remove.
8122
8123 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8124
8125         PR rtl-optimization/60030
8126         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
8127         lopart with paradoxical subreg before shifting it up by hprec.
8128
8129 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8130
8131         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
8132         Remove extra newline at end of file.
8133         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
8134         (arm_issue_rate): Handle cortexa57.
8135         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
8136         (cortex-a57.cortex-a53): Likewise.
8137
8138 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8139
8140         PR target/59575
8141         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
8142         don't record in REG_FRAME_RELATED_EXPR registers not set in that
8143         bitmask.
8144         (arm_expand_prologue): Adjust all callers.
8145         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
8146         info, registers also at the lowest numbered registers side.  Use
8147         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
8148         XEXP.
8149
8150         PR debug/59992
8151         * var-tracking.c (adjust_mems): Before adding a SET to
8152         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
8153
8154 2014-02-06  Alan Modra  <amodra@gmail.com>
8155
8156         PR target/60032
8157         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
8158         change SDmode to DDmode when lra_in_progress.
8159
8160 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
8161
8162         PR middle-end/59150
8163         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
8164         free_data_ref on the dr first, and before goto again also set dr
8165         to the next dr.  For simd_lane_access, free old datarefs[i] before
8166         overwriting it.  For get_vectype_for_scalar_type failure, don't
8167         free_data_ref if simd_lane_access.
8168
8169         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
8170
8171         PR target/60062
8172         * tree.h (opts_for_fn): New inline function.
8173         (opt_for_fn): Define.
8174         * config/i386/i386.c (ix86_function_regparm): Use
8175         opt_for_fn (decl, optimize) instead of optimize.
8176
8177 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8178
8179         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
8180         for SYMBOL_REF in large memory model.
8181
8182 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8183
8184         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
8185         and crypto support.
8186         (cortex-a57): Likewise.
8187         (cortex-a57.cortex-a53): Likewise.
8188
8189 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
8190             Kugan Vivekanandarajah  <kuganv@linaro.org>
8191
8192         * config/arm/arm.c (arm_vector_alignment_reachable): Check
8193         unaligned_access.
8194         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
8195
8196 2014-02-06  Richard Biener  <rguenther@suse.de>
8197
8198         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
8199         set_loop_copy and initialize_original_copy_tables.
8200
8201 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
8202
8203         * config/aarch64/aarch64-simd.md
8204         (aarch64_ashr_simddi): Change QI to SI.
8205
8206 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
8207             Jakub Jelinek  <jakub@redhat.com>
8208
8209         PR middle-end/60013
8210         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
8211         of the dataflow.
8212
8213 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8214
8215         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
8216         CODE_FOR_altivec_vpku[hw]um to
8217         CODE_FOR_altivec_vpku[hw]um_direct.
8218         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
8219         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
8220         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
8221         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
8222
8223 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8224
8225         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
8226         generation for -maltivec=be.
8227         (altivec_vsumsws): Simplify redundant test.
8228
8229 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8230
8231         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
8232         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
8233         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
8234         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
8235         gen_altivec_vpkuwum.
8236         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
8237         BYTES_BIG_ENDIAN.
8238         (altivec_vpks<VI_char>ss): Likewise.
8239         (altivec_vpks<VI_char>us): Likewise.
8240         (altivec_vpku<VI_char>us): Likewise.
8241         (altivec_vpku<VI_char>um): Likewise.
8242         (altivec_vpku<VI_char>um_direct): New (copy of
8243         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
8244         internal use).
8245         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
8246         target is little endian and -maltivec=be is not specified.
8247         (*altivec_vupkhs<VU_char>_direct): New (copy of
8248         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
8249         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
8250         target is little endian and -maltivec=be is not specified.
8251         (*altivec_vupkls<VU_char>_direct): New (copy of
8252         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
8253         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
8254         little endian and -maltivec=be is not specified.
8255         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
8256         little endian and -maltivec=be is not specified.
8257
8258 2014-02-05  Richard Henderson  <rth@redhat.com>
8259
8260         PR debug/52727
8261         * combine-stack-adj.c: Revert r206943.
8262         * sched-int.h (struct deps_desc): Add last_args_size.
8263         * sched-deps.c (init_deps): Initialize it.
8264         (sched_analyze_insn): Add OUTPUT dependencies between insns that
8265         contain REG_ARGS_SIZE notes.
8266
8267 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
8268
8269         * lto-cgraph.c (asm_nodes_output): Make global.
8270         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
8271         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
8272         (driver_handle_option): Handle OPT_fwpa.
8273
8274 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
8275
8276         PR ipa/59947
8277         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
8278         a comment typo and formatting issue.  If odr_hash hasn't been
8279         created, return vNULL and set *completep to false.
8280
8281         PR middle-end/57499
8282         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
8283         bb with no successors.
8284
8285 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
8286
8287         PR target/59718
8288         * doc/invoke.texi (-march): Clarify documentation for ARM.
8289         (-mtune): Likewise.
8290         (-mcpu): Likewise.
8291
8292 2014-02-05  Richard Biener  <rguenther@suse.de>
8293
8294         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
8295         when not vectorizing because of too many alias checks.
8296         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
8297         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
8298
8299 2014-02-05  Nick Clifton  <nickc@redhat.com>
8300
8301         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
8302         accept extended registers in any mode when compiling for the MN10300.
8303
8304 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
8305
8306         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
8307         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
8308         sanitization attributes.
8309         (can_inline_edge_p): Likewise.
8310         (sanitize_attrs_match_for_inline_p): New function.
8311
8312 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8313
8314         * ipa-prop.c (detect_type_change): Shor circuit testing of
8315         type changes on THIS pointer.
8316
8317 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
8318
8319         PR target/59777
8320         * config/pa/pa.c (legitimize_tls_address): Return original address
8321         if not passed a SYMBOL_REF rtx.
8322         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
8323         addresses.
8324         (pa_emit_move_sequence): Simplify TLS source operands.
8325         (pa_legitimate_constant_p): Reject all TLS constants.
8326         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
8327         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
8328
8329 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8330
8331         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
8332         groups when we know they are controlled by LTO.
8333         * varasm.c (default_binds_local_p_1): If object is in other partition,
8334         it will be resolved locally.
8335
8336 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8337
8338         * config/host-linux.c (linux_gt_pch_use_address): Don't
8339         use SSIZE_MAX because it is not always defined.
8340
8341 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
8342
8343         PR bootstrap/59913
8344         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
8345         threshold for pseudo splitting.
8346         (update_ebb_live_info): Process call argument hard registers and
8347         hard registers from insn definition too.
8348         (max_small_class_regs_num): New constant.
8349         (inherit_in_ebb): Update live hard regs through EBBs.  Update
8350         reloads_num only for small register classes.  Don't split for
8351         outputs of jumps.
8352
8353 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
8354
8355         PR ipa/60058
8356         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
8357         is non-null.
8358
8359 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
8360
8361         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
8362         visibility is safe.
8363
8364 2014-02-04  Marek Polacek  <polacek@redhat.com>
8365
8366         * gdbinit.in (pel): Define.
8367
8368 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8369
8370         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
8371         behavior.
8372
8373 2014-02-04  Richard Biener  <rguenther@suse.de>
8374
8375         PR lto/59723
8376         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
8377         in function context local.
8378         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
8379         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
8380         similar to LTO_imported_decl_ref.
8381
8382 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
8383
8384         PR tree-optimization/60002
8385         * cgraphclones.c (build_function_decl_skip_args): Clear
8386         DECL_LANG_SPECIFIC.
8387
8388         PR tree-optimization/60023
8389         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
8390         false to gsi_replace.
8391         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
8392         has been in some EH region and vec_stmt could throw, add
8393         vec_stmt into the same EH region.
8394         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
8395         has no lhs, ignore it.
8396         * internal-fn.c (expand_MASK_LOAD): Likewise.
8397
8398         PR ipa/60026
8399         * tree-inline.c (copy_forbidden): Fail for
8400         __attribute__((optimize (0))) functions.
8401
8402         PR other/58712
8403         * omp-low.c (simd_clone_struct_copy): If from->inbranch
8404         is set, copy one less argument.
8405         (expand_simd_clones): Don't subtract clone_info->inbranch
8406         from simd_clone_struct_alloc argument.
8407
8408         PR rtl-optimization/57915
8409         * recog.c (simplify_while_replacing): If all unary/binary/relational
8410         operation arguments are constant, attempt to simplify those.
8411
8412         PR middle-end/59261
8413         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
8414         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
8415
8416 2014-02-04  Richard Biener  <rguenther@suse.de>
8417
8418         PR tree-optimization/60012
8419         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
8420         TBAA disambiguation to all DDRs.
8421
8422 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8423
8424         PR target/59788
8425         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
8426         (LINK_SPEC): Use it for -shared, -shared-libgcc.
8427
8428 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8429
8430         PR ipa/59882
8431         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
8432
8433 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8434
8435         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
8436         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
8437
8438 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8439
8440         PR ipa/59831
8441         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
8442         to figure out targets of polymorphic calls with known decl.
8443         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8444         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
8445         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
8446         (get_polymorphic_call_info): ... here.
8447         (get_polymorphic_call_info_from_invariant): New function.
8448
8449 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8450
8451         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
8452         lookup via vtable pointer; check for type consistency
8453         and turn inconsitent facts into UNREACHABLE.
8454         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8455         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
8456         type inconsistent querries; return UNREACHABLE instead.
8457
8458 2014-02-03  Richard Henderson  <rth@twiddle.net>
8459
8460         PR tree-opt/59924
8461         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
8462         already processed this node.
8463         (normalize_one_pred_1): Pass along mark_set.
8464         (normalize_one_pred): Create and destroy a pointer_set_t.
8465         (normalize_one_pred_chain): Likewise.
8466
8467 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
8468
8469         PR gcov-profile/58602
8470         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
8471
8472 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
8473
8474         PR ipa/59831
8475         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
8476         -fno-devirtualize; try to devirtualize by the knowledge of
8477         virtual table pointer given by aggregate propagation.
8478         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
8479         (ipa_print_node_jump_functions): Dump also offset that
8480         is relevant for polymorphic calls.
8481         (determine_known_aggregate_parts): Add arg_type parameter; use it
8482         instead of determining the type from pointer type.
8483         (ipa_compute_jump_functions_for_edge): Update call of
8484         determine_known_aggregate_parts.
8485         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
8486         (gimple_get_virt_method_for_binfo): ... here; simplify using
8487         vtable_pointer_value_to_vtable.
8488         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
8489         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
8490         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
8491         (vtable_pointer_value_to_vtable): Break out from ...; handle also
8492         POINTER_PLUS_EXPR.
8493         (vtable_pointer_value_to_binfo): ... here.
8494         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
8495
8496 2014-02-03  Teresa Johnson  <tejohnson@google.com>
8497
8498         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
8499         redef of outer loop index variable.
8500
8501 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
8502
8503         PR c++/53017
8504         PR c++/59211
8505         * doc/extend.texi (Function Attributes): Typo.
8506
8507 2014-02-03  Cong Hou  <congh@google.com>
8508
8509         PR tree-optimization/60000
8510         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
8511         if the vectorized statement is a store.  A store statement can only
8512         appear at the end of pattern statements.
8513
8514 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
8515
8516         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
8517         (ix86_option_override_internal): Default long double to 64-bit for
8518         32-bit Bionic and to 128-bit for 64-bit Bionic.
8519
8520         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
8521         TARGET_LONG_DOUBLE_128 is true.
8522         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
8523
8524         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
8525         (mlong-double-64): Negate -mlong-double-128.
8526         (mlong-double-128): New option.
8527
8528         * config/i386/i386-c.c (ix86_target_macros): Define
8529         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
8530
8531         * doc/invoke.texi: Document -mlong-double-128.
8532
8533 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
8534
8535         PR rtl-optimization/60024
8536         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
8537
8538 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
8539
8540         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
8541
8542 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8543
8544         PR rtl-optimization/57662
8545         * sel-sched.c (code_motion_path_driver): Do not mark already not
8546         existing blocks in the visiting bitmap.
8547
8548 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8549
8550         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
8551         on the insn being emitted.
8552
8553 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
8554             Will Deacon  <will.deacon@arm.com>
8555
8556         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
8557
8558 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8559
8560         * config/arm/arm-tables.opt: Regenerate.
8561
8562 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8563
8564         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
8565         for vector types other than V16QImode.
8566         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
8567         define_expand, and call altivec_expand_vec_perm_le when producing
8568         code with little endian element order.
8569         (*altivec_vperm_<mode>_internal): New insn having previous
8570         behavior of altivec_vperm_<mode>.
8571         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
8572         altivec_expand_vec_perm_le when producing code with little endian
8573         element order.
8574         (*altivec_vperm_<mode>_uns_internal): New insn having previous
8575         behavior of altivec_vperm_<mode>_uns.
8576
8577 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8578
8579         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
8580         (altivec_vsumsws): Add handling for -maltivec=be with a little
8581         endian target.
8582         (altivec_vsumsws_direct): New.
8583         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
8584         gen_altivec_vsumsws.
8585
8586 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8587
8588         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
8589         vtable_pointer_value_to_binfo): New functions.
8590         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
8591         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
8592
8593 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
8594
8595         * config/nios2/nios2.md (load_got_register): Initialize GOT
8596         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
8597         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
8598
8599 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8600
8601         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
8602         preserverd by passthrough, do not propagate the type.
8603
8604 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8605
8606         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
8607         (mips_atomic_assign_expand_fenv): New function.
8608         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
8609
8610 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8611
8612         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
8613         (__builtin_mips_set_fcsr): Likewise.
8614         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
8615         MIPS_USI_FTYPE_VOID.
8616         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
8617         (mips16_expand_set_fcsr): Likewise.
8618         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
8619         (mips16_set_fcsr_stub): Likewise.
8620         (mips16_get_fcsr_one_only_stub): New class.
8621         (mips16_set_fcsr_one_only_stub): Likewise.
8622         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
8623         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
8624         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
8625         (hard_float): New availability predicate.
8626         (mips_builtins): Add get_fcsr and set_fcsr.
8627         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
8628         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
8629         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
8630         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
8631         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
8632         patterns.
8633
8634 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8635
8636         * config/mips/mips.c (mips_one_only_stub): New class.
8637         (mips_need_mips16_rdhwr_p): Replace with...
8638         (mips16_rdhwr_stub): ...this new variable.
8639         (mips16_stub_call_address): New function.
8640         (mips16_rdhwr_one_only_stub): New class.
8641         (mips_expand_thread_pointer): Use mips16_stub_call_address.
8642         (mips_output_mips16_rdhwr): Delete.
8643         (mips_finish_stub): New function.
8644         (mips_code_end): Use it to handle rdhwr stubs.
8645
8646 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
8647
8648         PR target/60017
8649         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
8650         when calculating size of integer atomic types.
8651
8652 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
8653
8654         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
8655
8656 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
8657
8658         PR tree-optimization/60003
8659         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
8660         * profile.c (branch_prob): Use gimple_call_builtin_p
8661         to check for BUILT_IN_SETJMP_RECEIVER.
8662         * tree-inline.c (copy_bb): Call notice_special_calls.
8663
8664 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
8665
8666         PR bootstrap/59985
8667         * lra-constraints.c (process_alt_operands): Update reload_sum only
8668         on the first pass.
8669
8670 2014-01-31  Richard Henderson  <rth@redhat.com>
8671
8672         PR middle-end/60004
8673         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
8674         until after else_eh is processed.
8675
8676 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8677
8678         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
8679         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
8680         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
8681         in smmintrin.h, remove them.
8682         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
8683         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
8684         * config/i386/i386.md (ROUND_SAE): Fix value.
8685         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
8686         (const48_operand): New.
8687         * config/i386/subst.md (round), (round_expand): Use
8688         const_4_or_8_to_11_operand.
8689         (round_saeonly), (round_saeonly_expand): Use const48_operand.
8690
8691 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8692
8693         * config/i386/constraints.md (Yk): Swap meaning with k.
8694         * config/i386/i386.md (movhi_internal): Change Yk to k.
8695         (movqi_internal): Ditto.
8696         (*k<logic><mode>): Ditto.
8697         (*andhi_1): Ditto.
8698         (*andqi_1): Ditto.
8699         (kandn<mode>): Ditto.
8700         (*<code>hi_1): Ditto.
8701         (*<code>qi_1): Ditto.
8702         (kxnor<mode>): Ditto.
8703         (kortestzhi): Ditto.
8704         (kortestchi): Ditto.
8705         (kunpckhi): Ditto.
8706         (*one_cmplhi2_1): Ditto.
8707         (*one_cmplqi2_1): Ditto.
8708         * config/i386/sse.md (): Change k to Yk.
8709         (avx512f_load<mode>_mask): Ditto.
8710         (avx512f_blendm<mode>): Ditto.
8711         (avx512f_store<mode>_mask): Ditto.
8712         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
8713         (avx512f_storedqu<mode>_mask): Ditto.
8714         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
8715         Ditto.
8716         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
8717         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
8718         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
8719         (avx512f_maskcmp<mode>3): Ditto.
8720         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
8721         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
8722         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
8723         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
8724         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
8725         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
8726         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
8727         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
8728         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
8729         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
8730         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
8731         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
8732         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
8733         (vec_extract_lo_<mode>_maskm): Ditto.
8734         (vec_extract_hi_<mode>_maskm): Ditto.
8735         (avx512f_vternlog<mode>_mask): Ditto.
8736         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
8737         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
8738         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
8739         (avx512f_<code>v8div16qi2_mask): Ditto.
8740         (avx512f_<code>v8div16qi2_mask_store): Ditto.
8741         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
8742         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
8743         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
8744         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
8745         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
8746         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8747         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
8748         (*avx512pf_scatterpf<mode>df_mask): Ditto.
8749         (avx512cd_maskb_vec_dupv8di): Ditto.
8750         (avx512cd_maskw_vec_dupv16si): Ditto.
8751         (avx512f_vpermi2var<mode>3_maskz): Ditto.
8752         (avx512f_vpermi2var<mode>3_mask): Ditto.
8753         (avx512f_vpermi2var<mode>3_mask): Ditto.
8754         (avx512f_vpermt2var<mode>3_maskz): Ditto.
8755         (*avx512f_gathersi<mode>): Ditto.
8756         (*avx512f_gathersi<mode>_2): Ditto.
8757         (*avx512f_gatherdi<mode>): Ditto.
8758         (*avx512f_gatherdi<mode>_2): Ditto.
8759         (*avx512f_scattersi<mode>): Ditto.
8760         (*avx512f_scatterdi<mode>): Ditto.
8761         (avx512f_compress<mode>_mask): Ditto.
8762         (avx512f_compressstore<mode>_mask): Ditto.
8763         (avx512f_expand<mode>_mask): Ditto.
8764         * config/i386/subst.md (mask): Change k to Yk.
8765         (mask_scalar_merge): Ditto.
8766         (sd): Ditto.
8767
8768 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
8769
8770         * doc/extend.texi (Vector Extensions): Document ?: in C++.
8771
8772 2014-01-31  Richard Biener  <rguenther@suse.de>
8773
8774         PR middle-end/59990
8775         * builtins.c (fold_builtin_memory_op): Make sure to not
8776         use a floating-point mode or a boolean or enumeral type for
8777         the copy operation.
8778
8779 2014-01-30  DJ Delorie  <dj@redhat.com>
8780
8781         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
8782         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
8783         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
8784         whenever main() has an epilogue.
8785
8786 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8787
8788         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
8789         unused variable "field".
8790         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
8791         (vsx_mergeh_<mode>): Likewise.
8792         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
8793         (altivec_vmrghh): Likewise.
8794         (altivec_vmrghw): Likewise.
8795         (altivec_vmrglb): Likewise.
8796         (altivec_vmrglh): Likewise.
8797         (altivec_vmrglw): Likewise.
8798         (altivec_vspltb): Add missing uses.
8799         (altivec_vsplth): Likewise.
8800         (altivec_vspltw): Likewise.
8801         (altivec_vspltsf): Likewise.
8802
8803 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8804
8805         PR target/59923
8806         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
8807         frame related instructions.
8808
8809 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
8810
8811         PR rtl-optimization/59959
8812         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
8813         any reload of register whose subreg is invalid.
8814
8815 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8816
8817         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
8818         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
8819         Add missing return type - void.
8820
8821 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8822
8823         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
8824         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
8825         remove element index adjustment for endian (now handled in vsx.md
8826         and altivec.md).
8827         (altivec_expand_vec_perm_const): Use
8828         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
8829         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
8830         (vsx_xxspltw_<mode>): Adjust element index for little endian.
8831         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
8832         define_expand and a new define_insn *altivec_vspltb_internal;
8833         adjust for -maltivec=be on a little endian target.
8834         (altivec_vspltb_direct): New.
8835         (altivec_vsplth): Divide into a define_expand and a new
8836         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
8837         little endian target.
8838         (altivec_vsplth_direct): New.
8839         (altivec_vspltw): Divide into a define_expand and a new
8840         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
8841         little endian target.
8842         (altivec_vspltw_direct): New.
8843         (altivec_vspltsf): Divide into a define_expand and a new
8844         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
8845         a little endian target.
8846
8847 2014-01-30  Richard Biener  <rguenther@suse.de>
8848
8849         PR tree-optimization/59993
8850         * tree-ssa-forwprop.c (associate_pointerplus): Check we
8851         can propagate form the earlier stmt and avoid the transform
8852         when the intermediate result is needed.
8853
8854 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
8855
8856         * README.Portability: Fix typo.
8857
8858 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
8859
8860         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
8861         comparison_operator with ordered_comparison_operator.
8862
8863 2014-01-30  Nick Clifton  <nickc@redhat.com>
8864
8865         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
8866         Rename to mn10300_store_multiple_regs.
8867         * config/mn10300/mn10300.c: Likewise.
8868         * config/mn10300/mn10300.md (store_movm): Fix typo: call
8869         store_multiple_regs.
8870         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
8871         Call mn10300_store_multiple_regs.
8872
8873 2014-01-30  Nick Clifton  <nickc@redhat.com>
8874             DJ Delorie  <dj@redhat.com>
8875
8876         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
8877         %fp 2 to keep registers after it properly word-aligned.
8878         (rl78_alloc_physical_registers_umul): Handle the case where both
8879         input operands are the same.
8880
8881 2014-01-30  Richard Biener  <rguenther@suse.de>
8882
8883         PR tree-optimization/59903
8884         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
8885         check properly.
8886
8887 2014-01-30  Jason Merrill  <jason@redhat.com>
8888
8889         PR c++/59633
8890         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
8891
8892         PR c++/59645
8893         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
8894
8895 2014-01-30  Richard Biener  <rguenther@suse.de>
8896
8897         PR tree-optimization/59951
8898         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
8899
8900 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
8901
8902         PR target/59784
8903         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
8904         SFmode to DFmode case.
8905
8906 2014-01-29  DJ Delorie  <dj@redhat.com>
8907
8908         * config/msp430/msp430.opt (-minrt): New.
8909         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
8910         if -minrt given.
8911         (ENDFILE_SPEC): Likewise.
8912
8913 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
8914
8915         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
8916         (estimate_function_body_sizes): Use it.
8917
8918 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
8919
8920         PR c++/58561
8921         * dwarf2out.c (is_cxx_auto): New.
8922         (is_base_type): Use it.
8923         (gen_type_die_with_usage): Likewise.
8924
8925 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8926
8927         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
8928         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
8929         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
8930         -maltivec=be with LE targets.
8931         (vsx_mergeh_<mode>): Likewise.
8932         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
8933         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
8934         (altivec_vmrghb): Replace with define_expand and new
8935         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
8936         (altivec_vmrghb_direct): New define_insn.
8937         (altivec_vmrghh): Replace with define_expand and new
8938         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
8939         (altivec_vmrghh_direct): New define_insn.
8940         (altivec_vmrghw): Replace with define_expand and new
8941         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
8942         (altivec_vmrghw_direct): New define_insn.
8943         (*altivec_vmrghsf): Adjust for endianness.
8944         (altivec_vmrglb): Replace with define_expand and new
8945         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
8946         (altivec_vmrglb_direct): New define_insn.
8947         (altivec_vmrglh): Replace with define_expand and new
8948         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
8949         (altivec_vmrglh_direct): New define_insn.
8950         (altivec_vmrglw): Replace with define_expand and new
8951         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
8952         (altivec_vmrglw_direct): New define_insn.
8953         (*altivec_vmrglsf): Adjust for endianness.
8954         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8955         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8956         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8957         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8958         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8959         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8960         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8961         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8962
8963 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8964
8965         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
8966         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
8967         whitespace.
8968
8969 2014-01-29  Richard Biener  <rguenther@suse.de>
8970
8971         PR tree-optimization/58742
8972         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
8973         associate_pointerplus_align.
8974         (associate_pointerplus_diff): New function.
8975         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
8976         and associate_pointerplus_diff.
8977
8978 2014-01-29  Richard Biener  <rguenther@suse.de>
8979
8980         * lto-streamer.h (LTO_major_version): Bump to 3.
8981         (LTO_minor_version): Reset to 0.
8982
8983 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
8984
8985         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
8986         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
8987         (arm_file_start): Generate correct asm header for armv7ve.
8988         * config/arm/bpabi.h: Add multilib support for armv7ve.
8989         * config/arm/driver-arm.c: Change the architectures of cortex-a7
8990         and cortex-a15 to armv7ve.
8991         * config/arm/t-aprofile: Add multilib support for armv7ve.
8992         * doc/invoke.texi: Document -march=armv7ve.
8993
8994 2014-01-29  Richard Biener  <rguenther@suse.de>
8995
8996         PR tree-optimization/58742
8997         * tree-ssa-forwprop.c (associate_plusminus): Return true
8998         if we changed sth, defer EH cleanup to ...
8999         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
9000         (simplify_mult): New function.
9001
9002 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
9003
9004         PR middle-end/59917
9005         PR tree-optimization/59920
9006         * tree.c (build_common_builtin_nodes): Remove
9007         __builtin_setjmp_dispatcher initialization.
9008         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
9009         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
9010         instead of gsi_after_labels + manually skipping debug stmts.
9011         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
9012         ignore bbs with IFN_ABNORMAL_DISPATCHER.
9013         * tree-inline.c (copy_edges_for_bb): Remove
9014         can_make_abnormal_goto argument, instead add abnormal_goto_dest
9015         argument.  Ignore computed_goto_p stmts.  Don't call
9016         make_abnormal_goto_edges.  If a call might need abnormal edges
9017         for non-local gotos, see if it already has an edge to
9018         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
9019         with true argument, don't do anything then, otherwise add
9020         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
9021         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
9022         caller.
9023         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
9024         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
9025         (lower_stmt): Don't set data->calls_builtin_setjmp.
9026         (lower_builtin_setjmp): Adjust comment.
9027         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
9028         * tree-cfg.c (found_computed_goto): Remove.
9029         (factor_computed_gotos): Remove.
9030         (make_goto_expr_edges): Return bool, true for computed gotos.
9031         Don't call make_abnormal_goto_edges.
9032         (build_gimple_cfg): Don't set found_computed_goto, don't call
9033         factor_computed_gotos.
9034         (computed_goto_p): No longer static.
9035         (make_blocks): Don't set found_computed_goto.
9036         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
9037         (make_edges): If make_goto_expr_edges returns true, push bb
9038         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
9039         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
9040         vector.  Record mapping between bbs and OpenMP regions if there
9041         are any, adjust make_gimple_omp_edges caller.  Call
9042         handle_abnormal_edges.
9043         (make_abnormal_goto_edges): Remove.
9044         * tree-cfg.h (make_abnormal_goto_edges): Remove.
9045         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
9046         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
9047         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
9048         * internal-fn.def (ABNORMAL_DISPATCHER): New.
9049         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
9050         filling *region also set *region_idx to (*region)->entry->index.
9051
9052         PR other/58712
9053         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
9054         For REGs set ORIGINAL_REGNO.
9055
9056 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
9057
9058         * doc/md.texi: Mention that a target shouldn't implement
9059         vec_widen_(s|u)mul_even/odd pair if it is less efficient
9060         than hi/lo pair.
9061
9062 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
9063
9064         PR tree-optimization/59594
9065         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
9066         a copy of the datarefs vector rather than the vector itself.
9067
9068 2014-01-28  Jason Merrill  <jason@redhat.com>
9069
9070         PR c++/53756
9071         * dwarf2out.c (auto_die): New static.
9072         (gen_type_die_with_usage): Handle C++1y 'auto'.
9073         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
9074         on definition.
9075
9076 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
9077
9078         PR target/59672
9079         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
9080         (SPEC_X32): Likewise.
9081         (SPEC_64): Likewise.
9082         * config/i386/i386.c (ix86_option_override_internal): Turn off
9083         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
9084         for TARGET_16BIT.
9085         (x86_file_start): Output .code16gcc for TARGET_16BIT.
9086         * config/i386/i386.h (TARGET_16BIT): New macro.
9087         (TARGET_16BIT_P): Likewise.
9088         * config/i386/i386.opt: Add m16.
9089         * doc/invoke.texi: Document -m16.
9090
9091 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
9092
9093         PR preprocessor/59935
9094         * input.c (location_get_source_line): Bail out on when line number
9095         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
9096
9097 2014-01-28  Richard Biener  <rguenther@suse.de>
9098
9099         PR tree-optimization/58742
9100         * tree-ssa-forwprop.c (associate_plusminus): Handle
9101         pointer subtraction of the form (T)(P + A) - (T)P.
9102
9103 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9104
9105         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
9106         at const_int_cost.
9107
9108 2014-01-28  Richard Biener  <rguenther@suse.de>
9109
9110         Revert
9111         2014-01-28  Richard Biener  <rguenther@suse.de>
9112
9113         PR rtl-optimization/45364
9114         PR rtl-optimization/59890
9115         * var-tracking.c (local_get_addr_clear_given_value): Handle
9116         already cleared slot.
9117         (val_reset): Handle not allocated local_get_addr_cache.
9118         (vt_find_locations): Use post-order on the inverted CFG.
9119
9120 2014-01-28  Richard Biener  <rguenther@suse.de>
9121
9122         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
9123
9124 2014-01-28  Richard Biener  <rguenther@suse.de>
9125
9126         PR rtl-optimization/45364
9127         PR rtl-optimization/59890
9128         * var-tracking.c (local_get_addr_clear_given_value): Handle
9129         already cleared slot.
9130         (val_reset): Handle not allocated local_get_addr_cache.
9131         (vt_find_locations): Use post-order on the inverted CFG.
9132
9133 2014-01-28  Alan Modra  <amodra@gmail.com>
9134
9135         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
9136         * configure.ac <recursive call for build != host>: Define
9137         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
9138         and LD_FOR_BUILD too.
9139         * configure: Regenerate.
9140
9141 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
9142
9143         * config/i386/i386.c (get_builtin_code_for_version): Separate
9144         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
9145         Broadwell from Haswell.
9146
9147 2014-01-27  Steve Ellcey  <sellcey@mips.com>
9148
9149         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
9150         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
9151         * config/mips/mips.c (mips_option_override): Change setting
9152         of TARGET_DSP.
9153         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
9154         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
9155         Change from Mask to Var.
9156
9157 2014-01-27  Jeff Law  <law@redhat.com>
9158
9159         * ipa-inline.c (inline_small_functions): Fix typo.
9160
9161 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
9162
9163         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
9164         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
9165         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
9166         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
9167         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
9168         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
9169         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
9170         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
9171         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
9172         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
9173         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
9174         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
9175         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
9176         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
9177         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
9178         (_mm512_storeu_epi64): Ditto.
9179         (_mm512_cmpge_epi32_mask): Ditto.
9180         (_mm512_cmpge_epu32_mask): Ditto.
9181         (_mm512_cmpge_epi64_mask): Ditto.
9182         (_mm512_cmpge_epu64_mask): Ditto.
9183         (_mm512_cmple_epi32_mask): Ditto.
9184         (_mm512_cmple_epu32_mask): Ditto.
9185         (_mm512_cmple_epi64_mask): Ditto.
9186         (_mm512_cmple_epu64_mask): Ditto.
9187         (_mm512_cmplt_epi32_mask): Ditto.
9188         (_mm512_cmplt_epu32_mask): Ditto.
9189         (_mm512_cmplt_epi64_mask): Ditto.
9190         (_mm512_cmplt_epu64_mask): Ditto.
9191         (_mm512_cmpneq_epi32_mask): Ditto.
9192         (_mm512_cmpneq_epu32_mask): Ditto.
9193         (_mm512_cmpneq_epi64_mask): Ditto.
9194         (_mm512_cmpneq_epu64_mask): Ditto.
9195         (_mm512_expand_pd): Ditto.
9196         (_mm512_expand_ps): Ditto.
9197         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
9198         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
9199         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
9200         * config/i386/i386.c (ix86_builtins): Add
9201         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
9202         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
9203         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
9204         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
9205         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
9206         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
9207         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
9208         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
9209         IX86_BUILTIN_PMOVUSQW512_MEM.
9210         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
9211         __builtin_ia32_pmovsqd512mem_mask,
9212         __builtin_ia32_pmovqd512mem_mask,
9213         __builtin_ia32_pmovusqw512mem_mask,
9214         __builtin_ia32_pmovsqw512mem_mask,
9215         __builtin_ia32_pmovqw512mem_mask,
9216         __builtin_ia32_pmovusdw512mem_mask,
9217         __builtin_ia32_pmovsdw512mem_mask,
9218         __builtin_ia32_pmovdw512mem_mask,
9219         __builtin_ia32_pmovqb512mem_mask,
9220         __builtin_ia32_pmovusqb512mem_mask,
9221         __builtin_ia32_pmovsqb512mem_mask,
9222         __builtin_ia32_pmovusdb512mem_mask,
9223         __builtin_ia32_pmovsdb512mem_mask,
9224         __builtin_ia32_pmovdb512mem_mask.
9225         (bdesc_args): Add __builtin_ia32_expanddf512,
9226         __builtin_ia32_expandsf512.
9227         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
9228         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
9229         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
9230         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
9231         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
9232         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
9233         (avx512f_<code>v8div16qi2_mask_store): This.
9234         (avx512f_expand<mode>): New.
9235
9236 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
9237
9238         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
9239         New.
9240         (_mm512_mask_prefetch_i64gather_pd): Ditto.
9241         (_mm512_prefetch_i32scatter_pd): Ditto.
9242         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
9243         (_mm512_prefetch_i64scatter_pd): Ditto.
9244         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
9245         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
9246         (_mm512_mask_prefetch_i64gather_ps): Ditto.
9247         (_mm512_prefetch_i32scatter_ps): Ditto.
9248         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
9249         (_mm512_prefetch_i64scatter_ps): Ditto.
9250         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
9251         * config/i386/i386-builtin-types.def: Define
9252         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
9253         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
9254         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
9255         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
9256         IX86_BUILTIN_SCATTERPFQPD.
9257         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
9258         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
9259         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
9260         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
9261         __builtin_ia32_scatterpfqps.
9262         (ix86_expand_builtin): Expand new built-ins.
9263         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
9264         fix memory access data type.
9265         (*avx512pf_gatherpf<mode>_mask): Ditto.
9266         (*avx512pf_gatherpf<mode>): Ditto.
9267         (avx512pf_scatterpf<mode>): Ditto.
9268         (*avx512pf_scatterpf<mode>_mask): Ditto.
9269         (*avx512pf_scatterpf<mode>): Ditto.
9270         (GATHER_SCATTER_SF_MEM_MODE): New.
9271         (avx512pf_gatherpf<mode>df): Ditto.
9272         (*avx512pf_gatherpf<mode>df_mask): Ditto.
9273         (*avx512pf_scatterpf<mode>df): Ditto.
9274
9275 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
9276
9277         PR bootstrap/59934
9278         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
9279         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
9280         reached.
9281
9282 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
9283
9284         * common/config/arm/arm-common.c
9285         (arm_rewrite_mcpu): Handle multiple names.
9286         * config/arm/arm.h
9287         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
9288
9289 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
9290
9291         * gimple-builder.h (create_gimple_tmp): Delete.
9292
9293 2014-01-27  Christian Bruel  <christian.bruel@st.com>
9294
9295         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
9296         words comparisons.
9297
9298 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
9299
9300         * config/pa/pa.md (call): Generate indirect long calls to non-local
9301         functions when outputing 32-bit code.
9302         (call_value): Likewise except for special call to buggy powf function.
9303
9304         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
9305         portable runtime and PIC indirect calls.
9306         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
9307         and PIC call sequences.  Use ldo instead of blr to set return register
9308         in PIC call sequence.
9309
9310 2014-01-25  Walter Lee  <walt@tilera.com>
9311
9312         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
9313         avoid clobbering a live register.
9314
9315 2014-01-25  Walter Lee  <walt@tilera.com>
9316
9317         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
9318         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
9319         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
9320         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
9321
9322 2014-01-25  Walter Lee  <walt@tilera.com>
9323
9324         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
9325         arguments on even registers.
9326         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
9327         STACK_BOUNDARY.
9328         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
9329         (BIGGEST_ALIGNMENT): Ditto.
9330         (BIGGEST_FIELD_ALIGNMENT): Ditto.
9331
9332 2014-01-25  Walter Lee  <walt@tilera.com>
9333
9334         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
9335         insns before bundling.
9336         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
9337
9338 2014-01-25  Walter Lee  <walt@tilera.com>
9339
9340         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
9341         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
9342         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
9343
9344 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
9345
9346         * config/mips/constraints.md (kl): Delete.
9347         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
9348         define expands, using...
9349         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
9350         instructions for MIPS16.
9351         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
9352         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
9353
9354 2014-01-25  Walter Lee  <walt@tilera.com>
9355
9356         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
9357         (clzdi2): Ditto.
9358         (ffsdi2): Ditto.
9359
9360 2014-01-25  Walter Lee  <walt@tilera.com>
9361
9362         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
9363         (TARGET_EXPAND_TO_RTL_HOOK): Define.
9364
9365 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
9366
9367         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
9368         Handle XOR.
9369
9370 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
9371
9372         * print-rtl.c (in_call_function_usage): New var.
9373         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
9374         EXPR_LIST mode as mode and not as reg note name.
9375
9376         PR middle-end/59561
9377         * cfgloopmanip.c (copy_loop_info): If
9378         loop->warned_aggressive_loop_optimizations, make sure
9379         the flag is set in target loop too.
9380
9381 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
9382
9383         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
9384         flag_cilkplus.
9385         * builtins.def: Likewise.
9386         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
9387         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
9388         * ira.c (ira_setup_eliminable_regset): Likewise.
9389         * omp-low.c (gate_expand_omp): Likewise.
9390         (execute_lower_omp): Likewise.
9391         (diagnose_sb_0): Likewise.
9392         (gate_diagnose_omp_blocks): Likewise.
9393         (simd_clone_clauses_extract): Likewise.
9394         (gate): Likewise.
9395
9396 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9397
9398         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
9399         correction for little endian...
9400         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
9401         here.
9402
9403 2014-01-24  Jeff Law  <law@redhat.com>
9404
9405         PR tree-optimization/59919
9406         * tree-vrp.c (find_assert_locations_1): Do not register asserts
9407         for non-returning calls.
9408
9409 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
9410
9411         * common/config/aarch64/aarch64-common.c
9412         (aarch64_rewrite_mcpu): Handle multiple names.
9413         * config/aarch64/aarch64.h
9414         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
9415
9416 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
9417
9418         * input.c (add_file_to_cache_tab): Handle the case where fopen
9419         returns NULL.
9420
9421 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
9422
9423         PR target/59929
9424         * config/i386/i386.md (pushsf splitter): Get stack adjustment
9425         from push operand if code of push isn't PRE_DEC.
9426
9427 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
9428
9429         PR target/59909
9430         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
9431         -mquad-memory-atomic.  Update -mquad-memory documentation to say
9432         it is only used for non-atomic loads/stores.
9433
9434         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
9435         -mquad-memory or -mquad-memory-atomic switches.
9436
9437         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
9438         -mquad-memory-atomic to ISA 2.07 support.
9439
9440         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
9441         to separate support of normal quad word memory operations (ldq, stq)
9442         from the atomic quad word memory operations.
9443
9444         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9445         support to separate non-atomic quad word operations from atomic
9446         quad word operations.  Disable non-atomic quad word operations in
9447         little endian mode so that we don't have to swap words after the
9448         load and before the store.
9449         (quad_load_store_p): Add comment about atomic quad word support.
9450         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
9451         options printed with -mdebug=reg.
9452
9453         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
9454         -mquad-memory-atomic as the test for whether we have quad word
9455         atomic instructions.
9456         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
9457         or -mp8-vector are used, allow byte/half-word atomic operations.
9458
9459         * config/rs6000/sync.md (load_lockedti): Insure that the address
9460         is a proper indexed or indirect address for the lqarx instruction.
9461         On little endian systems, swap the hi/lo registers after the lqarx
9462         instruction.
9463         (load_lockedpti): Use indexed_or_indirect_operand predicate to
9464         insure the address is valid for the lqarx instruction.
9465         (store_conditionalti): Insure that the address is a proper indexed
9466         or indirect address for the stqcrx. instruction.  On little endian
9467         systems, swap the hi/lo registers before doing the stqcrx.
9468         instruction.
9469         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
9470         insure the address is valid for the stqcrx. instruction.
9471
9472         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9473         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
9474         type of quad memory support is available.
9475
9476 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
9477
9478         PR regression/59915
9479         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
9480         there is a danger of looping.
9481
9482 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
9483
9484         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9485         force flag_ira_loop_pressure if set via command line.
9486
9487 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9488
9489         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
9490         (ashr_simd): New builtin handling DI mode.
9491         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
9492         (aarch64_sshr_simddi): New match pattern.
9493         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
9494         (vshrd_n_s64): Likewise.
9495         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
9496
9497 2014-01-23  Nick Clifton  <nickc@redhat.com>
9498
9499         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
9500         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
9501         favour of mcu specific scripts.
9502         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
9503         430x multilibs.
9504
9505 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
9506             Alex Velenko  <Alex.Velenko@arm.com>
9507
9508         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
9509         (vaddv_s16): Likewise.
9510         (vaddv_s32): Likewise.
9511         (vaddv_u8): Likewise.
9512         (vaddv_u16): Likewise.
9513         (vaddv_u32): Likewise.
9514         (vaddvq_s8): Likewise.
9515         (vaddvq_s16): Likewise.
9516         (vaddvq_s32): Likewise.
9517         (vaddvq_s64): Likewise.
9518         (vaddvq_u8): Likewise.
9519         (vaddvq_u16): Likewise.
9520         (vaddvq_u32): Likewise.
9521         (vaddvq_u64): Likewise.
9522         (vaddv_f32): Likewise.
9523         (vaddvq_f32): Likewise.
9524         (vaddvq_f64): Likewise.
9525         (vmaxv_f32): Likewise.
9526         (vmaxv_s8): Likewise.
9527         (vmaxv_s16): Likewise.
9528         (vmaxv_s32): Likewise.
9529         (vmaxv_u8): Likewise.
9530         (vmaxv_u16): Likewise.
9531         (vmaxv_u32): Likewise.
9532         (vmaxvq_f32): Likewise.
9533         (vmaxvq_f64): Likewise.
9534         (vmaxvq_s8): Likewise.
9535         (vmaxvq_s16): Likewise.
9536         (vmaxvq_s32): Likewise.
9537         (vmaxvq_u8): Likewise.
9538         (vmaxvq_u16): Likewise.
9539         (vmaxvq_u32): Likewise.
9540         (vmaxnmv_f32): Likewise.
9541         (vmaxnmvq_f32): Likewise.
9542         (vmaxnmvq_f64): Likewise.
9543         (vminv_f32): Likewise.
9544         (vminv_s8): Likewise.
9545         (vminv_s16): Likewise.
9546         (vminv_s32): Likewise.
9547         (vminv_u8): Likewise.
9548         (vminv_u16): Likewise.
9549         (vminv_u32): Likewise.
9550         (vminvq_f32): Likewise.
9551         (vminvq_f64): Likewise.
9552         (vminvq_s8): Likewise.
9553         (vminvq_s16): Likewise.
9554         (vminvq_s32): Likewise.
9555         (vminvq_u8): Likewise.
9556         (vminvq_u16): Likewise.
9557         (vminvq_u32): Likewise.
9558         (vminnmv_f32): Likewise.
9559         (vminnmvq_f32): Likewise.
9560         (vminnmvq_f64): Likewise.
9561
9562 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
9563
9564         * config/aarch64/aarch64-simd.md
9565         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
9566         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
9567         (*aarch64_mul3_elt<mode>): Likewise.
9568         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
9569         (*aarch64_mul3_elt_to_64v2df): Likewise.
9570         (*aarch64_mla_elt<mode>): Likewise.
9571         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
9572         (*aarch64_mls_elt<mode>): Likewise.
9573         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
9574         (*aarch64_fma4_elt<mode>): Likewise.
9575         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
9576         (*aarch64_fma4_elt_to_64v2df): Likewise.
9577         (*aarch64_fnma4_elt<mode>): Likewise.
9578         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
9579         (*aarch64_fnma4_elt_to_64v2df): Likewise.
9580         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
9581         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
9582         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9583         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9584         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
9585         (aarch64_sqdmull_lane<mode>_internal): Likewise.
9586         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
9587
9588 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9589
9590         * config/aarch64/aarch64-simd.md
9591         (aarch64_be_checked_get_lane<mode>): New define_expand.
9592         * config/aarch64/aarch64-simd-builtins.def
9593         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
9594         New builtin definition.
9595         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
9596         Use new safe be builtin.
9597
9598 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9599
9600         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
9601         New define_insn.
9602         (aarch64_be_st1<mode>): Likewise.
9603         (aarch_ld1<VALL:mode>): Define_expand modified.
9604         (aarch_st1<VALL:mode>): Likewise.
9605         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
9606         (UNSPEC_ST1): Likewise.
9607
9608 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
9609
9610         * config/microblaze/microblaze.md: Add trap insn and attribute
9611
9612 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
9613
9614         PR preprocessor/58580
9615         * input.h (location_get_source_line): Take an additional line_size
9616         parameter.
9617         (void diagnostics_file_cache_fini): Declare new function.
9618         * input.c (struct fcache): New type.
9619         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
9620         New static constants.
9621         (diagnostic_file_cache_init, total_lines_num)
9622         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
9623         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
9624         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
9625         (get_next_line, read_next_line, goto_next_line, read_line_num):
9626         New static function definitions.
9627         (diagnostic_file_cache_fini): New function.
9628         (location_get_source_line): Take an additional output line_len
9629         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
9630         read_line_num.
9631         * diagnostic.c (diagnostic_finish): Call
9632         diagnostic_file_cache_fini.
9633         (adjust_line): Take an additional input parameter for the length
9634         of the line, rather than calculating it with strlen.
9635         (diagnostic_show_locus): Adjust the use of
9636         location_get_source_line and adjust_line with respect to their new
9637         signature.  While displaying a line now, do not stop at the first
9638         null byte.  Rather, display the zero byte as a space and keep
9639         going until we reach the size of the line.
9640         * Makefile.in: Add vec.o to OBJS-libcommon
9641
9642 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9643             Ilya Tocar     <ilya.tocar@intel.com>
9644
9645         * config/i386/avx512fintrin.h (_mm512_kmov): New.
9646         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
9647         (__builtin_ia32_kmov16): Ditto.
9648         * config/i386/i386.md (UNSPEC_KMOV): New.
9649         (kmovw): Ditto.
9650
9651 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9652
9653         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
9654         (_mm512_storeu_si512): Ditto.
9655
9656 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
9657
9658         PR target/52125
9659         * rtl.h (get_referenced_operands): Declare.
9660         * recog.c (get_referenced_operands): New function.
9661         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
9662         operands have been referenced when recording LO_SUM references.
9663
9664 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
9665
9666         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
9667
9668 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9669
9670         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
9671         Enable for generic and recent AMD targets.
9672
9673 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9674
9675         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
9676         ARG_SIZE note when adjustment was eliminated.
9677
9678 2014-01-22  Jeff Law  <law@redhat.com>
9679
9680         PR tree-optimization/59597
9681         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
9682         in file.  Accept new argument REGISTERING and use it to modify
9683         dump output appropriately.
9684         (register_jump_thread): Corresponding changes.
9685         (mark_threaded_blocks): Reinstate code to cancel unprofitable
9686         thread paths involving joiner blocks.  Add code to dump cancelled
9687         jump threading paths.
9688
9689 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
9690
9691         PR rtl-optimization/59477
9692         * lra-constraints.c (inherit_in_ebb): Process call for living hard
9693         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
9694
9695 2014-01-22  Tom Tromey  <tromey@redhat.com>
9696
9697         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
9698         PARAMS.
9699         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
9700
9701 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9702
9703         PR rtl-optimization/59896
9704         * lra-constraints.c (process_alt_operands): Check unused note for
9705         matched operands of insn with no output reloads.
9706
9707 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
9708
9709         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
9710         (mips_move_from_gpr_cost): Likewise.
9711
9712 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9713
9714         PR rtl-optimization/59858
9715         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
9716         ira_class_hard_regs_num.
9717         (process_alt_operands): Increase reject for dying matched operand.
9718
9719 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
9720
9721         PR target/59003
9722         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
9723         smaller than size, perform several stores or loads and stores
9724         at dst + count - size to store or copy all of size bytes, rather
9725         than just last modesize bytes.
9726
9727 2014-01-20  DJ Delorie  <dj@redhat.com>
9728
9729         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
9730         that CLOBBERs are REGs before propogating their values.
9731
9732 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
9733
9734         PR middle-end/59789
9735         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
9736         (cgraph_inline_failed_type): New function.
9737         * cgraph.h (DEFCIFCODE): Add type.
9738         (cgraph_inline_failed_type_t): New enum.
9739         (cgraph_inline_failed_type): New prototype.
9740         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
9741         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
9742         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
9743         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
9744         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
9745         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
9746         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
9747         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
9748         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
9749         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
9750         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
9751         OPTIMIZATION_MISMATCH.
9752         * tree-inline.c (expand_call_inline): Emit errors during
9753         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
9754
9755 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9756
9757         PR target/59685
9758         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
9759         mode attribute in insn output.
9760
9761 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
9762
9763         * output.h (output_constant): Delete.
9764         * varasm.c (output_constant): Make private.
9765
9766 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
9767
9768         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
9769
9770 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9771
9772         PR middle-end/59860
9773         * tree.h (fold_builtin_strcat): New prototype.
9774         * builtins.c (fold_builtin_strcat): No longer static.  Add len
9775         argument, if non-NULL, don't call c_strlen.  Optimize
9776         directly into __builtin_memcpy instead of __builtin_strcpy.
9777         (fold_builtin_2): Adjust fold_builtin_strcat caller.
9778         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
9779
9780 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9781
9782         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9783         for SImode_address_operand operands, having only a REG argument.
9784
9785 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9786
9787         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
9788         loader name using mbig-endian.
9789         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
9790
9791 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9792
9793         * doc/invoke.texi (-march): Clarify documentation for AArch64.
9794         (-mtune): Likewise.
9795         (-mcpu): Likewise.
9796
9797 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
9798
9799         * config/aarch64/aarch64-protos.h
9800         (aarch64_cannot_change_mode_class_ptr): Declare.
9801         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
9802         aarch64_cannot_change_mode_class_ptr): New.
9803         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
9804         backend hook aarch64_cannot_change_mode_class.
9805
9806 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9807
9808         * common/config/aarch64/aarch64-common.c
9809         (aarch64_handle_option): Don't handle any option order logic here.
9810         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
9811         selected_cpu, warn on architecture version mismatch.
9812         (aarch64_override_options): Fix parsing order for option strings.
9813
9814 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9815             Iain Sandoe  <iain@codesourcery.com>
9816
9817         PR bootstrap/59496
9818         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
9819         warning.  Amend comment to reflect current functionality.
9820
9821 2014-01-20  Richard Biener  <rguenther@suse.de>
9822
9823         PR middle-end/59860
9824         * builtins.c (fold_builtin_strcat): Remove case better handled
9825         by tree-ssa-strlen.c.
9826
9827 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
9828
9829         * config/aarch64/aarch64.opt
9830         (mcpu, march, mtune): Make case-insensitive.
9831
9832 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9833
9834         PR target/59880
9835         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9836         if operands[1] is a REG or ZERO_EXTEND of a REG.
9837
9838 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
9839
9840         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
9841
9842 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
9843
9844         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
9845         long non-pic millicode calls.
9846
9847 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9848
9849         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
9850
9851 2014-01-19  Kito Cheng  <kito@0xlab.org>
9852
9853         * builtins.c (expand_movstr): Check movstr expand done or fail.
9854
9855 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9856             H.J. Lu  <hongjiu.lu@intel.com>
9857
9858         PR target/59379
9859         * config/i386/i386.md (*lea<mode>): Zero-extend return register
9860         to DImode for zero-extended addresses.
9861
9862 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
9863
9864         PR rtl-optimization/57763
9865         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
9866         on the new indirect jump_insn and increment LABEL_NUSES (label).
9867
9868 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
9869
9870         PR bootstrap/59580
9871         PR bootstrap/59583
9872         * config.gcc (x86_archs): New variable.
9873         (x86_64_archs): Likewise.
9874         (x86_cpus): Likewise.
9875         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
9876         --with-arch/--with-cpu= options.
9877         Support --with-arch=/--with-cpu={nehalem,westmere,
9878         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
9879
9880 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9881
9882         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
9883         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
9884
9885 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9886
9887         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
9888
9889 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
9890
9891         PR target/58944
9892         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
9893         clear cpp_get_options (parse_in)->warn_unused_macros for
9894         ix86_target_macros_internal with cpp_define.
9895
9896 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
9897
9898         * jump.c (delete_related_insns): Keep (use (insn))s.
9899         * reorg.c (redundant_insn): Check for barriers too.
9900
9901 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9902
9903         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
9904
9905 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
9906
9907         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
9908         call to $$dyncall when TARGET_LONG_CALLS is true.
9909
9910 2014-01-17  Jeff Law  <law@redhat.com>
9911
9912         * ree.c (combine_set_extension): Temporarily disable test for
9913         changing number of hard registers.
9914
9915 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9916
9917         PR middle-end/58125
9918         * ipa-inline-analysis.c (inline_free_summary):
9919         Do not free summary of aliases.
9920
9921 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9922
9923         PR middle-end/59706
9924         * gimplify.c (gimplify_expr): Use create_tmp_var
9925         instead of create_tmp_var_raw.  If cond doesn't have
9926         integral type, don't add the IFN_ANNOTATE builtin at all.
9927
9928 2014-01-17  Martin Jambor  <mjambor@suse.cz>
9929
9930         PR ipa/59736
9931         * ipa-cp.c (prev_edge_clone): New variable.
9932         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
9933         Also resize prev_edge_clone vector.
9934         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
9935         (ipcp_edge_removal_hook): New function.
9936         (ipcp_driver): Register ipcp_edge_removal_hook.
9937
9938 2014-01-17  Andrew Pinski  <apinski@cavium.com>
9939             Steve Ellcey  <sellcey@mips.com>
9940
9941         PR target/59462
9942         * config/mips/mips.c (mips_print_operand): Check operand mode instead
9943         of operator mode.
9944
9945 2014-01-17  Jeff Law  <law@redhat.com>
9946
9947         PR middle-end/57904
9948         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
9949         so that pass_ccp runs first.
9950
9951 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9952
9953         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
9954         (ix86_adjust_cost): Use !TARGET_XXX.
9955         (do_reorder_for_imul): Likewise.
9956         (swap_top_of_ready_list): Likewise.
9957         (ix86_sched_reorder): Likewise.
9958
9959 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9960
9961         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9962         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
9963         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
9964         (intel_memset): New.  Duplicate slm_memset.
9965         (intel_cost): New.  Duplicate slm_cost.
9966         (m_INTEL): New macro.
9967         (processor_target_table): Add "intel".
9968         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
9969         with PROCESSOR_INTEL for "intel".
9970         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
9971         PROCESSOR_SILVERMONT.
9972         (ix86_issue_rate): Likewise.
9973         (ix86_adjust_cost): Likewise.
9974         (ia32_multipass_dfa_lookahead): Likewise.
9975         (swap_top_of_ready_list): Likewise.
9976         (ix86_sched_reorder): Likewise.
9977         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
9978         instead of TARGET_OPT_AGU.
9979         * config/i386/i386.h (TARGET_INTEL): New.
9980         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
9981         (processor_type): Add PROCESSOR_INTEL.
9982         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
9983         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
9984
9985 2014-01-17  Marek Polacek  <polacek@redhat.com>
9986
9987         PR c/58346
9988         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
9989         size is zero.
9990
9991 2014-01-17  Richard Biener  <rguenther@suse.de>
9992
9993         PR tree-optimization/46590
9994         * opts.c (default_options_table): Add entries for
9995         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
9996         all enabled at -O1 but not for -Og.
9997         * common.opt (fbranch-count-reg): Remove Init(1).
9998         (fmove-loop-invariants): Likewise.
9999         (ftree-pta): Likewise.
10000
10001 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
10002
10003         * config/i386/i386.c (ix86_data_alignment): For compatibility with
10004         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
10005         decls to at least the GCC 4.8 used alignments.
10006
10007         PR fortran/59440
10008         * tree-nested.c (convert_nonlocal_reference_stmt,
10009         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
10010         of GIMPLE_BIND stmts, adjust associated decls.
10011
10012 2014-01-17  Richard Biener  <rguenther@suse.de>
10013
10014         PR tree-optimization/46590
10015         * vec.h (vec<>::bseach): New member function implementing
10016         binary search according to C89 bsearch.
10017         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
10018         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
10019         bitmap pointer again.  Make accesses_in_loop a flat array.
10020         (mem_ref_obstack): New global.
10021         (outermost_indep_loop): Adjust for mem_ref->stored changes.
10022         (mark_ref_stored): Likewise.
10023         (ref_indep_loop_p_2): Likewise.
10024         (set_ref_stored_in_loop): New helper function.
10025         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
10026         (memref_free): Adjust.
10027         (record_mem_ref_loc): Simplify.
10028         (gather_mem_refs_stmt): Adjust.
10029         (sort_locs_in_loop_postorder_cmp): New function.
10030         (analyze_memory_references): Sort accesses_in_loop after
10031         loop postorder number.
10032         (find_ref_loc_in_loop_cmp): New function.
10033         (for_all_locs_in_loop): Find relevant cluster of locs in
10034         accesses_in_loop and iterate without recursion.
10035         (execute_sm): Avoid uninit warning.
10036         (struct ref_always_accessed): Simplify.
10037         (ref_always_accessed::operator ()): Likewise.
10038         (ref_always_accessed_p): Likewise.
10039         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
10040         loop postorder numbers here.
10041         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
10042         numbers.
10043
10044 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
10045
10046         PR c++/57945
10047         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
10048         on decls for which assemble_alias has been called.
10049
10050 2014-01-17  Nick Clifton  <nickc@redhat.com>
10051
10052         * config/msp430/msp430.opt: (mcpu): New option.
10053         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
10054         (msp430_option_override): Parse target_cpu.  If the MCU name
10055         matches a generic string, clear target_mcu.
10056         (msp430_attr): Allow numeric interrupt values up to 63.
10057         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
10058         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
10059         option.
10060         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
10061         Add mcpu matches.
10062         * config/msp430/msp430.md (popm): Use %J rather than %I.
10063         (addsi3): Use msp430_nonimmediate_operand for operand 2.
10064         (addhi_cy_i): Use immediate_operand for operand 2.
10065         * doc/invoke.texi: Document -mcpu option.
10066
10067 2014-01-17  Richard Biener  <rguenther@suse.de>
10068
10069         PR rtl-optimization/38518
10070         * df.h (df_analyze_loop): Declare.
10071         * df-core.c: Include cfgloop.h.
10072         (df_analyze_1): Split out main part of df_analyze.
10073         (df_analyze): Adjust.
10074         (loop_inverted_post_order_compute): New function.
10075         (loop_post_order_compute): Likewise.
10076         (df_analyze_loop): New function avoiding whole-function
10077         postorder computes.
10078         * loop-invariant.c (find_defs): Use df_analyze_loop.
10079         (find_invariants): Adjust.
10080         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
10081
10082 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
10083
10084         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
10085         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
10086
10087 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
10088
10089         * ipa-ref.c (ipa_remove_stmt_references): Fix references
10090         traversal when removing references.
10091
10092 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
10093
10094         PR ipa/59775
10095         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
10096
10097 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
10098
10099         PR middle-end/56791
10100         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
10101         pushing a reload for an autoinc when we had previously reloaded an
10102         inner part of the address.
10103
10104 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
10105
10106         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
10107         field.
10108         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
10109         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
10110         when not giving up or versioning for alias only because of
10111         loop->safelen.
10112         (vect_analyze_data_ref_dependences): Set to true.
10113         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
10114         is a GIMPLE_PHI.
10115         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
10116         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
10117         to the condition.
10118
10119         PR middle-end/58344
10120         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
10121
10122         PR target/59839
10123         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
10124         operand 0 predicate for gathers, use a new pseudo as subtarget.
10125
10126 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
10127
10128         PR middle-end/59609
10129         * lra-constraints.c (process_alt_operands): Add printing debug info.
10130         Check absence of input/output reloads for matched operands too.
10131
10132 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
10133
10134         PR rtl-optimization/59835
10135         * ira.c (ira_init_register_move_cost): Increase cost for
10136         impossible modes.
10137
10138 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
10139
10140         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
10141
10142 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
10143
10144         PR target/59780
10145         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
10146         non-register objects.  Use gen_(high/low)part more consistently.
10147         Fix assertions.
10148
10149 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
10150
10151         PR target/59844
10152         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
10153         endian support, remove tests for WORDS_BIG_ENDIAN.
10154         (p8_mfvsrd_3_<mode>): Likewise.
10155         (reload_gpr_from_vsx<mode>): Likewise.
10156         (reload_gpr_from_vsxsf): Likewise.
10157         (p8_mfvsrd_4_disf): Likewise.
10158
10159 2014-01-16  Richard Biener  <rguenther@suse.de>
10160
10161         PR rtl-optimization/46590
10162         * lcm.c (compute_antinout_edge): Use postorder iteration.
10163         (compute_laterin): Use inverted postorder iteration.
10164
10165 2014-01-16  Nick Clifton  <nickc@redhat.com>
10166
10167         PR middle-end/28865
10168         * varasm.c (output_constant): Return the number of bytes actually
10169         emitted.
10170         (output_constructor_array_range): Update the field size with the
10171         number of bytes emitted by output_constant.
10172         (output_constructor_regular_field): Likewise.  Also do not
10173         complain if the total number of bytes emitted is now greater
10174         than the expected fieldpos.
10175         * output.h (output_constant): Update prototype and descriptive comment.
10176
10177 2014-01-16  Marek Polacek  <polacek@redhat.com>
10178
10179         PR middle-end/59827
10180         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
10181         it is error_mark_node.
10182
10183 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
10184
10185         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
10186         VALID_AVX256_REG_OR_OI_MODE.
10187
10188 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
10189
10190         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
10191         current procedure should be profiled.
10192
10193 2014-01-15  Andrew Pinski  <apinski@cavium.com>
10194
10195         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
10196         of moving from/to the STACK_REG register class.
10197
10198 2014-01-15  Richard Henderson  <rth@redhat.com>
10199
10200         PR debug/54694
10201         * reginfo.c (global_regs_decl): Globalize.
10202         * rtl.h (global_regs_decl): Declare.
10203         * ira.c (do_reload): Diagnose frame_pointer_needed and it
10204         reserved via global_regs.
10205
10206 2014-01-15  Teresa Johnson  <tejohnson@google.com>
10207
10208         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
10209
10210 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
10211
10212         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
10213         and vmulosh rather than call gen_vec_widen_smult_*.
10214         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
10215         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
10216         (vec_widen_smult_even_v16qi): Likewise.
10217         (vec_widen_umult_even_v8hi): Likewise.
10218         (vec_widen_smult_even_v8hi): Likewise.
10219         (vec_widen_umult_odd_v16qi): Likewise.
10220         (vec_widen_smult_odd_v16qi): Likewise.
10221         (vec_widen_umult_odd_v8hi): Likewise.
10222         (vec_widen_smult_odd_v8hi): Likewise.
10223         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
10224         vmuloub rather than call gen_vec_widen_umult_*.
10225         (vec_widen_umult_lo_v16qi): Likewise.
10226         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
10227         vmulosb rather than call gen_vec_widen_smult_*.
10228         (vec_widen_smult_lo_v16qi): Likewise.
10229         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
10230         rather than call gen_vec_widen_umult_*.
10231         (vec_widen_umult_lo_v8hi): Likewise.
10232         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
10233         rather than call gen_vec_widen_smult_*.
10234         (vec_widen_smult_lo_v8hi): Likewise.
10235
10236 2014-01-15  Jeff Law  <law@redhat.com>
10237
10238         PR tree-optimization/59747
10239         * ree.c (find_and_remove_re): Properly handle case where a second
10240         eliminated extension requires widening a copy created for elimination
10241         of a prior extension.
10242         (combine_set_extension): Ensure that the number of hard regs needed
10243         for a destination register does not change when we widen it.
10244
10245 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
10246
10247         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
10248         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
10249         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
10250         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
10251         (avr-*-rtems*): Likewise.
10252         (bfin*-rtems*): Likewise.
10253         (moxie-*-rtems*): Likewise.
10254         (h8300-*-rtems*): Likewise.
10255         (i[34567]86-*-rtems*): Likewise.
10256         (lm32-*-rtems*): Likewise.
10257         (m32r-*-rtems*): Likewise.
10258         (m68k-*-rtems*): Likewise.
10259         (microblaze*-*-rtems*): Likewise.
10260         (mips*-*-rtems*): Likewise.
10261         (powerpc-*-rtems*): Likewise.
10262         (sh-*-rtems*): Likewise.
10263         (sparc-*-rtems*): Likewise.
10264         (sparc64-*-rtems*): Likewise.
10265         (v850-*-rtems*): Likewise.
10266         (m32c-*-rtems*): Likewise.
10267
10268 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
10269
10270         PR rtl-optimization/59511
10271         * ira.c (ira_init_register_move_cost): Use memory costs for some
10272         cases of register move cost calculations.
10273         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
10274         instead of BB frequency.
10275         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
10276         * lra-assigns.c (find_hard_regno_for): Ditto.
10277
10278 2014-01-15  Richard Biener  <rguenther@suse.de>
10279
10280         PR tree-optimization/59822
10281         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
10282         (vectorizable_load): Use it to hoist defs of uses of invariant
10283         loads out of the loop.
10284
10285 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
10286             Kugan Vivekanandarajah  <kuganv@linaro.org>
10287
10288         PR target/59695
10289         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
10290         truncation.
10291
10292 2014-01-15  Richard Biener  <rguenther@suse.de>
10293
10294         PR rtl-optimization/59802
10295         * lcm.c (compute_available): Use inverted postorder to seed
10296         the initial worklist.
10297
10298 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10299
10300         PR target/59803
10301         * config/s390/s390.c (s390_preferred_reload_class): Don't return
10302         ADDR_REGS for invalid symrefs in non-PIC code.
10303
10304 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
10305
10306         PR other/58712
10307         * builtins.c (determine_block_size): Initialize *probable_max_size
10308         even if len_rtx is CONST_INT.
10309
10310 2014-01-14  Andrew Pinski  <apinski@cavium.com>
10311
10312         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
10313         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
10314         (cortexa53_tunings): Likewise.
10315         (aarch64_sched_issue_rate): New function.
10316         (TARGET_SCHED_ISSUE_RATE): Define.
10317
10318 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
10319
10320         * ira-costs.c (find_costs_and_classes): Add missed
10321         ira_init_register_move_cost_if_necessary.
10322
10323 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
10324
10325         PR target/59787
10326         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
10327
10328 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
10329
10330         PR target/59794
10331         * config/i386/i386.c (type_natural_mode): Add a bool parameter
10332         to indicate if type is used for function return value.  Warn ABI
10333         change if the vector mode isn't available for function return value.
10334         (ix86_function_arg_advance): Pass false to type_natural_mode.
10335         (ix86_function_arg): Likewise.
10336         (ix86_gimplify_va_arg): Likewise.
10337         (function_arg_32): Don't warn ABI change.
10338         (ix86_function_value): Pass true to type_natural_mode.
10339         (ix86_return_in_memory): Likewise.
10340         (ix86_struct_value_rtx): Removed.
10341         (TARGET_STRUCT_VALUE_RTX): Likewise.
10342
10343 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10344
10345         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
10346         converting a conditional jump into a conditional return.
10347
10348 2014-01-14  Richard Biener  <rguenther@suse.de>
10349
10350         PR tree-optimization/58921
10351         PR tree-optimization/59006
10352         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
10353         hoisting invariant stmts.
10354         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
10355         invariant loads on the preheader edge if possible.
10356
10357 2014-01-14  Joey Ye  <joey.ye@arm.com>
10358
10359         * doc/plugin.texi (Building GCC plugins): Update to C++.
10360
10361 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
10362
10363         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
10364         (_mm_rcp28_round_ss): Ditto.
10365         (_mm_rsqrt28_round_sd): Ditto.
10366         (_mm_rsqrt28_round_ss): Ditto.
10367         (_mm_rcp28_sd): Ditto.
10368         (_mm_rcp28_ss): Ditto.
10369         (_mm_rsqrt28_sd): Ditto.
10370         (_mm_rsqrt28_ss): Ditto.
10371         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
10372         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
10373         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
10374         (IX86_BUILTIN_RCP28SD): Ditto.
10375         (IX86_BUILTIN_RCP28SS): Ditto.
10376         (IX86_BUILTIN_RSQRT28SD): Ditto.
10377         (IX86_BUILTIN_RSQRT28SS): Ditto.
10378         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
10379         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
10380         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
10381         (ix86_expand_special_args_builtin): Expand new FTYPE.
10382         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
10383         (srcp14<mode>): Make insn unary.
10384         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
10385         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
10386         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
10387         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
10388         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
10389         Fix rounding: make it SAE only.
10390         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
10391         Ditto.
10392         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
10393         Ditto.
10394         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
10395         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
10396         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
10397         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
10398         (round_saeonly_mask_scalar_operand4): Ditto.
10399         (round_saeonly_mask_scalar_op3): Ditto.
10400         (round_saeonly_mask_scalar_op4): Ditto.
10401
10402 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10403
10404         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10405         Implement -maltivec=be for vec_insert and vec_extract.
10406
10407 2014-01-10  DJ Delorie  <dj@redhat.com>
10408
10409         * config/msp430/msp430.md (call_internal): Don't allow memory
10410         references with SP as the base register.
10411         (call_value_internal): Likewise.
10412         * config/msp430/constraints.md (Yc): New.  For memory references
10413         that don't use SP as a base register.
10414
10415         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
10416         "an integer without a # prefix"
10417         * config/msp430/msp430.md (epilogue_helper): Use it.
10418
10419 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
10420
10421         PR target/59617
10422         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
10423         AVX512F gather builtins.
10424         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
10425         on gather decls with INTEGER_TYPE masktype.
10426         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
10427         directly into the builtin rather than hoisting it before loop.
10428
10429         PR tree-optimization/59387
10430         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
10431         (scev_const_prop): If folded_casts and type has undefined overflow,
10432         use force_gimple_operand instead of force_gimple_operand_gsi and
10433         for each added stmt if it is assign with
10434         arith_code_with_undefined_signed_overflow, call
10435         rewrite_to_defined_overflow.
10436         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
10437         gimple-fold.h instead.
10438         (arith_code_with_undefined_signed_overflow,
10439         rewrite_to_defined_overflow): Moved to ...
10440         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
10441         rewrite_to_defined_overflow): ... here.  No longer static.
10442         Include gimplify-me.h.
10443         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
10444         rewrite_to_defined_overflow): New prototypes.
10445
10446 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10447
10448         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
10449
10450 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
10451
10452         * builtins.c (get_object_alignment_2): Minor tweak.
10453         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
10454
10455 2014-01-13  Christian Bruel  <christian.bruel@st.com>
10456
10457         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
10458         optimized non constant lengths.
10459
10460 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
10461
10462         PR libgomp/59194
10463         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
10464         load as __atomic_load_N if possible.
10465
10466 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
10467
10468         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
10469         target parameter.
10470         (rs6000_expand_builtin): Adjust call.
10471
10472 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
10473
10474         PR target/58115
10475         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
10476         * config/rs6000/rs6000.c: Include target-globals.h.
10477         (rs6000_set_current_function): Instead of doing target_reinit
10478         unconditionally, use save_target_globals_default_opts and
10479         restore_target_globals.
10480
10481         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
10482         FPSCR.
10483         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
10484         (rs6000_expand_builtin): Handle mffs and mtfsf.
10485         (rs6000_init_builtins): Define mffs and mtfsf.
10486         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
10487         (rs6000_mffs): New pattern.
10488         (rs6000_mtfsf): New pattern.
10489
10490 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
10491
10492         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
10493         Start narrowing with START.  Apply candidate-use pair
10494         and check overall cost in narrowing.
10495         (iv_ca_prune): Pass new argument.
10496
10497 2014-01-10  Jeff Law  <law@redhat.com>
10498
10499         PR middle-end/59743
10500         * ree.c (combine_reaching_defs): Ensure the defining statement
10501         occurs before the extension when optimizing extensions with
10502         different source and destination hard registers.
10503
10504 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
10505
10506         PR ipa/58585
10507         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
10508         vtables into the type inheritance graph.
10509
10510 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10511
10512         PR rtl-optimization/59754
10513         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
10514         modes in the REGNO != REGNO case.
10515
10516 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10517
10518         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
10519
10520 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10521
10522         PR tree-optimization/59745
10523         * tree-predcom.c (tree_predictive_commoning_loop): Call
10524         free_affine_expand_cache if giving up because components is NULL.
10525
10526         * target-globals.c (save_target_globals): Allocate < 4KB structs using
10527         GC in payload of target_globals struct instead of allocating them on
10528         the heap and the larger structs separately using GC.
10529         * target-globals.h (struct target_globals): Make regs, hard_regs,
10530         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
10531         of GTY((skip)) and change type to void *.
10532         (reset_target_globals): Cast loads from those fields to corresponding
10533         types.
10534
10535 2014-01-10  Steve Ellcey  <sellcey@mips.com>
10536
10537         PR plugins/59335
10538         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
10539         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
10540         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
10541
10542 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
10543
10544         PR target/59744
10545         * aarch64-modes.def (CC_Zmode): New flags mode.
10546         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
10547         represents an equality.
10548         (aarch64_get_condition_code): Handle CC_Zmode.
10549         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
10550
10551 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10552
10553         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
10554         extraction in good case.
10555
10556 2014-01-10  Richard Biener  <rguenther@suse.de>
10557
10558         PR tree-optimization/59374
10559         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
10560         checking after SLP discovery.  Mark stmts not participating
10561         in any SLP instance properly.
10562
10563 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10564
10565         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
10566         when handling a SET rtx.
10567
10568 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10569
10570         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
10571         (cortex-a57): Likewise.
10572         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
10573
10574 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10575
10576         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
10577         non-iwmmxt builtins.
10578
10579 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
10580
10581         PR ipa/58252
10582         PR ipa/59226
10583         * ipa-devirt.c record_target_from_binfo): Take as argument
10584         stack of binfos and lookup matching one for virtual inheritance.
10585         (possible_polymorphic_call_targets_1): Update.
10586
10587 2014-01-10  Huacai Chen  <chenhc@lemote.com>
10588
10589         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
10590         kernel strings for Loongson-2E/2F/3A.
10591
10592 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10593
10594         PR middle-end/59670
10595         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
10596         is_gimple_call before calling gimple_call_internal_p.
10597
10598 2014-01-09  Steve Ellcey  <sellcey@mips.com>
10599
10600         * Makefile.in (TREE_FLOW_H): Remove.
10601         (TREE_SSA_H): Add file names from tree-flow.h.
10602         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
10603         * tree.h: Remove tree-flow.h reference.
10604         * hash-table.h: Remove tree-flow.h reference.
10605         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
10606         reference with tree-ssa-loop.h.
10607
10608 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10609
10610         * doc/invoke.texi: Add -maltivec={be,le} options, and document
10611         default element-order behavior for -maltivec.
10612         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
10613         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
10614         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
10615         when targeting big endian, at least for now.
10616         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
10617
10618 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10619
10620         PR middle-end/47735
10621         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
10622         var satisfies use_register_for_decl, just take into account type
10623         alignment, rather than decl alignment.
10624
10625         PR tree-optimization/59622
10626         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
10627         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
10628         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
10629         Don't devirtualize for inplace at all.  For targets.length () == 1,
10630         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
10631
10632 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10633
10634         * config/i386/i386.md (cpu): Remove the unused btver1.
10635
10636 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10637
10638         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
10639
10640 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10641
10642         PR target/58115
10643         * tree-core.h (struct target_globals): New forward declaration.
10644         (struct tree_target_option): Add globals field.
10645         * tree.h (TREE_TARGET_GLOBALS): Define.
10646         (prepare_target_option_nodes_for_pch): New prototype.
10647         * target-globals.h (struct target_globals): Define even if
10648         !SWITCHABLE_TARGET.
10649         * tree.c (prepare_target_option_node_for_pch,
10650         prepare_target_option_nodes_for_pch): New functions.
10651         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
10652         * config/i386/i386.c: Include target-globals.h.
10653         (ix86_set_current_function): Instead of doing target_reinit
10654         unconditionally, use save_target_globals_default_opts and
10655         restore_target_globals.
10656
10657 2014-01-09  Richard Biener  <rguenther@suse.de>
10658
10659         PR tree-optimization/59715
10660         * tree-cfg.h (split_critical_edges): Declare.
10661         * tree-cfg.c (split_critical_edges): Export.
10662         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
10663
10664 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
10665
10666         * cfgexpand.c (expand_stack_vars): Optionally disable
10667         asan stack protection.
10668         (expand_used_vars): Likewise.
10669         (partition_stack_vars): Likewise.
10670         * asan.c (asan_emit_stack_protection): Optionally disable
10671         after return stack usage.
10672         (instrument_derefs): Optionally disable memory access instrumentation.
10673         (instrument_builtin_call): Likewise.
10674         (instrument_strlen_call): Likewise.
10675         (asan_protect_global): Optionally disable global variables protection.
10676         * doc/invoke.texi: Added doc for new options.
10677         * params.def: Added new options.
10678         * params.h: Likewise.
10679
10680 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10681
10682         PR rtl-optimization/59724
10683         * ifcvt.c (cond_exec_process_if_block): Don't call
10684         flow_find_head_matching_sequence with 0 longest_match.
10685         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
10686         non-active insns if !stop_after.
10687         (try_head_merge_bb): Revert 2014-01-07 changes.
10688
10689 2014-01-08  Jeff Law  <law@redhat.com>
10690
10691         * ree.c (get_sub_rtx): New function, extracted from...
10692         (merge_def_and_ext): Here.
10693         (combine_reaching_defs): Use get_sub_rtx.
10694
10695 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
10696
10697         * cgraph.h (varpool_variable_node): Do not choke on null node.
10698
10699 2014-01-08  Catherine Moore  <clm@codesourcery.com>
10700
10701         * config/mips/mips.md (simple_return): Attempt to use JRC
10702         for microMIPS.
10703         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
10704
10705 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10706
10707         PR rtl-optimization/59137
10708         * reorg.c (steal_delay_list_from_target): Call update_block for
10709         elided insns.
10710         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
10711
10712 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10713
10714         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10715         two duplicate entries.
10716
10717 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10718
10719         Revert:
10720         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
10721
10722         * config/mips/mips.c (mips_truncated_op_cost): New function.
10723         (mips_rtx_costs): Adjust test for BADDU.
10724         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
10725
10726         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
10727
10728         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
10729         (*baddu_si): ...this new pattern.
10730
10731 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10732
10733         PR ipa/59722
10734         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
10735
10736 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10737
10738         PR middle-end/57748
10739         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
10740         inner_reference_p.
10741         (expand_expr, expand_normal): Adjust.
10742         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
10743         inner_reference_p. Use inner_reference_p to expand inner references.
10744         (store_expr): Adjust.
10745         * cfgexpand.c (expand_call_stmt): Adjust.
10746
10747 2014-01-08  Rong Xu  <xur@google.com>
10748
10749         * gcov-io.c (gcov_var): Move from gcov-io.h.
10750         (gcov_position): Ditto.
10751         (gcov_is_error): Ditto.
10752         (gcov_rewrite): Ditto.
10753         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
10754         only part to libgcc/libgcov.h.
10755
10756 2014-01-08  Marek Polacek  <polacek@redhat.com>
10757
10758         PR middle-end/59669
10759         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
10760
10761 2014-01-08  Marek Polacek  <polacek@redhat.com>
10762
10763         PR sanitizer/59667
10764         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
10765
10766 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10767
10768         PR rtl-optimization/59649
10769         * stor-layout.c (get_mode_bounds): For BImode return
10770         0 and STORE_FLAG_VALUE.
10771
10772 2014-01-08  Richard Biener  <rguenther@suse.de>
10773
10774         PR middle-end/59630
10775         * gimple.h (is_gimple_builtin_call): Remove.
10776         (gimple_builtin_call_types_compatible_p): New.
10777         (gimple_call_builtin_p): New overload.
10778         * gimple.c (is_gimple_builtin_call): Remove.
10779         (validate_call): Rename to ...
10780         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
10781         check return types.
10782         (validate_type): New static function.
10783         (gimple_call_builtin_p): New overload and adjust.
10784         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
10785         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
10786         (gimple_fold_stmt_to_constant_1): Likewise.
10787         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
10788
10789 2014-01-08  Richard Biener  <rguenther@suse.de>
10790
10791         PR middle-end/59471
10792         * gimplify.c (gimplify_expr): Gimplify register-register type
10793         VIEW_CONVERT_EXPRs to separate stmts.
10794
10795 2014-01-07  Jeff Law  <law@redhat.com>
10796
10797         PR middle-end/53623
10798         * ree.c (combine_set_extension): Handle case where source
10799         and destination registers in an extension insn are different.
10800         (combine_reaching_defs): Allow source and destination registers
10801         in extension to be different under limited circumstances.
10802         (add_removable_extension): Remove restriction that the
10803         source and destination registers in the extension are the same.
10804         (find_and_remove_re): Emit a copy from the extension's
10805         destination to its source after the defining insn if
10806         the source and destination registers are different.
10807
10808         PR middle-end/59285
10809         * ifcvt.c (merge_if_block): If we are merging a block with more than
10810         one successor with a block with no successors, remove any BARRIER
10811         after the second block.
10812
10813 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
10814
10815         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
10816
10817 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
10818
10819         PR target/59652
10820         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
10821         for 14-bit register offsets when INT14_OK_STRICT is false.
10822
10823 2014-01-07  Roland Stigge  <stigge@antcom.de>
10824             Michael Meissner  <meissner@linux.vnet.ibm.com>
10825
10826         PR 57386/target
10827         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
10828         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
10829
10830 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
10831
10832         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
10833         -mcpu.
10834
10835 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
10836
10837         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
10838         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
10839         rtx is const0_rtx or not.
10840
10841 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
10842
10843         PR target/58115
10844         * target-globals.c (save_target_globals): Remove this_fn_optab
10845         handling.
10846         * toplev.c: Include optabs.h.
10847         (target_reinit): Temporarily restore the global options if another
10848         set of options are in force.
10849
10850 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
10851
10852         PR rtl-optimization/58668
10853         * cfgcleanup.c (flow_find_cross_jump): Don't count
10854         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
10855         to determine what is counted.
10856         (flow_find_head_matching_sequence): Use active_insn_p to determine
10857         what is counted.
10858         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
10859         counting change.
10860         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
10861         determine what is counted.
10862
10863         PR tree-optimization/59643
10864         * tree-predcom.c (split_data_refs_to_components): If one dr is
10865         read and one write, determine_offset fails and the write isn't
10866         in the bad component, just put the read into the bad component.
10867
10868 2014-01-07  Mike Stump  <mikestump@comcast.net>
10869             Jakub Jelinek  <jakub@redhat.com>
10870
10871         PR pch/59436
10872         * tree-core.h (struct tree_optimization_option): Change optabs
10873         type from unsigned char * to void *.
10874         * optabs.c (init_tree_optimization_optabs): Adjust
10875         TREE_OPTIMIZATION_OPTABS initialization.
10876
10877 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
10878
10879         PR target/59644
10880         * config/i386/i386.h (struct machine_function): Add
10881         no_drap_save_restore field.
10882         * config/i386/i386.c (ix86_save_reg): Use
10883         !cfun->machine->no_drap_save_restore instead of
10884         crtl->stack_realign_needed.
10885         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
10886         this function clears frame_pointer_needed.  Set
10887         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
10888         and DRAP reg is needed.
10889
10890 2014-01-06  Marek Polacek  <polacek@redhat.com>
10891
10892         PR c/57773
10893         * doc/implement-c.texi: Mention that other integer types are
10894         permitted as bit-field types in strictly conforming mode.
10895
10896 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
10897
10898         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
10899         is newly allocated.
10900
10901 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
10902
10903         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
10904
10905 2014-01-06  Martin Jambor  <mjambor@suse.cz>
10906
10907         PR ipa/59008
10908         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
10909         to int.
10910         * ipa-prop.c (ipa_print_node_params): Fix indentation.
10911
10912 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
10913
10914         PR debug/59350
10915         PR debug/59510
10916         * var-tracking.c (add_stores): Preserve the value of the source even if
10917         we don't record the store.
10918
10919 2014-01-06  Terry Guo  <terry.guo@arm.com>
10920
10921         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
10922
10923 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
10924
10925         PR bootstrap/59541
10926         * config/darwin.c (darwin_function_section): Adjust return values to
10927         correspond to optimisation changes made in r206070.
10928
10929 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
10930
10931         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
10932         from prefetch_block tune setting.
10933         (nocona_cost): Correct size of prefetch block to 64.
10934
10935 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
10936
10937         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
10938         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
10939         used to save the static chain register in the computation of the offset
10940         from which the FP registers need to be restored.
10941
10942 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
10943
10944         PR tree-optimization/59519
10945         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
10946         ICE if get_current_def (current_new_name) is already non-NULL, as long
10947         as it is a phi result of some other phi in *new_exit_bb that has
10948         the same argument.
10949
10950         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
10951         or vmovdqu* for misaligned_operand.
10952         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
10953         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
10954         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
10955         aligned_mem for AVX512F masked aligned load and store builtins and for
10956         non-temporal moves.
10957
10958 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
10959
10960         PR tree-optimization/59651
10961         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
10962         Address range for negative step should be added by TYPE_SIZE_UNIT.
10963
10964 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
10965
10966         * config/m68k/m68k.c (handle_move_double): Handle pushes with
10967         overlapping registers also for registers other than the stack pointer.
10968
10969 2014-01-03  Marek Polacek  <polacek@redhat.com>
10970
10971         PR other/59661
10972         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
10973         __builtin_FILE.
10974
10975 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
10976
10977         PR target/59625
10978         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
10979         asm goto as jump.
10980
10981         * config/i386/i386.md (MODE_SIZE): New mode attribute.
10982         (push splitter): Use <P:MODE_SIZE> instead of
10983         GET_MODE_SIZE (<P:MODE>mode).
10984         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
10985         (mov -1, reg peephole2): Likewise.
10986         * config/i386/sse.md (*mov<mode>_internal,
10987         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
10988         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
10989         *<code><mode>3, *andnot<mode>3<mask_name>,
10990         <mask_codefor><code><mode>3<mask_name>): Likewise.
10991         * config/i386/subst.md (mask_mode512bit_condition,
10992         sd_mask_mode512bit_condition): Likewise.
10993
10994 2014-01-02  Xinliang David Li  <davidxl@google.com>
10995
10996         PR tree-optimization/59303
10997         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
10998         (dump_predicates): Better output format.
10999         (pred_equal_p): New function.
11000         (is_neq_relop_p): Ditto.
11001         (is_neq_zero_form_p): Ditto.
11002         (pred_expr_equal_p): Ditto.
11003         (pred_neg_p): Ditto.
11004         (simplify_pred): Ditto.
11005         (simplify_preds_2): Ditto.
11006         (simplify_preds_3): Ditto.
11007         (simplify_preds_4): Ditto.
11008         (simplify_preds): Ditto.
11009         (push_pred): Ditto.
11010         (push_to_worklist): Ditto.
11011         (get_pred_info_from_cmp): Ditto.
11012         (is_degenerated_phi): Ditto.
11013         (normalize_one_pred_1): Ditto.
11014         (normalize_one_pred): Ditto.
11015         (normalize_one_pred_chain): Ditto.
11016         (normalize_preds): Ditto.
11017         (normalize_cond_1): Remove function.
11018         (normalize_cond): Ditto.
11019         (is_gcond_subset_of): Ditto.
11020         (is_subset_of_any): Ditto.
11021         (is_or_set_subset_of): Ditto.
11022         (is_and_set_subset_of): Ditto.
11023         (is_norm_cond_subset_of): Ditto.
11024         (pred_chain_length_cmp): Ditto.
11025         (convert_control_dep_chain_into_preds): Type change.
11026         (find_predicates): Ditto.
11027         (find_def_preds): Ditto.
11028         (destroy_predicates_vecs): Ditto.
11029         (find_matching_predicates_in_rest_chains): Ditto.
11030         (use_pred_not_overlap_with_undef_path_pred): Ditto.
11031         (is_pred_expr_subset): Ditto.
11032         (is_pred_chain_subset_of): Ditto.
11033         (is_included_in): Ditto.
11034         (is_superset_of): Ditto.
11035
11036 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
11037
11038         Update copyright years.
11039
11040 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
11041
11042         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
11043         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
11044         config/arc/arc.md, config/arc/arc.opt,
11045         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
11046         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
11047         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
11048         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
11049         config/linux-protos.h, config/linux.c, config/winnt-c.c,
11050         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
11051         vtable-verify.c, vtable-verify.h: Use the standard form for the
11052         copyright notice.
11053
11054 2014-01-02  Tobias Burnus  <burnus@net-b.de>
11055
11056         * gcc.c (process_command): Update copyright notice dates.
11057         * gcov-dump.c: Ditto.
11058         * gcov.c: Ditto.
11059         * doc/cpp.texi: Bump @copying's copyright year.
11060         * doc/cppinternals.texi: Ditto.
11061         * doc/gcc.texi: Ditto.
11062         * doc/gccint.texi: Ditto.
11063         * doc/gcov.texi: Ditto.
11064         * doc/install.texi: Ditto.
11065         * doc/invoke.texi: Ditto.
11066
11067 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11068
11069         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
11070
11071 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
11072
11073         * config/i386/sse.md (*mov<mode>_internal): Guard
11074         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
11075
11076         PR rtl-optimization/59647
11077         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
11078         new_rtx into UNSIGNED_FLOAT rtxes.
11079 \f
11080 Copyright (C) 2014 Free Software Foundation, Inc.
11081
11082 Copying and distribution of this file, with or without modification,
11083 are permitted in any medium without royalty provided the copyright
11084 notice and this notice are preserved.