gcc/ChangeLog

   1 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
   2
   3         * fold-const.c (optimize_bit_field_compare): Use wi:: operations
   4         instead of const_binop.
   5         (fold_binary_loc): Likewise.
   6
   7 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
   8
   9         * tree-dfa.h (get_addr_base_and_unit_offset_1): Update array index
  10         calculation to match get_ref_base_and_extent.
  11
  12 2014-05-13  Catherine Moore  <clm@codesourcery.com>
  13             Sandra Loosemore  <sandra@codesourcery.com>
  14
  15         * configure.ac: Fix assembly for explicit JALR relocation check.
  16         * configure: Regenerate.
  17
  18 2014-05-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  19
  20         * config/arm/arm.c (neon_itype): Remove NEON_RESULTPAIR.
  21         (arm_init_neon_builtins): Remove handling of NEON_RESULTPAIR.
  22         Remove associated type declarations and initialisations.
  23         (arm_expand_neon_builtin): Likewise.
  24         (neon_emit_pair_result_insn): Delete.
  25         * config/arm/arm_neon_builtins (vtrn, vzip, vuzp): Delete.
  26         * config/arm/neon.md (neon_vtrn<mode>): Delete.
  27         (neon_vzip<mode>): Likewise.
  28         (neon_vuzp<mode>): Likewise.
  29
  30 2014-05-13  Richard Biener  <rguenther@suse.de>
  31
  32         PR ipa/60973
  33         * tree-inline.c (remap_gimple_stmt): Clear tail call flag,
  34         it needs revisiting whether the call still may be tail-called.
  35
  36 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
  37
  38         * rtl.def (SYMBOL_REF): Remove middle "0" field.
  39         * rtl.h (block_symbol): Reduce number of fields to 2.
  40         (rtx_def): Add u2.symbol_ref_flags.
  41         (SYMBOL_REF_FLAGS): Use it.
  42         (SYMBOL_REF_DATA, SET_SYMBOL_REF_DECL, SYMBOL_REF_DECL)
  43         (SET_SYMBOL_REF_CONSTANT, SYMBOL_REF_CONSTANT): Lower index.
  44         * gengtype.c (adjust_field_rtx_def): Remove SYMBOL_REF_FLAGS handling.
  45         Lower index of SYMBOL_REF_DATA.
  46         * print-rtl.c (print_rtx): Lower index for SYMBOL_REF_DATA.
  47         Print SYMBOL_REF_FLAGS at the same time.
  48         * genattrtab.c (attr_rtx_1): Only initialize 1 "0" SYMBOL_REF field.
  49
  50 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
  51
  52         * rtl.def (VAR_LOCATION): Remove "i" field.
  53         * rtl.h (rtx_def): Add u2.var_location_status.
  54         (PAT_VAR_LOCATION_STATUS): Use it.
  55         (gen_rtx_VAR_LOCATION): Declare.
  56         * gengenrtl.c (excluded_rtx): Add VAR_LOCATION.
  57         * emit-rtl.c (gen_rtx_VAR_LOCATION): New function.
  58         * var-tracking.c (emit_note_insn_var_location): Remove casts.
  59
  60 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
  61
  62         * rtl.def (scratch): Fix outdated comment and remove "0" field.
  63         * gengtype.c (adjust_field_rtx_def): Update accordingly.
  64
  65 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
  66
  67         * rtl.def (DEBUG_INSN, INSN, JUMP_INSN, CALL_INSN, JUMP_TABLE_DATA)
  68         (BARRIER, CODE_LABEL, NOTE): Remove first "i" field.
  69         * rtl.h (rtx_def): Add insn_uid to u2 field.
  70         (RTX_FLAG_CHECK8): Delete in favor of...
  71         (RTL_INSN_CHAIN_FLAG_CHECK): ...this new macro.
  72         (INSN_DELETED_P): Update accordingly.
  73         (INSN_UID): Use u2.insn_uid.
  74         (INSN_CHAIN_CODE_P): Define.
  75         (PREV_INSN, NEXT_INSN, BLOCK_FOR_INSN, PATTERN, INSN_LOCATION)
  76         (INSN_CODE, REG_NOTES, CALL_INSN_FUNCTION_USAGE, CODE_LABEL_NUMBER)
  77         (NOTE_DATA, NOTE_DELETED_LABEL_NAME, NOTE_BLOCK, NOTE_EH_HANDLER)
  78         (NOTE_BASIC_BLOCK, NOTE_VAR_LOCATION, NOTE_CFI, NOTE_LABEL_NUMBER)
  79         (NOTE_KIND, LABEL_NAME, LABEL_NUSES, JUMP_LABEL, LABEL_REFS): Lower
  80         indices accordingly.
  81         * print-rtl.c (print_rtx): Print INSN_UIDs before the main loop.
  82         Update indices for insn-chain rtxes.
  83         * gengtype.c (gen_rtx_next): Adjust test for insn-chain rtxes.
  84         (adjust_field_rtx_def): Lower '0' indices for all insn-chain rtxes.
  85         * emit-rtl.c (gen_label_rtx): Update gen_rtx_LABEL call.
  86         * caller-save.c (init_caller_save): Update gen_rtx_INSN calls.
  87         * combine.c (try_combine): Likewise.
  88         * ira.c (setup_prohibited_mode_move_regs): Likewise.
  89
  90 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
  91
  92         * rtl.def (REG): Remove middle field.
  93         * rtl.h (rtx_def): Add orignal_regno to u2.
  94         (ORIGINAL_REGNO): Use it instead of field 1.
  95         (REG_ATTRS): Lower field index accordingly.
  96         * gengtype.c (adjust_field_rtx_def): Remove handling of
  97         ORIGINAL_REGNO.  Move REG_ATTRS index down.
  98         * print-rtl.c (print_rtx): Move ORIGINAL_REGNO handling to the
  99         code that prints the REGNO.
 100
 101 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 102
 103         * print-rtl.c (print_rtx): Guard whole '0' block with ifndef
 104         GENERATOR_FILE.
 105
 106 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
 107
 108         * rtl.h (rtx_def): Mark u2 as GTY ((skip)).
 109
 110 2014-05-13  Bin Cheng  <bin.cheng@arm.com>
 111
 112         * tree-ssa-loop-ivopts.c (contain_complex_addr_expr): New.
 113         (alloc_iv): Lower base expressions containing ADDR_EXPR.
 114
 115 2014-05-13  Ian Bolton  <ian.bolton@arm.com>
 116
 117         * config/aarch64/aarch64-protos.h
 118         (aarch64_hard_regno_caller_save_mode): New prototype.
 119         * config/aarch64/aarch64.c (aarch64_hard_regno_caller_save_mode):
 120         New function.
 121         * config/aarch64/aarch64.h (HARD_REGNO_CALLER_SAVE_MODE): New macro.
 122
 123 2014-05-13  Christian Bruel  <christian.bruel@st.com>
 124
 125         * target.def (mode_switching): New hook vector.
 126         (mode_emit, mode_needed, mode_after, mode_entry): New hooks.
 127         (mode_exit, modepriority_to_mode): Likewise.
 128         * mode-switching.c (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Hookify.
 129         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 130         (default_priority_to_mode): Define.
 131         * targhooks.h (default_priority_to_mode): Declare.
 132         * target.h: Include tm.h and hard-reg-set.h.
 133         * doc/tm.texi.in (EMIT_MODE_SET, MODE_NEEDED, MODE_AFTER, MODE_ENTRY)
 134         (MODE_EXIT, MODE_PRIORITY_TO_MODE): Delete and hookify.
 135         * doc/tm.texi Regenerate.
 136         * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 137         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 138         * config/sh/sh.c (sh_emit_mode_set, sh_mode_priority): Hookify.
 139         (sh_mode_needed, sh_mode_after, sh_mode_entry, sh_mode_exit): Likewise.
 140         * config/i386/i386.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 141         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 142         * config/i386/i386-protos.h (ix86_mode_needed, ix86_mode_after)
 143         (ix86_mode_entrym, ix86_emit_mode_set): Remove external declaration.
 144         * config/i386/i386.c (ix86_mode_needed, ix86_mode_after, ix86_mode_exit,
 145         (ix86_mode_entry, ix86_mode_priority, ix86_emit_mode_set): Hookify.
 146         * config/epiphany/epiphany.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY):
 147         Delete
 148         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 149         * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
 150         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
 151         * config/sh/sh.c (sh4_emit_mode_set, sh4_mode_needed): Hookify.
 152         (sh4_mode_after, sh4_mode_entry, sh4_mode_exit): Likewise.
 153         * config/epiphany/epiphany-protos.h (epiphany_mode_needed)
 154         (emit_set_fp_mode, epiphany_mode_entry_exit, epiphany_mode_after)
 155         (epiphany_mode_priority_to_mode): Remove declaration.
 156         * config/epiphany/epiphany.c (emit_set_fp_mode): Hookify.
 157         (epiphany_mode_needed, epiphany_mode_priority_to_mode): Likewise.
 158         (epiphany_mode_entry, epiphany_mode_exit, epiphany_mode_after):
 159         Likewise.
 160         (epiphany_mode_priority_to_mode): Change priority type. Hookify.
 161         (epiphany_mode_needed, epiphany_mode_entry_exit): Hookify.
 162         (epiphany_mode_after, epiphany_mode_entry, emit_set_fp_mode): Hookify.
 163
 164 2014-05-13  Jakub Jelinek  <jakub@redhat.com>
 165
 166         PR target/61060
 167         * config/i386/i386.c (ix86_expand_set_or_movmem): If count_exp
 168         is const0_rtx, return immediately.  Don't test count == 0 when
 169         it is always true.
 170
 171 2014-05-13  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
 172
 173         * Makefile.in: add shrink-wrap.o.
 174         * config/i386/i386.c: include "shrink-wrap.h"
 175         * function.c: Likewise.
 176         (requires_stack_frame_p, next_block_for_reg,
 177         move_insn_for_shrink_wrap, prepare_shrink_wrap,
 178         dup_block_and_redirect): Move to shrink-wrap.c
 179         (thread_prologue_and_epilogue_insns): Extract three code segments
 180         as functions in shrink-wrap.c
 181         * function.h: Move #ifdef HAVE_simple_return ... #endif block to
 182         shrink-wrap.h
 183         * shrink-wrap.c: New file.
 184         * shrink-wrap.h: New file.
 185
 186 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
 187
 188         * doc/extend.texi: Reflect current numbers of pragmas.  Remove
 189         reference to Solaris.
 190
 191 2014-05-12  Mike Stump  <mikestump@comcast.net>
 192
 193         PR other/31778
 194         * genattrtab.c (filename): Add.
 195         (convert_set_attr_alternative): Improve error message.
 196         (check_defs): Restore read_md_filename for error messages.
 197         (gen_insn): Save filename.
 198
 199 2014-05-12  Dimitris Papavasiliou  <dpapavas@gmail.com>
 200
 201         * doc/invoke.texi: Document new switches -Wno-shadow-ivar,
 202         -fno-local-ivars and -fivar-visibility.
 203         * c-family/c.opt: Make -Wshadow also implicitly enable
 204         -Wshadow-ivar.
 205
 206 2014-05-12  David Wohlferd <dw@LimeGreenSocks.com>
 207
 208         * doc/tm.texi: Remove reference to deleted macro.
 209         * doc/tm.texi.in: Likewise.
 210
 211 2014-05-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
 212
 213         PR target/60991
 214         * config/avr/avr.c (avr_out_store_psi): Use correct constant
 215         to restore Y.
 216
 217 2014-05-12  Georg-Johann Lay  <avr@gjlay.de>
 218
 219         * config/arm/arm.h (License): Add GCC Runtime Library Exception.
 220         * config/arm/aout.h (License): Same.
 221         * config/arm/bpabi.h (License): Same.
 222         * config/arm/elf.h (License): Same.
 223         * config/arm/linux-elf.h (License): Same.
 224         * config/arm/linux-gas.h (License): Same.
 225         * config/arm/netbsd-elf.h (License): Same.
 226         * config/arm/uclinux-eabi.h (License): Same.
 227         * config/arm/uclinux-elf.h (License): Same.
 228         * config/arm/vxworks.h (License): Same.
 229
 230 2014-05-11  Jakub Jelinek  <jakub@redhat.com>
 231
 232         * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
 233         * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
 234         number of operands to 3.
 235         (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
 236         * tree-nested.c (convert_nonlocal_omp_clauses,
 237         convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
 238         * gimplify.c (gimplify_scan_omp_clauses): Handle
 239         OMP_CLAUSE_LINEAR_STMT.
 240         * omp-low.c (lower_rec_input_clauses): Fix typo.
 241         (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
 242         cast between Fortran boolean_type_node and C _Bool if
 243         needed.
 244
 245 2014-05-11  Richard Sandiford  <rdsandiford@googlemail.com>
 246
 247         PR tree-optimization/61136
 248         * wide-int.h (multiple_of_p): Define a version that doesn't return
 249         the quotient.
 250         * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
 251         integer_zerop/const_binop pair.
 252         (multiple_of_p): Likewise, converting both operands to widest_int
 253         precision.
 254
 255 2014-05-09  Teresa Johnson  <tejohnson@google.com>
 256
 257         * cgraphunit.c (analyze_functions): Use correct dump file.
 258
 259 2014-05-09  Florian Weimer  <fweimer@redhat.com>
 260
 261         * cfgexpand.c (stack_protect_decl_p): New function, extracted from
 262         expand_used_vars.
 263         (stack_protect_return_slot_p): New function.
 264         (expand_used_vars): Call stack_protect_decl_p and
 265         stack_protect_return_slot_p for -fstack-protector-strong.
 266
 267 2014-05-09  Kenneth Zadeck  <zadeck@naturalbridge.com>
 268
 269         PR middle-end/61111
 270         * fold-const.c (fold_binary_loc): Changed width of mask.
 271
 272 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 273
 274         * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
 275         unsigned int initializers for regno_in, regno_out.
 276
 277 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
 278
 279         PR target/61055
 280         * config/avr/avr.md (cc): Add new attribute set_vzn.
 281         (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
 282         Set cc insn attribute to set_vzn instead of set_zn for alternatives
 283         with INC, DEC or NEG.
 284         * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
 285         (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
 286         INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
 287
 288 2014-05-09  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 289
 290         Revert:
 291         2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 292
 293         * wide-int.cc (UTItype): Define.
 294         (UDWtype): Define for appropriate W_TYPE_SIZE.
 295
 296 2014-05-09  Richard Biener  <rguenther@suse.de>
 297
 298         * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
 299         * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
 300         (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
 301         (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
 302         ssa_propagate): Adjust.
 303
 304 2014-05-08  Jeff Law  <law@redhat.com>
 305
 306         PR tree-optimization/61009
 307         * tree-ssa-threadedge.c (thread_through_normal_block): Return a
 308         tri-state rather than a boolean.  When a block is too big to
 309         thread through, inform caller via negative return value.
 310         (thread_across_edge): If a block was too big for normal threading,
 311         then it's too big for a joiner too, so remove temporary equivalences
 312         and return immediately.
 313
 314 2014-05-08  Manuel López-Ibáñez  <manu@gcc.gnu.org>
 315             Matthias Klose  <doko@ubuntu.com>
 316
 317         PR driver/61106
 318         * optc-gen.awk: Fix option handling for -Wunused-parameter.
 319
 320 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 321
 322         PR target/59952
 323         * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
 324
 325 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
 326
 327         PR target/61092
 328         * config/alpha/alpha.c: Include gimple-iterator.h.
 329         (alpha_gimple_fold_builtin): New function.  Move
 330         ALPHA_BUILTIN_UMULH folding from ...
 331         (alpha_fold_builtin): ... here.
 332         (TARGET_GIMPLE_FOLD_BUILTIN): New define.
 333
 334 2014-05-08  Wei Mi  <wmi@google.com>
 335
 336         PR target/58066
 337         * config/i386/i386.c (ix86_compute_frame_layout): Update
 338         preferred_stack_boundary for call, expanded from tls descriptor.
 339         * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
 340         to depend on SP register.
 341         (*tls_local_dynamic_base_32_gnu): Ditto.
 342         (*tls_local_dynamic_32_once): Ditto.
 343         (tls_global_dynamic_64_<mode>): Set
 344         ix86_tls_descriptor_calls_expanded_in_cfun.
 345         (tls_local_dynamic_base_64_<mode>): Ditto.
 346         (tls_global_dynamic_32): Set
 347         ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
 348         to depend on SP register.
 349         (tls_local_dynamic_base_32): Ditto.
 350
 351 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 352
 353         * config/arm/arm_neon.h: Update comment.
 354         * config/arm/neon-docgen.ml: Delete.
 355         * config/arm/neon-gen.ml: Delete.
 356         * doc/arm-neon-intrinsics.texi: Update comment.
 357
 358 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 359
 360         * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
 361         and v4sf versions.
 362         (vand, vorr, veor, vorn, vbic): Remove.
 363         * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
 364         iterator.
 365         (neon_vsub_unspec): Likewise.
 366         (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
 367
 368 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 369
 370         * config/arm/arm_neon.h (vadd_s8): GNU C implementation
 371         (vadd_s16): Likewise.
 372         (vadd_s32): Likewise.
 373         (vadd_f32): Likewise.
 374         (vadd_u8): Likewise.
 375         (vadd_u16): Likewise.
 376         (vadd_u32): Likewise.
 377         (vadd_s64): Likewise.
 378         (vadd_u64): Likewise.
 379         (vaddq_s8): Likewise.
 380         (vaddq_s16): Likewise.
 381         (vaddq_s32): Likewise.
 382         (vaddq_s64): Likewise.
 383         (vaddq_f32): Likewise.
 384         (vaddq_u8): Likewise.
 385         (vaddq_u16): Likewise.
 386         (vaddq_u32): Likewise.
 387         (vaddq_u64): Likewise.
 388         (vmul_s8): Likewise.
 389         (vmul_s16): Likewise.
 390         (vmul_s32): Likewise.
 391         (vmul_f32): Likewise.
 392         (vmul_u8): Likewise.
 393         (vmul_u16): Likewise.
 394         (vmul_u32): Likewise.
 395         (vmul_p8): Likewise.
 396         (vmulq_s8): Likewise.
 397         (vmulq_s16): Likewise.
 398         (vmulq_s32): Likewise.
 399         (vmulq_f32): Likewise.
 400         (vmulq_u8): Likewise.
 401         (vmulq_u16): Likewise.
 402         (vmulq_u32): Likewise.
 403         (vsub_s8): Likewise.
 404         (vsub_s16): Likewise.
 405         (vsub_s32): Likewise.
 406         (vsub_f32): Likewise.
 407         (vsub_u8): Likewise.
 408         (vsub_u16): Likewise.
 409         (vsub_u32): Likewise.
 410         (vsub_s64): Likewise.
 411         (vsub_u64): Likewise.
 412         (vsubq_s8): Likewise.
 413         (vsubq_s16): Likewise.
 414         (vsubq_s32): Likewise.
 415         (vsubq_s64): Likewise.
 416         (vsubq_f32): Likewise.
 417         (vsubq_u8): Likewise.
 418         (vsubq_u16): Likewise.
 419         (vsubq_u32): Likewise.
 420         (vsubq_u64): Likewise.
 421         (vand_s8): Likewise.
 422         (vand_s16): Likewise.
 423         (vand_s32): Likewise.
 424         (vand_u8): Likewise.
 425         (vand_u16): Likewise.
 426         (vand_u32): Likewise.
 427         (vand_s64): Likewise.
 428         (vand_u64): Likewise.
 429         (vandq_s8): Likewise.
 430         (vandq_s16): Likewise.
 431         (vandq_s32): Likewise.
 432         (vandq_s64): Likewise.
 433         (vandq_u8): Likewise.
 434         (vandq_u16): Likewise.
 435         (vandq_u32): Likewise.
 436         (vandq_u64): Likewise.
 437         (vorr_s8): Likewise.
 438         (vorr_s16): Likewise.
 439         (vorr_s32): Likewise.
 440         (vorr_u8): Likewise.
 441         (vorr_u16): Likewise.
 442         (vorr_u32): Likewise.
 443         (vorr_s64): Likewise.
 444         (vorr_u64): Likewise.
 445         (vorrq_s8): Likewise.
 446         (vorrq_s16): Likewise.
 447         (vorrq_s32): Likewise.
 448         (vorrq_s64): Likewise.
 449         (vorrq_u8): Likewise.
 450         (vorrq_u16): Likewise.
 451         (vorrq_u32): Likewise.
 452         (vorrq_u64): Likewise.
 453         (veor_s8): Likewise.
 454         (veor_s16): Likewise.
 455         (veor_s32): Likewise.
 456         (veor_u8): Likewise.
 457         (veor_u16): Likewise.
 458         (veor_u32): Likewise.
 459         (veor_s64): Likewise.
 460         (veor_u64): Likewise.
 461         (veorq_s8): Likewise.
 462         (veorq_s16): Likewise.
 463         (veorq_s32): Likewise.
 464         (veorq_s64): Likewise.
 465         (veorq_u8): Likewise.
 466         (veorq_u16): Likewise.
 467         (veorq_u32): Likewise.
 468         (veorq_u64): Likewise.
 469         (vbic_s8): Likewise.
 470         (vbic_s16): Likewise.
 471         (vbic_s32): Likewise.
 472         (vbic_u8): Likewise.
 473         (vbic_u16): Likewise.
 474         (vbic_u32): Likewise.
 475         (vbic_s64): Likewise.
 476         (vbic_u64): Likewise.
 477         (vbicq_s8): Likewise.
 478         (vbicq_s16): Likewise.
 479         (vbicq_s32): Likewise.
 480         (vbicq_s64): Likewise.
 481         (vbicq_u8): Likewise.
 482         (vbicq_u16): Likewise.
 483         (vbicq_u32): Likewise.
 484         (vbicq_u64): Likewise.
 485         (vorn_s8): Likewise.
 486         (vorn_s16): Likewise.
 487         (vorn_s32): Likewise.
 488         (vorn_u8): Likewise.
 489         (vorn_u16): Likewise.
 490         (vorn_u32): Likewise.
 491         (vorn_s64): Likewise.
 492         (vorn_u64): Likewise.
 493         (vornq_s8): Likewise.
 494         (vornq_s16): Likewise.
 495         (vornq_s32): Likewise.
 496         (vornq_s64): Likewise.
 497         (vornq_u8): Likewise.
 498         (vornq_u16): Likewise.
 499         (vornq_u32): Likewise.
 500         (vornq_u64): Likewise.
 501
 502 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 503
 504         * wide-int.cc (UTItype): Define.
 505         (UDWtype): Define for appropriate W_TYPE_SIZE.
 506
 507 2014-05-08  Marc Glisse  <marc.glisse@inria.fr>
 508
 509         PR tree-optimization/59100
 510         * tree-ssa-phiopt.c: Include tree-inline.h.
 511         (neutral_element_p, absorbing_element_p): New functions.
 512         (value_replacement): Handle conditional binary operations with a
 513         neutral or absorbing element.
 514
 515 2014-05-08  Richard Biener  <rguenther@suse.de>
 516
 517         * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
 518         folding the expression.
 519         (valueize_expr): Remove.
 520         (visit_reference_op_load): Do not valueize the result of
 521         vn_get_expr_for.
 522         (simplify_binary_expression): Likewise.
 523         (simplify_unary_expression): Likewise.
 524
 525 2014-05-08  Richard Biener  <rguenther@suse.de>
 526
 527         * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
 528         looking at TYPE_ARG_TYPES.
 529
 530 2014-05-08  Richard Biener  <rguenther@suse.de>
 531
 532         * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
 533         pointer propagation special-case.
 534
 535 2014-05-08  Bin Cheng  <bin.cheng@arm.com>
 536
 537         * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
 538         core part of address expressions.
 539
 540 2014-05-08  Alan Modra  <amodra@gmail.com>
 541
 542         PR target/60737
 543         * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
 544         loads and stores when -mno-strict-align at any alignment.
 545         (expand_block_clear): Similarly.  Also correct calculation of
 546         instruction count.
 547
 548 2014-05-07  Thomas Preud'homme  <thomas.preudhomme@arm.com>
 549
 550         PR middle-end/39246
 551         * tree-complex.c (expand_complex_move): Keep line info when expanding
 552         complex move.
 553         * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
 554         of complex expression. Use new argument to display correct location
 555         for values coming from phi statement.
 556         (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
 557         (warn_uninitialized_phi): Pass location of phi argument to
 558         warn_uninit.
 559         * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
 560         COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
 561
 562 2014-05-07  Segher Boessenkool  <segher@kernel.crashing.org>
 563
 564         * config/rs6000/predicates.md (indexed_address_mem): New.
 565         * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
 566         load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
 567         fpstore_ux, fpstore_u.
 568         (sign_extend, indexed, update): New.
 569         (cell_micro): Adjust.
 570         (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
 571         *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
 572         *movsi_internal1, *movsi_internal1_single, *movhi_internal,
 573         *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
 574         *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
 575         *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
 576         *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
 577         *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
 578         *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
 579         *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
 580         *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
 581         *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
 582         *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
 583         *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
 584         *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
 585
 586         * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
 587         * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
 588         *vsx_extract_<mode>_store): Adjust.
 589         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
 590         is_cracked_insn, insn_must_be_first_in_group,
 591         insn_must_be_last_in_group): Adjust.
 592
 593         * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
 594         Adjust.
 595         * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
 596         ppc440-fpstore): Adjust.
 597         * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
 598         ppc476-fpstore): Adjust.
 599         * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
 600         ppc601-fpstore): Adjust.
 601         * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
 602         Adjust.
 603         * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
 604         Adjust.
 605         * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
 606         ppc7450-fpstore): Adjust.
 607         * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
 608         * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
 609         * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
 610         Adjust.
 611         * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
 612         cell-fpload, cell-fpload-update, cell-store, cell-store-update,
 613         cell-fpstore, cell-fpstore-update): Adjust.
 614         * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
 615         ppce300c3_store, ppce300c3_fpstore): Adjust.
 616         * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
 617         e500mc_fpstore): Adjust.
 618         * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
 619         e500mc64_store, e500mc64_fpstore): Adjust.
 620         * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
 621         e5500_fpstore): Adjust.
 622         * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
 623         e6500_fpstore): Adjust.
 624         * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
 625         Adjust.
 626         * config/rs6000/power4.md (power4-load, power4-load-ext,
 627         power4-load-ext-update, power4-load-ext-update-indexed,
 628         power4-load-update-indexed, power4-load-update, power4-fpload,
 629         power4-fpload-update, power4-store, power4-store-update,
 630         power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
 631         Adjust.
 632         * config/rs6000/power5.md (power5-load, power5-load-ext,
 633         power5-load-ext-update, power5-load-ext-update-indexed,
 634         power5-load-update-indexed, power5-load-update, power5-fpload,
 635         power5-fpload-update, power5-store, power5-store-update,
 636         power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
 637         Adjust.
 638         * config/rs6000/power6.md (power6-load, power6-load-ext,
 639         power6-load-update, power6-load-update-indexed,
 640         power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
 641         power6-fpload-update, power6-store, power6-store-update,
 642         power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
 643         Adjust.
 644         * config/rs6000/power7.md (power7-load, power7-load-ext,
 645         power7-load-update, power7-load-update-indexed,
 646         power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
 647         power7-fpload-update, power7-store, power7-store-update,
 648         power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
 649         Adjust.
 650         * config/rs6000/power8.md (power8-load, power8-load-update,
 651         power8-load-ext, power8-load-ext-update, power8-fpload,
 652         power8-fpload-update, power8-store, power8-store-update-indexed,
 653         power8-fpstore, power8-fpstore-update): Adjust.
 654         * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
 655         Adjust.
 656         * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
 657         titan_lsu_store, titan_lsu_fpstore): Adjust.
 658         * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
 659
 660 2014-05-07  Oleg Endo  <olegendo@gcc.gnu.org>
 661
 662         PR target/60884
 663         * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
 664         unrolled byte insns.  Emit address increments after move insns.
 665
 666 2014-05-07  David Malcolm  <dmalcolm@redhat.com>
 667
 668         * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
 669         const_gimple, rather than a gimple.
 670         (gimple_call_builtin_p): Likewise, for the three variants.
 671
 672         * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
 673         (gimple_call_builtin_p): Likewise, for the three variants.
 674
 675 2014-05-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
 676
 677         PR tree-optimization/61095
 678         * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
 679
 680 2014-05-07  Richard Biener  <rguenther@suse.de>
 681
 682         PR tree-optimization/61034
 683         * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
 684         (maybe_skip_until): Use translate to take into account
 685         lattices when trying to do disambiguations.
 686         (get_continuation_for_phi_1): Likewise.
 687         (get_continuation_for_phi): Adjust for added translate arguments.
 688         (walk_non_aliased_vuses): Likewise.
 689         * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
 690         (walk_non_aliased_vuses): Likewise.
 691         (call_may_clobber_ref_p_1): Declare.
 692         * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
 693         calls.  Stop early if we are only supposed to disambiguate.
 694         * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
 695
 696 2014-05-07  Joern Rennecke  <joern.rennecke@embecosm.com>
 697
 698         * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
 699         Emit an error when the function has arguments.
 700
 701 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
 702
 703         * cfgloop.h (unswitch_loops): Remove.
 704         * doc/passes.texi: Remove references to loop-unswitch.c
 705         * timevar.def (TV_LOOP_UNSWITCH): Remove.
 706
 707 2014-05-07  Evgeny Stupachenko  <evstupac@gmail.com>
 708
 709         * tree-vect-data-refs.c (vect_grouped_load_supported): New
 710         check for loads group of length 3.
 711         (vect_permute_load_chain): New permutations for loads group of
 712         length 3.
 713         * tree-vect-stmts.c (vect_model_load_cost): Change cost
 714         of vec_perm_shuffle for the new permutations.
 715
 716 2014-05-07  Alan Lawrence  <alan.lawrence@arm.com>
 717
 718         * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
 719         vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
 720         vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
 721         vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
 722         vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
 723         vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
 724         vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
 725         vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
 726
 727 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
 728
 729         * loop-unswitch.c: Delete.
 730
 731 2014-05-07  Richard Biener  <rguenther@suse.de>
 732
 733         * config.gcc: Always set need_64bit_hwint to yes.
 734
 735 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
 736
 737         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
 738         of using optimize_size.
 739
 740 2014-05-06  Mike Stump  <mikestump@comcast.net>
 741
 742         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
 743
 744 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
 745
 746         * config/i386/sse.md (*mov<mode>_internal)
 747         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
 748         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
 749         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
 750         (*<code><mode>3, *andnot<mode>3<mask_name>)
 751         (<mask_codefor><code><mode>3<mask_name>): Only consider
 752         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
 753
 754 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
 755
 756         Revert:
 757         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
 758
 759         * lra-constraints.c (valid_address_p): Move earlier in file.
 760         Add a constraint argument to the address_info version.
 761         (satisfies_memory_constraint_p): New function.
 762         (satisfies_address_constraint_p): Likewise.
 763         (process_alt_operands, curr_insn_transform): Use them.
 764         (process_address): Pass the constraint to valid_address_p when
 765         checking address operands.
 766
 767 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
 768
 769         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
 770         to their respective blocks.  Fix inadvertent use of "node".
 771
 772 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
 773
 774         * emit-rtl.c (init_derived_machine_modes): New functionm, split
 775         out from...
 776         (init_emit_once): ...here.
 777         * rtl.h (init_derived_machine_modes): Declare.
 778         * toplev.c (do_compile): Call it even if no_backend.
 779
 780 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
 781             Mike Stump  <mikestump@comcast.net>
 782             Richard Sandiford  <rdsandiford@googlemail.com>
 783             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 784
 785         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
 786         (rtx_equal_for_memref_p): Update comment.
 787         (adjust_offset_for_component_ref): Use wide-int interfaces.
 788         * builtins.c (get_object_alignment_2): Likewise.
 789         (c_readstr): Likewise.
 790         (target_char_cast): Add comment.
 791         (determine_block_size): Use wide-int interfaces.
 792         (expand_builtin_signbit): Likewise.
 793         (fold_builtin_int_roundingfn): Likewise.
 794         (fold_builtin_bitop): Likewise.
 795         (fold_builtin_bswap): Likewise.
 796         (fold_builtin_logarithm): Use signop.
 797         (fold_builtin_pow): Likewise.
 798         (fold_builtin_memory_op): Use wide-int interfaces.
 799         (fold_builtin_object_size): Likewise.
 800         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
 801         nb_iterations_estimate.
 802         (record_niter_bound): Use wide-int interfaces.
 803         (get_estimated_loop_iterations_int): Likewise.
 804         (get_estimated_loop_iterations): Likewise.
 805         (get_max_loop_iterations): Likewise.
 806         * cfgloop.h: Include wide-int.h.
 807         (struct nb_iter_bound): Change bound to widest_int.
 808         (struct loop): Change nb_iterations_upper_bound and
 809         nb_iterations_estimate to widest_int.
 810         (record_niter_bound): Switch to use widest_int.
 811         (get_estimated_loop_iterations): Likewise.
 812         (get_max_loop_iterations): Likewise.
 813         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
 814         update for wide-int.
 815         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
 816         * combine.c (try_combine): Likewise.
 817         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
 818         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
 819         interfaces.
 820         (aarch64_float_const_representable_p): Likewise.
 821         * config/arc/arc.c: Include wide-int.h.
 822         (arc_can_use_doloop_p): Use wide-int interfaces.
 823         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
 824         (vfp3_const_double_index): Likewise.
 825         * config/avr/avr.c (avr_out_round): Likewise.
 826         (avr_fold_builtin): Likewise.
 827         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
 828         (bfin_can_use_doloop_p): Likewise.
 829         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
 830         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
 831         * config/i386/i386.c: Include wide-int.h.
 832         (ix86_data_alignment): Use wide-int interfaces.
 833         (ix86_local_alignment): Likewise.
 834         (ix86_emit_swsqrtsf): Update real_from_integer.
 835         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
 836         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
 837         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
 838         (zero_constant): Likewise.
 839         (input_operand): Likewise.
 840         (splat_input_operand): Likewise.
 841         (non_logical_cint_operand): Change const_double to const_wide_int.
 842         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
 843         (easy_altivec_constant): Remove comment.
 844         (paired_expand_vector_init): Use CONSTANT_P.
 845         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
 846         (rs6000_emit_move): Update checks.
 847         (rs6000_aggregate_candidate): Use wide-int interfaces.
 848         (rs6000_expand_ternop_builtin): Likewise.
 849         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
 850         (rs6000_assemble_integer): Likewise.
 851         (rs6000_hash_constant): Likewise.
 852         (output_toc): Likewise.
 853         (rs6000_rtx_costs): Likewise.
 854         (rs6000_emit_swrsqrt); Update call to real_from_integer.
 855         * config/rs6000/rs6000-c.c: Include wide-int.h.
 856         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
 857         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
 858         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
 859         Handle CONST_WIDE_INT.
 860         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
 861         Use tree_fits_uhwi_p.
 862         * config/sparc/sparc.c: Include wide-int.h.
 863         (sparc_fold_builtin): Use wide-int interfaces.
 864         * config/vax/vax.c: Include wide-int.h.
 865         (vax_float_literal): Use real_from_integer.
 866         * coretypes.h (struct hwivec_def): New.
 867         (hwivec): New.
 868         (const_hwivec): New.
 869         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
 870         (equiv_constant): Handle CONST_WIDE_INT.
 871         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
 872         (cselib_hash_rtx): Handle CONST_WIDE_INT.
 873         * dbxout.c (stabstr_U): Use wide-int interfaces.
 874         (dbxout_type): Update to use cst_fits_shwi_p.
 875         * defaults.h (LOG2_BITS_PER_UNIT): Define.
 876         (TARGET_SUPPORTS_WIDE_INT): Add default.
 877         * dfp.c: Include wide-int.h.
 878         (decimal_real_to_integer2): Use wide-int interfaces and rename to
 879         decimal_real_to_integer.
 880         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
 881         decimal_real_to_integer.
 882         * doc/generic.texi (Constant expressions): Update for wide_int.
 883         * doc/rtl.texi (const_double): Likewise.
 884         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
 885         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
 886         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
 887         (REAL_VALUE_FROM_INT): Remove.
 888         (TARGET_SUPPORTS_WIDE_INT): New.
 889         * doc/tm.texi: Regenerate.
 890         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
 891         * double-int.h: Include wide-int.h.
 892         (struct wi::int_traits): New.
 893         * dwarf2out.c (get_full_len): New.
 894         (dw_val_equal_p): Add case dw_val_class_wide_int.
 895         (size_of_loc_descr): Likewise.
 896         (output_loc_operands): Likewise.
 897         (insert_double): Remove.
 898         (insert_wide_int): New.
 899         (add_AT_wide): New.
 900         (print_die): Add case dw_val_class_wide_int.
 901         (attr_checksum): Likewise.
 902         (attr_checksum_ordered): Likewise.
 903         (same_dw_val_p): Likewise.
 904         (size_of_die): Likewise.
 905         (value_format): Likewise.
 906         (output_die): Likewise.
 907         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
 908         Use wide-int.
 909         (clz_loc_descriptor): Use wide-int interfaces.
 910         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
 911         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
 912         (round_up_to_align): Use wide-int interfaces.
 913         (field_byte_offset): Likewise.
 914         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
 915         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
 916         CONST_DOUBLE handling.  Use wide-int interfaces.
 917         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
 918         (gen_enumeration_type_die): Use add_AT_wide.
 919         (hash_loc_operands): Add case dw_val_class_wide_int.
 920         (compare_loc_operands): Likewise.
 921         * dwarf2out.h: Include wide-int.h.
 922         (wide_int_ptr): New.
 923         (enum dw_val_class): Add dw_val_class_wide_int.
 924         (struct dw_val_struct): Add val_wide.
 925         * emit-rtl.c (const_wide_int_htab): New.
 926         (const_wide_int_htab_hash): New.
 927         (const_wide_int_htab_eq): New.
 928         (lookup_const_wide_int): New.
 929         (const_double_htab_hash): Use wide-int interfaces.
 930         (const_double_htab_eq): Likewise.
 931         (rtx_to_double_int): Conditionally compile for wide-int.
 932         (immed_double_int_const): Rename to immed_wide_int_const and
 933         update for wide-int.
 934         (immed_double_const): Conditionally compile for wide-int.
 935         (init_emit_once): Use wide-int interfaces.
 936         * explow.c (plus_constant): Likewise.
 937         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
 938         (lshift_value): Use wide-int interfaces.
 939         (expand_mult): Likewise.
 940         (choose_multiplier): Likewise.
 941         (expand_smod_pow2): Likewise.
 942         (make_tree): Likewise.
 943         * expr.c (convert_modes): Consolidate handling of constants.
 944         Use wide-int interfaces.
 945         (emit_group_load_1): Add note.
 946         (store_expr): Update comment.
 947         (get_inner_reference): Use wide-int interfaces.
 948         (expand_constructor): Update comment.
 949         (expand_expr_real_2): Use wide-int interfaces.
 950         (expand_expr_real_1): Likewise.
 951         (reduce_to_bit_field_precision): Likewise.
 952         (const_vector_from_tree): Likewise.
 953         * final.c: Include wide-int-print.h.
 954         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
 955         * fixed-value.c: Include wide-int.h.
 956         (fixed_from_string): Use wide-int interfaces.
 957         (fixed_to_decimal): Likewise.
 958         (fixed_convert_from_real): Likewise.
 959         (real_convert_from_fixed): Likewise.
 960         * fold-const.h (mem_ref_offset): Return an offset_int.
 961         (div_if_zero_remainder): Remove code parameter.
 962         * fold-const.c (div_if_zero_remainder): Remove code parameter.
 963         Use wide-int interfaces.
 964         (may_negate_without_overflow_p): Use wide-int interfaces.
 965         (negate_expr_p): Likewise.
 966         (fold_negate_expr): Likewise.
 967         (int_const_binop_1): Likewise.
 968         (const_binop): Likewise.
 969         (fold_convert_const_int_from_int): Likewise.
 970         (fold_convert_const_int_from_real): Likewise.
 971         (fold_convert_const_int_from_fixed): Likewise.
 972         (fold_convert_const_fixed_from_int): Likewise.
 973         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
 974         (sign_bit_p): Use wide-int interfaces.
 975         (make_range_step): Likewise.
 976         (build_range_check): Likewise.  Pass an integer of the correct type
 977         instead of using integer_one_node.
 978         (range_predecessor): Pass an integer of the correct type instead
 979         of using integer_one_node.
 980         (range_successor): Likewise.
 981         (merge_ranges): Likewise.
 982         (unextend): Use wide-int interfaces.
 983         (extract_muldiv_1): Likewise.
 984         (fold_div_compare): Likewise.
 985         (fold_single_bit_test): Likewise.
 986         (fold_sign_changed_comparison): Likewise.
 987         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
 988         (fold_plusminus_mult_expr): Use wide-int interfaces.
 989         (native_encode_int): Likewise.
 990         (native_interpret_int): Likewise.
 991         (fold_unary_loc): Likewise.
 992         (pointer_may_wrap_p): Likewise.
 993         (size_low_cst): Likewise.
 994         (mask_with_tz): Likewise.
 995         (fold_binary_loc): Likewise.
 996         (fold_ternary_loc): Likewise.
 997         (multiple_of_p): Likewise.
 998         (tree_call_nonnegative_warnv_p): Update calls to
 999         tree_int_cst_min_precision and real_from_integer.
1000         (fold_negate_const): Use wide-int interfaces.
1001         (fold_abs_const): Likewise.
1002         (fold_relational_const): Use tree_int_cst_lt.
1003         (round_up_loc): Use wide-int interfaces.
1004         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
1005         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
1006         * gengtype.c: Remove include of double-int.h.
1007         (do_typedef): Use wide-int interfaces.
1008         (open_base_files): Add wide-int.h.
1009         (main): Add offset_int and widest_int typedefs.
1010         * gengtype-lex.l: Handle "^".
1011         (CXX_KEYWORD): Add "static".
1012         * gengtype-parse.c (require3): New.
1013         (require_template_declaration): Handle constant template arguments
1014         and nested templates.
1015         * gengtype-state.c: Don't include "double-int.h".
1016         * genpreds.c (write_one_predicate_function): Update comment.
1017         (write_tm_constrs_h): Add check for hval and lval use in
1018         CONST_WIDE_INT.
1019         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
1020         (add_to_sequence): Likewise.
1021         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
1022         and const_double_operand.
1023         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
1024         interfaces.
1025         * gimple-fold.c (get_base_constructor): Likewise.
1026         (fold_array_ctor_reference): Likewise.
1027         (fold_nonarray_ctor_reference): Likewise.
1028         (fold_const_aggregate_ref_1): Likewise.
1029         (gimple_val_nonnegative_real_p): Likewise.
1030         (gimple_fold_indirect_ref): Likewise.
1031         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
1032         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
1033         (struct slsr_cand_d): Change index to be widest_int.
1034         (struct incr_info_d): Change incr to be widest_int.
1035         (alloc_cand_and_find_basis): Use wide-int interfaces.
1036         (slsr_process_phi): Likewise.
1037         (backtrace_base_for_ref): Likewise.  Return a widest_int.
1038         (restructure_reference): Take a widest_int instead of a double_int.
1039         (slsr_process_ref): Use wide-int interfaces.
1040         (create_mul_ssa_cand): Likewise.
1041         (create_mul_imm_cand): Likewise.
1042         (create_add_ssa_cand): Likewise.
1043         (create_add_imm_cand): Take a widest_int instead of a double_int.
1044         (slsr_process_add): Use wide-int interfaces.
1045         (slsr_process_cast): Likewise.
1046         (slsr_process_copy): Likewise.
1047         (dump_candidate): Likewise.
1048         (dump_incr_vec): Likewise.
1049         (replace_ref): Likewise.
1050         (cand_increment): Likewise.  Return a widest_int.
1051         (cand_abs_increment): Likewise.
1052         (replace_mult_candidate): Take a widest_int instead of a double_int.
1053         (replace_unconditional_candidate): Use wide-int interfaces.
1054         (incr_vec_index): Take a widest_int instead of a double_int.
1055         (create_add_on_incoming_edge): Likewise.
1056         (create_phi_basis): Use wide-int interfaces.
1057         (replace_conditional_candidate): Likewise.
1058         (record_increment): Take a widest_int instead of a double_int.
1059         (record_phi_increments): Use wide-int interfaces.
1060         (phi_incr_cost): Take a widest_int instead of a double_int.
1061         (lowest_cost_path): Likewise.
1062         (total_savings): Likewise.
1063         (analyze_increments): Use wide-int interfaces.
1064         (ncd_with_phi): Take a widest_int instead of a double_int.
1065         (ncd_of_cand_and_phis): Likewise.
1066         (nearest_common_dominator_for_cands): Likewise.
1067         (insert_initializers): Use wide-int interfaces.
1068         (all_phi_incrs_profitable): Likewise.
1069         (replace_one_candidate): Likewise.
1070         (replace_profitable_candidates): Likewise.
1071         * godump.c: Include wide-int-print.h.
1072         (go_output_typedef): Use wide-int interfaces.
1073         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
1074         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1075         (build_loop_iteration_domains): Likewise.
1076         * hooks.h: Include wide-int.h rather than double-int.h.
1077         (hook_bool_dint_dint_uint_bool_true): Delete.
1078         (hook_bool_wint_wint_uint_bool_true): Declare.
1079         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
1080         (hook_bool_wint_wint_uint_bool_true): New.
1081         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
1082         interfaces.
1083         (ubsan_expand_si_overflow_mul_check): Likewise.
1084         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
1085         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1086         (get_ancestor_addr_info): Likewise.
1087         (ipa_modify_call_arguments): Likewise.
1088         * loop-doloop.c (doloop_modify): Likewise.
1089         (doloop_optimize): Likewise.
1090         * loop-iv.c (iv_number_of_iterations): Likewise.
1091         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
1092         (unroll_loop_constant_iterations): Likewise.
1093         (decide_unroll_runtime_iterations): Likewise.
1094         (unroll_loop_runtime_iterations): Likewise.
1095         (decide_peel_simple): Likewise.
1096         (decide_unroll_stupid): Likewise.
1097         * lto-streamer-in.c (streamer_read_wi): Add.
1098         (input_cfg): Use wide-int interfaces.
1099         (lto_input_tree_1): Likewise.
1100         * lto-streamer-out.c (streamer_write_wi): Add.
1101         (hash_tree): Use wide-int interfaces.
1102         (output_cfg): Likewise.
1103         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
1104         (GTFILES): Add wide-int.h and signop.h.
1105         (TAGS): Look for .cc files too.
1106         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
1107         * optabs.c (expand_subword_shift): Likewise.
1108         (expand_doubleword_shift): Likewise.
1109         (expand_absneg_bit): Likewise.
1110         (expand_copysign_absneg): Likewise.
1111         (expand_copysign_bit): Likewise.
1112         * postreload.c (reload_cse_simplify_set): Likewise.
1113         * predict.c (predict_iv_comparison): Likewise.
1114         * pretty-print.h: Include wide-int-print.h.
1115         (pp_wide_int) New.
1116         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
1117         * print-tree.c: Include wide-int-print.h.
1118         (print_node_brief): Use wide-int interfaces.
1119         (print_node): Likewise.
1120         * read-rtl.c (validate_const_wide_int): New.
1121         (read_rtx_code): Add CONST_WIDE_INT case.
1122         * real.c: Include wide-int.h.
1123         (real_to_integer2): Delete.
1124         (real_to_integer): New function, returning a wide_int.
1125         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1126         (ten_to_ptwo): Update call to real_from_integer.
1127         (real_digit): Likewise.
1128         * real.h: Include signop.h, wide-int.h and insn-modes.h.
1129         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
1130         (REAL_VALUE_TO_INT): Delete.
1131         (real_to_integer): Declare a wide-int form.
1132         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
1133         * recog.c (const_int_operand): Improve comment.
1134         (const_scalar_int_operand): New.
1135         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
1136         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
1137         (split_double): Likewise.
1138         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
1139         (rtx_size): Likewise.
1140         (rtx_alloc_stat_v): New.
1141         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
1142         (cwi_output_hex): New.
1143         (iterative_hash_rtx): Handle CONST_WIDE_INT.
1144         (cwi_check_failed_bounds): New.
1145         * rtl.def (CONST_WIDE_INT): New.
1146         * rtl.h: Include <utility> and wide-int.h.
1147         (struct hwivec_def): New.
1148         (CWI_GET_NUM_ELEM): New.
1149         (CWI_PUT_NUM_ELEM): New.
1150         (struct rtx_def): Add num_elem and hwiv.
1151         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
1152         (CASE_CONST_UNIQUE): Likewise.
1153         (CASE_CONST_ANY): Likewise.
1154         (CONST_SCALAR_INT_P): Likewise.
1155         (CONST_WIDE_INT_P): New.
1156         (CWI_ELT): New.
1157         (HWIVEC_CHECK): New.
1158         (cwi_check_failed_bounds): New.
1159         (CWI_ELT): New.
1160         (HWIVEC_CHECK): New.
1161         (CONST_WIDE_INT_VEC) New.
1162         (CONST_WIDE_INT_NUNITS) New.
1163         (CONST_WIDE_INT_ELT) New.
1164         (rtx_mode_t): New type.
1165         (wi::int_traits <rtx_mode_t>): New.
1166         (wi::shwi): New.
1167         (wi::min_value): New.
1168         (wi::max_value): New.
1169         (rtx_alloc_v) New.
1170         (const_wide_int_alloc): New.
1171         (immed_wide_int_const): New.
1172         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
1173         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
1174         * signop.h: New file.
1175         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
1176         (simplify_const_unary_operation): Use wide-int interfaces.
1177         (simplify_binary_operation_1): Likewise.
1178         (simplify_const_binary_operation): Likewise.
1179         (simplify_const_relational_operation): Likewise.
1180         (simplify_immed_subreg): Likewise.
1181         * stmt.c (expand_case): Likewise.
1182         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
1183         signop rather than a bool.
1184         * stor-layout.c (layout_type): Use wide-int interfaces.
1185         (initialize_sizetypes): Update calls to
1186         set_min_and_max_values_for_integral_type.
1187         (set_min_and_max_values_for_integral_type): Take a signop rather
1188         than a bool.  Use wide-int interfaces.
1189         (fixup_signed_type): Update accordingly.  Remove
1190         HOST_BITS_PER_DOUBLE_INT limit.
1191         (fixup_unsigned_type): Likewise.
1192         * system.h (STATIC_CONSTANT_P): New.
1193         (STATIC_ASSERT): New.
1194         * target.def (can_use_doloop_p): Take widest_ints rather than
1195         double_ints.
1196         * target.h: Include wide-int.h rather than double-int.h.
1197         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
1198         than double_ints.
1199         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
1200         rather than INT_CST_LT_UNSIGNED.
1201         (can_use_doloop_if_innermost): Take widest_ints rather than
1202         double_ints.
1203         * tree-affine.c: Include wide-int-print.h.
1204         (double_int_ext_for_comb): Delete.
1205         (wide_int_ext_for_comb): New.
1206         (aff_combination_zero): Use wide-int interfaces.
1207         (aff_combination_const): Take a widest_int instead of a double_int.
1208         (aff_combination_elt): Use wide-int interfaces.
1209         (aff_combination_scale): Take a widest_int instead of a double_int.
1210         (aff_combination_add_elt): Likewise.
1211         (aff_combination_add_cst): Likewise.
1212         (aff_combination_add): Use wide-int interfaces.
1213         (aff_combination_convert): Likewise.
1214         (tree_to_aff_combination): Likewise.
1215         (add_elt_to_tree): Take a widest_int instead of a double_int.
1216         (aff_combination_to_tree): Use wide-int interfaces.
1217         (aff_combination_remove_elt): Likewise.
1218         (aff_combination_add_product): Take a widest_int instead of
1219         a double_int.
1220         (aff_combination_mult): Use wide-int interfaces.
1221         (aff_combination_expand): Likewise.
1222         (double_int_constant_multiple_p): Delete.
1223         (wide_int_constant_multiple_p): New.
1224         (aff_combination_constant_multiple_p): Take a widest_int pointer
1225         instead of a double_int pointer.
1226         (print_aff): Use wide-int interfaces.
1227         (get_inner_reference_aff): Take a widest_int pointer
1228         instead of a double_int pointer.
1229         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
1230         * tree-affine.h: Include wide-int.h.
1231         (struct aff_comb_elt): Change type of coef to widest_int.
1232         (struct affine_tree_combination): Change type of offset to widest_int.
1233         (double_int_ext_for_comb): Delete.
1234         (wide_int_ext_for_comb): New.
1235         (aff_combination_const): Use widest_int instead of double_int.
1236         (aff_combination_scale): Likewise.
1237         (aff_combination_add_elt): Likewise.
1238         (aff_combination_constant_multiple_p): Likewise.
1239         (get_inner_reference_aff): Likewise.
1240         (aff_comb_cannot_overlap_p): Likewise.
1241         (aff_combination_zero_p): Use wide-int interfaces.
1242         * tree.c: Include tree.h.
1243         (init_ttree): Use make_int_cst.
1244         (tree_code_size): Removed code for INTEGER_CST case.
1245         (tree_size): Add INTEGER_CST case.
1246         (make_node_stat): Update comment.
1247         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
1248         (build_int_cst_type): Use wide-int interfaces.
1249         (double_int_to_tree): Likewise.
1250         (double_int_fits_to_tree_p): Delete.
1251         (force_fit_type_double): Delete.
1252         (force_fit_type): New.
1253         (int_cst_hash_hash): Use wide-int interfaces.
1254         (int_cst_hash_eq): Likewise.
1255         (build_int_cst_wide): Delete.
1256         (wide_int_to_tree): New.
1257         (cache_integer_cst): Use wide-int interfaces.
1258         (build_low_bits_mask): Likewise.
1259         (cst_and_fits_in_hwi): Likewise.
1260         (real_value_from_int_cst): Likewise.
1261         (make_int_cst_stat): New.
1262         (integer_zerop): Use wide_int interfaces.
1263         (integer_onep): Likewise.
1264         (integer_all_onesp): Likewise.
1265         (integer_pow2p): Likewise.
1266         (integer_nonzerop): Likewise.
1267         (tree_log2): Likewise.
1268         (tree_floor_log2): Likewise.
1269         (tree_ctz): Likewise.
1270         (int_size_in_bytes): Likewise.
1271         (mem_ref_offset): Return an offset_int rather than a double_int.
1272         (build_type_attribute_qual_variant): Use wide_int interfaces.
1273         (type_hash_eq): Likewise
1274         (tree_int_cst_equal): Likewise.
1275         (tree_int_cst_lt): Delete.
1276         (tree_int_cst_compare): Likewise.
1277         (tree_fits_shwi_p): Use wide_int interfaces.
1278         (tree_fits_uhwi_p): Likewise.
1279         (tree_int_cst_sign_bit): Likewise.
1280         (tree_int_cst_sgn): Likewise.
1281         (tree_int_cst_min_precision): Take a signop rather than a bool.
1282         (simple_cst_equal): Use wide_int interfaces.
1283         (compare_tree_int): Likewise.
1284         (iterative_hash_expr): Likewise.
1285         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
1286         INT_CST_LT.
1287         (get_type_static_bounds): Use wide_int interfaces.
1288         (tree_int_cst_elt_check_failed): New.
1289         (build_common_tree_nodes): Reordered to set prec before filling in
1290         value.
1291         (int_cst_value): Check cst_and_fits_in_hwi.
1292         (widest_int_cst_value): Use wide_int interfaces.
1293         (upper_bound_in_type): Likewise.
1294         (lower_bound_in_type): Likewise.
1295         (num_ending_zeros): Likewise.
1296         (drop_tree_overflow): Likewise.
1297         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
1298         (gen_conditions_for_pow_cst_base): Likewise.
1299         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
1300         (group_case_labels_stmt): Use wide-int interfaces.
1301         (verify_gimple_assign_binary): Likewise.
1302         (print_loop): Likewise.
1303         * tree-chrec.c (tree_fold_binomial): Likewise.
1304         * tree-core.h (struct tree_base): Add int_length.
1305         (struct tree_int_cst): Change rep of value.
1306         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
1307         (dr_may_alias_p): Likewise.
1308         (max_stmt_executions_tree): Likewise.
1309         * tree.def (INTEGER_CST): Update comment.
1310         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
1311         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
1312         * tree-dump.c: Include wide-int.h and wide-int-print.h.
1313         (dequeue_and_dump): Use wide-int interfaces.
1314         * tree.h: Include wide-int.h.
1315         (NULL_TREE): Moved to earlier loc in file.
1316         (TREE_INT_CST_ELT_CHECK): New.
1317         (tree_int_cst_elt_check_failed): New.
1318         (TYPE_SIGN): New.
1319         (TREE_INT_CST): Delete.
1320         (TREE_INT_CST_LOW): Use wide-int interfaces.
1321         (TREE_INT_CST_HIGH): Delete.
1322         (TREE_INT_CST_NUNITS): New.
1323         (TREE_INT_CST_EXT_NUNITS): Likewise.
1324         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
1325         (TREE_INT_CST_ELT): Likewise.
1326         (INT_CST_LT): Delete.
1327         (tree_int_cst_elt_check): New (two forms).
1328         (type_code_size): Update comment.
1329         (make_int_cst_stat, make_int_cst): New.
1330         (tree_to_double_int): Delete.
1331         (double_int_fits_to_tree_p): Delete.
1332         (force_fit_type_double): Delete.
1333         (build_int_cstu): Replace with out-of-line function.
1334         (build_int_cst_wide): Delete.
1335         (tree_int_cst_lt): Define inline.
1336         (tree_int_cst_le): New.
1337         (tree_int_cst_compare): Define inline.
1338         (tree_int_cst_min_precision): Take a signop rather than a bool.
1339         (wi::int_traits <const_tree>): New.
1340         (wi::int_traits <tree>): New.
1341         (wi::extended_tree): New.
1342         (wi::int_traits <wi::extended_tree>): New.
1343         (wi::to_widest): New.
1344         (wi::to_offset): New.
1345         (wi::fits_to_tree_p): New.
1346         (wi::min_value): New.
1347         (wi::max_value): New.
1348         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
1349         (copy_tree_body_r): Likewise.
1350         * tree-object-size.c (compute_object_offset): Likewise.
1351         (addr_object_size): Likewise.
1352         * tree-predcom.c: Include wide-int-print.h.
1353         (struct dref_d): Change type of offset to widest_int.
1354         (dump_dref): Call wide-int printer.
1355         (aff_combination_dr_offset): Use wide-int interfaces.
1356         (determine_offset): Take a widest_int pointer rather than a
1357         double_int pointer.
1358         (split_data_refs_to_components): Use wide-int interfaces.
1359         (suitable_component_p): Likewise.
1360         (order_drefs): Likewise.
1361         (add_ref_to_chain): Likewise.
1362         (valid_initializer_p): Likewise.
1363         (determine_roots_comp): Likewise.
1364         * tree-pretty-print.c: Include wide-int-print.h.
1365         (dump_generic_node): Use wide-int interfaces.
1366         * tree-sra.c (sra_ipa_modify_expr): Likewise.
1367         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1368         (move_fixed_address_to_symbol): Likewise.
1369         (move_hint_to_base): Likewise.
1370         (move_pointer_to_base): Likewise.
1371         (move_variant_to_index): Likewise.
1372         (most_expensive_mult_to_index): Likewise.
1373         (addr_to_parts): Likewise.
1374         (copy_ref_info): Likewise.
1375         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1376         (indirect_refs_may_alias_p): Likewise.
1377         (stmt_kills_ref_p_1): Likewise.
1378         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
1379         * tree-ssa-ccp.c: Update comment at top of file.  Include
1380         wide-int-print.h.
1381         (struct prop_value_d): Change type of mask to widest_int.
1382         (extend_mask): New function.
1383         (dump_lattice_value): Use wide-int interfaces.
1384         (get_default_value): Likewise.
1385         (set_constant_value): Likewise.
1386         (set_value_varying): Likewise.
1387         (valid_lattice_transition): Likewise.
1388         (set_lattice_value): Likewise.
1389         (value_to_double_int): Delete.
1390         (value_to_wide_int): New.
1391         (get_value_from_alignment): Use wide-int interfaces.
1392         (get_value_for_expr): Likewise.
1393         (do_dbg_cnt): Likewise.
1394         (ccp_finalize): Likewise.
1395         (ccp_lattice_meet): Likewise.
1396         (bit_value_unop_1): Use widest_ints rather than double_ints.
1397         (bit_value_binop_1): Likewise.
1398         (bit_value_unop): Use wide-int interfaces.
1399         (bit_value_binop): Likewise.
1400         (bit_value_assume_aligned): Likewise.
1401         (evaluate_stmt): Likewise.
1402         (ccp_fold_stmt): Likewise.
1403         (visit_cond_stmt): Likewise.
1404         (ccp_visit_stmt): Likewise.
1405         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1406         (constant_pointer_difference): Likewise.
1407         (associate_pointerplus): Likewise.
1408         (combine_conversions): Likewise.
1409         * tree-ssa-loop.h: Include wide-int.h.
1410         (struct tree_niter_desc): Change type of max to widest_int.
1411         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
1412         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
1413         (remove_redundant_iv_tests): Likewise.
1414         (canonicalize_loop_induction_variables): Likewise.
1415         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
1416         (constant_multiple_of): Take a widest_int pointer instead of
1417         a double_int pointer.
1418         (get_computation_aff): Use wide-int interfaces.
1419         (ptr_difference_cost): Likewise.
1420         (difference_cost): Likewise.
1421         (get_loop_invariant_expr_id): Likewise.
1422         (get_computation_cost_at): Likewise.
1423         (iv_elimination_compare_lt): Likewise.
1424         (may_eliminate_iv): Likewise.
1425         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
1426         instead of double_int.
1427         (max_loop_iterations): Likewise.
1428         (max_stmt_executions): Likewise.
1429         (estimated_stmt_executions): Likewise.
1430         * tree-ssa-loop-niter.c: Include wide-int-print.h.
1431         (split_to_var_and_offset): Use wide-int interfaces.
1432         (determine_value_range): Likewise.
1433         (bound_difference_of_offsetted_base): Likewise.
1434         (bounds_add): Take a widest_int instead of a double_int.
1435         (number_of_iterations_ne_max): Use wide-int interfaces.
1436         (number_of_iterations_ne): Likewise.
1437         (number_of_iterations_lt_to_ne): Likewise.
1438         (assert_loop_rolls_lt): Likewise.
1439         (number_of_iterations_lt): Likewise.
1440         (number_of_iterations_le): Likewise.
1441         (number_of_iterations_cond): Likewise.
1442         (number_of_iterations_exit): Likewise.
1443         (finite_loop_p): Likewise.
1444         (derive_constant_upper_bound_assign): Likewise.
1445         (derive_constant_upper_bound): Return a widest_int.
1446         (derive_constant_upper_bound_ops): Likewise.
1447         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
1448         (record_estimate): Take a widest_int rather than a double_int.
1449         (record_nonwrapping_iv): Use wide-int interfaces.
1450         (double_int_cmp): Delete.
1451         (wide_int_cmp): New.
1452         (bound_index): Take a widest_int rather than a double_int.
1453         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
1454         (maybe_lower_iteration_bound): Likewise.
1455         (estimate_numbers_of_iterations_loop): Likewise.
1456         (estimated_loop_iterations): Take a widest_int pointer than than
1457         a double_int pointer.
1458         (estimated_loop_iterations_int): Use wide-int interfaces.
1459         (max_loop_iterations): Take a widest_int pointer than than
1460         a double_int pointer.
1461         (max_loop_iterations_int): Use wide-int interfaces.
1462         (max_stmt_executions): Take a widest_int pointer than than
1463         a double_int pointer.
1464         (estimated_stmt_executions): Likewise.
1465         (n_of_executions_at_most): Use wide-int interfaces.
1466         (scev_probably_wraps_p): Likewise.
1467         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
1468         to real_to_integer.
1469         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
1470         interfaces.
1471         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
1472         double_ints.  Adjust for trailing_wide_ints <3> representation.
1473         (set_nonzero_bits): Likewise.
1474         (get_range_info): Return wide_ints rather than double_ints.
1475         Adjust for trailing_wide_ints <3> representation.
1476         (get_nonzero_bits): Likewise.
1477         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
1478         representation.
1479         * tree-ssanames.h (struct range_info_def): Replace min, max and
1480         nonzero_bits with a trailing_wide_ints <3>.
1481         (set_range_info): Use wide_int_refs rather than double_ints.
1482         (set_nonzero_bits): Likewise.
1483         (get_range_info): Return wide_ints rather than double_ints.
1484         (get_nonzero_bits): Likewise.
1485         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
1486         * tree-ssa-pre.c (phi_translate_1): Likewise.
1487         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
1488         (acceptable_pow_call): Likewise.
1489         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
1490         interfaces.
1491         (vn_reference_fold_indirect): Likewise.
1492         (vn_reference_maybe_forwprop_address): Likewise.
1493         (valueize_refs_1): Likewise.
1494         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1495         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
1496         tree_int_cst_lt and tree_int_cst_le.
1497         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
1498         interfaces.
1499         (streamer_alloc_tree): Likewise.
1500         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
1501         (streamer_write_tree_header): Likewise.
1502         (streamer_write_integer_cst): Likewise.
1503         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
1504         (build_constructors): Likewise.
1505         (array_value_type): Likewise.
1506         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
1507         (vect_check_gather): Likewise.
1508         * tree-vect-generic.c (build_replicated_const): Likewise.
1509         (expand_vector_divmod): Likewise.
1510         * tree-vect-loop.c (vect_transform_loop): Likewise.
1511         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
1512         (vect_do_peeling_for_alignment): Likewise.
1513         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
1514         * tree-vrp.c: Include wide-int.h.
1515         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
1516         (extract_range_from_assert): Use wide-int interfaces.
1517         (vrp_int_const_binop): Likewise.
1518         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
1519         double_int pointers.
1520         (ranges_from_anti_range): Use wide-int interfaces.
1521         (quad_int_cmp): Delete.
1522         (quad_int_pair_sort): Likewise.
1523         (extract_range_from_binary_expr_1): Use wide-int interfaces.
1524         (extract_range_from_unary_expr_1): Likewise.
1525         (adjust_range_with_scev): Likewise.
1526         (masked_increment): Take and return wide_ints rather than double_ints.
1527         (register_edge_assert_for_2): Use wide-int interfaces.
1528         (check_array_ref): Likewise.
1529         (search_for_addr_array): Likewise.
1530         (maybe_set_nonzero_bits): Likewise.
1531         (union_ranges): Pass an integer of the correct type instead of
1532         using integer_one_node.
1533         (intersect_ranges): Likewise.
1534         (simplify_truth_ops_using_ranges): Likewise.
1535         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
1536         (range_fits_type_p): Likewise.
1537         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
1538         a bool.
1539         (simplify_conversion_using_ranges): Use wide-int interfaces.
1540         (simplify_float_conversion_using_ranges): Likewise.
1541         (vrp_finalize): Likewise.
1542         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
1543         (gimple_stringops_transform): Likewise.
1544         * varasm.c (decode_addr_const): Likewise.
1545         (const_hash_1): Likewise.
1546         (const_rtx_hash_1): Likewise
1547         (output_constant): Likewise.
1548         (array_size_for_constructor): Likewise.
1549         (output_constructor_regular_field): Likewise.
1550         (output_constructor_bitfield): Likewise.
1551         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
1552         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
1553         GENERATOR_FILEs.
1554         * gencheck.c: Define BITS_PER_UNIT.
1555         * wide-int.cc: New.
1556         * wide-int.h: New.
1557         * wide-int-print.cc: New.
1558         * wide-int-print.h: New.
1559
1560 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1561
1562         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
1563
1564 2014-05-06  Richard Biener  <rguenther@suse.de>
1565
1566         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
1567         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
1568         (TODO_verify_all): Adjust.
1569         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
1570         TODO_verify_stmts and TODO_verify_rtl_sharing.
1571         * bb-reorder.c: Likewise.
1572         * cfgexpand.c: Likewise.
1573         * cprop.c: Likewise.
1574         * cse.c: Likewise.
1575         * function.c: Likewise.
1576         * fwprop.c: Likewise.
1577         * gcse.c: Likewise.
1578         * gimple-ssa-isolate-paths.c: Likewise.
1579         * gimple-ssa-strength-reduction.c: Likewise.
1580         * ipa-split.c: Likewise.
1581         * loop-init.c: Likewise.
1582         * loop-unroll.c: Likewise.
1583         * lower-subreg.c: Likewise.
1584         * modulo-sched.c: Likewise.
1585         * postreload-gcse.c: Likewise.
1586         * predict.c: Likewise.
1587         * recog.c: Likewise.
1588         * sched-rgn.c: Likewise.
1589         * store-motion.c: Likewise.
1590         * tracer.c: Likewise.
1591         * trans-mem.c: Likewise.
1592         * tree-call-cdce.c: Likewise.
1593         * tree-cfg.c: Likewise.
1594         * tree-cfgcleanup.c: Likewise.
1595         * tree-complex.c: Likewise.
1596         * tree-eh.c: Likewise.
1597         * tree-emutls.c: Likewise.
1598         * tree-if-conv.c: Likewise.
1599         * tree-into-ssa.c: Likewise.
1600         * tree-loop-distribution.c: Likewise.
1601         * tree-object-size.c: Likewise.
1602         * tree-parloops.c: Likewise.
1603         * tree-pass.h: Likewise.
1604         * tree-sra.c: Likewise.
1605         * tree-ssa-ccp.c: Likewise.
1606         * tree-ssa-copy.c: Likewise.
1607         * tree-ssa-copyrename.c: Likewise.
1608         * tree-ssa-dce.c: Likewise.
1609         * tree-ssa-dom.c: Likewise.
1610         * tree-ssa-dse.c: Likewise.
1611         * tree-ssa-forwprop.c: Likewise.
1612         * tree-ssa-ifcombine.c: Likewise.
1613         * tree-ssa-loop-ch.c: Likewise.
1614         * tree-ssa-loop-ivcanon.c: Likewise.
1615         * tree-ssa-loop.c: Likewise.
1616         * tree-ssa-math-opts.c: Likewise.
1617         * tree-ssa-phiopt.c: Likewise.
1618         * tree-ssa-phiprop.c: Likewise.
1619         * tree-ssa-pre.c: Likewise.
1620         * tree-ssa-reassoc.c: Likewise.
1621         * tree-ssa-sink.c: Likewise.
1622         * tree-ssa-strlen.c: Likewise.
1623         * tree-ssa-tail-merge.c: Likewise.
1624         * tree-ssa-uncprop.c: Likewise.
1625         * tree-switch-conversion.c: Likewise.
1626         * tree-tailcall.c: Likewise.
1627         * tree-vect-generic.c: Likewise.
1628         * tree-vectorizer.c: Likewise.
1629         * tree-vrp.c: Likewise.
1630         * tsan.c: Likewise.
1631         * var-tracking.c: Likewise.
1632         * bt-load.c: Likewise.
1633         * cfgcleanup.c: Likewise.
1634         * combine-stack-adj.c: Likewise.
1635         * combine.c: Likewise.
1636         * compare-elim.c: Likewise.
1637         * config/epiphany/resolve-sw-modes.c: Likewise.
1638         * config/i386/i386.c: Likewise.
1639         * config/mips/mips.c: Likewise.
1640         * config/s390/s390.c: Likewise.
1641         * config/sh/sh_treg_combine.cc: Likewise.
1642         * config/sparc/sparc.c: Likewise.
1643         * dce.c: Likewise.
1644         * dse.c: Likewise.
1645         * final.c: Likewise.
1646         * ifcvt.c: Likewise.
1647         * mode-switching.c: Likewise.
1648         * passes.c: Likewise.
1649         * postreload.c: Likewise.
1650         * ree.c: Likewise.
1651         * reg-stack.c: Likewise.
1652         * regcprop.c: Likewise.
1653         * regrename.c: Likewise.
1654         * web.c: Likewise.
1655
1656 2014-05-06  Richard Biener  <rguenther@suse.de>
1657
1658         PR middle-end/61070
1659         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
1660         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
1661
1662 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
1663
1664         PR ipa/60965
1665         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
1666
1667 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
1668             Tom de Vries  <tom@codesourcery.com>
1669
1670         * target.def (call_fusage_contains_non_callee_clobbers): New
1671         DEFHOOKPOD.
1672         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
1673         Hooks to @menu.
1674         (@node Miscellaneous Register Hooks): New node.
1675         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
1676         * doc/tm.texi: Regenerate.
1677
1678 2014-05-05  Marek Polacek  <polacek@redhat.com>
1679
1680         PR driver/61065
1681         * opts.c (common_handle_option): Call error_at instead of warning_at.
1682
1683 2014-05-05  Richard Biener  <rguenther@suse.de>
1684
1685         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
1686         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
1687         under the TODO_verify_il umbrella.
1688
1689 2014-05-05  Richard Biener  <rguenther@suse.de>
1690
1691         * passes.c (execute_function_todo): Move TODO_verify_flow under
1692         the TODO_verify_ul umbrella.
1693
1694 2014-05-05  Richard Biener  <rguenther@suse.de>
1695
1696         PR middle-end/61010
1697         * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
1698         X & CST away from a CST that is the mask of a mode.
1699
1700 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1701
1702         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
1703         int argument to enum machine_mode.
1704         (picochip_class_max_nregs): Ditto.
1705         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
1706         (picochip_class_max_nregs): Ditto.
1707
1708 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1709
1710         * target.def: Add new target hook.
1711         * doc/tm.texi: Regenerate.
1712         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
1713         * targhooks.c (default_keep_leaf_when_profiled): New function.
1714
1715         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
1716         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
1717
1718 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
1719
1720         PR tree-optimization/60363
1721         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
1722         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
1723         (update_destination_phis): New parameter.
1724         (create_edge_and_update_destination_phis): Ditto.
1725         (ssa_fix_duplicate_block_edges): Pass new arguments.
1726         (thread_single_edge): Ditto.
1727
1728 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
1729
1730         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
1731         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
1732         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
1733         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
1734         Use RS6000_BTM_HARD_FLOAT.
1735         (BU_MISC_2): Likewise.
1736         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
1737         RS6000_BTM_HARD_FLOAT.
1738         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
1739         is explicitly used.
1740         (rs6000_invalid_builtin): Add hard floating builtin support.
1741         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
1742         hard float builtins.
1743         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
1744
1745 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1746
1747         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
1748         Add missing function* argument.
1749
1750 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1751
1752         * lra-constraints.c (valid_address_p): Move earlier in file.
1753         Add a constraint argument to the address_info version.
1754         (satisfies_memory_constraint_p): New function.
1755         (satisfies_address_constraint_p): Likewise.
1756         (process_alt_operands, curr_insn_transform): Use them.
1757         (process_address): Pass the constraint to valid_address_p when
1758         checking address operands.
1759
1760 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1761
1762         * config/mips/mips.c (mips_isa_rev): New variable.
1763         (mips_set_architecture): Set it.
1764         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
1765         from mips_isa_rev.
1766         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
1767         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
1768         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
1769         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
1770         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
1771         conditions in terms of mips_isa_rev.
1772         (mips_isa_rev): Declare.
1773
1774 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1775
1776         * config/sh/sh-mem.cc: Use tabs instead of spaces.
1777         (prob_unlikely, prob_likely): Make variables const.
1778
1779 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
1780
1781         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
1782
1783 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1784
1785         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
1786
1787 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1788
1789         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
1790         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
1791         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
1792         functions.
1793         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
1794         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
1795         sh_pass_in_reg_p.
1796         Replace usage of ROUND_REG with sh_round_reg.
1797         Use CEIL instead of ROUND_ADVANCE.
1798
1799 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1800
1801         PR target/61026
1802         * config/sh/sh.c: Include stdlib headers before everything else.
1803
1804 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
1805
1806         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
1807         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
1808         (gimplify_adjust_omp_clauses): Simd region is never
1809         directly nested in combined parallel.  Instead, for linear
1810         with copyin/copyout, if in combined for simd loop, make decl
1811         firstprivate/lastprivate on OMP_FOR.
1812         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
1813         expand_omp_for_static_chunk): When setting endvar, also set
1814         fd->loop.v to the same value.
1815
1816 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1817
1818         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
1819
1820 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
1821
1822         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
1823         expression.
1824
1825 2014-05-02  Marek Polacek  <polacek@redhat.com>
1826
1827         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
1828
1829 2014-05-02  Kito Cheng  <kito@0xlab.org>
1830
1831         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
1832         to a C expression marco.
1833         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
1834         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
1835         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
1836         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
1837         HONOR_REG_ALLOC_ORDER.
1838         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
1839
1840 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1841
1842         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
1843
1844 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1845
1846         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
1847
1848 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
1849
1850         * tree-if-conv.c (is_cond_scalar_reduction): New function.
1851         (convert_scalar_cond_reduction): Likewise.
1852         (predicate_scalar_phi): Add recognition and transformation
1853         of simple conditioanl reduction to be vectorizable.
1854
1855 2014-05-01  Marek Polacek  <polacek@redhat.com>
1856
1857         PR c/43245
1858         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
1859
1860 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
1861
1862         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
1863         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
1864         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
1865         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
1866         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
1867         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
1868         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
1869         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
1870
1871 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
1872
1873         * config/arc/arc.opt (mlra): Move comment above option name
1874         to avoid mis-parsing as language options.
1875
1876 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1877
1878         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
1879         * config/sol2.h: ... here.
1880         * config/sol2-10.h: Remove.
1881
1882         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
1883         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
1884         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
1885         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
1886         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
1887         * config/sol2.h: ... here.
1888         (SECTION_NAME_FORMAT): Don't redefine.
1889         (STARTFILE_ARCH32_SPEC): Rename to ...
1890         (STARTFILE_ARCH_SPEC): ... this.
1891         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
1892         * config/sparc/sol2.h: ... here.
1893         (SECTION_NAME_FORMAT): Don't undef.
1894         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
1895         (SUBTARGET_EXTRA_SPECS): Remove.
1896         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
1897
1898         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
1899         (MD_STARTFILE_PREFIX): Remove.
1900         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
1901         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
1902         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
1903         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
1904         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
1905         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
1906         * config/i386/sol2.h: ... here.
1907         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
1908         * config/i386/sol2-bi.h: Remove.
1909         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
1910         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
1911
1912         * config/i386/t-sol2-64: Rename to ...
1913         * config/i386/t-sol2: ... this.
1914         * config/sparc/t-sol2-64: Rename to ...
1915         * config/sparc/t-sol2: ... this.
1916
1917         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
1918         sol2_tm_file_head, sol2_tm_file_tail.
1919         Include ${cpu_type}/sol2.h before sol2.h.
1920         Remove sol2-10.h.
1921         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
1922         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
1923         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
1924         Reflect i386/t-sol2-64 renaming.
1925         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
1926         Reflect sparc/t-sol2-64 renaming.
1927
1928 2014-04-30  Richard Biener  <rguenther@suse.de>
1929
1930         * passes.c (execute_function_todo): Move TODO_verify_stmts
1931         and TODO_verify_ssa under the TODO_verify_il umbrella.
1932         * tree-ssa.h (verify_ssa): Adjust prototype.
1933         * tree-ssa.c (verify_ssa): Add parameter to tell whether
1934         we should verify SSA operands.
1935         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
1936         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
1937         whether we should verify whether not throwing stmts have EH info.
1938         * graphite-scop-detection.c (create_sese_edges): Adjust.
1939         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
1940         * tree-eh.c (lower_try_finally_switch): Do not add the
1941         default case label twice.
1942
1943 2014-04-30  Marek Polacek  <polacek@redhat.com>
1944
1945         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
1946         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
1947         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
1948         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
1949
1950 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
1951
1952         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
1953         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
1954         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
1955         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
1956         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
1957         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
1958         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
1959         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
1960
1961 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
1962
1963         * tree-cfg.c (dump_function_to_file): Dump the return type of
1964         functions, in a line to itself before the function body, mimicking
1965         the layout of a C function.
1966
1967 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
1968
1969         PR tree-optimization/60971
1970         * tree-tailcall.c (process_assignment): Reject conversions which
1971         reduce precision.
1972
1973 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
1974
1975         * calls.c (initialize_argument_information): Always treat
1976         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
1977         (expand_call): Likewise.
1978         (emit_library_call_calue_1): Likewise.
1979         * expr.c (PUSH_ARGS_REVERSED): Do not define.
1980         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
1981         code accordingly.
1982
1983 2014-04-29  Nick Clifton  <nickc@redhat.com>
1984
1985         * config/msp430/msp430.md (umulsidi): Fix typo.
1986         (mulhisi3): Enable even inside interrupt handlers.
1987         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
1988         bigger return address pushed in large mode.
1989
1990 2014-04-29  Nick Clifton  <nickc@redhat.com>
1991
1992         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
1993         (arc_init_reg_tables): Use a machine_mode enum to iterate over
1994         available modes.
1995         * config/m32r/m32r.c (init_reg_tables): Likewise.
1996         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
1997         enum to hold the modes.
1998
1999 2014-04-29  Richard Biener  <rguenther@suse.de>
2000
2001         * dominance.c (free_dominance_info): Add overload with
2002         function parameter.
2003         (dom_info_state): Likewise.
2004         (dom_info_available_p): Likewise.
2005         * basic-block.h (free_dominance_info, dom_info_state,
2006         dom_info_available_p): Declare overloads.
2007         * passes.c (execute_function_todo): Verify that verifiers
2008         don't change dominator info state.  Drop dominator info
2009         for IPA pass invocations.
2010         * cgraph.c (release_function_body): Restore asserts that
2011         dominator information is released.
2012
2013 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
2014
2015         * doc/invoke.texi: Fix typo.
2016         * tree-vrp.c: Fix typos.
2017         * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
2018
2019 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2020
2021         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
2022
2023 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2024
2025         * config/aarch64/aarch64-builtins.c
2026         (aarch64_types_storestruct_lane_qualifiers): New.
2027         (TYPES_STORESTRUCT_LANE): Likewise.
2028         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
2029         (st3_lane): Likewise.
2030         (st4_lane): Likewise.
2031         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
2032         (vec_store_lanesci_lane<mode>): Likewise.
2033         (vec_store_lanesxi_lane<mode>): Likewise.
2034         (aarch64_st2_lane<VQ:mode>): Likewise.
2035         (aarch64_st3_lane<VQ:mode>): Likewise.
2036         (aarch64_st4_lane<VQ:mode>): Likewise.
2037         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
2038         * config/aarch64/arm_neon.h
2039         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
2040         use new macro arguments.
2041         (__ST3_LANE_FUNC): Likewise.
2042         (__ST4_LANE_FUNC): Likewise.
2043         * config/aarch64/iterators.md (V_TWO_ELEM): New.
2044         (V_THREE_ELEM): Likewise.
2045         (V_FOUR_ELEM): Likewise.
2046
2047 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
2048
2049         * doc/gimple.texi: Replace the description of the now-defunct
2050         union gimple_statement_d with a diagram showing the
2051         gimple_statement_base class hierarchy and its relationships to
2052         the GSS_ and GIMPLE_ enums.
2053
2054 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
2055
2056         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
2057         * config/aarch64/aarch64.c
2058         (aarch64_cannot_change_mode_class): Weaken conditions.
2059         (aarch64_modes_tieable_p): New.
2060         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
2061
2062 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
2063
2064         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2065         (loadsync_<mode>): Change mode.
2066         (load_quadpti, store_quadpti): New.
2067         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2068         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2069
2070 2014-04-28  Martin Jambor  <mjambor@suse.cz>
2071
2072         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
2073         same alias type as the original statement.
2074         (subreplacement_assignment_data): New type.
2075         (handle_unscalarized_data_in_subtree): New type of parameter,
2076         generate new memory accesses with same alias type as the original
2077         statement.
2078         (load_assign_lhs_subreplacements): Likewise.
2079         (sra_modify_constructor_assign): Generate new memory accesses with
2080         same alias type as the original statement.
2081
2082 2014-04-28  Richard Biener  <rguenther@suse.de>
2083
2084         * tree-pass.h (TODO_verify_il): Define.
2085         (TODO_verify_all): Complete properly.
2086         * passes.c (execute_function_todo): Move existing loop-closed
2087         SSA verification under TODO_verify_il.
2088         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
2089         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
2090         Fix tree sharing issue.
2091
2092 2014-04-28  Richard Biener  <rguenther@suse.de>
2093
2094         PR middle-end/60092
2095         * builtins.def (DEF_C11_BUILTIN): Add.
2096         (BUILT_IN_ALIGNED_ALLOC): Likewise.
2097         * coretypes.h (enum function_class): Add function_c11_misc.
2098         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
2099         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
2100         (call_may_clobber_ref_p_1): Likewise.
2101         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2102         (mark_all_reaching_defs_necessary_1): Likewise.
2103         (propagate_necessity): Likewise.
2104         (eliminate_unnecessary_stmts): Likewise.
2105         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
2106
2107 2014-04-28  Richard Biener  <rguenther@suse.de>
2108
2109         * tree-vrp.c (vrp_var_may_overflow): Remove.
2110         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
2111         with overflow immediately bump to one before that value and
2112         let iteration figure out overflow status.
2113
2114 2014-04-28  Richard Biener  <rguenther@suse.de>
2115
2116         * configure.ac: Do valgrind header checks unconditionally.
2117         Add --enable-valgrind-annotations.
2118         * system.h: Guard valgrind header inclusion with
2119         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
2120         * alloc-pool.c (pool_alloc, pool_free): Use
2121         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
2122         to guard possibly dead code.
2123         * config.in: Regenerated.
2124         * configure: Likewise.
2125
2126 2014-04-28  Jeff Law  <law@redhat.com>
2127
2128         PR tree-optimization/60902
2129         * tree-ssa-threadedge.c
2130         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
2131         over real defs when invalidating outputs from statements that do not
2132         produce useful outputs for threading.
2133
2134 2014-04-28  Richard Biener  <rguenther@suse.de>
2135
2136         PR tree-optimization/60979
2137         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
2138         SCOPs that end in a block with a successor with abnormal
2139         predecessors.
2140
2141 2014-04-28  Richard Biener  <rguenther@suse.de>
2142
2143         * tree-pass.h (execute_pass_list): Adjust prototype.
2144         * passes.c (pass_manager::execute_early_local_passes): Adjust.
2145         (do_per_function): Change callback signature, push all actual
2146         work to the callbals.
2147         (do_per_function_toporder): Likewise.
2148         (execute_function_dump): Adjust.
2149         (execute_function_todo): Likewise.
2150         (clear_last_verified): Likewise.
2151         (verify_curr_properties): Likewise.
2152         (update_properties_after_pass): Likewise.
2153         (execute_pass_list_1): Split out from ...
2154         (execute_pass_list): ... here.  Adjust.
2155         (execute_ipa_pass_list): Likewise.
2156         * cgraphunit.c (cgraph_add_new_function): Adjust.
2157         (analyze_function): Likewise.
2158         (expand_function): Likewise.
2159         * cgraph.c (release_function_body): Free dominance info
2160         here instead of asserting it was magically freed elsewhere.
2161
2162 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
2163
2164         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
2165         * configure: Regenerate.
2166         * config/sparc/sparc.opt (muser-mode): New option.
2167         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
2168         for LEON3.
2169         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
2170         * doc/invoke.texi (SPARC options): Document -muser-mode.
2171
2172 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
2173
2174         * cselib.c (find_slot_memmode): Delete.
2175         (cselib_hasher): Change compare_type to a struct.
2176         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
2177         constants.
2178         (preserve_constants_and_equivs): Adjust for new compare_type.
2179         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
2180         (wrap_constant): Delete.
2181         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
2182
2183 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
2184
2185         * doc/install.texi (Building with profile feedback): Remove
2186         outdated sentence.
2187
2188 2014-04-26  Tom de Vries  <tom@codesourcery.com>
2189
2190         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
2191         array accesses.
2192
2193 2014-04-25  Cary Coutant  <ccoutant@google.com>
2194
2195         PR debug/60929
2196         * dwarf2out.c (should_move_die_to_comdat): A type definition
2197         can contain a subprogram definition, but don't move it to a
2198         comdat unit.
2199         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
2200         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
2201         from original DIE.
2202         (clone_tree_hash): Rename to...
2203         (clone_tree_partial): ...this; change callers.  Copy
2204         DW_TAG_subprogram DIEs as declarations.
2205         (copy_decls_walk): Don't copy children of a declaration into a
2206         type unit.
2207
2208 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
2209
2210         PR target/60969
2211         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
2212         alternative 12.
2213
2214 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
2215
2216         * config/arm/predicates.md (call_insn_operand): Add long_call check.
2217         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
2218         reg for long_call.
2219         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
2220         restriction.
2221
2222 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2223
2224         * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
2225
2226 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2227
2228         PR tree-optimization/60930
2229         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
2230         creating a multiply candidate by folding two constant
2231         multiplicands when the result overflows.
2232
2233 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
2234
2235         PR tree-optimization/60960
2236         * tree-vect-generic.c (expand_vector_operation): Only call
2237         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
2238
2239 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2240
2241         * expr.c (clobber_reg_mode): New function.
2242         * expr.h (clobber_reg): New function.
2243
2244 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2245
2246         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
2247         clobbers.
2248
2249 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
2250             Tom de Vries  <tom@codesourcery.com>
2251
2252         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
2253         handle.
2254         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
2255         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
2256         new argument to find_all_hard_reg_sets call.
2257
2258 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2259
2260         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
2261         Use HOST_WIDE_INT_C for mask literal.
2262         (aarch_rev16_shleft_mask_imm_p): Likewise.
2263
2264 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
2265
2266         PR target/60941
2267         * config/sparc/sparc.md (ashlsi3_extend): Delete.
2268
2269 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
2270
2271         PR preprocessor/56540
2272         * config/i386/i386-c.c (ix86_target_macros): Define
2273         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
2274
2275 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2276
2277         * configure.ac (tga_func): Remove.
2278         (LIB_TLS_SPEC): Remove.
2279         * configure: Regenerate.
2280         * config.in: Regenerate.
2281         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
2282
2283 2014-04-25  Richard Biener  <rguenther@suse.de>
2284
2285         PR ipa/60912
2286         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
2287         call stmt use/clobber sets during stmt walk instead of
2288         walking the possibly incomplete set of caller edges.
2289
2290 2014-04-25  Richard Biener  <rguenther@suse.de>
2291
2292         PR ipa/60911
2293         * passes.c (apply_ipa_transforms): Inline into only caller ...
2294         (execute_one_pass): ... here.  Properly bring in function
2295         bodies for nodes we want to apply IPA transforms to.
2296
2297 2014-04-24  Cong Hou  <congh@google.com>
2298
2299         PR tree-optimization/60896
2300         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
2301         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
2302         (vect_mark_pattern_stmts): Set the def type of all statements in
2303         PATTERN_DEF_SEQ as vect_internal_def.
2304
2305 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
2306
2307         * doc/extend.texi (PowerPC Built-in Functions): Document new
2308         powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
2309         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
2310
2311         * config/rs6000/predicates.md (const_0_to_3_operand): New
2312         predicate to match 0..3 integer constants.
2313
2314         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
2315         to support adding miscellaneous builtin functions.
2316         (BU_DFP_MISC_2): Likewise.
2317         (BU_P7_MISC_1): Likewise.
2318         (BU_P7_MISC_2): Likewise.
2319         (BU_P8V_MISC_3): Likewise.
2320         (BU_MISC_1): Likewise.
2321         (BU_MISC_2): Likewise.
2322         (DIVWE): Add extended divide builtin functions.
2323         (DIVWEO): Likewise.
2324         (DIVWEU): Likewise.
2325         (DIVWEUO): Likewise.
2326         (DIVDE): Likewise.
2327         (DIVDEO): Likewise.
2328         (DIVDEU): Likewise.
2329         (DIVDEUO): Likewise.
2330         (DXEX): Add decimal floating-point builtin functions.
2331         (DXEXQ): Likewise.
2332         (DDEDPD): Likewise.
2333         (DDEDPDQ): Likewise.
2334         (DENBCD): Likewise.
2335         (DENBCDQ): Likewise.
2336         (DIEX): Likewise.
2337         (DIEXQ): Likewise.
2338         (DSCLI): Likewise.
2339         (DSCLIQ): Likewise.
2340         (DSCRI): Likewise.
2341         (DSCRIQ): Likewise.
2342         (CDTBCD): Add new BCD builtin functions.
2343         (CBCDTD): Likewise.
2344         (ADDG6S): Likewise.
2345         (BCDADD): Likewise.
2346         (BCDADD_LT): Likewise.
2347         (BCDADD_EQ): Likewise.
2348         (BCDADD_GT): Likewise.
2349         (BCDADD_OV): Likewise.
2350         (BCDSUB): Likewise.
2351         (BCDSUB_LT): Likewise.
2352         (BCDSUB_EQ): Likewise.
2353         (BCDSUB_GT): Likewise.
2354         (BCDSUB_OV): Likewise.
2355         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
2356         (UNPACK_TD): Likewise.
2357         (PACK_TF): Likewise.
2358         (UNPACK_TF): Likewise.
2359         (UNPACK_TF_0): Likewise.
2360         (UNPACK_TF_1): Likewise.
2361         (PACK_V1TI): Likewise.
2362         (UNPACK_V1TI): Likewise.
2363
2364         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2365         support for decimal floating point builtin functions.
2366         (rs6000_expand_ternop_builtin): Add checks for the new builtin
2367         functions that take constant arguments.
2368         (rs6000_invalid_builtin): Add decimal floating point builtin support.
2369         (rs6000_init_builtins): Setup long double, _Decimal64, and
2370         _Decimal128 types for new builtin functions.
2371         (builtin_function_type): Set the unsigned flags appropriately for
2372         the new builtin functions.
2373         (rs6000_opt_masks): Add support for decimal floating point builtin
2374         functions.
2375
2376         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
2377         floating point builtin functions.
2378         (RS6000_BTM_COMMON): Likewise.
2379         (RS6000_BTI_long_double): Likewise.
2380         (RS6000_BTI_dfloat64): Likewise.
2381         (RS6000_BTI_dfloat128): Likewise.
2382         (long_double_type_internal_node): Likewise.
2383         (dfloat64_type_internal_node): Likewise.
2384         (dfloat128_type_internal_node): Likewise.
2385
2386         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
2387         2.07 bcd arithmetic instructions.
2388         (UNSPEC_BCDSUB): Likewise.
2389         (UNSPEC_BCD_OVERFLOW): Likewise.
2390         (UNSPEC_BCD_ADD_SUB): Likewise.
2391         (bcd_add_sub): Likewise.
2392         (BCD_TEST): Likewise.
2393         (bcd<bcd_add_sub>): Likewise.
2394         (bcd<bcd_add_sub>_test): Likewise.
2395         (bcd<bcd_add_sub>_test2): Likewise.
2396         (bcd<bcd_add_sub>_<code>): Likewise.
2397         (peephole2 for combined bcd ops): Likewise.
2398
2399         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
2400         decimal floating point builtin functions.
2401         (UNSPEC_DENBCD): Likewise.
2402         (UNSPEC_DXEX): Likewise.
2403         (UNSPEC_DIEX): Likewise.
2404         (UNSPEC_DSCLI): Likewise.
2405         (UNSPEC_DSCRI): Likewise.
2406         (D64_D128): Likewise.
2407         (dfp_suffix): Likewise.
2408         (dfp_ddedpd_<mode>): Likewise.
2409         (dfp_denbcd_<mode>): Likewise.
2410         (dfp_dxex_<mode>): Likewise.
2411         (dfp_diex_<mode>): Likewise.
2412         (dfp_dscli_<mode>): Likewise.
2413         (dfp_dscri_<mode>): Likewise.
2414
2415         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
2416         builtin functions.
2417         (UNSPEC_CDTBCD): Likewise.
2418         (UNSPEC_CBCDTD): Likewise.
2419         (UNSPEC_DIVE): Add support for new extended divide builtin functions.
2420         (UNSPEC_DIVEO): Likewise.
2421         (UNSPEC_DIVEU): Likewise.
2422         (UNSPEC_DIVEUO): Likewise.
2423         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
2424         pack/unpack 128-bit types.
2425         (UNSPEC_PACK_128BIT): Likewise.
2426         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
2427         (udiv<mode>3): Use idiv_ldiv mode attribute.
2428         (div<mode>3): Likewise.
2429         (addg6s): Add new BCD builtin functions.
2430         (cdtbcd): Likewise.
2431         (cbcdtd): Likewise.
2432         (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
2433         (div_extend): Likewise.
2434         (div<div_extend>_<mode>"): Likewise.
2435         (FP128_64): Add support for new builtin functions to pack/unpack
2436         128-bit types.
2437         (unpack<mode>): Likewise.
2438         (unpacktf_0): Likewise.
2439         (unpacktf_1): Likewise.
2440         (unpack<mode>_dm): Likewise.
2441         (unpack<mode>_nodm): Likewise.
2442         (pack<mode>): Likewise.
2443         (unpackv1ti): Likewise.
2444         (packv1ti): Likewise.
2445
2446 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
2447
2448         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
2449         is disabled.
2450
2451 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
2452
2453         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
2454         * gimplify.c (omp_is_private): Change last argument's type to int.
2455         Only diagnose lastprivate if the simd argument is 1, only diagnose
2456         linear if the simd argument is 2.
2457         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
2458         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
2459         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
2460         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
2461         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
2462         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2463         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
2464         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2465         * tree-nested.c (convert_nonlocal_omp_clauses,
2466         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
2467
2468 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
2469
2470         PR target/60822
2471         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
2472         operand 1.
2473
2474 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
2475
2476         * flag-types.h (enum ivar_visibility): Add.
2477
2478 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
2479
2480         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
2481         function * argument.
2482
2483 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
2484
2485         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
2486
2487 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2488             Tom de Vries  <tom@codesourcery.com>
2489
2490         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
2491         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
2492         reg-note.
2493         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
2494         * emit-rtl.c (try_split): Same.
2495
2496 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2497             Tom de Vries  <tom@codesourcery.com>
2498
2499         * common.opt (fuse-caller-save): New option.
2500
2501 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
2502
2503         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
2504         elements for big-endian.
2505
2506 2014-04-24  Richard Biener  <rguenther@suse.de>
2507
2508         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
2509         during TER and instead use the sepops interface for expanding
2510         non-GIMPLE_SINGLE_RHS.
2511
2512 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2513
2514         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
2515         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
2516
2517 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2518
2519         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
2520         assembler 64-bit option.
2521         * configure: Regenerate.
2522
2523 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2524
2525         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
2526         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
2527         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
2528         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
2529         (TARGET_CRYPTO): Take TARGET_SIMD into account.
2530
2531 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2532
2533         * config/aarch64/aarch64-builtins.c
2534         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
2535         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
2536         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
2537         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
2538         builtins.
2539         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
2540         (Vrevsuff): New mode attribute.
2541
2542 2014-04-24  Terry Guo  <terry.guo@arm.com>
2543
2544         * config/arm/arm.h (machine_function): Define variable
2545         after_arm_reorg here.
2546         * config/arm/arm.c (after_arm_reorg): Remove the definition.
2547         (arm_split_constant): Update the way to access variable
2548         after_arm_reorg.
2549         (arm_reorg): Ditto.
2550         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
2551
2552 2014-04-23  Tom de Vries  <tom@codesourcery.com>
2553
2554         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
2555
2556 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
2557
2558         * is-a.h: Update comments to reflect the following changes to the
2559         "pointerness" of the API, making the template parameter match the
2560         return type, allowing use of is-a.h with typedefs of pointers.
2561         (is_a_helper::cast): Return a T rather then a pointer to a T, so
2562         that the return type matches the parameter to the is_a_helper.
2563         (as_a): Likewise.
2564         (dyn_cast): Likewise.
2565
2566         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
2567         pointer from the is-a.h API.
2568
2569         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
2570         (is_a_helper <cgraph_node *>::test): ...this, matching change to
2571         is-a.h API.
2572         (is_a_helper <varpool_node>::test): Likewise, convert to...
2573         (is_a_helper <varpool_node *>::test): ...this.
2574
2575         (varpool_first_variable): Update for removal of implicit pointer
2576         from the is-a.h API.
2577         (varpool_next_variable): Likewise.
2578         (varpool_first_static_initializer): Likewise.
2579         (varpool_next_static_initializer): Likewise.
2580         (varpool_first_defined_variable): Likewise.
2581         (varpool_next_defined_variable): Likewise.
2582         (cgraph_first_defined_function): Likewise.
2583         (cgraph_next_defined_function): Likewise.
2584         (cgraph_first_function): Likewise.
2585         (cgraph_next_function): Likewise.
2586         (cgraph_first_function_with_gimple_body): Likewise.
2587         (cgraph_next_function_with_gimple_body): Likewise.
2588         (cgraph_alias_target): Likewise.
2589         (varpool_alias_target): Likewise.
2590         (cgraph_function_or_thunk_node): Likewise.
2591         (varpool_variable_node): Likewise.
2592         (symtab_real_symbol_p): Likewise.
2593         * cgraphunit.c (referred_to_p): Likewise.
2594         (analyze_functions): Likewise.
2595         (handle_alias_pairs): Likewise.
2596         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
2597         * gimple-ssa.h (gimple_vuse_op): Likewise.
2598         (gimple_vdef_op): Likewise.
2599         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
2600         * gimple.c (gimple_build_asm_1): Likewise.
2601         (gimple_build_try): Likewise.
2602         (gimple_build_resx): Likewise.
2603         (gimple_build_eh_dispatch): Likewise.
2604         (gimple_build_omp_for): Likewise.
2605         (gimple_omp_for_set_clauses): Likewise.
2606
2607         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
2608         (is_a_helper <gimple_statement_asm *>::test): ...this.
2609         (is_a_helper <gimple_statement_bind>::test): Convert to...
2610         (is_a_helper <gimple_statement_bind *>::test): ...this.
2611         (is_a_helper <gimple_statement_call>::test): Convert to...
2612         (is_a_helper <gimple_statement_call *>::test): ...this.
2613         (is_a_helper <gimple_statement_catch>::test): Convert to...
2614         (is_a_helper <gimple_statement_catch *>::test): ...this.
2615         (is_a_helper <gimple_statement_resx>::test): Convert to...
2616         (is_a_helper <gimple_statement_resx *>::test): ...this.
2617         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
2618         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
2619         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
2620         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
2621         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
2622         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
2623         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
2624         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
2625         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
2626         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
2627         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
2628         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
2629         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
2630         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
2631         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
2632         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
2633         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
2634         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
2635         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
2636         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
2637         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
2638         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
2639         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
2640         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
2641         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
2642         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
2643         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
2644         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
2645         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
2646         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
2647         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
2648         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
2649         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
2650         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
2651         (is_a_helper <gimple_statement_phi>::test): Convert to...
2652         (is_a_helper <gimple_statement_phi *>::test): ...this.
2653         (is_a_helper <gimple_statement_transaction>::test): Convert to...
2654         (is_a_helper <gimple_statement_transaction *>::test): ...this.
2655         (is_a_helper <gimple_statement_try>::test): Convert to...
2656         (is_a_helper <gimple_statement_try *>::test): ...this.
2657         (is_a_helper <gimple_statement_wce>::test): Convert to...
2658         (is_a_helper <gimple_statement_wce *>::test): ...this.
2659         (is_a_helper <const gimple_statement_asm>::test): Convert to...
2660         (is_a_helper <const gimple_statement_asm *>::test): ...this.
2661         (is_a_helper <const gimple_statement_bind>::test): Convert to...
2662         (is_a_helper <const gimple_statement_bind *>::test): ...this.
2663         (is_a_helper <const gimple_statement_call>::test): Convert to...
2664         (is_a_helper <const gimple_statement_call *>::test): ...this.
2665         (is_a_helper <const gimple_statement_catch>::test): Convert to...
2666         (is_a_helper <const gimple_statement_catch *>::test): ...this.
2667         (is_a_helper <const gimple_statement_resx>::test): Convert to...
2668         (is_a_helper <const gimple_statement_resx *>::test): ...this.
2669         (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
2670         (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
2671         (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
2672         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
2673         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
2674         Convert to...
2675         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
2676         ...this.
2677         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
2678         Convert to...
2679         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
2680         ...this.
2681         (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
2682         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
2683         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
2684         to...
2685         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
2686         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
2687         to...
2688         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
2689         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
2690         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
2691         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
2692         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
2693         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
2694         to...
2695         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
2696         (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
2697         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
2698         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
2699         to...
2700         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
2701         (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
2702         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
2703         (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
2704         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
2705         (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
2706         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
2707         (is_a_helper <const gimple_statement_phi>::test): Convert to...
2708         (is_a_helper <const gimple_statement_phi *>::test): ...this.
2709         (is_a_helper <const gimple_statement_transaction>::test): Convert to...
2710         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
2711         (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
2712         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
2713         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
2714         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
2715         (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
2716         to...
2717         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
2718         ...this.
2719         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
2720         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
2721
2722         (gimple_use_ops): Update for removal of implicit pointer from the
2723         is-a.h API.
2724         (gimple_set_use_ops): Likewise.
2725         (gimple_vuse): Likewise.
2726         (gimple_vdef): Likewise.
2727         (gimple_vuse_ptr): Likewise.
2728         (gimple_vdef_ptr): Likewise.
2729         (gimple_set_vuse): Likewise.
2730         (gimple_set_vdef): Likewise.
2731         (gimple_omp_return_set_lhs): Likewise.
2732         (gimple_omp_return_lhs): Likewise.
2733         (gimple_omp_return_lhs_ptr): Likewise.
2734         (gimple_call_fntype): Likewise.
2735         (gimple_call_set_fntype): Likewise.
2736         (gimple_call_set_internal_fn): Likewise.
2737         (gimple_call_use_set): Likewise.
2738         (gimple_call_clobber_set): Likewise.
2739         (gimple_bind_vars): Likewise.
2740         (gimple_bind_set_vars): Likewise.
2741         (gimple_bind_body_ptr): Likewise.
2742         (gimple_bind_set_body): Likewise.
2743         (gimple_bind_add_stmt): Likewise.
2744         (gimple_bind_block): Likewise.
2745         (gimple_bind_set_block): Likewise.
2746         (gimple_asm_ninputs): Likewise.
2747         (gimple_asm_noutputs): Likewise.
2748         (gimple_asm_nclobbers): Likewise.
2749         (gimple_asm_nlabels): Likewise.
2750         (gimple_asm_input_op): Likewise.
2751         (gimple_asm_input_op_ptr): Likewise.
2752         (gimple_asm_output_op): Likewise.
2753         (gimple_asm_output_op_ptr): Likewise.
2754         (gimple_asm_set_output_op): Likewise.
2755         (gimple_asm_clobber_op): Likewise.
2756         (gimple_asm_set_clobber_op): Likewise.
2757         (gimple_asm_label_op): Likewise.
2758         (gimple_asm_set_label_op): Likewise.
2759         (gimple_asm_string): Likewise.
2760         (gimple_catch_types): Likewise.
2761         (gimple_catch_types_ptr): Likewise.
2762         (gimple_catch_handler_ptr): Likewise.
2763         (gimple_catch_set_types): Likewise.
2764         (gimple_catch_set_handler): Likewise.
2765         (gimple_eh_filter_types): Likewise.
2766         (gimple_eh_filter_types_ptr): Likewise.
2767         (gimple_eh_filter_failure_ptr): Likewise.
2768         (gimple_eh_filter_set_types): Likewise.
2769         (gimple_eh_filter_set_failure): Likewise.
2770         (gimple_eh_must_not_throw_fndecl): Likewise.
2771         (gimple_eh_must_not_throw_set_fndecl): Likewise.
2772         (gimple_eh_else_n_body_ptr): Likewise.
2773         (gimple_eh_else_e_body_ptr): Likewise.
2774         (gimple_eh_else_set_n_body): Likewise.
2775         (gimple_eh_else_set_e_body): Likewise.
2776         (gimple_try_eval_ptr): Likewise.
2777         (gimple_try_cleanup_ptr): Likewise.
2778         (gimple_try_set_eval): Likewise.
2779         (gimple_try_set_cleanup): Likewise.
2780         (gimple_wce_cleanup_ptr): Likewise.
2781         (gimple_wce_set_cleanup): Likewise.
2782         (gimple_phi_capacity): Likewise.
2783         (gimple_phi_num_args): Likewise.
2784         (gimple_phi_result): Likewise.
2785         (gimple_phi_result_ptr): Likewise.
2786         (gimple_phi_set_result): Likewise.
2787         (gimple_phi_arg): Likewise.
2788         (gimple_phi_set_arg): Likewise.
2789         (gimple_resx_region): Likewise.
2790         (gimple_resx_set_region): Likewise.
2791         (gimple_eh_dispatch_region): Likewise.
2792         (gimple_eh_dispatch_set_region): Likewise.
2793         (gimple_omp_critical_name): Likewise.
2794         (gimple_omp_critical_name_ptr): Likewise.
2795         (gimple_omp_critical_set_name): Likewise.
2796         (gimple_omp_for_clauses): Likewise.
2797         (gimple_omp_for_clauses_ptr): Likewise.
2798         (gimple_omp_for_set_clauses): Likewise.
2799         (gimple_omp_for_collapse): Likewise.
2800         (gimple_omp_for_index): Likewise.
2801         (gimple_omp_for_index_ptr): Likewise.
2802         (gimple_omp_for_set_index): Likewise.
2803         (gimple_omp_for_initial): Likewise.
2804         (gimple_omp_for_initial_ptr): Likewise.
2805         (gimple_omp_for_set_initial): Likewise.
2806         (gimple_omp_for_final): Likewise.
2807         (gimple_omp_for_final_ptr): Likewise.
2808         (gimple_omp_for_set_final): Likewise.
2809         (gimple_omp_for_incr): Likewise.
2810         (gimple_omp_for_incr_ptr): Likewise.
2811         (gimple_omp_for_set_incr): Likewise.
2812         (gimple_omp_for_pre_body_ptr): Likewise.
2813         (gimple_omp_for_set_pre_body): Likewise.
2814         (gimple_omp_parallel_clauses): Likewise.
2815         (gimple_omp_parallel_clauses_ptr): Likewise.
2816         (gimple_omp_parallel_set_clauses): Likewise.
2817         (gimple_omp_parallel_child_fn): Likewise.
2818         (gimple_omp_parallel_child_fn_ptr): Likewise.
2819         (gimple_omp_parallel_set_child_fn): Likewise.
2820         (gimple_omp_parallel_data_arg): Likewise.
2821         (gimple_omp_parallel_data_arg_ptr): Likewise.
2822         (gimple_omp_parallel_set_data_arg): Likewise.
2823         (gimple_omp_task_clauses): Likewise.
2824         (gimple_omp_task_clauses_ptr): Likewise.
2825         (gimple_omp_task_set_clauses): Likewise.
2826         (gimple_omp_task_child_fn): Likewise.
2827         (gimple_omp_task_child_fn_ptr): Likewise.
2828         (gimple_omp_task_set_child_fn): Likewise.
2829         (gimple_omp_task_data_arg): Likewise.
2830         (gimple_omp_task_data_arg_ptr): Likewise.
2831         (gimple_omp_task_set_data_arg): Likewise.
2832         (gimple_omp_taskreg_clauses): Likewise.
2833         (gimple_omp_taskreg_clauses_ptr): Likewise.
2834         (gimple_omp_taskreg_set_clauses): Likewise.
2835         (gimple_omp_taskreg_child_fn): Likewise.
2836         (gimple_omp_taskreg_child_fn_ptr): Likewise.
2837         (gimple_omp_taskreg_set_child_fn): Likewise.
2838         (gimple_omp_taskreg_data_arg): Likewise.
2839         (gimple_omp_taskreg_data_arg_ptr): Likewise.
2840         (gimple_omp_taskreg_set_data_arg): Likewise.
2841         (gimple_omp_task_copy_fn): Likewise.
2842         (gimple_omp_task_copy_fn_ptr): Likewise.
2843         (gimple_omp_task_set_copy_fn): Likewise.
2844         (gimple_omp_task_arg_size): Likewise.
2845         (gimple_omp_task_arg_size_ptr): Likewise.
2846         (gimple_omp_task_set_arg_size): Likewise.
2847         (gimple_omp_task_arg_align): Likewise.
2848         (gimple_omp_task_arg_align_ptr): Likewise.
2849         (gimple_omp_task_set_arg_align): Likewise.
2850         (gimple_omp_single_clauses): Likewise.
2851         (gimple_omp_single_clauses_ptr): Likewise.
2852         (gimple_omp_single_set_clauses): Likewise.
2853         (gimple_omp_target_clauses): Likewise.
2854         (gimple_omp_target_clauses_ptr): Likewise.
2855         (gimple_omp_target_set_clauses): Likewise.
2856         (gimple_omp_target_child_fn): Likewise.
2857         (gimple_omp_target_child_fn_ptr): Likewise.
2858         (gimple_omp_target_set_child_fn): Likewise.
2859         (gimple_omp_target_data_arg): Likewise.
2860         (gimple_omp_target_data_arg_ptr): Likewise.
2861         (gimple_omp_target_set_data_arg): Likewise.
2862         (gimple_omp_teams_clauses): Likewise.
2863         (gimple_omp_teams_clauses_ptr): Likewise.
2864         (gimple_omp_teams_set_clauses): Likewise.
2865         (gimple_omp_sections_clauses): Likewise.
2866         (gimple_omp_sections_clauses_ptr): Likewise.
2867         (gimple_omp_sections_set_clauses): Likewise.
2868         (gimple_omp_sections_control): Likewise.
2869         (gimple_omp_sections_control_ptr): Likewise.
2870         (gimple_omp_sections_set_control): Likewise.
2871         (gimple_omp_for_set_cond): Likewise.
2872         (gimple_omp_for_cond): Likewise.
2873         (gimple_omp_atomic_store_set_val): Likewise.
2874         (gimple_omp_atomic_store_val): Likewise.
2875         (gimple_omp_atomic_store_val_ptr): Likewise.
2876         (gimple_omp_atomic_load_set_lhs): Likewise.
2877         (gimple_omp_atomic_load_lhs): Likewise.
2878         (gimple_omp_atomic_load_lhs_ptr): Likewise.
2879         (gimple_omp_atomic_load_set_rhs): Likewise.
2880         (gimple_omp_atomic_load_rhs): Likewise.
2881         (gimple_omp_atomic_load_rhs_ptr): Likewise.
2882         (gimple_omp_continue_control_def): Likewise.
2883         (gimple_omp_continue_control_def_ptr): Likewise.
2884         (gimple_omp_continue_set_control_def): Likewise.
2885         (gimple_omp_continue_control_use): Likewise.
2886         (gimple_omp_continue_control_use_ptr): Likewise.
2887         (gimple_omp_continue_set_control_use): Likewise.
2888         (gimple_transaction_body_ptr): Likewise.
2889         (gimple_transaction_label): Likewise.
2890         (gimple_transaction_label_ptr): Likewise.
2891         (gimple_transaction_set_body): Likewise.
2892         (gimple_transaction_set_label): Likewise.
2893
2894         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
2895         * ipa-inline-analysis.c (inline_write_summary): Likewise.
2896         * ipa-ref.c (ipa_record_reference): Likewise.
2897         * ipa-reference.c (analyze_function): Likewise.
2898         (ipa_reference_write_optimization_summary): Likewise.
2899         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
2900         (address_taken_from_non_vtable_p): Likewise.
2901         (comdat_can_be_unshared_p_1): Likewise.
2902         * lto-cgraph.c (lto_output_ref): Likewise.
2903         (add_references): Likewise.
2904         (compute_ltrans_boundary): Likewise.
2905         (output_symtab): Likewise.
2906         (input_ref): Likewise.
2907         (input_cgraph_1): Likewise.
2908         (output_cgraph_opt_summary): Likewise.
2909         * lto-streamer-out.c (lto_output): Likewise.
2910         (output_symbol_p): Likewise.
2911         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
2912         (lsei_start_function_in_partition): Likewise.
2913         (lsei_next_variable_in_partition): Likewise.
2914         (lsei_start_variable_in_partition): Likewise.
2915         * symtab.c (insert_to_assembler_name_hash): Likewise.
2916         (unlink_from_assembler_name_hash): Likewise.
2917         (symtab_unregister_node): Likewise.
2918         (symtab_remove_node): Likewise.
2919         (dump_symtab_node): Likewise.
2920         (verify_symtab_base): Likewise.
2921         (verify_symtab_node): Likewise.
2922         (symtab_make_decl_local): Likewise.
2923         (symtab_alias_ultimate_target): Likewise.
2924         (symtab_resolve_alias): Likewise.
2925         (symtab_get_symbol_partitioning_class): Likewise.
2926         * tree-phinodes.c (allocate_phi_node): Likewise.
2927         (reserve_phi_args_for_new_edge): Likewise.
2928         (remove_phi_args): Likewise.
2929         * varpool.c (varpool_node_for_asm): Likewise.
2930         (varpool_remove_unreferenced_decls): Likewise.
2931
2932 2014-04-23  Jeff Law  <law@redhat.com>
2933
2934         PR tree-optimization/60902
2935         * tree-ssa-threadedge.c
2936         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
2937         invalidate outputs from statements that do not produce useful
2938         outputs for threading.
2939
2940 2014-04-23  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
2941
2942         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
2943         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
2944         machine descriptions for Stack Smashing Protector.
2945
2946 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
2947
2948         * aarch64.md (<optab>_rol<mode>3): New pattern.
2949         (<optab>_rolsi3_uxtw): Likewise.
2950         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
2951
2952 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
2953
2954         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
2955         (arm_cortex_a12_tune): Likewise.
2956
2957 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2958
2959         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
2960
2961 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2962
2963         * config/arm/arm.md (arm_rev16si2): New pattern.
2964         (arm_rev16si2_alt): Likewise.
2965         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
2966
2967 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2968
2969         * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
2970         (rev16<mode>2_alt): Likewise.
2971         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
2972         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
2973         (aarch_rev16_shleft_mask_imm_p): Likewise.
2974         (aarch_rev16_p_1): Likewise.
2975         (aarch_rev16_p): Likewise.
2976         * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
2977         (aarch_rev16_shright_mask_imm_p): Likewise.
2978         (aarch_rev16_shleft_mask_imm_p): Likewise.
2979
2980 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2981
2982         * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
2983         * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
2984         rev cost.
2985         (cortex_a53_extra_costs): Likewise.
2986         (cortex_a57_extra_costs): Likewise.
2987         * config/arm/arm.c (cortexa9_extra_costs): Likewise.
2988         (cortexa7_extra_costs): Likewise.
2989         (cortexa8_extra_costs): Likewise.
2990         (cortexa12_extra_costs): Likewise.
2991         (cortexa15_extra_costs): Likewise.
2992         (v7m_extra_costs): Likewise.
2993         (arm_new_rtx_costs): Handle BSWAP.
2994
2995 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2996
2997         * config/arm/arm.c (cortexa8_extra_costs): New table.
2998         (arm_cortex_a8_tune): New tuning struct.
2999         * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
3000
3001 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3002
3003         * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
3004
3005 2014-04-23  Richard Biener  <rguenther@suse.de>
3006
3007         * Makefile.in (OBJS): Remove loop-unswitch.o.
3008         * tree-pass.h (make_pass_rtl_unswitch): Remove.
3009         * passes.def (pass_rtl_unswitch): Likewise.
3010         * loop-init.c (gate_rtl_unswitch): Likewise.
3011         (rtl_unswitch): Likewise.
3012         (pass_data_rtl_unswitch): Likewise.
3013         (pass_rtl_unswitch): Likewise.
3014         (make_pass_rtl_unswitch): Likewise.
3015         * rtl.h (reversed_condition): Likewise.
3016         (compare_and_jump_seq): Likewise.
3017         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
3018         and make static.
3019         * loop-unroll.c (compare_and_jump_seq): Likewise.
3020
3021 2014-04-23  Richard Biener  <rguenther@suse.de>
3022
3023         PR tree-optimization/60903
3024         * tree-ssa-loop-im.c (analyze_memory_references): Remove
3025         commented code block.
3026         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
3027         loop flags to newly created BBs and edges.
3028
3029 2014-04-23  Nick Clifton  <nickc@redhat.com>
3030
3031         * config/msp430/msp430.c (msp430_handle_option): Move function
3032         to msp430-common.c
3033         (msp430_option_override): Simplify mcu and mcpu option handling.
3034         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
3035         support for -mhwmult command line option.
3036         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
3037         -mhwmult command line option.
3038         (msp430_hwmult_enabled): Delete.
3039         (msp43o_output_labelref): Add support for -mhwmult command line option.
3040         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
3041         (umulsidi3): Likewise.
3042         * config/msp430/msp430.opt (mmcu): Add Report attribute.
3043         (mcpu, mlarge, msmall): Likewise.
3044         (mhwmult): New option.
3045         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
3046         prototype.
3047         (msp430_is_f5_mcu): Remove prototype.
3048         (msp430_use_f5_series_hwmult): Add prototype.
3049         * config/msp430/msp430-opts.h: New file.
3050         * common/config/msp430: New directory.
3051         * common/config/msp430/msp430-common.c: New file.
3052         * config.gcc (msp430): Remove target_has_targetm_common.
3053         * doc/invoke.texi: Document -mhwmult command line option.
3054
3055 2014-04-23  Nick Clifton  <nickc@redhat.com>
3056
3057         * config/i386/cygwin.h (ENDFILE_SPEC): Include
3058         default-manifest.o if it can be found in the search path.
3059         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
3060
3061 2014-04-23  Terry Guo  <terry.guo@arm.com>
3062
3063         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
3064
3065 2014-04-23  Richard Biener  <rguenther@suse.de>
3066
3067         PR middle-end/60895
3068         * tree-inline.c (declare_return_variable): Use mark_addressable.
3069
3070 2014-04-23  Richard Biener  <rguenther@suse.de>
3071
3072         PR middle-end/60891
3073         * loop-init.c (loop_optimizer_init): Make sure to apply
3074         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
3075
3076 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3077
3078         PR sanitizer/60275
3079         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
3080         New options.
3081         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
3082         if flag_sanitize_undefined_trap_on_error.
3083         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
3084         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
3085         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
3086         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3087         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
3088         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
3089         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
3090         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
3091         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
3092         * ubsan.c (ubsan_instrument_unreachable): Return
3093         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
3094         (ubsan_expand_null_ifn): Emit __builtin_trap ()
3095         if flag_sanitize_undefined_trap_on_error and
3096         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
3097         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
3098         instrument_bool_enum_load): Emit __builtin_trap () if
3099         flag_sanitize_undefined_trap_on_error and
3100         __builtin_handle_*_abort () if !flag_sanitize_recover.
3101         * doc/invoke.texi (-fsanitize-recover,
3102         -fsanitize-undefined-trap-on-error): Document.
3103
3104 2014-04-22  Christian Bruel  <christian.bruel@st.com>
3105
3106         * config/sh/sh.md (mov<mode>): Replace movQIHI.
3107          Force immediates to SImode.
3108
3109 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
3110
3111         * config/nios2/nios2.md (UNSPEC_ROUND): New.
3112         (lroundsfsi2): New.
3113         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
3114         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
3115         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
3116         (nios2_fpu_insn): Add entry for round.
3117         (N2FPU_NO_ERRNO_P): Define.
3118         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
3119         flag_errno_math.
3120         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
3121
3122 2014-04-22  Richard Henderson  <rth@redhat.com>
3123
3124         * config/aarch64/aarch64 (addti3, subti3): New expanders.
3125         (add<GPI>3_compare0): Remove leading * from name.
3126         (add<GPI>3_carryin): Likewise.
3127         (sub<GPI>3_compare0): Likewise.
3128         (sub<GPI>3_carryin): Likewise.
3129         (<su_optab>mulditi3): New expander.
3130         (multi3): New expander.
3131         (madd<GPI>): Remove leading * from name.
3132
3133 2014-04-22  Martin Jambor  <mjambor@suse.cz>
3134
3135         * cgraphclones.c (cgraph_function_versioning): Copy
3136         ipa_transforms_to_apply instead of asserting it is empty.
3137
3138 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
3139
3140         PR target/60868
3141         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
3142         on count_exp to get mode.
3143
3144 2014-04-22  Andrew Pinski  <apinski@cavium.com>
3145
3146         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3147         Handle TLS for ILP32.
3148         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
3149         (tlsie_small_<mode>): this and handle PTR.
3150         (tlsie_small_sidi): New pattern.
3151         (tlsle_small): Change to an expand to handle ILP32.
3152         (tlsle_small_<mode>): New pattern.
3153         (tlsdesc_small): Rename to ...
3154         (tlsdesc_small_<mode>): this and handle PTR.
3155
3156 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3157
3158         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
3159
3160 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3161
3162         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3163         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
3164         (aarch64_types_signed_poly_qualifiers): Likewise.
3165         (aarch64_types_unsigned_signed_qualifiers): Likewise.
3166         (aarch64_types_poly_signed_qualifiers): Likewise.
3167         (TYPES_REINTERP_SS): Type macro added.
3168         (TYPES_REINTERP_SU): Likewise.
3169         (TYPES_REINTERP_SP): Likewise.
3170         (TYPES_REINTERP_US): Likewise.
3171         (TYPES_REINTERP_PS): Likewise.
3172         (aarch64_fold_builtin): New expression folding added.
3173         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
3174         Declarations removed.
3175         (REINTERP_SS): Declarations added.
3176         (REINTERP_US): Likewise.
3177         (REINTERP_PS): Likewise.
3178         (REINTERP_SU): Likewise.
3179         (REINTERP_SP): Likewise.
3180         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
3181         (vreinterpretq_p8_f64): Likewise.
3182         (vreinterpret_p16_f64): Likewise.
3183         (vreinterpretq_p16_f64): Likewise.
3184         (vreinterpret_f32_f64): Likewise.
3185         (vreinterpretq_f32_f64): Likewise.
3186         (vreinterpret_f64_f32): Likewise.
3187         (vreinterpret_f64_p8): Likewise.
3188         (vreinterpret_f64_p16): Likewise.
3189         (vreinterpret_f64_s8): Likewise.
3190         (vreinterpret_f64_s16): Likewise.
3191         (vreinterpret_f64_s32): Likewise.
3192         (vreinterpret_f64_s64): Likewise.
3193         (vreinterpret_f64_u8): Likewise.
3194         (vreinterpret_f64_u16): Likewise.
3195         (vreinterpret_f64_u32): Likewise.
3196         (vreinterpret_f64_u64): Likewise.
3197         (vreinterpretq_f64_f32): Likewise.
3198         (vreinterpretq_f64_p8): Likewise.
3199         (vreinterpretq_f64_p16): Likewise.
3200         (vreinterpretq_f64_s8): Likewise.
3201         (vreinterpretq_f64_s16): Likewise.
3202         (vreinterpretq_f64_s32): Likewise.
3203         (vreinterpretq_f64_s64): Likewise.
3204         (vreinterpretq_f64_u8): Likewise.
3205         (vreinterpretq_f64_u16): Likewise.
3206         (vreinterpretq_f64_u32): Likewise.
3207         (vreinterpretq_f64_u64): Likewise.
3208         (vreinterpret_s64_f64): Likewise.
3209         (vreinterpretq_s64_f64): Likewise.
3210         (vreinterpret_u64_f64): Likewise.
3211         (vreinterpretq_u64_f64): Likewise.
3212         (vreinterpret_s8_f64): Likewise.
3213         (vreinterpretq_s8_f64): Likewise.
3214         (vreinterpret_s16_f64): Likewise.
3215         (vreinterpretq_s16_f64): Likewise.
3216         (vreinterpret_s32_f64): Likewise.
3217         (vreinterpretq_s32_f64): Likewise.
3218         (vreinterpret_u8_f64): Likewise.
3219         (vreinterpretq_u8_f64): Likewise.
3220         (vreinterpret_u16_f64): Likewise.
3221         (vreinterpretq_u16_f64): Likewise.
3222         (vreinterpret_u32_f64): Likewise.
3223         (vreinterpretq_u32_f64): Likewise.
3224
3225 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3226
3227         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3228         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
3229         (vreinterpret_p8_s8): Likewise.
3230         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
3231         (vreinterpret_p8_s16): Likewise.
3232         (vreinterpret_p8_s32): Likewise.
3233         (vreinterpret_p8_s64): Likewise.
3234         (vreinterpret_p8_f32): Likewise.
3235         (vreinterpret_p8_u8): Likewise.
3236         (vreinterpret_p8_u16): Likewise.
3237         (vreinterpret_p8_u32): Likewise.
3238         (vreinterpret_p8_u64): Likewise.
3239         (vreinterpret_p8_p16): Likewise.
3240         (vreinterpretq_p8_s8): Likewise.
3241         (vreinterpretq_p8_s16): Likewise.
3242         (vreinterpretq_p8_s32): Likewise.
3243         (vreinterpretq_p8_s64): Likewise.
3244         (vreinterpretq_p8_f32): Likewise.
3245         (vreinterpretq_p8_u8): Likewise.
3246         (vreinterpretq_p8_u16): Likewise.
3247         (vreinterpretq_p8_u32): Likewise.
3248         (vreinterpretq_p8_u64): Likewise.
3249         (vreinterpretq_p8_p16): Likewise.
3250         (vreinterpret_p16_s8): Likewise.
3251         (vreinterpret_p16_s16): Likewise.
3252         (vreinterpret_p16_s32): Likewise.
3253         (vreinterpret_p16_s64): Likewise.
3254         (vreinterpret_p16_f32): Likewise.
3255         (vreinterpret_p16_u8): Likewise.
3256         (vreinterpret_p16_u16): Likewise.
3257         (vreinterpret_p16_u32): Likewise.
3258         (vreinterpret_p16_u64): Likewise.
3259         (vreinterpret_p16_p8): Likewise.
3260         (vreinterpretq_p16_s8): Likewise.
3261         (vreinterpretq_p16_s16): Likewise.
3262         (vreinterpretq_p16_s32): Likewise.
3263         (vreinterpretq_p16_s64): Likewise.
3264         (vreinterpretq_p16_f32): Likewise.
3265         (vreinterpretq_p16_u8): Likewise.
3266         (vreinterpretq_p16_u16): Likewise.
3267         (vreinterpretq_p16_u32): Likewise.
3268         (vreinterpretq_p16_u64): Likewise.
3269         (vreinterpretq_p16_p8): Likewise.
3270         (vreinterpret_f32_s8): Likewise.
3271         (vreinterpret_f32_s16): Likewise.
3272         (vreinterpret_f32_s32): Likewise.
3273         (vreinterpret_f32_s64): Likewise.
3274         (vreinterpret_f32_u8): Likewise.
3275         (vreinterpret_f32_u16): Likewise.
3276         (vreinterpret_f32_u32): Likewise.
3277         (vreinterpret_f32_u64): Likewise.
3278         (vreinterpret_f32_p8): Likewise.
3279         (vreinterpret_f32_p16): Likewise.
3280         (vreinterpretq_f32_s8): Likewise.
3281         (vreinterpretq_f32_s16): Likewise.
3282         (vreinterpretq_f32_s32): Likewise.
3283         (vreinterpretq_f32_s64): Likewise.
3284         (vreinterpretq_f32_u8): Likewise.
3285         (vreinterpretq_f32_u16): Likewise.
3286         (vreinterpretq_f32_u32): Likewise.
3287         (vreinterpretq_f32_u64): Likewise.
3288         (vreinterpretq_f32_p8): Likewise.
3289         (vreinterpretq_f32_p16): Likewise.
3290         (vreinterpret_s64_s8): Likewise.
3291         (vreinterpret_s64_s16): Likewise.
3292         (vreinterpret_s64_s32): Likewise.
3293         (vreinterpret_s64_f32): Likewise.
3294         (vreinterpret_s64_u8): Likewise.
3295         (vreinterpret_s64_u16): Likewise.
3296         (vreinterpret_s64_u32): Likewise.
3297         (vreinterpret_s64_u64): Likewise.
3298         (vreinterpret_s64_p8): Likewise.
3299         (vreinterpret_s64_p16): Likewise.
3300         (vreinterpretq_s64_s8): Likewise.
3301         (vreinterpretq_s64_s16): Likewise.
3302         (vreinterpretq_s64_s32): Likewise.
3303         (vreinterpretq_s64_f32): Likewise.
3304         (vreinterpretq_s64_u8): Likewise.
3305         (vreinterpretq_s64_u16): Likewise.
3306         (vreinterpretq_s64_u32): Likewise.
3307         (vreinterpretq_s64_u64): Likewise.
3308         (vreinterpretq_s64_p8): Likewise.
3309         (vreinterpretq_s64_p16): Likewise.
3310         (vreinterpret_u64_s8): Likewise.
3311         (vreinterpret_u64_s16): Likewise.
3312         (vreinterpret_u64_s32): Likewise.
3313         (vreinterpret_u64_s64): Likewise.
3314         (vreinterpret_u64_f32): Likewise.
3315         (vreinterpret_u64_u8): Likewise.
3316         (vreinterpret_u64_u16): Likewise.
3317         (vreinterpret_u64_u32): Likewise.
3318         (vreinterpret_u64_p8): Likewise.
3319         (vreinterpret_u64_p16): Likewise.
3320         (vreinterpretq_u64_s8): Likewise.
3321         (vreinterpretq_u64_s16): Likewise.
3322         (vreinterpretq_u64_s32): Likewise.
3323         (vreinterpretq_u64_s64): Likewise.
3324         (vreinterpretq_u64_f32): Likewise.
3325         (vreinterpretq_u64_u8): Likewise.
3326         (vreinterpretq_u64_u16): Likewise.
3327         (vreinterpretq_u64_u32): Likewise.
3328         (vreinterpretq_u64_p8): Likewise.
3329         (vreinterpretq_u64_p16): Likewise.
3330         (vreinterpret_s8_s16): Likewise.
3331         (vreinterpret_s8_s32): Likewise.
3332         (vreinterpret_s8_s64): Likewise.
3333         (vreinterpret_s8_f32): Likewise.
3334         (vreinterpret_s8_u8): Likewise.
3335         (vreinterpret_s8_u16): Likewise.
3336         (vreinterpret_s8_u32): Likewise.
3337         (vreinterpret_s8_u64): Likewise.
3338         (vreinterpret_s8_p8): Likewise.
3339         (vreinterpret_s8_p16): Likewise.
3340         (vreinterpretq_s8_s16): Likewise.
3341         (vreinterpretq_s8_s32): Likewise.
3342         (vreinterpretq_s8_s64): Likewise.
3343         (vreinterpretq_s8_f32): Likewise.
3344         (vreinterpretq_s8_u8): Likewise.
3345         (vreinterpretq_s8_u16): Likewise.
3346         (vreinterpretq_s8_u32): Likewise.
3347         (vreinterpretq_s8_u64): Likewise.
3348         (vreinterpretq_s8_p8): Likewise.
3349         (vreinterpretq_s8_p16): Likewise.
3350         (vreinterpret_s16_s8): Likewise.
3351         (vreinterpret_s16_s32): Likewise.
3352         (vreinterpret_s16_s64): Likewise.
3353         (vreinterpret_s16_f32): Likewise.
3354         (vreinterpret_s16_u8): Likewise.
3355         (vreinterpret_s16_u16): Likewise.
3356         (vreinterpret_s16_u32): Likewise.
3357         (vreinterpret_s16_u64): Likewise.
3358         (vreinterpret_s16_p8): Likewise.
3359         (vreinterpret_s16_p16): Likewise.
3360         (vreinterpretq_s16_s8): Likewise.
3361         (vreinterpretq_s16_s32): Likewise.
3362         (vreinterpretq_s16_s64): Likewise.
3363         (vreinterpretq_s16_f32): Likewise.
3364         (vreinterpretq_s16_u8): Likewise.
3365         (vreinterpretq_s16_u16): Likewise.
3366         (vreinterpretq_s16_u32): Likewise.
3367         (vreinterpretq_s16_u64): Likewise.
3368         (vreinterpretq_s16_p8): Likewise.
3369         (vreinterpretq_s16_p16): Likewise.
3370         (vreinterpret_s32_s8): Likewise.
3371         (vreinterpret_s32_s16): Likewise.
3372         (vreinterpret_s32_s64): Likewise.
3373         (vreinterpret_s32_f32): Likewise.
3374         (vreinterpret_s32_u8): Likewise.
3375         (vreinterpret_s32_u16): Likewise.
3376         (vreinterpret_s32_u32): Likewise.
3377         (vreinterpret_s32_u64): Likewise.
3378         (vreinterpret_s32_p8): Likewise.
3379         (vreinterpret_s32_p16): Likewise.
3380         (vreinterpretq_s32_s8): Likewise.
3381         (vreinterpretq_s32_s16): Likewise.
3382         (vreinterpretq_s32_s64): Likewise.
3383         (vreinterpretq_s32_f32): Likewise.
3384         (vreinterpretq_s32_u8): Likewise.
3385         (vreinterpretq_s32_u16): Likewise.
3386         (vreinterpretq_s32_u32): Likewise.
3387         (vreinterpretq_s32_u64): Likewise.
3388         (vreinterpretq_s32_p8): Likewise.
3389         (vreinterpretq_s32_p16): Likewise.
3390         (vreinterpret_u8_s8): Likewise.
3391         (vreinterpret_u8_s16): Likewise.
3392         (vreinterpret_u8_s32): Likewise.
3393         (vreinterpret_u8_s64): Likewise.
3394         (vreinterpret_u8_f32): Likewise.
3395         (vreinterpret_u8_u16): Likewise.
3396         (vreinterpret_u8_u32): Likewise.
3397         (vreinterpret_u8_u64): Likewise.
3398         (vreinterpret_u8_p8): Likewise.
3399         (vreinterpret_u8_p16): Likewise.
3400         (vreinterpretq_u8_s8): Likewise.
3401         (vreinterpretq_u8_s16): Likewise.
3402         (vreinterpretq_u8_s32): Likewise.
3403         (vreinterpretq_u8_s64): Likewise.
3404         (vreinterpretq_u8_f32): Likewise.
3405         (vreinterpretq_u8_u16): Likewise.
3406         (vreinterpretq_u8_u32): Likewise.
3407         (vreinterpretq_u8_u64): Likewise.
3408         (vreinterpretq_u8_p8): Likewise.
3409         (vreinterpretq_u8_p16): Likewise.
3410         (vreinterpret_u16_s8): Likewise.
3411         (vreinterpret_u16_s16): Likewise.
3412         (vreinterpret_u16_s32): Likewise.
3413         (vreinterpret_u16_s64): Likewise.
3414         (vreinterpret_u16_f32): Likewise.
3415         (vreinterpret_u16_u8): Likewise.
3416         (vreinterpret_u16_u32): Likewise.
3417         (vreinterpret_u16_u64): Likewise.
3418         (vreinterpret_u16_p8): Likewise.
3419         (vreinterpret_u16_p16): Likewise.
3420         (vreinterpretq_u16_s8): Likewise.
3421         (vreinterpretq_u16_s16): Likewise.
3422         (vreinterpretq_u16_s32): Likewise.
3423         (vreinterpretq_u16_s64): Likewise.
3424         (vreinterpretq_u16_f32): Likewise.
3425         (vreinterpretq_u16_u8): Likewise.
3426         (vreinterpretq_u16_u32): Likewise.
3427         (vreinterpretq_u16_u64): Likewise.
3428         (vreinterpretq_u16_p8): Likewise.
3429         (vreinterpretq_u16_p16): Likewise.
3430         (vreinterpret_u32_s8): Likewise.
3431         (vreinterpret_u32_s16): Likewise.
3432         (vreinterpret_u32_s32): Likewise.
3433         (vreinterpret_u32_s64): Likewise.
3434         (vreinterpret_u32_f32): Likewise.
3435         (vreinterpret_u32_u8): Likewise.
3436         (vreinterpret_u32_u16): Likewise.
3437         (vreinterpret_u32_u64): Likewise.
3438         (vreinterpret_u32_p8): Likewise.
3439         (vreinterpret_u32_p16): Likewise.
3440         (vreinterpretq_u32_s8): Likewise.
3441         (vreinterpretq_u32_s16): Likewise.
3442         (vreinterpretq_u32_s32): Likewise.
3443         (vreinterpretq_u32_s64): Likewise.
3444         (vreinterpretq_u32_f32): Likewise.
3445         (vreinterpretq_u32_u8): Likewise.
3446         (vreinterpretq_u32_u16): Likewise.
3447         (vreinterpretq_u32_u64): Likewise.
3448         (vreinterpretq_u32_p8): Likewise.
3449         (vreinterpretq_u32_p16): Likewise.
3450
3451 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3452
3453         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
3454         Pattern extended.
3455         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
3456         (sqabs): Likewise.
3457         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
3458         (vqnegd_s64): Likewise.
3459         (vqabs_s64): Likewise.
3460         (vqabsd_s64): Likewise.
3461
3462 2014-04-22  Richard Henderson  <rth@redhat.com>
3463
3464         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
3465         computation to the top of the loop.
3466
3467 2014-04-22  Renlin  <renlin.li@arm.com>
3468             Jiong Wang  <jiong.wang@arm.com>
3469
3470         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
3471         * config/aarch64/aarch64.c (aarch64_layout_frame)
3472         (aarch64_initial_elimination_offset): Likewise.
3473
3474 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3475
3476         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
3477         Fix indentation.
3478
3479 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
3480
3481         * machmode.h (bitwise_mode_for_mode): Declare.
3482         * stor-layout.h (bitwise_type_for_mode): Likewise.
3483         * stor-layout.c (bitwise_mode_for_mode): New function.
3484         (bitwise_type_for_mode): Likewise.
3485         * builtins.c (fold_builtin_memory_op): Use it instead of
3486         int_mode_for_mode and build_nonstandard_integer_type.
3487
3488 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3489
3490         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
3491         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
3492         (*-*-solaris2*): Simplify.
3493         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
3494         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
3495         *-*-solaris2.9* handling.
3496
3497         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
3498         as bug.
3499         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
3500         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
3501         handling, simplify.
3502         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
3503         * configure: Regenerate.
3504
3505         * config/i386/sol2-9.h: Remove.
3506
3507         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
3508         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
3509         Remove Solaris 9 references.
3510
3511 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
3512
3513         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
3514         (floatuns<GPI:mode><GPF:mode>2): Remove.
3515         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
3516         and floatuns conversions.
3517         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
3518         and floatuns conversions.
3519         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
3520         (w1,w2): New mode attributes for inequal width conversions.
3521
3522 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
3523
3524         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
3525         the output asm format.
3526
3527 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
3528
3529         * config/aarch64/aarch64-simd.md
3530         (aarch64_cm<optab>di): Always split.
3531         (*aarch64_cm<optab>di): New.
3532         (aarch64_cmtstdi): Always split.
3533         (*aarch64_cmtstdi): New.
3534
3535 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3536
3537         PR tree-optimization/60823
3538         * omp-low.c (ipa_simd_modify_function_body): Go through
3539         all SSA_NAMEs and for those refering to vector arguments
3540         which are going to be replaced adjust SSA_NAME_VAR and,
3541         if it is a default definition, change it into a non-default
3542         definition assigned at the beginning of function from new_decl.
3543         (ipa_simd_modify_stmt_ops): Rewritten.
3544         * tree-dfa.c (set_ssa_default_def): When removing default def,
3545         check for NULL loc instead of NULL *loc.
3546
3547 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3548
3549         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
3550         restrictions on core registers for DImode values in Thumb2.
3551
3552 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3553
3554         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
3555         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
3556
3557 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3558
3559         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
3560         (*iordi_notzesidi_di): Likewise.
3561         (*iordi_notsesidi_di): Likewise.
3562
3563 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3564
3565         * config/arm/arm-protos.h (tune_params): New struct members.
3566         * config/arm/arm.c: Initialise tune_params per processor.
3567         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
3568         for speed, based on new tune_params.
3569
3570 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3571
3572         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
3573         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
3574         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
3575         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
3576         * config/aarch64/arm_neon.h (vrnd_f64): Added.
3577         (vrnda_f64): Likewise.
3578         (vrndi_f64): Likewise.
3579         (vrndm_f64): Likewise.
3580         (vrndn_f64): Likewise.
3581         (vrndp_f64): Likewise.
3582         (vrndx_f64): Likewise.
3583
3584 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3585
3586         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
3587         GET_MODE_SIZE argument is enum machine_mode.
3588
3589 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3590
3591         PR target/60910
3592         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
3593         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
3594
3595 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
3596
3597         PR middle-end/60281
3598         * asan.c (asan_emit_stack_protection): Force the base to align to
3599         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
3600         appropriate bits if STRICT_ALIGNMENT.
3601         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
3602         when asan is on.
3603         (expand_used_vars): Leave a space in the stack frame for alignment
3604         if STRICT_ALIGNMENT.
3605
3606 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
3607
3608         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
3609         than a gimple.
3610         (gimple_store_p): Likewise.
3611         (gimple_assign_load_p): Likewise.
3612         (gimple_assign_cast_p): Likewise.
3613         (gimple_clobber_p): Likewise.
3614
3615         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
3616         rather than a gimple.
3617         (gimple_assign_cast_p): Likewise.
3618
3619 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
3620
3621         PR target/60735
3622         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
3623         If mode is DDmode and TARGET_E500_DOUBLE allow move.
3624
3625         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
3626         more debug information for E500 if -mdebug=reg.
3627
3628 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
3629
3630         PR target/60909
3631         * config/i386/i386.c (ix86_expand_builtin)
3632         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
3633         register for target RTX.
3634         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
3635
3636 2014-04-18  Cong Hou  <congh@google.com>
3637
3638         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
3639         the widen-mult pattern by handling two operands with different sizes,
3640         and operands whose size is smaller than half of the result type.
3641
3642 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
3643
3644         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
3645         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
3646         (do_estimate_edge_time): Compute it.
3647         * ipa-inline.c (want_inline_small_function_p): Bypass
3648         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
3649
3650 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
3651
3652         * ipa-inline.c (spec_rem): New static variable.
3653         (dump_overall_stats): New function.
3654         (dump_inline_stats): New function.
3655
3656 2014-04-18  Richard Henderson  <rth@redhat.com>
3657
3658         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
3659         to GET_MODE_SIZE, not a reg_class_t.
3660
3661 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3662
3663         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
3664         (vsx_xxmrglw_<mode>): Likewise.
3665
3666 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3667
3668         PR target/60876
3669         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
3670         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
3671         (rs6000_init_hard_regno_mode_ok): Likewise.
3672
3673 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
3674
3675         * ipa-inline.c (inline_small_functions): Account only non-cold
3676         functions.
3677         * doc/invoke.texi (inline-unit-growth): Update documentation.
3678
3679 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
3680
3681         * config/rs6000/rs6000.md (addti3, subti3): New.
3682
3683 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
3684
3685         PR target/60863
3686         * config/i386/i386.c (ix86_expand_clear): Remove outdated
3687         comment.  Check optimize_insn_for_size_p instead of
3688         optimize_insn_for_speed_p.
3689
3690 2014-04-17  Martin Jambor  <mjambor@suse.cz>
3691
3692         * gimple-iterator.c (gsi_start_edge): New function.
3693         * gimple-iterator.h (gsi_start_edge): Declare.
3694         * tree-sra.c (single_non_eh_succ): New function.
3695         (disqualify_ops_if_throwing_stmt): Renamed to
3696         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
3697         having one non-EH successor BB.
3698         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
3699         generate loads into replacements.
3700         (sra_modify_assign): Likewise and and also use the simple path for
3701         such statements.
3702         (sra_modify_function_body): Commit statements on edges.
3703
3704 2014-04-17  Richard Biener  <rguenther@suse.de>
3705
3706         PR middle-end/60849
3707         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
3708         comparison results and add clarifying comment.
3709
3710 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3711
3712         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
3713         (blank_mode): Initialize it.
3714         (emit_mode_size_inline, emit_mode_nunits_inline,
3715         emit_mode_inner_inline): New functions.
3716         (emit_insn_modes_h): Call them and surround their output with
3717         #if GCC_VERSION >= 4001 ... #endif.
3718         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
3719         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
3720         mode_* arrays if the argument is __builtin_constant_p.
3721         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
3722         is enum machine_mode.
3723
3724 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3725
3726         * passes.c (opt_pass::execute): Adjust.
3727         (pass_manager::execute_pass_mode_switching): Likewise.
3728         (early_local_passes::execute): Likewise.
3729         (execute_one_pass): Pass cfun to the pass's execute method.
3730         * tree-pass.h (opt_pass::execute): Add function * argument.
3731         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3732         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3733         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3734         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3735         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3736         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
3737         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
3738         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3739         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3740         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
3741         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
3742         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
3743         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
3744         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
3745         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3746         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3747         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3748         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
3749         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
3750         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3751         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3752         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3753         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3754         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3755         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3756         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3757         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3758         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3759         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3760         Adjust.
3761
3762 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3763
3764         * passes.c (opt_pass::gate): Take function * argument.
3765         (gate_all_early_local_passes): Merge into
3766         (early_local_passes::gate): this.
3767         (gate_all_early_optimizations): Merge into
3768         (all_early_optimizations::gate): this.
3769         (gate_all_optimizations): Mege into
3770         (all_optimizations::gate): this.
3771         (gate_all_optimizations_g): Merge into
3772         (all_optimizations_g::gate): this.
3773         (gate_rest_of_compilation): Mege into
3774         (rest_of_compilation::gate): this.
3775         (gate_postreload): Merge into
3776         (postreload::gate): this.
3777         (dump_one_pass): Pass cfun to the pass's gate method.
3778         (execute_ipa_summary_passes): Likewise.
3779         (execute_one_pass): Likewise.
3780         (ipa_write_summaries_2): Likewise.
3781         (ipa_write_optimization_summaries_1): Likewise.
3782         (ipa_read_summaries_1): Likewise.
3783         (ipa_read_optimization_summaries_1): Likewise.
3784         (execute_ipa_stmt_fixups): Likewise.
3785         * tree-pass.h (opt_pass::gate): Add function * argument.
3786         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
3787         combine-stack-adj.c, combine.c, compare-elim.c,
3788         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3789         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
3790         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
3791         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
3792         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3793         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3794         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3795         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
3796         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
3797         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
3798         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3799         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3800         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
3801         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3802         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3803         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3804         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3805         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3806         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3807         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3808         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3809         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3810         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3811         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
3812         var-tracking.c, vtable-verify.c, web.c: Adjust.
3813
3814 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3815
3816         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
3817         * configure: Regenerate.
3818
3819 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3820
3821         * passes.c (dump_one_pass): don't check pass->has_gate.
3822         (execute_ipa_summary_passes): Likewise.
3823         (execute_one_pass): Likewise.
3824         (ipa_write_summaries_2): Likewise.
3825         (ipa_write_optimization_summaries_1): Likewise.
3826         (ipa_read_optimization_summaries_1): Likewise.
3827         (execute_ipa_stmt_fixups): Likewise.
3828         * tree-pass.h (pass_data::has_gate): Remove.
3829         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3830         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3831         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3832         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3833         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3834         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
3835         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
3836         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
3837         gimple-low.c, gimple-ssa-isolate-paths.c,
3838         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
3839         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3840         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3841         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
3842         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
3843         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
3844         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
3845         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
3846         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
3847         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3848         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3849         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3850         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3851         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3852         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3853         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3854         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3855         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3856         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3857         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3858         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3859         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3860         Adjust.
3861
3862 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3863
3864         * pass_manager.h (pass_manager::register_dump_files_1): Remove
3865         declaration.
3866         * passes.c (pass_manager::register_dump_files_1): Merge into
3867         (pass_manager::register_dump_files): this, and remove its handling of
3868         properties since the pass always has the properties anyway.
3869         (pass_manager::pass_manager): Adjust.
3870
3871 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3872
3873         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
3874         * passes.c (pass_manager::register_dump_files_1): Remove dead code
3875         dealing with properties.
3876         (pass_manager::register_dump_files): Adjust.
3877
3878 2014-03-20  Mark Wielaard  <mjw@redhat.com>
3879
3880         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
3881         then represent the bound as normal constant value.
3882
3883 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3884
3885         PR target/60847
3886         Forward port from 4.8 branch
3887         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
3888
3889         * config/i386/bmiintrin.h (_blsi_u32): New.
3890         (_blsi_u64): Ditto.
3891         (_blsr_u32): Ditto.
3892         (_blsr_u64): Ditto.
3893         (_blsmsk_u32): Ditto.
3894         (_blsmsk_u64): Ditto.
3895         (_tzcnt_u32): Ditto.
3896         (_tzcnt_u64): Ditto.
3897
3898 2014-04-17  Kito Cheng  <kito@0xlab.org>
3899
3900         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
3901
3902 2014-04-17  Richard Biener  <rguenther@suse.de>
3903
3904         PR middle-end/60849
3905         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
3906         boolean results for comparisons.
3907
3908 2014-04-17  Richard Biener  <rguenther@suse.de>
3909
3910         PR tree-optimization/60836
3911         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
3912         initial PHI args to be gimple values.
3913
3914 2014-04-17   Richard Biener  <rguenther@suse.de>
3915
3916         PR tree-optimization/60841
3917         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
3918         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
3919         of stmts to SLP build.
3920         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3921         (vect_analyze_slp): Likewise.
3922         (vect_analyze_slp_instance): Likewise.
3923         (vect_build_slp_tree): Limit overall SLP tree growth.
3924         * tree-vectorizer.h (vect_analyze_data_refs,
3925         vect_analyze_slp): Adjust prototypes.
3926
3927 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3928
3929         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
3930         Silvermont.
3931
3932 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3933
3934         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
3935         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
3936         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
3937         for TARGET_SLOW_PSHUFB
3938
3939 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3940
3941         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
3942         * config/i386/i386.c (intel_cost): Ditto.
3943
3944 2014-04-17  Joey Ye  <joey.ye@arm.com>
3945
3946         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
3947
3948 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3949
3950         * opts.c (common_handle_option): Disable -fipa-reference coorectly
3951         with -fuse-profile.
3952
3953 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3954
3955         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
3956         (type_all_derivations_known_p): New predicate.
3957         (type_all_ctors_visible_p): New predicate.
3958         (type_possibly_instantiated_p): New predicate.
3959         (get_odr_type): Compute all_derivations_known.
3960         (dump_odr_type): Dump the flag.
3961         (maybe_record_type): Cleanup.
3962         (record_target_from_binfo): Add bases_to_consider array;
3963         record bases for types w/o instances and skip CXX destructor.
3964         (possible_polymorphic_call_targets_1): Add bases_to_consider
3965         and consider_construction parameters; check if type may have instance.
3966         (get_polymorphic_call_info): Set maybe_in_construction to true
3967         when we know nothing.
3968         (record_targets_from_bases): Skip CXX destructors; they are
3969         never called for types in construction.
3970         (possible_polymorphic_call_targets): Do not record target when
3971         type may not have instance.
3972
3973 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3974
3975         PR ipa/60854
3976         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
3977         external aliases alive, too.
3978
3979 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
3980
3981         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
3982         definition.
3983
3984 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3985
3986         * final.c (compute_alignments): Do not apply loop alignment to a block
3987         falling through to the exit.
3988
3989 2014-04-16  Catherine Moore  <clm@codesourcery.com>
3990
3991         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
3992         Adjust constraints for microMIPS store patterns.
3993
3994 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
3995
3996         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
3997
3998 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3999
4000         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
4001         (append_use): Run at -O0.
4002         (append_vdef): Likewise.
4003         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
4004         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
4005
4006 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
4007
4008         PR tree-optimization/60844
4009         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
4010         (propagate_op_to_single_use, remove_visited_stmt_chain,
4011         linearize_expr, repropagate_negates, reassociate_bb): Use it
4012         instead of gsi_remove.
4013
4014 2014-04-16  Martin Jambor  <mjambor@suse.cz>
4015
4016         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
4017         ipa_transforms_to_apply.
4018         (cgraph_function_versioning): Assert that old_node has empty
4019         ipa_transforms_to_apply.
4020         * trans-mem.c (ipa_tm_create_version): Likewise.
4021         * tree-inline.c (tree_function_versioning): Do not duplicate
4022         ipa_transforms_to_apply.
4023
4024 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4025
4026         PR target/60817
4027         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
4028         x86_64-*-* cases.
4029         Pass necessary as flags on 64-bit Solaris/x86.
4030         Use lowercase relocs for x86_64-*-*.
4031         * configure: Regenerate.
4032
4033 2014-04-15  Jan Hubicka  <jh@suse.cz>
4034
4035         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
4036         (maybe_record_node, likely_target_p): Use it.
4037
4038 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4039
4040         PR target/60839
4041         Revert following patch
4042
4043         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4044
4045         PR target/60735
4046         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4047         software floating point or no floating point registers, do not
4048         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4049         in GPRs that occurs after we tested for GPRs that would never be
4050         true.
4051
4052         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4053         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4054         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4055         specifically allow DDmode, since that does not use the SPE SIMD
4056         instructions.
4057
4058 2014-03-21  Mark Wielaard  <mjw@redhat.com>
4059
4060         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
4061         as unsigned or int depending on type and value used.
4062
4063 2014-04-15  Richard Biener  <rguenther@suse.de>
4064
4065         PR rtl-optimization/56965
4066         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
4067         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
4068         ... here.
4069         * alias.c (true_dependence_1): Do not call
4070         nonoverlapping_component_refs_p.
4071         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
4072         nonoverlapping_component_refs_p.
4073         (indirect_refs_may_alias_p): Likewise.
4074
4075 2014-04-15  Teresa Johnson  <tejohnson@google.com>
4076
4077         * cfg.c (dump_bb_info): Fix flags check.
4078         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
4079
4080 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4081
4082         PR rtl-optimization/60663
4083         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
4084         avoid 0 cost.
4085
4086 2014-04-15  Richard Biener  <rguenther@suse.de>
4087
4088         * lto-streamer.h (LTO_major_version): Bump to 4.
4089
4090 2014-04-15  Richard Biener  <rguenther@suse.de>
4091
4092         * common.opt (lto_partition_model): New enum.
4093         (flto-partition=): Merge separate options with a single with argument,
4094         add -flto-partition=one support.
4095         * flag-types.h (enum lto_partition_model): Declare.
4096         * opts.c (finish_options): Remove duplicate -flto-partition=
4097         option check.
4098         * lto-wrapper.c (run_gcc): Adjust.
4099
4100 2014-04-15  Richard Biener  <rguenther@suse.de>
4101
4102         * alias.c (ncr_compar): New function.
4103         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
4104
4105 2014-04-15  Richard Biener  <rguenther@suse.de>
4106
4107         * alias.c (record_component_aliases): Do not walk BINFOs.
4108
4109 2014-04-15  Richard Biener  <rguenther@suse.de>
4110
4111         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4112         Add struct function argument and adjust.
4113         (find_func_aliases_for_call): Likewise.
4114         (find_func_aliases): Likewise.
4115         (find_func_clobbers): Likewise.
4116         (intra_create_variable_infos): Likewise.
4117         (compute_points_to_sets): Likewise.
4118         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
4119
4120 2014-04-15  Richard Biener  <rguenther@suse.de>
4121
4122         * tree.c (iterative_hash_expr): Use enum tree_code_class
4123         to store TREE_CODE_CLASS.
4124         (tree_block): Likewise.
4125         (tree_set_block): Likewise.
4126         * tree.h (fold_build_pointer_plus_loc): Use
4127         convert_to_ptrofftype_loc.
4128
4129 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
4130
4131         PR plugins/59335
4132         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
4133         added in 4.9.
4134
4135 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
4136
4137         * cfgloop.h (struct loop): Move force_vectorize down.
4138         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
4139         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
4140         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
4141         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
4142         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
4143         * tree-core.h (enum annot_expr_kind): Add new kind values.
4144         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
4145         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
4146         kinds.
4147         * tree.def (ANNOTATE_EXPR): Tweak comment.
4148
4149 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4150
4151         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
4152         cxa_pure_virtual).
4153
4154 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
4155
4156         * tree.h (TYPE_IDENTIFIER): Declare.
4157         * tree.c (subrange_type_for_debug_p): Use it.
4158         * godump.c (go_format_type): Likewise.
4159         * dwarf2out.c (is_cxx_auto, modified_type_die,
4160         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
4161         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
4162
4163 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
4164
4165         PR lto/60820
4166         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
4167
4168 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
4169
4170         * config/i386/i386.c (examine_argument): Return bool.  Return true if
4171         parameter should be passed in memory.
4172         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
4173         (construct_container): Update calls to examine_argument.
4174         (function_arg_advance_64): Ditto.
4175         (return_in_memory_32): Merge with ix86_return_in_memory.
4176         (return_in_memory_64): Ditto.
4177         (return_in_memory_ms_64): Ditto.
4178
4179 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
4180
4181         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
4182         * coverage.c (coverage_compute_profile_id): Handle externally visible
4183         symbols.
4184
4185 2014-04-14  Martin Jambor  <mjambor@suse.cz>
4186
4187         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
4188         DECL_DISREGARD_INLINE_LIMITS functions.
4189
4190 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4191
4192         PR target/60827
4193         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
4194
4195 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
4196
4197         PR target/60827
4198         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
4199         optimize_insn_for_speed_p instead of
4200         optimize_function_for_speed_p.
4201
4202 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
4203
4204         * doc/invoke.texi (free): Document AArch64.
4205
4206 2014-04-14  Richard Biener  <rguenther@suse.de>
4207
4208         PR tree-optimization/60042
4209         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
4210         (insert_into_preds_of_block): Do not prevent PHI insertion
4211         for REFERENCE exprs here ...
4212         (eliminate_dom_walker::before_dom_children): ... but prevent
4213         their use here under similar conditions when applied to the
4214         IL after PRE optimizations.
4215
4216 2014-04-14  Richard Biener  <rguenther@suse.de>
4217
4218         * passes.def: Move early points-to after early SRA.
4219
4220 2014-04-14  Richard Biener  <rguenther@suse.de>
4221
4222         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
4223         check for which sign-changes we allow when forwarding
4224         a converted value into a switch.
4225
4226 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4227
4228         * stor-layout.c (place_field): Finalize non-constant offset for the
4229         field, if any.
4230
4231 2014-04-14  Richard Biener  <rguenther@suse.de>
4232
4233         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
4234         as argument.
4235         (expand_switch_using_bit_tests_p): Likewise.
4236         (process_switch): Compute and pass on speed_p based on the
4237         switch stmt.
4238         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
4239         optimize_bb_for_speed_p.
4240
4241 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4242
4243         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
4244         * function.h (struct function): Rename has_force_vect_loops into
4245         has_force_vectorize_loops.
4246         * lto-streamer-in.c (input_cfg): Adjust for renaming.
4247         (input_struct_function_base): Likewise.
4248         * lto-streamer-out.c (output_cfg): Likewise.
4249         (output_struct_function_base): Likewise.
4250         * omp-low.c (expand_omp_simd): Likewise.
4251         * tree-cfg.c (move_sese_region_to_fn): Likewise.
4252         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4253         (version_loop_for_if_conversion): Likewise.
4254         (tree_if_conversion): Likewise.
4255         (main_tree_if_conversion): Likewise.
4256         (gate_tree_if_conversion): Likewise.
4257         * tree-inline.c (copy_loops): Likewise.
4258         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
4259         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
4260         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
4261         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
4262         * tree-vectorizer.c (vectorize_loops): Likewise.
4263         * tree-vectorizer.h (unlimited_cost_model): Likewise.
4264
4265 2014-04-14  Richard Biener  <rguenther@suse.de>
4266
4267         PR lto/60720
4268         * lto-streamer-out.c (wrap_refs): New function.
4269         (lto_output): Wrap symbol references in global initializes in
4270         type-preserving MEM_REFs.
4271
4272 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4273
4274         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
4275
4276 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4277
4278         * config/sh/sh.md (setmemqi): New expand pattern.
4279         * config/sh/sh.h (CLEAR_RATIO): Define.
4280         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
4281         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
4282
4283 2014-04-14  Richard Biener  <rguenther@suse.de>
4284
4285         PR middle-end/55022
4286         * fold-const.c (negate_expr_p): Don't negate directional rounding
4287         division.
4288         (fold_negate_expr): Likewise.
4289
4290 2014-04-14  Richard Biener  <rguenther@suse.de>
4291
4292         PR tree-optimization/59817
4293         PR tree-optimization/60453
4294         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
4295         recursion to catch all CHRECs in the scalar evolution and restrict
4296         the predicate for the remains appropriately.
4297
4298 2014-04-12  Catherine Moore  <clm@codesourcery.com>
4299
4300         * config/mips/constraints.md: Add new register constraint "kb".
4301         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
4302         (*movhi_internal): Likewise.
4303         (*movqi_internal): Likewise.
4304         * config/mips/mips.h (M16_STORE_REGS): New register class.
4305         (REG_CLASS_NAMES): Add M16_STORE_REGS.
4306         (REG_CLASS_CONTENTS): Likewise.
4307         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
4308
4309 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4310
4311         PR c/60194
4312         * doc/invoke.texi (-Wformat-signedness): Document it.
4313         (Wformat=2): Mention that this enables -Wformat-signedness.
4314
4315 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4316
4317         * common/config/epiphany/epiphany-common.c
4318         (epiphany_option_optimization_table): Enable section anchors by
4319         default at -O1 or higher.
4320         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
4321         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
4322         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
4323         carries no extra cost.
4324         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
4325         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
4326         * config/epiphany/predicates.md (memclob_operand): New predicate.
4327         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
4328         Use memclob_operand predicate and X constraint for operand 3.
4329
4330 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4331
4332         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
4333         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
4334         its operands.
4335
4336 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4337
4338         PR rtl-optimization/60651
4339         * mode-switching.c (optimize_mode_switching): Make sure to emit
4340         sets of a lower numbered entity before sets of a higher numbered
4341         entity to a mode of the same or lower priority.
4342         When creating a seginfo for a basic block that starts with a code
4343         label, move the insertion point past the code label.
4344         (new_seginfo): Document and enforce requirement that
4345         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
4346         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
4347         * doc/tm.texi: Regenerate.
4348
4349 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4350
4351         PR target/60811
4352         * config/arc/arc.c (arc_save_restore): Fix assert typo.
4353
4354 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
4355
4356         * BASE-VER: Set to 4.10.0.
4357
4358 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4359
4360         PR other/59055
4361         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
4362         * doc/gcc.texi (Service): Update description in the @menu
4363         * doc/invoke.texi (Option Summary): Remove misplaced and
4364         duplicated @menu.
4365
4366 2014-04-11  Steve Ellcey  <sellcey@mips.com>
4367             Jakub Jelinek  <jakub@redhat.com>
4368
4369         PR middle-end/60556
4370         * expr.c (convert_move): Use emit_store_flag_force instead of
4371         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
4372         argument to it.
4373
4374 2014-04-11  Richard Biener  <rguenther@suse.de>
4375
4376         PR middle-end/60797
4377         * varasm.c (assemble_alias): Avoid endless error reporting
4378         recursion by setting TREE_ASM_WRITTEN.
4379
4380 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4381
4382         * config/s390/s390.md: Add a splitter for NOT rtx.
4383
4384 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
4385
4386         PR rtl-optimization/60663
4387         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
4388
4389 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
4390             Jakub Jelinek  <jakub@redhat.com>
4391
4392         PR lto/60567
4393         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
4394         flag from decl_node to node.
4395
4396 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4397
4398         PR debug/60655
4399         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
4400         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
4401         ameliorating the cases where it can be.
4402
4403 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
4404
4405         Revert
4406         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4407
4408         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4409         (loadsync_<mode>): Change mode.
4410         (load_quadpti, store_quadpti): New.
4411         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4412         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4413         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
4414
4415 2014-04-09  Cong Hou  <congh@google.com>
4416
4417         PR testsuite/60773
4418         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
4419         documentation.
4420
4421 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4422
4423         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
4424         instead of vnor to exploit possible fusion opportunity in the
4425         future.
4426         (altivec_expand_vec_perm_const_le): Likewise.
4427
4428 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4429
4430         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4431         (loadsync_<mode>): Change mode.
4432         (load_quadpti, store_quadpti): New.
4433         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4434         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4435
4436 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
4437
4438         PR target/60763
4439         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
4440         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
4441         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
4442
4443 2014-04-08  Richard Biener  <rguenther@suse.de>
4444
4445         PR middle-end/60706
4446         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
4447         a 64bit widest int print double-int similar to on HWI64 hosts.
4448
4449 2014-04-08  Richard Biener  <rguenther@suse.de>
4450
4451         PR tree-optimization/60785
4452         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
4453         default defs properly.
4454
4455 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
4456
4457         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
4458         (Weffc++): Likewise.
4459
4460 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
4461
4462         * ipa-devirt.c (maybe_record_node): When node is not recorded,
4463         set completep to false rather than true.
4464
4465 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
4466
4467         PR target/60504
4468         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
4469         ARM_TARGET2_DWARF_FORMAT.
4470
4471 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
4472
4473         PR target/60609
4474         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
4475         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
4476         ADDR_DIFF_VEC.
4477
4478 2014-04-07  Richard Biener  <rguenther@suse.de>
4479
4480         PR tree-optimization/60766
4481         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
4482         (may_eliminate_iv): Convert cand_value_at result to desired type.
4483
4484 2014-04-07  Jason Merrill  <jason@redhat.com>
4485
4486         PR c++/60731
4487         * common.opt (-fno-gnu-unique): Add.
4488         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
4489
4490 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4491
4492         * haifa-sched.c: Fix outdated function reference and minor
4493         grammar errors in introductory comment.
4494
4495 2014-04-07  Richard Biener  <rguenther@suse.de>
4496
4497         PR middle-end/60750
4498         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
4499         for noreturn calls.
4500         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
4501
4502 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
4503
4504         PR debug/55794
4505         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
4506         size accounting for thunks.
4507         (pa_asm_output_mi_thunk): Use final_start_function() and
4508         final_end_function() to output function start and end directives.
4509
4510 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
4511
4512         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
4513         device specific ISA/ feature information. Remove short_sp and
4514         errata_skip ds.  Add avr_device_specific_features enum to have device
4515         specific info.
4516         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
4517         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
4518         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
4519         updated device specific info.
4520         * config/avr/avr-mcus.def: Merge device specific details to
4521         dev_attribute field.
4522         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
4523         errata_skip.
4524         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
4525         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
4526         assembler if RMW isa supported by current device.
4527         * config/avr/genmultilib.awk: Update as device info structure changed.
4528         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
4529
4530 2014-04-04  Cong Hou  <congh@google.com>
4531
4532         PR tree-optimization/60656
4533         * tree-vect-stmts.c (supportable_widening_operation):
4534         Fix a bug that elements in a vector with vect_used_by_reduction
4535         property are incorrectly reordered when the operation on it is not
4536         consistant with the one in reduction operation.
4537
4538 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
4539
4540         PR rtl-optimization/60155
4541         * gcse.c (record_set_data): New function.
4542         (single_set_gcse): New function.
4543         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
4544         (hoist_code): Likewise.
4545         (get_pressure_class_and_nregs): Likewise.
4546
4547 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
4548
4549         * explow.c (probe_stack_range): Emit a final optimization blockage.
4550
4551 2014-04-04  Anthony Green  <green@moxielogic.com>
4552
4553         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
4554         typos.
4555
4556 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
4557
4558         PR ipa/59626
4559         * lto-cgraph.c (input_overwrite_node): Check that partitioning
4560         flags are set only during streaming.
4561         * ipa.c (process_references, walk_polymorphic_call_targets,
4562         symtab_remove_unreachable_nodes): Drop bodies of always inline
4563         after early inlining.
4564         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
4565
4566 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
4567         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4568
4569         PR debug/60655
4570         * dwarf2out.c (const_ok_for_output_1): Reject expressions
4571         containing a NOT.
4572
4573 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4574
4575         PR bootstrap/60743
4576         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
4577         duration.
4578         (cortex_a53_fdivd): Likewise.
4579
4580 2014-04-04  Martin Jambor  <mjambor@suse.cz>
4581
4582         PR ipa/60640
4583         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
4584         Adjust all callers.
4585         * cgraph.c (clone_of_p): Also return true if thunks match.
4586         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
4587         cgraph_function_or_thunk_node and an obsolete comment.
4588         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
4589         file.
4590         (build_function_decl_skip_args): Likewise.
4591         (set_new_clone_decl_and_node_flags): New function.
4592         (duplicate_thunk_for_node): Likewise.
4593         (redirect_edge_duplicating_thunks): Likewise.
4594         (cgraph_clone_node): New parameter args_to_skip, pass it to
4595         redirect_edge_duplicating_thunks which is called instead of
4596         cgraph_redirect_edge_callee.
4597         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
4598         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
4599
4600 2014-04-04  Jeff Law  <law@redhat.com>
4601
4602         PR target/60657
4603         * config/arm/predicates.md (const_int_I_operand): New predicate.
4604         (const_int_M_operand): Similarly.
4605         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
4606         const_int_operand.
4607         (insv_t2, extv_reg, extzv_t2): Likewise.
4608         (load_multiple_with_writeback): Similarly for const_int_I_operand.
4609         (pop_multiple_with_writeback_and_return): Likewise.
4610         (vfp_pop_multiple_with_writeback): Likewise
4611
4612 2014-04-04  Richard Biener  <rguenther@suse.de>
4613
4614         PR ipa/60746
4615         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
4616         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
4617         non-GIMPLE_LABELs.
4618         * gimplify.h (gimple_add_tmp_var_fn): Declare.
4619         * gimplify.c (gimple_add_tmp_var_fn): New function.
4620         * gimple-expr.h (create_tmp_reg_fn): Declare.
4621         * gimple-expr.c (create_tmp_reg_fn): New function.
4622         * gimple-low.c (record_vars_into): Don't change cfun.
4623         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
4624         code generation without cfun.
4625
4626 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
4627
4628         PR bootstrap/60719
4629         * Makefile.in (install-driver): Fix shell scripting.
4630
4631 2014-04-03  Cong Hou  <congh@google.com>
4632
4633         PR tree-optimization/60505
4634         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
4635         threshold of number of iterations below which no vectorization
4636         will be done.
4637         * tree-vect-loop.c (new_loop_vec_info):
4638         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
4639         * tree-vect-loop.c (vect_analyze_loop_operations):
4640         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
4641         * tree-vect-loop.c (vect_transform_loop):
4642         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
4643         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
4644         of iterations of the loop and see if we should build the epilogue.
4645
4646 2014-04-03  Richard Biener  <rguenther@suse.de>
4647
4648         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
4649         (streamer_tree_cache_create): Adjust.
4650         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
4651         to allow optional nodes array.
4652         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
4653         (streamer_tree_cache_append): Likewise.
4654         (streamer_tree_cache_create): Create nodes array optionally
4655         as specified by parameter.
4656         * lto-streamer-out.c (create_output_block): Avoid maintaining
4657         the node array in the writer cache.
4658         (DFS_write_tree): Remove assertion.
4659         (produce_asm_for_decls): Free the out decl state hash table early.
4660         * lto-streamer-in.c (lto_data_in_create): Adjust for
4661         streamer_tree_cache_create prototype change.
4662
4663 2014-04-03  Richard Biener  <rguenther@suse.de>
4664
4665         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
4666         set TREE_CHAIN to NULL_TREE.
4667
4668 2014-04-03  Richard Biener  <rguenther@suse.de>
4669
4670         PR tree-optimization/60740
4671         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
4672         over all GIMPLE_COND operands.
4673
4674 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
4675
4676         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
4677         (Weffc++): Remove Scott's numbering, merge lists and reference
4678         Wnon-virtual-dtor.
4679
4680 2014-04-03  Nick Clifton  <nickc@redhat.com>
4681
4682         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
4683         properly.
4684
4685 2014-04-03  Martin Jambor  <mjambor@suse.cz>
4686
4687         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
4688         mention gcc_unreachable before failing.
4689         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
4690         removed symbols.
4691
4692 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
4693
4694         PR ipa/60659
4695         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
4696         inconsistent code and instead mark the context inconsistent.
4697         (possible_polymorphic_call_targets): For inconsistent contexts
4698         return empty complete list.
4699
4700 2014-04-02  Anthony Green  <green@moxielogic.com>
4701
4702         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
4703         (extendqisi2, extendhisi2): Define.
4704         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
4705         (WCHAR_TYPE): Change to unsigned int.
4706
4707 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4708
4709         PR tree-optimization/60733
4710         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
4711         insertion point for PHI candidates to be the end of the feeding
4712         block for the PHI argument.
4713
4714 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
4715
4716         PR rtl-optimization/60650
4717         * lra-constraints.c (process_alt_operands): Decrease reject for
4718         earlyclobber matching.
4719
4720 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4721
4722         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
4723
4724 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4725
4726         * config/spu/spu.c (pad_bb): Do not crash when the last
4727         insn is CODE_FOR_blockage.
4728
4729 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4730
4731         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
4732         lies outside the target mode.
4733
4734 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4735
4736         PR target/60735
4737         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4738         software floating point or no floating point registers, do not
4739         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4740         in GPRs that occurs after we tested for GPRs that would never be
4741         true.
4742
4743         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4744         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4745         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4746         specifically allow DDmode, since that does not use the SPE SIMD
4747         instructions.
4748
4749 2014-04-02  Richard Biener  <rguenther@suse.de>
4750
4751         PR middle-end/60729
4752         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
4753         MODE_INTs.  Properly use negv_optab.
4754         (expand_abs): Likewise.
4755
4756 2014-04-02  Richard Biener  <rguenther@suse.de>
4757
4758         PR bootstrap/60719
4759         * Makefile.in (install-driver): Guard extra installs with special
4760         names properly.
4761
4762 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
4763
4764         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4765         Document vec_vgbbd.
4766
4767 2014-04-01  Richard Henderson  <rth@redhat.com>
4768
4769         PR target/60704
4770         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
4771         alternative enabled before register allocation.
4772
4773 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
4774
4775         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
4776         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
4777         typo.
4778         (nios2_large_got_address): Remove unneeded 'sym' parameter.
4779         (nios2_got_address): Update nios2_large_got_address call site.
4780         (nios2_delegitimize_address): New function.
4781         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
4782         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
4783         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
4784
4785 2014-04-01  Martin Husemann  <martin@duskware.de>
4786
4787         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
4788         for -mabi=32.
4789
4790 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
4791
4792         PR rtl-optimization/60604
4793         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
4794         check from register_operand.
4795         (register_operand): Redefine in terms of general_operand.
4796         (nonmemory_operand): Use register_operand for the non-constant cases.
4797
4798 2014-04-01  Richard Biener  <rguenther@suse.de>
4799
4800         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
4801
4802 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4803
4804         * doc/invoke.texi (mapp-regs): Clarify.
4805
4806 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
4807
4808         * config/i386/avx512fintrin.h (__v32hi): Define type.
4809         (__v64qi): Likewise.
4810         (_mm512_set1_epi8): Define.
4811         (_mm512_set1_epi16): Define.
4812         (_mm512_set4_epi32): Define.
4813         (_mm512_set4_epi64): Define.
4814         (_mm512_set4_pd): Define.
4815         (_mm512_set4_ps): Define.
4816         (_mm512_setr4_epi64): Define.
4817         (_mm512_setr4_epi32): Define.
4818         (_mm512_setr4_pd): Define.
4819         (_mm512_setr4_ps): Define.
4820         (_mm512_setzero_epi32): Define.
4821
4822 2014-03-31  Martin Jambor  <mjambor@suse.cz>
4823
4824         PR middle-end/60647
4825         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
4826         callsite_arguments_match_p.  Updated all callers.  Also check types of
4827         corresponding formal parameters and actual arguments.
4828         (not_all_callers_have_enough_arguments_p) Renamed to
4829         some_callers_have_mismatched_arguments_p.
4830
4831 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
4832
4833         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
4834
4835 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
4836
4837         PR target/60034
4838         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
4839         section anchor.
4840
4841 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
4842
4843         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
4844         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
4845         Split out
4846         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
4847         Use FMAMODE_NOVF512 mode iterator.
4848         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
4849         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
4850         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
4851         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
4852         Split out
4853         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
4854         Use VF_128_256 mode iterator.
4855         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
4856         Ditto.
4857
4858 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4859
4860         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
4861         static chain if needed.
4862
4863 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4864
4865         PR target/60697
4866         * lra-constraints.c (index_part_to_reg): New.
4867         (process_address): Use it.
4868
4869 2014-03-27  Jeff Law  <law@redhat.com>
4870             Jakub Jelinek  <jakub@redhat.com>
4871
4872         PR target/60648
4873         * expr.c (do_tablejump): Use simplify_gen_binary rather than
4874         gen_rtx_{PLUS,MULT} to build up the address expression.
4875
4876         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
4877         creating non-canonical RTL.
4878
4879 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4880
4881         PR ipa/60243
4882         * ipa-inline.c (want_inline_small_function_p): Short circuit large
4883         functions; reorganize to make cheap checks first.
4884         (inline_small_functions): Do not estimate growth when dumping;
4885         it is expensive.
4886         * ipa-inline.h (inline_summary): Add min_size.
4887         (growth_likely_positive): New function.
4888         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
4889         (set_cond_stmt_execution_predicate): Cleanup.
4890         (estimate_edge_size_and_time): Compute min_size.
4891         (estimate_calls_size_and_time): Likewise.
4892         (estimate_node_size_and_time): Likewise.
4893         (inline_update_overall_summary): Update min_size.
4894         (do_estimate_edge_time): Likewise.
4895         (do_estimate_edge_size): Update.
4896         (do_estimate_edge_hints): Update.
4897         (growth_likely_positive): New function.
4898
4899 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
4900
4901         PR target/60693
4902         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
4903         also if addr has VOIDmode.
4904
4905 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4906
4907         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
4908         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
4909         Declare extern.
4910         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
4911         instructions as well as AdvancedSIMD loads.
4912
4913 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4914
4915         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
4916         Use crypto_aese type.
4917         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
4918         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
4919         crypto_aese, crypto_aesmc.  Move to types.md.
4920         * config/arm/types.md (crypto_aes): Split into crypto_aese,
4921         crypto_aesmc.
4922         * config/arm/iterators.md (crypto_type): Likewise.
4923
4924 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4925
4926         * cgraph.c: Include expr.h and tree-dfa.h.
4927         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
4928         remove LHS.
4929
4930 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4931
4932         PR target/60675
4933         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
4934         regs from checking multi-reg pseudos.
4935
4936 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4937
4938         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
4939
4940 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4941
4942         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
4943         if it would clobber the stack pointer, even temporarily.
4944
4945 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
4946
4947         * mode-switching.c: Make small adjustments to the top comment.
4948
4949 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
4950
4951         * config/rs6000/constraints.md (wD constraint): New constraint to
4952         match the constant integer to get the top DImode/DFmode out of a
4953         vector in a VSX register.
4954
4955         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
4956         match the constant integer to get the top DImode/DFmode out of a
4957         vector in a VSX register.
4958
4959         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
4960         for ISA 2.07.
4961
4962         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4963         vbpermq builtins.
4964
4965         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
4966         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
4967
4968         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
4969         Optimize vec_extract of 64-bit values, where the value being
4970         extracted is in the top word, where we can use scalar
4971         instructions.  Add direct move and store support.  Combine the big
4972         endian/little endian vector select load support into a single insn.
4973         (vsx_extract_<mode>_internal1): Likewise.
4974         (vsx_extract_<mode>_internal2): Likewise.
4975         (vsx_extract_<mode>_load): Likewise.
4976         (vsx_extract_<mode>_store): Likewise.
4977         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
4978         combined into vsx_extract_<mode>_load.
4979         (vsx_extract_<mode>_one_le): Likewise.
4980
4981         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
4982         define the top 64-bit vector element.
4983
4984         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
4985         constraint.
4986
4987         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4988         Document vec_vbpermq builtin.
4989
4990         PR target/60672
4991         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
4992         enable use of xxsldwi and xxpermdi builtin functions.
4993         (vec_xxpermdi): Likewise.
4994
4995         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4996         Document use of vec_xxsldwi and vec_xxpermdi builtins.
4997
4998 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
4999
5000         PR rtl-optimization/60650
5001         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
5002         first_p.  Use it.
5003         (find_spills_for): New.
5004         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
5005         Spill all pseudos on the second iteration.
5006
5007 2014-03-27  Marek Polacek  <polacek@redhat.com>
5008
5009         PR c/50347
5010         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
5011         types.
5012
5013 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5014
5015         * config/s390/s390.c (s390_can_use_return_insn): Check for
5016         call-saved FPRs on 31 bit.
5017
5018 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
5019
5020         PR middle-end/60682
5021         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
5022         if they need regimplification, just drop them instead of
5023         calling gimple_regimplify_operands on them.
5024
5025 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5026
5027         PR target/60580
5028         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
5029         (aarch64_frame_pointer_required): Adjust logic.
5030         (aarch64_can_eliminate): Adjust logic.
5031         (aarch64_override_options_after_change): Adjust logic.
5032
5033 2014-03-27  Dehao Chen  <dehao@google.com>
5034
5035         * ipa-inline.c (early_inliner): Update node's inline info.
5036
5037 2014-03-26  Dehao Chen  <dehao@google.com>
5038
5039         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
5040         compiler inserted conditional jumps for NAN float check.
5041
5042 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5043
5044         * ubsan.h (ubsan_create_data): Change second argument's type
5045         to const location_t *.
5046         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
5047         _("<unknown>").
5048         (ubsan_create_data): Change second argument to const location_t *PLOC.
5049         Create Loc field whenever PLOC is non-NULL.
5050         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
5051         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
5052         callers.
5053
5054         PR other/59545
5055         * real.c (real_to_integer2): Change type of low to UHWI.
5056
5057 2014-03-26  Tobias Burnus  <burnus@net-b.de>
5058
5059         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
5060         (CILK_SELF_SPECS): New define.
5061         (driver_self_specs): Use it.
5062
5063 2014-03-26  Richard Biener  <rguenther@suse.de>
5064
5065         * tree-pretty-print.c (percent_K_format): Implement special
5066         case for LTO and its stripped down BLOCK tree.
5067
5068 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
5069
5070         PR sanitizer/60636
5071         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
5072
5073         * tree-vrp.c (simplify_internal_call_using_ranges): If only
5074         one range is range_int_cst_p, but not both, at least optimize
5075         addition/subtraction of 0 and multiplication by 0 or 1.
5076         * gimple-fold.c (gimple_fold_call): Fold
5077         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
5078         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
5079         INTEGER_CSTs, try to fold at least x * 0 and y - y.
5080
5081 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
5082
5083         PR rtl-optimization/60452
5084         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
5085         <case REG>: Return 1 for invalid offsets from the frame pointer.
5086
5087 2014-03-26  Marek Polacek  <polacek@redhat.com>
5088
5089         PR c/37428
5090         * doc/extend.texi (C Extensions): Mention variable-length arrays in
5091         a structure/union.
5092
5093 2014-03-26  Marek Polacek  <polacek@redhat.com>
5094
5095         PR c/39525
5096         * doc/extend.texi (Designated Inits): Describe what happens to omitted
5097         field members.
5098
5099 2014-03-26  Marek Polacek  <polacek@redhat.com>
5100
5101         PR other/59545
5102         * ira-color.c (update_conflict_hard_regno_costs): Perform the
5103         multiplication in unsigned type.
5104
5105 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5106
5107         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
5108
5109 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
5110
5111         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
5112
5113 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5114
5115         PR ipa/60315
5116         * cif-code.def (UNREACHABLE) New code.
5117         * ipa-inline.c (inline_small_functions): Skip edges to
5118         __builtlin_unreachable.
5119         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
5120         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
5121         predicate to __bulitin_unreachable.
5122         (set_cond_stmt_execution_predicate): Fix issue when
5123         invert_tree_comparison returns ERROR_MARK.
5124         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
5125         propagate to inline clones.
5126         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
5127         to unreachable.
5128         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
5129         * cgraphclones.c (cgraph_clone_node): If call destination is already
5130         ureachable, do not redirect it back.
5131         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
5132         unreachable.
5133
5134 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
5135
5136         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
5137         Do not modify inline clones.
5138
5139 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5140
5141         * config/i386/i386.md (general_sext_operand): New mode attr.
5142         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
5143         don't generate (sign_extend (const_int)).
5144         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
5145         operands[2].  Use We constraint instead of <i> and
5146         <general_sext_operand> predicate instead of <general_operand>.
5147         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
5148         * config/i386/constraints.md (We): New constraint.
5149         * config/i386/predicates.md (x86_64_sext_operand,
5150         sext_operand): New predicates.
5151
5152 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5153
5154         PR ipa/60600
5155         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
5156         inconsistent devirtualizations to __builtin_unreachable.
5157
5158 2014-03-25  Marek Polacek  <polacek@redhat.com>
5159
5160         PR c/35449
5161         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
5162
5163 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
5164
5165         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
5166         order of elements for big-endian.
5167
5168 2014-03-25  Richard Biener  <rguenther@suse.de>
5169
5170         PR middle-end/60635
5171         * gimplify-me.c (gimple_regimplify_operands): Update the
5172         re-gimplifed stmt.
5173
5174 2014-03-25  Martin Jambor  <mjambor@suse.cz>
5175
5176         PR ipa/59176
5177         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
5178         (lto_output_varpool_node): Likewise.
5179         (input_overwrite_node): Likewise.
5180         (input_varpool_node): Likewise.
5181
5182 2014-03-25  Richard Biener  <rguenther@suse.de>
5183
5184         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
5185         (run_gcc): Likewise.
5186
5187 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
5188
5189         * combine.c (simplify_compare_const): Add MODE argument.
5190         Handle mode_width 0 as very large mode_width.
5191         (try_combine, simplify_comparison): Adjust callers.
5192
5193         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
5194         type to avoid signed integer overflow.
5195         * explow.c (plus_constant): Likewise.
5196
5197 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5198
5199         * doc/generic.texi: Correct typos.
5200
5201 2014-03-24  Tobias Burnus  <burnus@net-b.de>
5202
5203         * doc/invoke.texi (-flto): Expand section about
5204         using static libraries with LTO.
5205
5206 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5207
5208         PR rtl-optimization/60501
5209         * optabs.def (addptr3_optab): New optab.
5210         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
5211         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
5212         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
5213
5214         * lra.c (emit_add3_insn): Use the addptr pattern if available.
5215
5216         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
5217
5218 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
5219
5220         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
5221         _mm512_set1_pd.
5222
5223         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
5224         (_mm256_undefined_ps): Define.
5225         (_mm256_undefined_pd): Define.
5226         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
5227         (_mm_undefined_pd): Define.
5228         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
5229         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
5230         (_mm512_undefined_ps): Define.
5231         (_mm512_undefined_pd): Define.
5232         Use _mm*_undefined_*.
5233         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
5234
5235 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
5236
5237         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
5238         (lshr_simd): DI mode added.
5239         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
5240         (aarch64_ushr_simddi): Likewise.
5241         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
5242         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
5243         (vshrd_n_u64): Likewise.
5244
5245 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5246
5247         * Makefile.in (s-macro_list): Depend on cc1.
5248
5249 2014-03-23  Teresa Johnson  <tejohnson@google.com>
5250
5251         * ipa-utils.c (ipa_print_order): Use specified dump file.
5252
5253 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
5254
5255         PR rtl-optimization/60601
5256         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
5257
5258         * gcc.c (eval_spec_function): Initialize save_growing_value.
5259
5260 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
5261
5262         PR sanitizer/60613
5263         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
5264         code == MINUS_EXPR, never swap op0 with op1.
5265
5266         * toplev.c (init_local_tick): Avoid signed integer multiplication
5267         overflow.
5268         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
5269         shift by first operand's bitsize.
5270
5271 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
5272
5273         PR target/60610
5274         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
5275         redefine to 1 or 0.
5276         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
5277         TARGET_ISA_64BIT_P(x).
5278
5279 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5280
5281         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
5282         pattern for vector nor instead of subtract from splat(-1).
5283         (altivec_expand_vec_perm_const_le): Likewise.
5284
5285 2014-03-21  Richard Henderson  <rth@twiddle.net>
5286
5287         PR target/60598
5288         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
5289         related insns after epilogue_completed.
5290
5291 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5292
5293         PR ipa/59176
5294         * cgraph.h (symtab_node): New flag body_removed.
5295         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
5296         when removing bodies.
5297         * symtab.c (dump_symtab_base): Dump body_removed flag.
5298         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
5299         had their bodies removed.
5300
5301 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5302
5303         PR ipa/60419
5304         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
5305         in the border.
5306
5307 2014-03-21  Richard Biener  <rguenther@suse.de>
5308
5309         PR tree-optimization/60577
5310         * tree-core.h (struct tree_base): Document nothrow_flag use
5311         in VAR_DECL_NONALIASED.
5312         * tree.h (VAR_DECL_NONALIASED): New.
5313         (may_be_aliased): Adjust.
5314         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
5315
5316 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5317
5318         * expr.c (expand_expr_real_1): Remove outdated comment.
5319
5320 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
5321
5322         PR middle-end/60597
5323         * ira.c (adjust_cleared_regs): Call copy_rtx on
5324         *reg_equiv[REGNO (loc)].src_p before passing it to
5325         simplify_replace_fn_rtx.
5326
5327         PR target/60568
5328         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
5329         into CONST, put pic register as first operand of PLUS.  Use
5330         gen_const_mem for both 32-bit and 64-bit PIC got loads.
5331
5332 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5333
5334         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
5335
5336 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5337
5338         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
5339         around for store forwarding issue in the FPU on the UT699.
5340         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
5341         loads and operations if -mfix-ut699 is specified.
5342         (divtf3_hq): Tweak attribute.
5343         (sqrttf2_hq): Likewise.
5344
5345 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5346
5347         * calls.c (store_one_arg): Remove incorrect const qualification on the
5348         type of the temporary.
5349         * cfgexpand.c (expand_return): Likewise.
5350         * expr.c (expand_constructor): Likewise.
5351         (expand_expr_real_1): Likewise.
5352
5353 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5354
5355         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
5356         of parts.
5357
5358 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
5359
5360         PR target/60039
5361         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
5362
5363 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
5364
5365         * config/arm/aarch-common-protos.h
5366         (alu_cost_table): Fix spelling of "extend".
5367         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
5368
5369 2014-03-19  Richard Biener  <rguenther@suse.de>
5370
5371         PR middle-end/60553
5372         * tree-core.h (tree_type_common): Re-order pointer members
5373         to reduce recursion depth during GC walks.
5374
5375 2014-03-19  Marek Polacek  <polacek@redhat.com>
5376
5377         PR sanitizer/60569
5378         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
5379         before accessing it.
5380
5381 2014-03-19  Richard Biener  <rguenther@suse.de>
5382
5383         PR lto/59543
5384         * lto-streamer-in.c (input_function): In WPA stage do not drop
5385         debug stmts.
5386
5387 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
5388
5389         PR tree-optimization/60559
5390         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
5391         with build_zero_cst assignment.
5392
5393 2014-03-18  Kai Tietz  <ktietz@redhat.com>
5394
5395         PR rtl-optimization/56356
5396         * sdbout.c (sdbout_parms): Verify that parms'
5397         incoming argument is valid.
5398         (sdbout_reg_parms): Likewise.
5399
5400 2014-03-18  Richard Henderson  <rth@redhat.com>
5401
5402         PR target/60562
5403         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
5404         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
5405         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
5406
5407 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
5408
5409         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
5410         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
5411         Italicize plugin event names in description.  Explain that
5412         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
5413         Remind that no GCC functions should be called after PLUGIN_FINISH.
5414         Explain what pragmas with expansion are.
5415
5416 2014-03-18  Martin Liska  <mliska@suse.cz>
5417
5418         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
5419         gimple call statement is update.
5420         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
5421         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
5422
5423 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5424
5425         PR sanitizer/60557
5426         * ubsan.c (ubsan_instrument_unreachable): Call
5427         initialize_sanitizer_builtins.
5428         (ubsan_pass): Likewise.
5429
5430         PR sanitizer/60535
5431         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
5432         varpool_finalize_decl instead of rest_of_decl_compilation.
5433
5434 2014-03-18  Richard Biener  <rguenther@suse.de>
5435
5436         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
5437         by using bitmap_and_compl instead of bitmap_and_compl_into.
5438         (df_rd_transfer_function): Likewise.
5439
5440 2014-03-18  Richard Biener  <rguenther@suse.de>
5441
5442         * doc/lto.texi (fresolution): Fix typo.
5443
5444 2014-03-18  Richard Biener  <rguenther@suse.de>
5445
5446         * doc/invoke.texi (flto): Update for changes in 4.9.
5447
5448 2014-03-18  Richard Biener  <rguenther@suse.de>
5449
5450         * doc/loop.texi: Remove section on the removed lambda framework.
5451         Update loop docs with recent changes in preserving loop structure.
5452
5453 2014-03-18  Richard Biener  <rguenther@suse.de>
5454
5455         * doc/lto.texi (-fresolution): Document.
5456
5457 2014-03-18  Richard Biener  <rguenther@suse.de>
5458
5459         * doc/contrib.texi: Adjust my name.
5460
5461 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5462
5463         PR ipa/58721
5464         * internal-fn.c: Include diagnostic-core.h.
5465         (expand_BUILTIN_EXPECT): New function.
5466         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
5467         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
5468         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
5469         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
5470         IFN_BUILTIN_EXPECT.
5471         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
5472         Revert 3 argument __builtin_expect code.
5473         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
5474         * gimple-fold.c (gimple_fold_call): Likewise.
5475         * tree.h (fold_builtin_expect): New prototype.
5476         * builtins.c (build_builtin_expect_predicate): Add predictor
5477         argument, if non-NULL, create 3 argument __builtin_expect.
5478         (fold_builtin_expect): No longer static.  Add ARG2 argument,
5479         pass it through to build_builtin_expect_predicate.
5480         (fold_builtin_2): Adjust caller.
5481         (fold_builtin_3): Handle BUILT_IN_EXPECT.
5482         * internal-fn.def (BUILTIN_EXPECT): New.
5483
5484 2014-03-18  Tobias Burnus  <burnus@net-b.de>
5485
5486         PR ipa/58721
5487         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
5488         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
5489         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
5490
5491 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
5492
5493         PR ipa/58721
5494         * predict.c (combine_predictions_for_bb): Fix up formatting.
5495         (expr_expected_value_1, expr_expected_value): Add predictor argument,
5496         fill what it points to if non-NULL.
5497         (tree_predict_by_opcode): Adjust caller, use the predictor.
5498         * predict.def (PRED_COMPARE_AND_SWAP): Add.
5499
5500 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
5501
5502         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
5503         proper constant for the store mode.
5504
5505 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
5506
5507         * symtab.c (change_decl_assembler_name): Fix transparent alias
5508         chain construction.
5509
5510 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
5511
5512         * config/aarch64/aarch64.c: Correct the comments about the
5513         aarch64 stack layout.
5514
5515 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
5516
5517         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
5518         check for GF_OMP_FOR_KIND_FOR.
5519
5520 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
5521
5522         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
5523         ymm and zmm register names.
5524
5525 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
5526
5527         PR target/60516
5528         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
5529         note creation for the 2010-08-31 changes.
5530
5531 2014-03-17  Marek Polacek  <polacek@redhat.com>
5532
5533         PR middle-end/60534
5534         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
5535         as -fno-tree-loop-vectorize.
5536         (expand_omp_simd): Likewise.
5537
5538 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
5539
5540         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
5541         (eligible_for_call_delay): New prototype.
5542         * config/sparc/sparc.c (tls_call_delay): Rename into...
5543         (eligible_for_call_delay): ...this.  Return false if the instruction
5544         cannot be put in the delay slot of a branch.
5545         (eligible_for_restore_insn): Simplify.
5546         (eligible_for_return_delay): Return false if the instruction cannot be
5547         put in the delay slot of a branch and simplify.
5548         (eligible_for_sibcall_delay): Return false if the instruction cannot be
5549         put in the delay slot of a branch.
5550         * config/sparc/sparc.md (fix_ut699): New attribute.
5551         (tls_call_delay): Delete.
5552         (in_call_delay): Reimplement.
5553         (eligible_for_sibcall_delay): Rename into...
5554         (in_sibcall_delay): ...this.
5555         (eligible_for_return_delay): Rename into...
5556         (in_return_delay): ...this.
5557         (in_branch_delay): Reimplement.
5558         (in_uncond_branch_delay): Delete.
5559         (in_annul_branch_delay): Delete.
5560
5561 2014-03-14  Richard Henderson  <rth@redhat.com>
5562
5563         PR target/60525
5564         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
5565         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
5566         (*floathi<X87MODEF>2_i387_with_temp): Remove.
5567         (floathi splitters): Remove.
5568         (float<SWI48x>xf2): New pattern.
5569         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
5570         code that tried to handle DImode for 32-bit, but which was excluded
5571         by the pattern's condition.  Drop allocation of stack temporary.
5572         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
5573         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
5574         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
5575         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
5576         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
5577         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
5578         (*float<SWI48><MODEF>2_sse_interunit): Remove.
5579         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
5580         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
5581         (*float<SWI48x><X87MODEF>2_i387): Remove.
5582         (all float _with_temp splitters): Remove.
5583         (*float<SWI48x><MODEF>2_i387): New pattern.
5584         (*float<SWI48><MODEF>2_sse): New pattern.
5585         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
5586         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
5587
5588 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
5589             Marek Polacek  <polacek@redhat.com>
5590
5591         PR middle-end/60484
5592         * common.opt (dump_base_name_prefixed): New Variable.
5593         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
5594         if x_dump_base_name_prefixed is already set, set it at the end.
5595
5596 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
5597
5598         PR rtl-optimization/60508
5599         * lra-constraints.c (get_reload_reg): Add new parameter
5600         in_subreg_p.
5601         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
5602         Pass the new parameter values.
5603
5604 2014-03-14  Richard Biener  <rguenther@suse.de>
5605
5606         * common.opt: Revert unintented changes from r205065.
5607         * opts.c: Likewise.
5608
5609 2014-03-14  Richard Biener  <rguenther@suse.de>
5610
5611         PR middle-end/60518
5612         * cfghooks.c (split_block): Properly adjust all loops the
5613         block was a latch of.
5614
5615 2014-03-14  Martin Jambor  <mjambor@suse.cz>
5616
5617         PR lto/60461
5618         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
5619         and simplify it.
5620
5621 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
5622
5623         PR target/59396
5624         * config/avr/avr.c (avr_set_current_function): Pass function name
5625         through default_strip_name_encoding before sanity checking instead
5626         of skipping the first char of the assembler name.
5627
5628 2014-03-13  Richard Henderson  <rth@redhat.com>
5629
5630         PR debug/60438
5631         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
5632         (ix86_force_to_memory, ix86_free_from_memory): Remove.
5633         * config/i386/i386-protos.h: Likewise.
5634         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
5635         in the expander instead of a splitter.
5636         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
5637         any possibility of requiring a memory.
5638         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
5639         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
5640         (fp branch splitters): Update for ix86_split_fp_branch.
5641         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
5642         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
5643         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
5644         (*fop_<MODEF>_2_i387): Remove f/r alternative.
5645         (*fop_<MODEF>_3_i387): Likewise.
5646         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
5647         (splitters for the fop_* register patterns): Remove.
5648         (fscalexf4_i387): Rename from *fscalexf4_i387.
5649         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
5650
5651 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
5652
5653         PR tree-optimization/59779
5654         * tree-dfa.c (get_ref_base_and_extent): Use double_int
5655         type for bitsize and maxsize instead of HOST_WIDE_INT.
5656
5657 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
5658
5659         PR rtl-optimization/57320
5660         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
5661         the CFG after thread_prologue_and_epilogue_insns.
5662
5663 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
5664
5665         PR rtl-optimization/57189
5666         * lra-constraints.c (process_alt_operands): Disfavor spilling
5667         vector pseudos.
5668
5669 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
5670
5671         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
5672
5673 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
5674
5675         PR tree-optimization/59025
5676         PR middle-end/60418
5677         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
5678         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
5679
5680 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
5681
5682         PR target/60486
5683         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
5684         calls of avr_out_plus_1.
5685
5686 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
5687
5688         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
5689         BB's single pred and update the father loop's latch info later.
5690
5691 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5692
5693         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
5694         (VEC_M): Likewise.
5695         (VEC_N): Likewise.
5696         (VEC_R): Likewise.
5697         (VEC_base): Likewise.
5698         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
5699         registers, we need to swap double words in little endian mode.
5700
5701         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
5702         to be a container mode for 128-bit integer operations added in ISA
5703         2.07.  Unlike TImode and PTImode, the preferred register set is
5704         the Altivec/VMX registers for the 128-bit operations.
5705
5706         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
5707         declarations.
5708         (rs6000_split_128bit_ok_p): Likewise.
5709
5710         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
5711         macros for creating ISA 2.07 normal and overloaded builtin
5712         functions with 3 arguments.
5713         (BU_P8V_OVERLOAD_3): Likewise.
5714         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
5715         for use as overloaded functions.
5716         (VPERM_1TI_UNS): Likewise.
5717         (VSEL_1TI): Likewise.
5718         (VSEL_1TI_UNS): Likewise.
5719         (ST_INTERNAL_1ti): Likewise.
5720         (LD_INTERNAL_1ti): Likewise.
5721         (XXSEL_1TI): Likewise.
5722         (XXSEL_1TI_UNS): Likewise.
5723         (VPERM_1TI): Likewise.
5724         (VPERM_1TI_UNS): Likewise.
5725         (XXPERMDI_1TI): Likewise.
5726         (SET_1TI): Likewise.
5727         (LXVD2X_V1TI): Likewise.
5728         (STXVD2X_V1TI): Likewise.
5729         (VEC_INIT_V1TI): Likewise.
5730         (VEC_SET_V1TI): Likewise.
5731         (VEC_EXT_V1TI): Likewise.
5732         (EQV_V1TI): Likewise.
5733         (NAND_V1TI): Likewise.
5734         (ORC_V1TI): Likewise.
5735         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
5736         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
5737         overloaded builtin.
5738         (VADDUQM): Likewise.
5739         (VSUBCUQ): Likewise.
5740         (VADDEUQM): Likewise.
5741         (VADDECUQ): Likewise.
5742         (VSUBEUQM): Likewise.
5743         (VSUBECUQ): Likewise.
5744
5745         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
5746         __int128_t and __uint128_t types.
5747         (__uint128_type): Likewise.
5748         (altivec_categorize_keyword): Add support for vector __int128_t,
5749         vector __uint128_t, vector __int128, and vector unsigned __int128
5750         as a container type for TImode operations that need to be done in
5751         VSX/Altivec registers.
5752         (rs6000_macro_to_expand): Likewise.
5753         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
5754         to support 128-bit integer instructions vaddcuq, vadduqm,
5755         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
5756         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
5757
5758         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
5759         for V1TImode, and set up preferences to use VSX/Altivec registers.
5760         Setup VSX reload handlers.
5761         (rs6000_debug_reg_global): Likewise.
5762         (rs6000_init_hard_regno_mode_ok): Likewise.
5763         (rs6000_preferred_simd_mode): Likewise.
5764         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
5765         (easy_altivec_constant): Likewise.
5766         (output_vec_const_move): Likewise.
5767         (rs6000_expand_vector_set): Convert V1TImode set and extract to
5768         simple move.
5769         (rs6000_expand_vector_extract): Likewise.
5770         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
5771         addressing.
5772         (rs6000_const_vec): Add support for V1TImode.
5773         (rs6000_emit_le_vsx_load): Swap double words when loading or
5774         storing TImode/V1TImode.
5775         (rs6000_emit_le_vsx_store): Likewise.
5776         (rs6000_emit_le_vsx_move): Likewise.
5777         (rs6000_emit_move): Add support for V1TImode.
5778         (altivec_expand_ld_builtin): Likewise.
5779         (altivec_expand_st_builtin): Likewise.
5780         (altivec_expand_vec_init_builtin): Likewise.
5781         (altivec_expand_builtin): Likewise.
5782         (rs6000_init_builtins): Add support for V1TImode type.  Add
5783         support for ISA 2.07 128-bit integer builtins.  Define type names
5784         for the VSX/Altivec vector types.
5785         (altivec_init_builtins): Add support for overloaded vector
5786         functions with V1TImode type.
5787         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
5788         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
5789         external function.
5790         (rs6000_split_128bit_ok_p): Likewise.
5791         (rs6000_handle_altivec_attribute): Create V1TImode from vector
5792         __int128_t and vector __uint128_t.
5793
5794         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
5795         and mode attributes.
5796         (VSX_M): Likewise.
5797         (VSX_M2): Likewise.
5798         (VSm): Likewise.
5799         (VSs): Likewise.
5800         (VSr): Likewise.
5801         (VSv): Likewise.
5802         (VS_scalar): Likewise.
5803         (VS_double): Likewise.
5804         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
5805
5806         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
5807         we support the ISA 2.07 128-bit integer arithmetic instructions.
5808         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
5809         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
5810         and TImode types for use with the builtin functions.
5811         (V1TI_type_node): Likewise.
5812         (unsigned_V1TI_type_node): Likewise.
5813         (intTI_type_internal_node): Likewise.
5814         (uintTI_type_internal_node): Likewise.
5815
5816         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
5817         128-bit builtin functions.
5818         (UNSPEC_VADDEUQM): Likewise.
5819         (UNSPEC_VADDECUQ): Likewise.
5820         (UNSPEC_VSUBCUQ): Likewise.
5821         (UNSPEC_VSUBEUQM): Likewise.
5822         (UNSPEC_VSUBECUQ): Likewise.
5823         (VM): Add V1TImode to vector mode iterators.
5824         (VM2): Likewise.
5825         (VI_unit): Likewise.
5826         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
5827         (altivec_vaddcuq): Likewise.
5828         (altivec_vsubuqm): Likewise.
5829         (altivec_vsubcuq): Likewise.
5830         (altivec_vaddeuqm): Likewise.
5831         (altivec_vaddecuq): Likewise.
5832         (altivec_vsubeuqm): Likewise.
5833         (altivec_vsubecuq): Likewise.
5834
5835         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
5836         mode iterators.
5837         (BOOL_128): Likewise.
5838         (BOOL_REGS_OUTPUT): Likewise.
5839         (BOOL_REGS_OP1): Likewise.
5840         (BOOL_REGS_OP2): Likewise.
5841         (BOOL_REGS_UNARY): Likewise.
5842         (BOOL_REGS_AND_CR0): Likewise.
5843
5844         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
5845         128-bit integer builtin support.
5846         (vec_vadduqm): Likewise.
5847         (vec_vaddecuq): Likewise.
5848         (vec_vaddeuqm): Likewise.
5849         (vec_vsubecuq): Likewise.
5850         (vec_vsubeuqm): Likewise.
5851         (vec_vsubcuq): Likewise.
5852         (vec_vsubuqm): Likewise.
5853
5854         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5855         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
5856         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
5857         128-bit integer add/subtract to ISA 2.07.
5858
5859 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
5860
5861         * config/arc/arc.c (arc_predicate_delay_insns):
5862         Fix third argument passed to conditionalize_nonjump.
5863
5864 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
5865
5866         * config/aarch64/aarch64-builtins.c
5867         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
5868         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
5869         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
5870         instead of __builtin_lfloor.
5871         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
5872
5873 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5874
5875         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
5876         (tree_ssa_ifcombine_bb_1): New function.
5877         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
5878         is an empty forwarder block to then_bb or vice versa and then_bb
5879         and else_bb are effectively swapped.
5880
5881 2014-03-12  Christian Bruel  <christian.bruel@st.com>
5882
5883         PR target/60264
5884         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
5885         REG_CFA_DEF_CFA note.
5886         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
5887         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
5888
5889 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
5890
5891         PR tree-optimization/60454
5892         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
5893
5894 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5895
5896         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
5897         Do not define target_cpu_default2 to generic.
5898         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
5899         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
5900         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
5901
5902 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5903             Marc Glisse  <marc.glisse@inria.fr>
5904
5905         PR tree-optimization/60502
5906         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
5907         instead of build_low_bits_mask.
5908
5909 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5910
5911         PR middle-end/60482
5912         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
5913         if there are multiple uses, but op doesn't live on E edge.
5914         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
5915         clobber stmts before __builtin_unreachable.
5916
5917 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
5918
5919         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
5920         hard_frame_pointer_rtx.
5921         * cse.c (cse_insn): Remove volatile check.
5922         * cselib.c (cselib_process_insn): Likewise.
5923         * dse.c (scan_insn): Likewise.
5924
5925 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5926
5927         * config/arc/arc.c (conditionalize_nonjump): New function,
5928         broken out of ...
5929         (arc_ifcvt): ... this.
5930         (arc_predicate_delay_insns): Use it.
5931
5932 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5933
5934         * config/arc/predicates.md (extend_operand): During/after reload,
5935         allow const_int_operand.
5936         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
5937         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
5938         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
5939         to "i".
5940         (umulsi3_highpart_i): Likewise.
5941
5942 2014-03-11  Richard Biener  <rguenther@suse.de>
5943
5944         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
5945         Add asserts to guard possible wrong-code bugs.
5946
5947 2014-03-11  Richard Biener  <rguenther@suse.de>
5948
5949         PR tree-optimization/60429
5950         PR tree-optimization/60485
5951         * tree-ssa-structalias.c (set_union_with_increment): Properly
5952         take into account all fields that overlap the shifted vars.
5953         (do_sd_constraint): Likewise.
5954         (do_ds_constraint): Likewise.
5955         (get_constraint_for_ptr_offset): Likewise.
5956
5957 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
5958
5959         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
5960         (nios2_compute_frame_layout):
5961         Add calculation of cfun->machine->fp_save_offset.
5962         (nios2_expand_prologue): Correct setting of frame pointer register
5963         in prologue.
5964         (nios2_expand_epilogue): Update recovery of stack pointer from
5965         frame pointer accordingly.
5966         (nios2_initial_elimination_offset): Update calculation of offset
5967         for eliminating to HARD_FRAME_POINTER_REGNUM.
5968
5969 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
5970
5971         PR ipa/60457
5972         * ipa.c (symtab_remove_unreachable_nodes): Don't call
5973         cgraph_get_create_node on VAR_DECLs.
5974
5975 2014-03-10  Richard Biener  <rguenther@suse.de>
5976
5977         PR middle-end/60474
5978         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
5979
5980 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
5981
5982         * config/vms/vms.opt (vms_float_format): New variable.
5983
5984 2014-03-08  Tobias Burnus  <burnus@net-b.de>
5985
5986         * doc/invoke.texi (-fcilkplus): Update implementation status.
5987
5988 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
5989             Richard Biener  <rguenther@suse.de>
5990
5991         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
5992         consistently accross all TUs.
5993         (run_gcc): Enable -fshort-double automatically at link at link-time
5994         and disallow override.
5995
5996 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
5997
5998         PR target/58271
5999         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
6000         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
6001         if they can't be used.
6002
6003 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6004
6005         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
6006         for Solaris 11/x86 ld.
6007         * configure: Regenerate.
6008
6009 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6010
6011         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
6012         (LIB_TLS_SPEC): Save as ld_tls_libs.
6013         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
6014         (HAVE_AS_IX86_TLSLDM): New test.
6015         * configure, config.in: Regenerate.
6016         * config/i386/i386.c (legitimize_tls_address): Fall back to
6017         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
6018         cannot support TLS_MODEL_LOCAL_DYNAMIC.
6019         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
6020         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
6021
6022 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
6023
6024         * common.opt (fira-loop-pressure): Mark as optimization.
6025
6026 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
6027
6028         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
6029         an OpenMP mappable type.
6030
6031 2014-03-06  Matthias Klose  <doko@ubuntu.com>
6032
6033         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
6034         MULTILIB_OSDIRNAMES is not defined.
6035
6036 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
6037             Meador Inge  <meadori@codesourcery.com>
6038
6039         PR target/58595
6040         * config/arm/arm.c (arm_tls_symbol_p): Remove.
6041         (arm_legitimize_address): Call legitimize_tls_address for any
6042         arm_tls_referenced_p expression, handle constant addend.  Call it
6043         before testing for !TARGET_ARM.
6044         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
6045
6046 2014-03-06  Richard Biener  <rguenther@suse.de>
6047
6048         PR middle-end/60445
6049         PR lto/60424
6050         PR lto/60427
6051         Revert
6052         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6053
6054         * tree-streamer.c (record_common_node): Assert we don't record
6055         nodes with type double.
6056         (preload_common_node): Skip type double, complex double and double
6057         pointer since it is now frontend dependent due to fshort-double option.
6058
6059 2014-03-06  Richard Biener  <rguenther@suse.de>
6060
6061         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
6062         or -fno-lto is specified and the linker has full plugin support.
6063         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
6064         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
6065         * lto-wrapper.c (merge_and_complain): Merge compile-time
6066         optimization levels.
6067         (run_gcc): And pass it through to the link options.
6068
6069 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
6070
6071         PR debug/60381
6072         Revert:
6073         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6074         PR debug/59992
6075         * cselib.c (remove_useless_values): Skip to avoid quadratic
6076         behavior if the condition moved from...
6077         (cselib_process_insn): ... here holds.
6078
6079 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6080
6081         PR plugins/59335
6082         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
6083         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
6084
6085         PR plugins/59335
6086         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
6087         (TM_H): Add x86-tune.def.
6088
6089 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6090
6091         * config/aarch64/aarch64.c (generic_tunings):
6092         Use cortexa57_extra_costs.
6093
6094 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
6095
6096         PR lto/60404
6097         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
6098         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
6099         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
6100         cost for in_lto_p.
6101
6102 2014-03-04  Heiher  <r@hev.cc>
6103
6104         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
6105         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
6106
6107 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
6108
6109         * config/i386/predicates.md (const2356_operand): Change to ...
6110         (const2367_operand): ... this.
6111         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
6112         const2367_operand.
6113         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6114         (*avx512pf_scatterpf<mode>sf): Ditto.
6115         (avx512pf_scatterpf<mode>df): Ditto.
6116         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6117         (*avx512pf_scatterpf<mode>df): Ditto.
6118         * config/i386/i386.c (ix86_expand_builtin): Update
6119         incorrect hint operand error message.
6120
6121 2014-03-04  Richard Biener  <rguenther@suse.de>
6122
6123         * lto-section-in.c (lto_get_section_data): Fix const cast.
6124
6125 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
6126
6127         * tree-streamer.c (record_common_node): Assert we don't record
6128         nodes with type double.
6129         (preload_common_node): Skip type double, complex double and double
6130         pointer since it is now frontend dependent due to fshort-double option.
6131
6132 2014-03-04  Richard Biener  <rguenther@suse.de>
6133
6134         PR lto/60405
6135         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
6136         (lto_input_toplevel_asms): Likewise.
6137         * lto-section-in.c (lto_get_section_data): Instead do it here
6138         for every section.
6139
6140 2014-03-04  Richard Biener  <rguenther@suse.de>
6141
6142         PR tree-optimization/60382
6143         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
6144         dead PHIs a reduction.
6145
6146 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
6147
6148         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
6149         hint value.
6150         (_mm_prefetch): Move out of GCC target("sse") pragma.
6151         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
6152         GCC target("prfchw") pragma.
6153         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
6154         for locality <= 2.
6155         * config/i386/i386.c (ix86_option_override_internal): Enable
6156         -mprfchw with -mprefetchwt1.
6157
6158 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6159
6160         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
6161         Mark as varying.
6162
6163 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
6164
6165         * opts.h (CL_PCH_IGNORE): Define.
6166         * targhooks.c (option_affects_pch_p):
6167         Return false for options that have CL_PCH_IGNORE set.
6168         * opt-functions.awk: Process PchIgnore.
6169         * doc/options.texi: Document PchIgnore.
6170
6171         * config/arc/arc.opt (misize): Add PchIgnore property.
6172
6173 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6174
6175         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
6176         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
6177         constraint on constants to permit them being loaded into
6178         GENERAL_REGS or BASE_REGS.
6179
6180 2014-03-03  Nick Clifton  <nickc@redhat.com>
6181
6182         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
6183         anti-cacnonical alternatives.
6184         (negandhi3_real): New pattern.
6185         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
6186
6187 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
6188
6189         * config/avr/avr-mcus.def: Remove atxmega16x1.
6190         * config/avr/avr-tables.opt: Regenerate.
6191         * config/avr/t-multilib: Regenerate.
6192         * doc/avr-mmcu.texi: Regenerate.
6193
6194 2014-03-03  Tobias Grosser  <tobias@grosser.es>
6195             Mircea Namolaru  <mircea.namolaru@inria.fr>
6196
6197         PR tree-optimization/58028
6198         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
6199         scalar dimensions.
6200
6201 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6202
6203         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
6204         not handled by recognizers.
6205
6206 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
6207
6208         PR middle-end/60175
6209         * function.c (expand_function_end): Don't emit
6210         clobber_return_register sequence if clobber_after is a BARRIER.
6211         * cfgexpand.c (construct_exit_block): Append instructions before
6212         return_label to prev_bb.
6213
6214 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6215
6216         * config/rs6000/constraints.md: Document reserved use of "wc".
6217
6218 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6219
6220         PR ipa/60150
6221         * ipa.c (function_and_variable_visibility): When dissolving comdat
6222         group, also set all symbols to local.
6223
6224 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6225
6226         PR ipa/60306
6227
6228         Revert:
6229         2013-12-14   Jan Hubicka  <jh@suse.cz>
6230         PR middle-end/58477
6231         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
6232
6233 2014-03-02  Jon Beniston  <jon@beniston.com>
6234
6235         PR bootstrap/48230
6236         PR bootstrap/50927
6237         PR bootstrap/52466
6238         PR target/46898
6239         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
6240           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
6241         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
6242         (simple_return, *simple_return): New patterns
6243         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
6244         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
6245
6246 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
6247
6248         * dwarf2out.c (gen_subprogram_die): Tidy.
6249
6250 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
6251
6252         PR target/60071
6253         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
6254         (*mov_t_msb_neg_negc): ... this new insn.
6255
6256 2014-02-28  Jason Merrill  <jason@redhat.com>
6257
6258         PR c++/58678
6259         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
6260         function.
6261
6262 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
6263
6264         PR c++/60314
6265         * dwarf2out.c (decltype_auto_die): New static.
6266         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
6267         (gen_type_die_with_usage): Handle 'decltype(auto)'.
6268         (is_cxx_auto): Likewise.
6269
6270 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
6271
6272         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
6273         we are not using general regs only.
6274
6275 2014-02-28  Richard Biener  <rguenther@suse.de>
6276
6277         PR target/60280
6278         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
6279         previous fix and only allow to remove trivial pre-headers
6280         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
6281         (remove_forwarder_block): Properly update the latch of a loop.
6282
6283 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6284
6285         PR debug/59992
6286         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
6287         (cselib_preserved_hash_table): New.
6288         (preserve_constants_and_equivs): Move preserved vals to it.
6289         (cselib_find_slot): Look it up first.
6290         (cselib_init): Initialize it.
6291         (cselib_finish): Release it.
6292         (dump_cselib_table): Dump it.
6293
6294 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6295
6296         PR debug/59992
6297         * cselib.c (remove_useless_values): Skip to avoid quadratic
6298         behavior if the condition moved from...
6299         (cselib_process_insn): ... here holds.
6300
6301 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6302
6303         PR debug/57232
6304         * var-tracking.c (vt_initialize): Apply the same condition to
6305         preserve the CFA base value.
6306
6307 2014-02-28  Joey Ye  <joey.ye@arm.com>
6308
6309         PR target/PR60169
6310         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
6311         if reload in progress or completed.
6312
6313 2014-02-28  Tobias Burnus  <burnus@net-b.de>
6314
6315         PR middle-end/60147
6316         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
6317         NAMELIST_DECL.
6318
6319 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
6320
6321         * doc/tm.texi.in (Condition Code Status): Update documention for
6322         relative locations of cc0-setter and cc0-user.
6323
6324 2014-02-27  Jeff Law  <law@redhat.com>
6325
6326         PR rtl-optimization/52714
6327         * combine.c (try_combine): When splitting an unrecognized PARALLEL
6328         into two independent simple sets, if I3 is a jump, ensure the
6329         pattern we place into I3 is a (set (pc) ...).
6330
6331 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
6332             Jeff Law  <law@redhat.com>
6333
6334         PR rtl-optimization/49847
6335         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
6336         are in different blocks.
6337         * doc/tm.texi (Condition Code Status): Update documention for
6338         relative locations of cc0-setter and cc0-user.
6339
6340 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
6341
6342         PR target/59222
6343         * lra.c (lra_emit_add): Check SUBREG too.
6344
6345 2014-02-27  Andreas Schwab  <schwab@suse.de>
6346
6347         * config/m68k/m68k.c (m68k_option_override): Disable
6348         -flive-range-shrinkage for classic m68k.
6349         (m68k_override_options_after_change): Likewise.
6350
6351 2014-02-27  Marek Polacek  <polacek@redhat.com>
6352
6353         PR middle-end/59223
6354         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
6355         -Wmaybe-uninitialized.
6356
6357 2014-02-27  Alan Modra  <amodra@gmail.com>
6358
6359         PR target/57936
6360         * reload1.c (emit_input_reload_insns): When reload_override_in,
6361         set old to rl->in_reg when rl->in_reg is a subreg.
6362
6363 2014-02-26  Richard Biener  <rguenther@suse.de>
6364
6365         PR bootstrap/60343
6366         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
6367
6368 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6369
6370         * common/config/i386/predicates.md (const1256_operand): Remove.
6371         (const2356_operand): New.
6372         (const_1_to_2_operand): Remove.
6373         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
6374         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6375         (*avx512pf_gatherpf<mode>sf): Ditto.
6376         (avx512pf_gatherpf<mode>df): Ditto.
6377         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6378         (*avx512pf_gatherpf<mode>df): Ditto.
6379         (avx512pf_scatterpf<mode>sf): Ditto.
6380         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6381         (*avx512pf_scatterpf<mode>sf): Ditto.
6382         (avx512pf_scatterpf<mode>df): Ditto.
6383         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6384         (*avx512pf_scatterpf<mode>df): Ditto.
6385         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
6386
6387 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
6388
6389         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
6390         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
6391         (_mm512_mask_testn_epi64_mask): Move to ...
6392         * config/i386/avx512cdintrin.h: Here.
6393         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
6394         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
6395         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
6396         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
6397         TARGET_AVX512F from TARGET_AVX512CD.
6398
6399 2014-02-26  Richard Biener  <rguenther@suse.de>
6400
6401         PR ipa/60327
6402         * ipa.c (walk_polymorphic_call_targets): Properly guard
6403         call to inline_update_overall_summary.
6404
6405 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
6406
6407         PR target/60280
6408         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
6409         and latches only if requested.  Fix latch if it is removed.
6410         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
6411         LOOPS_HAVE_PREHEADERS.
6412
6413 2014-02-25  Andrew Pinski  <apinski@cavium.com>
6414
6415         * builtins.c (expand_builtin_thread_pointer): Create a new target
6416         when the target is NULL.
6417
6418 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
6419
6420         PR rtl-optimization/60317
6421         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6422         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6423         * lra-assigns.c: Include params.h.
6424         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
6425         other reload pseudos considerations.
6426
6427 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6428
6429         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
6430         to use canonical form for nor<mode>3.
6431
6432 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6433
6434         PR target/55426
6435         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
6436         conversions.
6437
6438 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6439
6440         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
6441         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
6442         (ix86_handle_option): Handle OPT_mprefetchwt1.
6443         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
6444         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
6445         PREFETCHWT1 CPUID.
6446         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6447         OPTION_MASK_ISA_PREFETCHWT1.
6448         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
6449         (PTA_PREFETCHWT1): New.
6450         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
6451         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
6452         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
6453         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
6454         (*prefetch_avx512pf_<mode>_: Change into ...
6455         (*prefetch_prefetchwt1_<mode>: This.
6456         * config/i386/i386.opt (mprefetchwt1): New.
6457         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
6458         (_mm_prefetch): Handle intent to write.
6459         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
6460
6461 2014-02-25  Richard Biener  <rguenther@suse.de>
6462
6463         PR middle-end/60291
6464         * emit-rtl.c (mem_attrs_htab): Remove.
6465         (mem_attrs_htab_hash): Likewise.
6466         (mem_attrs_htab_eq): Likewise.
6467         (set_mem_attrs): Always allocate new mem-attrs when something changed.
6468         (init_emit_once): Do not allocate mem_attrs_htab.
6469
6470 2014-02-25  Richard Biener  <rguenther@suse.de>
6471
6472         PR lto/60319
6473         * lto-opts.c (lto_write_options): Output non-explicit conservative
6474         -fwrapv, -fno-trapv and -fno-strict-overflow.
6475         * lto-wrapper.c (merge_and_complain): Handle merging those options.
6476         (run_gcc): And pass them through.
6477
6478 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6479
6480         * sel-sched.c (calculate_new_fences): New parameter ptime.
6481         Calculate it as a maximum over all fence cycles.
6482         (sel_sched_region_2): Adjust the call to calculate_new_fences.
6483         Print the final schedule timing when sched_verbose.
6484
6485 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6486
6487         PR rtl-optimization/60292
6488         * sel-sched.c (fill_vec_av_set): Do not reset target availability
6489         bit fot the fence instruction.
6490
6491 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
6492
6493         * calls.h: Fix typo in comment.
6494
6495 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
6496
6497         * config/pa/pa.c (pa_output_move_double): Don't valididate when
6498         adjusting offsetable addresses.
6499
6500 2014-02-24  Guozhi Wei  <carrot@google.com>
6501
6502         * sparseset.h (sparseset_pop): Fix the wrong index.
6503
6504 2014-02-24  Walter Lee  <walt@tilera.com>
6505
6506         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
6507         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
6508         triplet.
6509         * common/config/tilegx/tilegx-common.c
6510         (TARGET_DEFAULT_TARGET_FLAGS): Define.
6511         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
6512         (LINK_SPEC): Ditto.
6513         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
6514         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
6515         (tilegx_gimplify_va_arg_expr): Handle big endian.
6516         (tilegx_expand_unaligned_load): Ditto.
6517         (tilegx_expand_unaligned_store): Ditto.
6518         (TARGET_RETURN_IN_MSB): New.
6519         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
6520         (TARGET_ENDIAN_DEFAULT): New.
6521         (TARGET_BIG_ENDIAN): Handle big endian.
6522         (BYTES_BIG_ENDIAN): Ditto.
6523         (WORDS_BIG_ENDIAN): Ditto.
6524         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
6525         (ENDIAN_SPEC): New.
6526         (EXTRA_SPECS): New.
6527         * config/tilegx/tilegx.md (extv): Handle big endian.
6528         (extzv): Ditto.
6529         (insn_st<n>): Ditto.
6530         (insn_st<n>_add<bitsuffix>): Ditto.
6531         (insn_stnt<n>): Ditto.
6532         (insn_stnt<n>_add<bitsuffix>):Ditto.
6533         (vec_interleave_highv8qi): Handle big endian.
6534         (vec_interleave_highv8qi_be): New.
6535         (vec_interleave_highv8qi_le): New.
6536         (insn_v1int_h): Handle big endian.
6537         (vec_interleave_lowv8qi): Handle big endian.
6538         (vec_interleave_lowv8qi_be): New.
6539         (vec_interleave_lowv8qi_le): New.
6540         (insn_v1int_l): Handle big endian.
6541         (vec_interleave_highv4hi): Handle big endian.
6542         (vec_interleave_highv4hi_be): New.
6543         (vec_interleave_highv4hi_le): New.
6544         (insn_v2int_h): Handle big endian.
6545         (vec_interleave_lowv4hi): Handle big endian.
6546         (vec_interleave_lowv4hi_be): New.
6547         (vec_interleave_lowv4hi_le): New.
6548         (insn_v2int_l): Handle big endian.
6549         (vec_interleave_highv2si): Handle big endian.
6550         (vec_interleave_highv2si_be): New.
6551         (vec_interleave_highv2si_le): New.
6552         (insn_v4int_h): Handle big endian.
6553         (vec_interleave_lowv2si): Handle big endian.
6554         (vec_interleave_lowv2si_be): New.
6555         (vec_interleave_lowv2si_le): New.
6556         (insn_v4int_l): Handle big endian.
6557         * config/tilegx/tilegx.opt (mbig-endian): New option.
6558         (mlittle-endian): New option.
6559         * doc/install.texi: Document tilegxbe-linux.
6560         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
6561
6562 2014-02-24  Martin Jambor  <mjambor@suse.cz>
6563
6564         PR ipa/60266
6565         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
6566         there are no parameter descriptors.
6567
6568 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
6569
6570         PR rtl-optimization/60268
6571         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
6572         initialization to ...
6573         (sched_rgn_init): ... here.
6574         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
6575
6576 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6577
6578         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
6579         names.
6580
6581 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
6582
6583         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
6584         definition.
6585
6586 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6587
6588         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
6589         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
6590
6591 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6592
6593         * config/microblaze/predicates.md: Add cmp_op predicate.
6594         * config/microblaze/microblaze.md: Add branch_compare instruction
6595         which uses cmp_op predicate and emits cmp insn before branch.
6596         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
6597         to microblaze_expand_conditional_branch and consolidate logic.
6598         (microblaze_expand_conditional_branch): emit branch_compare
6599         insn instead of handling cmp op separate from branch insn.
6600
6601 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6602
6603         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
6604         to permit subregs.
6605
6606 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6607
6608         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
6609         define_insn with define_expand and new define_insn
6610         *altivec_lve<VI_char>x_internal.
6611         (altivec_stve<VI_char>x): Replace define_insn with define_expand
6612         and new define_insn *altivec_stve<VI_char>x_internal.
6613         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
6614         prototype.
6615         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
6616         lve*x built-ins.
6617         (altivec_expand_stvex_be): New function.
6618
6619 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
6620
6621         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
6622         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
6623         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
6624         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
6625
6626 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
6627
6628         PR target/60298
6629         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
6630         instead of emit_move_insn.
6631
6632 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6633
6634         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
6635         vspltw with vsldoi.
6636         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
6637         gen_altivec_vsumsws.
6638
6639 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6640
6641         * config/rs6000/altivec.md (altivec_lvxl): Rename as
6642         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
6643         (altivec_lvxl_<mode>): New define_expand incorporating
6644         -maltivec=be semantics where needed.
6645         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
6646         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
6647         semantics where needed.
6648         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
6649         (altivec_stvx_<mode>): New define_expand incorporating
6650         -maltivec=be semantics where needed.
6651         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
6652         VM2 iterator instead of V4SI.
6653         (altivec_stvxl_<mode>): New define_expand incorporating
6654         -maltivec=be semantics where needed.
6655         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
6656         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
6657         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
6658         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
6659         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
6660         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
6661         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
6662         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
6663         ALTIVEC_BUILTIN_STVXL.
6664         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
6665         (altivec_expand_stvx_be): Likewise.
6666         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
6667         (altivec_expand_lvx_be): Likewise.
6668         (altivec_expand_stvx_be): Likewise.
6669         (altivec_expand_builtin): Add cases for
6670         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
6671         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
6672         (altivec_init_builtins): Add definitions for
6673         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
6674         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
6675
6676 2014-02-21  Catherine Moore  <clm@codesourcery.com>
6677
6678         * doc/invoke.texi (mvirt, mno-virt): Document.
6679         * config/mips/mips.opt (mvirt): New option.
6680         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
6681
6682 2014-02-21  Richard Biener  <rguenther@suse.de>
6683
6684         PR tree-optimization/60276
6685         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
6686         (STMT_VINFO_MIN_NEG_DIST): New macro.
6687         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
6688         STMT_VINFO_MIN_NEG_DIST.
6689         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
6690         made for negative dependence distances still hold.
6691
6692 2014-02-21  Richard Biener  <rguenther@suse.de>
6693
6694         PR middle-end/60291
6695         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
6696         DECL_INITIAL for globals not in the current function context.
6697
6698 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
6699
6700         PR tree-optimization/56490
6701         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
6702         * tree-ssa-uninit.c: Include params.h.
6703         (compute_control_dep_chain): Add num_calls argument, return false
6704         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
6705         num_calls to recursive call.
6706         (find_predicates): Change dep_chain into normal array,
6707         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
6708         variable and adjust compute_control_dep_chain caller.
6709         (find_def_preds): Likewise.
6710
6711 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
6712
6713         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
6714         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
6715
6716 2014-02-21  Nick Clifton  <nickc@redhat.com>
6717
6718         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
6719         (pushhi1): Likewise.
6720         (popqi1): Add mode to pre_dec.
6721         (pophi1): Likewise.
6722
6723 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
6724
6725         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
6726         mode for mask of V8SFmode permutation.
6727
6728 2014-02-20  Richard Henderson  <rth@redhat.com>
6729
6730         PR c++/60272
6731         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
6732         a new pseudo for OLDVAL.
6733
6734 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
6735
6736         PR target/57896
6737         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
6738         gen_reg_rtx if d->testing_p.
6739         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
6740         if d->testing_p and we will certainly return true.
6741         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
6742         if d->testing_p.
6743
6744 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
6745
6746         * emit-rtl.c (gen_reg_rtx): Assert that
6747         crtl->emit.regno_pointer_align_length is non-zero.
6748
6749 2014-02-20  Richard Henderson  <rth@redhat.com>
6750
6751         PR c++/60272
6752         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
6753         on failure the store back into EXPECT.
6754
6755 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
6756             Sandra Loosemore  <sandra@codesourcery.com>
6757
6758         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
6759         * config/nios2/nios2.c (nios2_function_profiler): Add
6760         -fPIC (flag_pic == 2) support.
6761         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
6762         (nios2_large_offset_p): New function.
6763         (nios2_unspec_reloc_p): Move up position, update to use
6764         nios2_large_offset_p.
6765         (nios2_unspec_address): Remove function.
6766         (nios2_unspec_offset): New function.
6767         (nios2_large_got_address): New function.
6768         (nios2_got_address): Add large offset support.
6769         (nios2_legitimize_tls_address): Update usage of removed and new
6770         functions.
6771         (nios2_symbol_binds_local_p): New function.
6772         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
6773         (nios2_legitimize_address): Update to use nios2_large_offset_p.
6774         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
6775         (nios2_print_operand): Merge H/L processing, add hiadj/lo
6776         processing for (const (unspec ...)).
6777         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
6778
6779 2014-02-20  Richard Biener  <rguenther@suse.de>
6780
6781         * tree-cfg.c (replace_uses_by): Mark altered BBs before
6782         doing the substitution.
6783         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
6784
6785 2014-02-20  Martin Jambor  <mjambor@suse.cz>
6786
6787         PR ipa/55260
6788         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
6789         info when checking whether lattices are bottom.
6790
6791 2014-02-20  Richard Biener  <rguenther@suse.de>
6792
6793         PR middle-end/60221
6794         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
6795         regions at -O0.
6796
6797 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
6798
6799         PR ipa/58555
6800         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
6801         parameter specifying the scaling.
6802         (inline_call): Update.
6803         (want_inline_recursively): Guard division by zero.
6804         (recursive_inlining): Update.
6805         * ipa-inline.h (clone_inlined_nodes): Update.
6806
6807 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6808
6809         PR target/60204
6810         * config/i386/i386.c (classify_argument): Pass structures of size
6811         64 bytes or less in register.
6812
6813 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6814             Kirill Yukhin  <kirill.yukhin@intel.com>
6815
6816         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
6817         (_mm_rcp28_round_ss): Ditto.
6818         (_mm_rsqrt28_round_sd): Ditto.
6819         (_mm_rsqrt28_round_ss): Ditto.
6820         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
6821         (_mm_rcp14_round_ss): Ditto.
6822         (_mm_rsqrt14_round_sd): Ditto.
6823         (_mm_rsqrt14_round_ss): Ditto.
6824         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
6825         the first input operand, get rid of match_dup.
6826         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
6827         attribute to sse.
6828         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6829         Ditto.
6830         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
6831         operand as the first input operand, set type attribute.
6832         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6833         Set type attribute.
6834         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
6835         operand as the first input operand, set type attribute.
6836
6837 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6838
6839         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
6840         bit of zero.
6841
6842 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
6843
6844         PR target/60207
6845         * config/i386/i386.c (construct_container): Remove TFmode check
6846         for X86_64_INTEGER_CLASS.
6847
6848 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
6849
6850         PR target/59794
6851         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
6852         only when -Wpsabi is enabled.
6853
6854 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
6855
6856          PR target/59799
6857         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
6858         passing arrays in registers are the same as for structs, so remove the
6859         special case for them.
6860
6861 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
6862
6863         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
6864         destination type, extract only the valid bits if the source type is not
6865         integral and has a different mode.
6866
6867 2014-02-19  Richard Biener  <rguenther@suse.de>
6868
6869         PR ipa/60243
6870         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
6871         for all calls.
6872
6873 2014-02-19  Richard Biener  <rguenther@suse.de>
6874
6875         PR ipa/60243
6876         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
6877         (ipa_modify_call_arguments): Emit an argument load explicitely and
6878         preserve virtual SSA form there and for the replacement call.
6879         Do not update SSA form nor free dominance info.
6880
6881 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6882
6883         * ipa.c (function_and_variable_visibility): Also clear WEAK
6884         flag when disolving COMDAT_GROUP.
6885
6886 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6887
6888         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
6889         * ipa-prop.c (ipa_set_jf_known_type): Return early when
6890         not devirtualizing.
6891         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
6892         do more sanity checks.
6893         (detect_type_change): Return true when giving up early.
6894         (compute_complex_assign_jump_func): Fix type parameter of
6895         ipa_set_ancestor_jf.
6896         (compute_complex_ancestor_jump_func): Likewise.
6897         (update_jump_functions_after_inlining): Fix updating of
6898         ancestor function.
6899         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
6900
6901 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6902
6903         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
6904         inline clones when edge disappears.
6905
6906 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
6907
6908         PR target/60203
6909         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
6910         Split 64-bit moves into 2 patterns.  Do not allow the use of
6911         direct move for TDmode in little endian, since the decimal value
6912         has little endian bytes within a word, but the 64-bit pieces are
6913         ordered in a big endian fashion, and normal subreg's of TDmode are
6914         not allowed.
6915         (mov<mode>_64bit_dm): Likewise.
6916         (movtd_64bit_nodm): Likewise.
6917
6918 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6919
6920         PR tree-optimization/60174
6921         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
6922         statement of an SSA_NAME that occurs in an abnormal PHI node.
6923
6924 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6925
6926         PR sanitizer/60142
6927         * final.c (SEEN_BB): Remove.
6928         (SEEN_NOTE, SEEN_EMITTED): Renumber.
6929         (final_scan_insn): Don't force_source_line on second
6930         NOTE_INSN_BASIC_BLOCK.
6931
6932 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
6933
6934         PR target/60205
6935         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
6936         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
6937         (type_natural_mode): Warn ABI change when %zmm register is not
6938         available for AVX512F vector value passing.
6939
6940 2014-02-18  Kai Tietz  <ktietz@redhat.com>
6941
6942         PR target/60193
6943         * config/i386/i386.c (ix86_expand_prologue): Use value in
6944         rax register as displacement when restoring %r10 or %rax.
6945         Fix wrong offset when restoring both registers.
6946
6947 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6948
6949         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
6950         assertion with conditional return.
6951
6952 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6953             Uros Bizjak  <ubizjak@gmail.com>
6954
6955         PR driver/60233
6956         * config/i386/driver-i386.c (host_detect_local_cpu): If
6957         YMM state is not saved by the OS, also clear has_f16c.  Move
6958         CPUID 0x80000001 handling before YMM state saving checking.
6959
6960 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
6961
6962         PR rtl-optimization/58960
6963         * haifa-sched.c (alloc_global_sched_pressure_data): New,
6964         factored out from ...
6965         (sched_init): ... here.
6966         (free_global_sched_pressure_data): New, factored out from ...
6967         (sched_finish): ... here.
6968         * sched-int.h (free_global_sched_pressure_data): Declare.
6969         * sched-rgn.c (nr_regions_initial): New static global.
6970         (haifa_find_rgns): Initialize it.
6971         (schedule_region): Disable sched-pressure for the newly
6972         generated regions.
6973
6974 2014-02-17  Richard Biener  <rguenther@suse.de>
6975
6976         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
6977         release SSA defs of pattern stmts.
6978
6979 2014-02-17  Richard Biener  <rguenther@suse.de>
6980
6981         * tree-inline.c (expand_call_inline): Release the virtual
6982         operand defined by the call we are about to inline.
6983
6984 2014-02-17  Richard Biener  <rguenther@suse.de>
6985
6986         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
6987
6988 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
6989             Ilya Tocar  <ilya.tocar@intel.com>
6990
6991         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
6992         arguments order in builtin.
6993         (_mm512_permutexvar_epi64): Ditto.
6994         (_mm512_mask_permutexvar_epi64): Ditto
6995         (_mm512_maskz_permutexvar_epi32): Ditto
6996         (_mm512_permutexvar_epi32): Ditto
6997         (_mm512_mask_permutexvar_epi32): Ditto
6998
6999 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7000
7001         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
7002         (p8_vmrgow): Likewise.
7003
7004 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7005
7006         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
7007         endian targets.
7008
7009 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
7010
7011         PR target/60203
7012         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
7013         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
7014         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
7015         using direct move instructions on ISA 2.07.  Also adjust
7016         instruction length for 64-bit.
7017         (mov<mode>_64bit, TFmode/TDmode): Likewise.
7018         (mov<mode>_32bit, TFmode/TDmode): Likewise.
7019
7020 2014-02-15  Alan Modra  <amodra@gmail.com>
7021
7022         PR target/58675
7023         PR target/57935
7024         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
7025         find_replacement on parts of insn rtl that might be reloaded.
7026
7027 2014-02-15  Richard Biener  <rguenther@suse.de>
7028
7029         PR tree-optimization/60183
7030         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
7031         (tree_ssa_phiprop): Calculate and free post-dominators.
7032
7033 2014-02-14  Jeff Law  <law@redhat.com>
7034
7035         PR rtl-optimization/60131
7036         * ree.c (get_extended_src_reg): New function.
7037         (combine_reaching_defs): Use it rather than assuming location of REG.
7038         (find_and_remove_re): Verify first operand of extension is
7039         a REG before adding the insns to the copy list.
7040
7041 2014-02-14  Roland McGrath  <mcgrathr@google.com>
7042
7043         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
7044         * configure: Regenerated.
7045         * config.in: Regenerated.
7046         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
7047         instead of ASM_SHORT.
7048
7049 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
7050             Richard Earnshaw  <rearnsha@arm.com>
7051
7052         PR rtl-optimization/59535
7053         * lra-constraints.c (process_alt_operands): Encourage alternative
7054         when unassigned pseudo class is superset of the alternative class.
7055         (inherit_reload_reg): Don't inherit when optimizing for code size.
7056         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
7057         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
7058         modes not less than 4 for Thumb1.
7059
7060 2014-02-14  Kyle McMartin  <kyle@redhat.com>
7061
7062         PR pch/60010
7063         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
7064
7065 2014-02-14  Richard Biener  <rguenther@suse.de>
7066
7067         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
7068         (get_frame_arg): Drop the assert with langhook types_compatible_p.
7069         Do not strip INDIRECT_REFs.
7070
7071 2014-02-14  Richard Biener  <rguenther@suse.de>
7072
7073         PR lto/60179
7074         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
7075         DECL_FUNCTION_SPECIFIC_TARGET.
7076         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
7077         * tree-streamer-out.c (pack_ts_target_option): Remove.
7078         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
7079         (write_ts_function_decl_tree_pointers): Do not stream
7080         DECL_FUNCTION_SPECIFIC_TARGET.
7081         * tree-streamer-in.c (unpack_ts_target_option): Remove.
7082         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
7083         (lto_input_ts_function_decl_tree_pointers): Do not stream
7084         DECL_FUNCTION_SPECIFIC_TARGET.
7085
7086 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
7087
7088         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
7089         (get_initial_def_for_induction, vectorizable_induction): Ignore
7090         debug stmts when looking for exit_phi.
7091         (vectorizable_live_operation): Fix up condition.
7092
7093 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7094
7095         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
7096         nreverse() because it changes the content of original tree list.
7097
7098 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7099
7100         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
7101         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
7102
7103 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
7104
7105         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
7106         GNU coding standards.
7107
7108 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7109
7110         PR debug/60152
7111         * dwarf2out.c (gen_subprogram_die): Don't call
7112         add_calling_convention_attribute if subr_die is old_die.
7113
7114 2014-02-13  Sharad Singhai  <singhai@google.com>
7115
7116         * doc/optinfo.texi: Fix order of nodes.
7117
7118 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
7119
7120         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
7121         operands[2], not operands[3].
7122
7123 2014-02-13  Richard Biener  <rguenther@suse.de>
7124
7125         PR bootstrap/59878
7126         * doc/install.texi (ISL): Update recommended version to 0.12.2,
7127         mention the possibility of an in-tree build.
7128         (CLooG): Update recommended version to 0.18.1, mention the
7129         possibility of an in-tree build and clarify that the ISL
7130         bundled with CLooG does not work.
7131
7132 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
7133
7134         PR target/43546
7135         * expr.c (compress_float_constant): If x is a hard register,
7136         extend into a pseudo and then move to x.
7137
7138 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7139
7140         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
7141         caused by bad second argument to warning_at() with -mhotpatch and
7142         nested functions (e.g. with gfortran).
7143
7144 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
7145
7146         * opts.c (option_name): Remove "enabled by default" rider.
7147
7148 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
7149
7150         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
7151
7152 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
7153             Uros Bizjak  <ubizjak@gmail.com>
7154
7155         PR target/60151
7156         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
7157         * configure: Regenerated.
7158
7159 2014-02-12  Richard Biener  <rguenther@suse.de>
7160
7161         * vec.c (vec_prefix::calculate_allocation): Move as
7162         inline variant to vec.h.
7163         (vec_prefix::calculate_allocation_1): New out-of-line version.
7164         * vec.h (vec_prefix::calculate_allocation_1): Declare.
7165         (vec_prefix::m_has_auto_buf): Rename to ...
7166         (vec_prefix::m_using_auto_storage): ... this.
7167         (vec_prefix::calculate_allocation): Inline the easy cases
7168         and dispatch to calculate_allocation_1 which doesn't need the
7169         prefix address.
7170         (va_heap::reserve): Use gcc_checking_assert.
7171         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
7172         m_using_auto_storage.
7173         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
7174         member and adjust.
7175         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
7176         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
7177         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
7178
7179 2014-02-12  Richard Biener  <rguenther@suse.de>
7180
7181         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
7182         when we found a dependence.
7183
7184 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
7185
7186         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
7187         common code...
7188         (maybe_fold_stmt): ... into this new function.
7189         * omp-low.c (lower_omp): Update comment.
7190
7191         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
7192         last use.
7193
7194         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
7195         dereference.
7196
7197 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
7198
7199         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
7200         identifiers in comments.
7201         (cortexa53_extra_costs): Likewise.
7202         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
7203         (cortexa7_extra_costs): Likewise.
7204         (cortexa12_extra_costs): Likewise.
7205         (cortexa15_extra_costs): Likewise.
7206         (v7m_extra_costs): Likewise.
7207
7208 2014-02-12  Richard Biener  <rguenther@suse.de>
7209
7210         PR middle-end/60092
7211         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
7212         of posix_memalign being successful.
7213         (lower_stmt): Restrict lowering of posix_memalign to when
7214         -ftree-bit-ccp is enabled.
7215
7216 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
7217
7218         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
7219         arg_loc.
7220         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
7221
7222 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
7223
7224         PR rtl-optimization/60116
7225         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
7226         other_insn once the combination has been validated.
7227
7228 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
7229
7230         PR lto/59468
7231         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
7232         and wrapper.
7233         * ipa-devirt.c: Include demangle.h
7234         (odr_violation_reported): New static variable.
7235         (add_type_duplicate): Update odr_violations.
7236         (maybe_record_node): Add completep parameter; update it.
7237         (record_target_from_binfo): Add COMPLETEP parameter;
7238         update it as needed.
7239         (possible_polymorphic_call_targets_1): Likewise.
7240         (struct polymorphic_call_target_d): Add nonconstruction_targets;
7241         rename FINAL to COMPLETE.
7242         (record_targets_from_bases): Sanity check we found the binfo;
7243         fix COMPLETEP updating.
7244         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
7245         parameter, fix computing of COMPLETEP.
7246         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
7247         at LTO time do demangling.
7248         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
7249         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
7250         parameter.
7251         (gimple_get_virt_method_for_binfo): Likewise.
7252         * gimple-fold.h (gimple_get_virt_method_for_binfo,
7253         gimple_get_virt_method_for_vtable): Update prototypes.
7254
7255 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
7256
7257         PR target/49008
7258         * genautomata.c (add_presence_absence): Fix typo with
7259         {final_}presence_list.
7260
7261 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7262
7263         PR target/60137
7264         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
7265         for VSX/Altivec vectors that land in GPR registers.
7266
7267 2014-02-11  Richard Henderson  <rth@redhat.com>
7268             Jakub Jelinek  <jakub@redhat.com>
7269
7270         PR debug/59776
7271         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
7272         around drhs if type conversion to lacc->type is not useless.
7273
7274 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7275
7276         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
7277         tuning struct.
7278         (cortex-a57.cortex-a53): Likewise.
7279         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
7280
7281 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7282
7283         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
7284         arm_restrict_it.
7285
7286 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
7287
7288         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
7289         add_options_for_arm_vfp3.
7290
7291 2014-02-11  Jeff Law  <law@redhat.com>
7292
7293         PR middle-end/54041
7294         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
7295         object with an undesirable mode.
7296
7297 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7298
7299         PR libgomp/60107
7300         * config/i386/sol2-9.h: New file.
7301         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
7302         *-*-solaris2.9*): Use it.
7303
7304 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7305
7306         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
7307         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
7308
7309 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7310
7311         * config/microblaze/microblaze.c: Extend mcpu version format
7312
7313 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
7314
7315         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
7316
7317 2014-02-10  Richard Henderson  <rth@redhat.com>
7318
7319         PR target/59927
7320         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
7321         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
7322         ms-abi vs -mno-accumulate-outgoing-args.
7323         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
7324         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
7325         respect to ms-abi.
7326
7327 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7328
7329         PR middle-end/60080
7330         * cfgexpand.c (expand_asm_operands): Attach source location to
7331         ASM_INPUT rtx objects.
7332         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
7333
7334 2014-02-10  Nick Clifton  <nickc@redhat.com>
7335
7336         * config/mn10300/mn10300.c (popcount): New function.
7337         (mn10300_expand_prologue): Include saved registers in stack usage
7338         count.
7339
7340 2014-02-10  Jeff Law  <law@redhat.com>
7341
7342         PR middle-end/52306
7343         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
7344         when changing the SET_DEST of a prior insn to avoid an input reload.
7345
7346 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7347
7348         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
7349         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
7350         -mcall-openbsd, or -mcall-linux.
7351         (CC1_ENDIAN_BIG_SPEC): Remove.
7352         (CC1_ENDIAN_LITTLE_SPEC): Remove.
7353         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7354         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
7355         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
7356         and %cc1_endian_default.
7357         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7358
7359 2014-02-10  Richard Biener  <rguenther@suse.de>
7360
7361         PR tree-optimization/60115
7362         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
7363         MEM_REF handling.  Properly verify that the accesses are not
7364         out of the objects bound.
7365
7366 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7367
7368         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
7369         coretex to cortex.
7370
7371 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
7372
7373         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
7374         proper constants and fix formatting.
7375         (possible_polymorphic_call_targets): Fix formatting.
7376
7377 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
7378             Ilya Tocar  <ilya.tocar@intel.com>
7379
7380         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
7381         (_mm512_loadu_epi32): Renamed into...
7382         (_mm512_loadu_si512): This.
7383         (_mm512_storeu_epi32): Renamed into...
7384         (_mm512_storeu_si512): This.
7385         (_mm512_maskz_ceil_ps): Removed.
7386         (_mm512_maskz_ceil_pd): Ditto.
7387         (_mm512_maskz_floor_ps): Ditto.
7388         (_mm512_maskz_floor_pd): Ditto.
7389         (_mm512_floor_round_ps): Ditto.
7390         (_mm512_floor_round_pd): Ditto.
7391         (_mm512_ceil_round_ps): Ditto.
7392         (_mm512_ceil_round_pd): Ditto.
7393         (_mm512_mask_floor_round_ps): Ditto.
7394         (_mm512_mask_floor_round_pd): Ditto.
7395         (_mm512_mask_ceil_round_ps): Ditto.
7396         (_mm512_mask_ceil_round_pd): Ditto.
7397         (_mm512_maskz_floor_round_ps): Ditto.
7398         (_mm512_maskz_floor_round_pd): Ditto.
7399         (_mm512_maskz_ceil_round_ps): Ditto.
7400         (_mm512_maskz_ceil_round_pd): Ditto.
7401         (_mm512_expand_pd): Ditto.
7402         (_mm512_expand_ps): Ditto.
7403         * config/i386/i386.c (ix86_builtins): Remove
7404         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
7405         (bdesc_args): Ditto.
7406         * config/i386/predicates.md (const1256_operand): New.
7407         (const_1_to_2_operand): Ditto.
7408         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
7409         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
7410         (*avx512pf_gatherpf<mode>sf): Ditto.
7411         (avx512pf_gatherpf<mode>df): Ditto.
7412         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7413         (*avx512pf_gatherpf<mode>df): Ditto.
7414         (avx512pf_scatterpf<mode>sf): Ditto.
7415         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7416         (*avx512pf_scatterpf<mode>sf): Ditto.
7417         (avx512pf_scatterpf<mode>df): Ditto.
7418         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7419         (*avx512pf_scatterpf<mode>df): Ditto.
7420         (avx512f_expand<mode>): Removed.
7421         (<shift_insn><mode>3<mask_name>): Change predicate type.
7422
7423 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7424
7425         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
7426         not at the end of datarefs vector use ordered_remove to avoid
7427         reordering datarefs vector.
7428
7429         PR c/59984
7430         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
7431         mark local addressable non-static vars as GOVD_PRIVATE
7432         instead of GOVD_LOCAL.
7433         * omp-low.c (lower_omp_for): Move gimple_bind_vars
7434         and BLOCK_VARS of gimple_bind_block to new_stmt rather
7435         than copying them.
7436
7437         PR middle-end/60092
7438         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
7439         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
7440         assume_aligned or alloc_align attributes.
7441         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
7442         arguments.  Handle also assume_aligned and alloc_align attributes.
7443         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
7444         calls to functions with assume_aligned or alloc_align attributes.
7445         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
7446
7447 2014-02-08  Terry Guo  <terry.guo@arm.com>
7448
7449         * doc/invoke.texi: Document ARM -march=armv7e-m.
7450
7451 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7452
7453         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
7454         flag on __cilkrts_rethrow builtin.
7455
7456         PR ipa/60026
7457         * ipa-cp.c (determine_versionability): Fail at -O0
7458         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
7459         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
7460
7461         Revert:
7462         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7463
7464         PR ipa/60026
7465         * tree-inline.c (copy_forbidden): Fail for
7466         __attribute__((optimize (0))) functions.
7467
7468 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7469
7470         * varpool.c: Include pointer-set.h.
7471         (varpool_remove_unreferenced_decls): Variables in other partitions
7472         will not be output; be however careful to not lose information
7473         about partitioning.
7474
7475 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7476
7477         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
7478         lookup in the vtable constructor.
7479
7480 2014-02-07  Jeff Law  <law@redhat.com>
7481
7482         PR target/40977
7483         * config/m68k/m68k.md (ashldi_extsi): Turn into a
7484         define_insn_and_split.
7485
7486         * ipa-inline.c (inline_small_functions): Fix typos.
7487
7488 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7489
7490         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
7491         (s390_can_use_return_insn): Declare.
7492         * config/s390/s390.h (EPILOGUE_USES): Define.
7493         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
7494         instructions.
7495         (s390_chunkify_start): Handle return JUMP_LABELs.
7496         (s390_early_mach): Emit a main_pool instruction on the entry edge.
7497         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
7498         (s390_can_use_return_insn): New functions.
7499         (s390_fix_long_loop_prediction): Handle conditional returns.
7500         (TARGET_SET_UP_BY_PROLOGUE): Define.
7501         * config/s390/s390.md (ANY_RETURN): New code iterator.
7502         (*creturn, *csimple_return, return, simple_return): New patterns.
7503
7504 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7505
7506         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
7507         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
7508         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
7509         REG_CFA_RESTORE list when deciding not to restore a register.
7510
7511 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7512
7513         * config/s390/s390.c: Include tree-pass.h and context.h.
7514         (s390_early_mach): New function, split out from...
7515         (s390_emit_prologue): ...here.
7516         (pass_data_s390_early_mach): New pass structure.
7517         (pass_s390_early_mach): New class.
7518         (s390_option_override): Create and register early_mach pass.
7519         Move to end of file.
7520
7521 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7522
7523         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
7524         to match for the exit block.
7525
7526 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7527
7528         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
7529         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
7530         Reject misaligned operands.
7531
7532 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7533
7534         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
7535
7536 2014-02-07  Richard Biener  <rguenther@suse.de>
7537
7538         PR middle-end/60092
7539         * gimple-low.c (lower_builtin_posix_memalign): New function.
7540         (lower_stmt): Call it to lower posix_memalign in a way
7541         to make alignment info accessible.
7542
7543 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
7544
7545         PR c++/60082
7546         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
7547         __builtin_setjmp_receiver.
7548
7549 2014-02-07  Richard Biener  <rguenther@suse.de>
7550
7551         PR middle-end/60092
7552         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
7553         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
7554         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
7555         Handle BUILT_IN_POSIX_MEMALIGN.
7556         (find_func_clobbers): Likewise.
7557         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7558         (call_may_clobber_ref_p_1): Likewise.
7559
7560 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7561
7562         PR ipa/59918
7563         * ipa-devirt.c (record_target_from_binfo): Remove overactive
7564         sanity check.
7565
7566 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7567
7568         PR ipa/59469
7569         * lto-cgraph.c (lto_output_node): Use
7570         symtab_get_symbol_partitioning_class.
7571         (lto_output_varpool_node): likewise.
7572         (symtab_get_symbol_partitioning_class): Move here from
7573         lto/lto-partition.c
7574         * cgraph.h (symbol_partitioning_class): Likewise.
7575         (symtab_get_symbol_partitioning_class): Declare.
7576
7577 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7578
7579         * ggc.h (ggc_internal_cleared_alloc): New macro.
7580         * vec.h (vec_safe_copy): Handle memory stats.
7581         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
7582         * target-globals.c (save_target_globals): Likewise.
7583
7584 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7585
7586         PR target/60077
7587         * expr.c (emit_move_resolve_push): Export; be bit more selective
7588         on when to clear alias set.
7589         * expr.h (emit_move_resolve_push): Declare.
7590         * function.h (struct function): Add tail_call_marked.
7591         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
7592         * config/i386/i386-protos.h (ix86_expand_push): Remove.
7593         * config/i386/i386.md (TImode move expander): De not call
7594         ix86_expand_push.
7595         (FP push expanders): Preserve memory attributes.
7596         * config/i386/sse.md (push<mode>1): Remove.
7597         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
7598         (ix86_expand_push): Remove.
7599         * config/i386/mmx.md (push<mode>1): Remove.
7600
7601 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7602
7603         PR rtl-optimization/60030
7604         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
7605         lopart with paradoxical subreg before shifting it up by hprec.
7606
7607 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7608
7609         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
7610         Remove extra newline at end of file.
7611         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
7612         (arm_issue_rate): Handle cortexa57.
7613         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
7614         (cortex-a57.cortex-a53): Likewise.
7615
7616 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7617
7618         PR target/59575
7619         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
7620         don't record in REG_FRAME_RELATED_EXPR registers not set in that
7621         bitmask.
7622         (arm_expand_prologue): Adjust all callers.
7623         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
7624         info, registers also at the lowest numbered registers side.  Use
7625         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
7626         XEXP.
7627
7628         PR debug/59992
7629         * var-tracking.c (adjust_mems): Before adding a SET to
7630         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
7631
7632 2014-02-06  Alan Modra  <amodra@gmail.com>
7633
7634         PR target/60032
7635         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
7636         change SDmode to DDmode when lra_in_progress.
7637
7638 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7639
7640         PR middle-end/59150
7641         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
7642         free_data_ref on the dr first, and before goto again also set dr
7643         to the next dr.  For simd_lane_access, free old datarefs[i] before
7644         overwriting it.  For get_vectype_for_scalar_type failure, don't
7645         free_data_ref if simd_lane_access.
7646
7647         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
7648
7649         PR target/60062
7650         * tree.h (opts_for_fn): New inline function.
7651         (opt_for_fn): Define.
7652         * config/i386/i386.c (ix86_function_regparm): Use
7653         opt_for_fn (decl, optimize) instead of optimize.
7654
7655 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7656
7657         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
7658         for SYMBOL_REF in large memory model.
7659
7660 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7661
7662         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
7663         and crypto support.
7664         (cortex-a57): Likewise.
7665         (cortex-a57.cortex-a53): Likewise.
7666
7667 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
7668             Kugan Vivekanandarajah  <kuganv@linaro.org>
7669
7670         * config/arm/arm.c (arm_vector_alignment_reachable): Check
7671         unaligned_access.
7672         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
7673
7674 2014-02-06  Richard Biener  <rguenther@suse.de>
7675
7676         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
7677         set_loop_copy and initialize_original_copy_tables.
7678
7679 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
7680
7681         * config/aarch64/aarch64-simd.md
7682         (aarch64_ashr_simddi): Change QI to SI.
7683
7684 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
7685             Jakub Jelinek  <jakub@redhat.com>
7686
7687         PR middle-end/60013
7688         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
7689         of the dataflow.
7690
7691 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7692
7693         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
7694         CODE_FOR_altivec_vpku[hw]um to
7695         CODE_FOR_altivec_vpku[hw]um_direct.
7696         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
7697         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
7698         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
7699         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
7700
7701 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7702
7703         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
7704         generation for -maltivec=be.
7705         (altivec_vsumsws): Simplify redundant test.
7706
7707 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7708
7709         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
7710         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
7711         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
7712         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
7713         gen_altivec_vpkuwum.
7714         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
7715         BYTES_BIG_ENDIAN.
7716         (altivec_vpks<VI_char>ss): Likewise.
7717         (altivec_vpks<VI_char>us): Likewise.
7718         (altivec_vpku<VI_char>us): Likewise.
7719         (altivec_vpku<VI_char>um): Likewise.
7720         (altivec_vpku<VI_char>um_direct): New (copy of
7721         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
7722         internal use).
7723         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
7724         target is little endian and -maltivec=be is not specified.
7725         (*altivec_vupkhs<VU_char>_direct): New (copy of
7726         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
7727         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
7728         target is little endian and -maltivec=be is not specified.
7729         (*altivec_vupkls<VU_char>_direct): New (copy of
7730         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
7731         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
7732         little endian and -maltivec=be is not specified.
7733         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
7734         little endian and -maltivec=be is not specified.
7735
7736 2014-02-05  Richard Henderson  <rth@redhat.com>
7737
7738         PR debug/52727
7739         * combine-stack-adj.c: Revert r206943.
7740         * sched-int.h (struct deps_desc): Add last_args_size.
7741         * sched-deps.c (init_deps): Initialize it.
7742         (sched_analyze_insn): Add OUTPUT dependencies between insns that
7743         contain REG_ARGS_SIZE notes.
7744
7745 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
7746
7747         * lto-cgraph.c (asm_nodes_output): Make global.
7748         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
7749         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
7750         (driver_handle_option): Handle OPT_fwpa.
7751
7752 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
7753
7754         PR ipa/59947
7755         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
7756         a comment typo and formatting issue.  If odr_hash hasn't been
7757         created, return vNULL and set *completep to false.
7758
7759         PR middle-end/57499
7760         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
7761         bb with no successors.
7762
7763 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
7764
7765         PR target/59718
7766         * doc/invoke.texi (-march): Clarify documentation for ARM.
7767         (-mtune): Likewise.
7768         (-mcpu): Likewise.
7769
7770 2014-02-05  Richard Biener  <rguenther@suse.de>
7771
7772         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
7773         when not vectorizing because of too many alias checks.
7774         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
7775         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
7776
7777 2014-02-05  Nick Clifton  <nickc@redhat.com>
7778
7779         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
7780         accept extended registers in any mode when compiling for the MN10300.
7781
7782 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
7783
7784         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
7785         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
7786         sanitization attributes.
7787         (can_inline_edge_p): Likewise.
7788         (sanitize_attrs_match_for_inline_p): New function.
7789
7790 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7791
7792         * ipa-prop.c (detect_type_change): Shor circuit testing of
7793         type changes on THIS pointer.
7794
7795 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
7796
7797         PR target/59777
7798         * config/pa/pa.c (legitimize_tls_address): Return original address
7799         if not passed a SYMBOL_REF rtx.
7800         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
7801         addresses.
7802         (pa_emit_move_sequence): Simplify TLS source operands.
7803         (pa_legitimate_constant_p): Reject all TLS constants.
7804         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
7805         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
7806
7807 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7808
7809         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
7810         groups when we know they are controlled by LTO.
7811         * varasm.c (default_binds_local_p_1): If object is in other partition,
7812         it will be resolved locally.
7813
7814 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7815
7816         * config/host-linux.c (linux_gt_pch_use_address): Don't
7817         use SSIZE_MAX because it is not always defined.
7818
7819 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
7820
7821         PR bootstrap/59913
7822         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
7823         threshold for pseudo splitting.
7824         (update_ebb_live_info): Process call argument hard registers and
7825         hard registers from insn definition too.
7826         (max_small_class_regs_num): New constant.
7827         (inherit_in_ebb): Update live hard regs through EBBs.  Update
7828         reloads_num only for small register classes.  Don't split for
7829         outputs of jumps.
7830
7831 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
7832
7833         PR ipa/60058
7834         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
7835         is non-null.
7836
7837 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7838
7839         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
7840         visibility is safe.
7841
7842 2014-02-04  Marek Polacek  <polacek@redhat.com>
7843
7844         * gdbinit.in (pel): Define.
7845
7846 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7847
7848         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
7849         behavior.
7850
7851 2014-02-04  Richard Biener  <rguenther@suse.de>
7852
7853         PR lto/59723
7854         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
7855         in function context local.
7856         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
7857         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
7858         similar to LTO_imported_decl_ref.
7859
7860 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7861
7862         PR tree-optimization/60002
7863         * cgraphclones.c (build_function_decl_skip_args): Clear
7864         DECL_LANG_SPECIFIC.
7865
7866         PR tree-optimization/60023
7867         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
7868         false to gsi_replace.
7869         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
7870         has been in some EH region and vec_stmt could throw, add
7871         vec_stmt into the same EH region.
7872         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
7873         has no lhs, ignore it.
7874         * internal-fn.c (expand_MASK_LOAD): Likewise.
7875
7876         PR ipa/60026
7877         * tree-inline.c (copy_forbidden): Fail for
7878         __attribute__((optimize (0))) functions.
7879
7880         PR other/58712
7881         * omp-low.c (simd_clone_struct_copy): If from->inbranch
7882         is set, copy one less argument.
7883         (expand_simd_clones): Don't subtract clone_info->inbranch
7884         from simd_clone_struct_alloc argument.
7885
7886         PR rtl-optimization/57915
7887         * recog.c (simplify_while_replacing): If all unary/binary/relational
7888         operation arguments are constant, attempt to simplify those.
7889
7890         PR middle-end/59261
7891         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
7892         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
7893
7894 2014-02-04  Richard Biener  <rguenther@suse.de>
7895
7896         PR tree-optimization/60012
7897         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
7898         TBAA disambiguation to all DDRs.
7899
7900 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7901
7902         PR target/59788
7903         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
7904         (LINK_SPEC): Use it for -shared, -shared-libgcc.
7905
7906 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7907
7908         PR ipa/59882
7909         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
7910
7911 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7912
7913         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
7914         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
7915
7916 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7917
7918         PR ipa/59831
7919         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
7920         to figure out targets of polymorphic calls with known decl.
7921         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7922         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
7923         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
7924         (get_polymorphic_call_info): ... here.
7925         (get_polymorphic_call_info_from_invariant): New function.
7926
7927 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7928
7929         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
7930         lookup via vtable pointer; check for type consistency
7931         and turn inconsitent facts into UNREACHABLE.
7932         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7933         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
7934         type inconsistent querries; return UNREACHABLE instead.
7935
7936 2014-02-03  Richard Henderson  <rth@twiddle.net>
7937
7938         PR tree-opt/59924
7939         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
7940         already processed this node.
7941         (normalize_one_pred_1): Pass along mark_set.
7942         (normalize_one_pred): Create and destroy a pointer_set_t.
7943         (normalize_one_pred_chain): Likewise.
7944
7945 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
7946
7947         PR gcov-profile/58602
7948         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
7949
7950 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7951
7952         PR ipa/59831
7953         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
7954         -fno-devirtualize; try to devirtualize by the knowledge of
7955         virtual table pointer given by aggregate propagation.
7956         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7957         (ipa_print_node_jump_functions): Dump also offset that
7958         is relevant for polymorphic calls.
7959         (determine_known_aggregate_parts): Add arg_type parameter; use it
7960         instead of determining the type from pointer type.
7961         (ipa_compute_jump_functions_for_edge): Update call of
7962         determine_known_aggregate_parts.
7963         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
7964         (gimple_get_virt_method_for_binfo): ... here; simplify using
7965         vtable_pointer_value_to_vtable.
7966         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
7967         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
7968         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
7969         (vtable_pointer_value_to_vtable): Break out from ...; handle also
7970         POINTER_PLUS_EXPR.
7971         (vtable_pointer_value_to_binfo): ... here.
7972         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
7973
7974 2014-02-03  Teresa Johnson  <tejohnson@google.com>
7975
7976         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
7977         redef of outer loop index variable.
7978
7979 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
7980
7981         PR c++/53017
7982         PR c++/59211
7983         * doc/extend.texi (Function Attributes): Typo.
7984
7985 2014-02-03  Cong Hou  <congh@google.com>
7986
7987         PR tree-optimization/60000
7988         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
7989         if the vectorized statement is a store.  A store statement can only
7990         appear at the end of pattern statements.
7991
7992 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
7993
7994         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
7995         (ix86_option_override_internal): Default long double to 64-bit for
7996         32-bit Bionic and to 128-bit for 64-bit Bionic.
7997
7998         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
7999         TARGET_LONG_DOUBLE_128 is true.
8000         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
8001
8002         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
8003         (mlong-double-64): Negate -mlong-double-128.
8004         (mlong-double-128): New option.
8005
8006         * config/i386/i386-c.c (ix86_target_macros): Define
8007         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
8008
8009         * doc/invoke.texi: Document -mlong-double-128.
8010
8011 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
8012
8013         PR rtl-optimization/60024
8014         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
8015
8016 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
8017
8018         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
8019
8020 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8021
8022         PR rtl-optimization/57662
8023         * sel-sched.c (code_motion_path_driver): Do not mark already not
8024         existing blocks in the visiting bitmap.
8025
8026 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
8027
8028         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
8029         on the insn being emitted.
8030
8031 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
8032             Will Deacon  <will.deacon@arm.com>
8033
8034         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
8035
8036 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8037
8038         * config/arm/arm-tables.opt: Regenerate.
8039
8040 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8041
8042         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
8043         for vector types other than V16QImode.
8044         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
8045         define_expand, and call altivec_expand_vec_perm_le when producing
8046         code with little endian element order.
8047         (*altivec_vperm_<mode>_internal): New insn having previous
8048         behavior of altivec_vperm_<mode>.
8049         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
8050         altivec_expand_vec_perm_le when producing code with little endian
8051         element order.
8052         (*altivec_vperm_<mode>_uns_internal): New insn having previous
8053         behavior of altivec_vperm_<mode>_uns.
8054
8055 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8056
8057         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
8058         (altivec_vsumsws): Add handling for -maltivec=be with a little
8059         endian target.
8060         (altivec_vsumsws_direct): New.
8061         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
8062         gen_altivec_vsumsws.
8063
8064 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8065
8066         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
8067         vtable_pointer_value_to_binfo): New functions.
8068         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
8069         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
8070
8071 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
8072
8073         * config/nios2/nios2.md (load_got_register): Initialize GOT
8074         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
8075         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
8076
8077 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
8078
8079         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
8080         preserverd by passthrough, do not propagate the type.
8081
8082 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8083
8084         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
8085         (mips_atomic_assign_expand_fenv): New function.
8086         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
8087
8088 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8089
8090         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
8091         (__builtin_mips_set_fcsr): Likewise.
8092         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
8093         MIPS_USI_FTYPE_VOID.
8094         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
8095         (mips16_expand_set_fcsr): Likewise.
8096         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
8097         (mips16_set_fcsr_stub): Likewise.
8098         (mips16_get_fcsr_one_only_stub): New class.
8099         (mips16_set_fcsr_one_only_stub): Likewise.
8100         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
8101         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
8102         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
8103         (hard_float): New availability predicate.
8104         (mips_builtins): Add get_fcsr and set_fcsr.
8105         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
8106         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
8107         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
8108         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
8109         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
8110         patterns.
8111
8112 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
8113
8114         * config/mips/mips.c (mips_one_only_stub): New class.
8115         (mips_need_mips16_rdhwr_p): Replace with...
8116         (mips16_rdhwr_stub): ...this new variable.
8117         (mips16_stub_call_address): New function.
8118         (mips16_rdhwr_one_only_stub): New class.
8119         (mips_expand_thread_pointer): Use mips16_stub_call_address.
8120         (mips_output_mips16_rdhwr): Delete.
8121         (mips_finish_stub): New function.
8122         (mips_code_end): Use it to handle rdhwr stubs.
8123
8124 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
8125
8126         PR target/60017
8127         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
8128         when calculating size of integer atomic types.
8129
8130 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
8131
8132         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
8133
8134 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
8135
8136         PR tree-optimization/60003
8137         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
8138         * profile.c (branch_prob): Use gimple_call_builtin_p
8139         to check for BUILT_IN_SETJMP_RECEIVER.
8140         * tree-inline.c (copy_bb): Call notice_special_calls.
8141
8142 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
8143
8144         PR bootstrap/59985
8145         * lra-constraints.c (process_alt_operands): Update reload_sum only
8146         on the first pass.
8147
8148 2014-01-31  Richard Henderson  <rth@redhat.com>
8149
8150         PR middle-end/60004
8151         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
8152         until after else_eh is processed.
8153
8154 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8155
8156         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
8157         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
8158         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
8159         in smmintrin.h, remove them.
8160         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
8161         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
8162         * config/i386/i386.md (ROUND_SAE): Fix value.
8163         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
8164         (const48_operand): New.
8165         * config/i386/subst.md (round), (round_expand): Use
8166         const_4_or_8_to_11_operand.
8167         (round_saeonly), (round_saeonly_expand): Use const48_operand.
8168
8169 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
8170
8171         * config/i386/constraints.md (Yk): Swap meaning with k.
8172         * config/i386/i386.md (movhi_internal): Change Yk to k.
8173         (movqi_internal): Ditto.
8174         (*k<logic><mode>): Ditto.
8175         (*andhi_1): Ditto.
8176         (*andqi_1): Ditto.
8177         (kandn<mode>): Ditto.
8178         (*<code>hi_1): Ditto.
8179         (*<code>qi_1): Ditto.
8180         (kxnor<mode>): Ditto.
8181         (kortestzhi): Ditto.
8182         (kortestchi): Ditto.
8183         (kunpckhi): Ditto.
8184         (*one_cmplhi2_1): Ditto.
8185         (*one_cmplqi2_1): Ditto.
8186         * config/i386/sse.md (): Change k to Yk.
8187         (avx512f_load<mode>_mask): Ditto.
8188         (avx512f_blendm<mode>): Ditto.
8189         (avx512f_store<mode>_mask): Ditto.
8190         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
8191         (avx512f_storedqu<mode>_mask): Ditto.
8192         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
8193         Ditto.
8194         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
8195         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
8196         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
8197         (avx512f_maskcmp<mode>3): Ditto.
8198         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
8199         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
8200         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
8201         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
8202         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
8203         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
8204         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
8205         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
8206         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
8207         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
8208         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
8209         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
8210         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
8211         (vec_extract_lo_<mode>_maskm): Ditto.
8212         (vec_extract_hi_<mode>_maskm): Ditto.
8213         (avx512f_vternlog<mode>_mask): Ditto.
8214         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
8215         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
8216         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
8217         (avx512f_<code>v8div16qi2_mask): Ditto.
8218         (avx512f_<code>v8div16qi2_mask_store): Ditto.
8219         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
8220         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
8221         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
8222         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
8223         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
8224         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8225         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
8226         (*avx512pf_scatterpf<mode>df_mask): Ditto.
8227         (avx512cd_maskb_vec_dupv8di): Ditto.
8228         (avx512cd_maskw_vec_dupv16si): Ditto.
8229         (avx512f_vpermi2var<mode>3_maskz): Ditto.
8230         (avx512f_vpermi2var<mode>3_mask): Ditto.
8231         (avx512f_vpermi2var<mode>3_mask): Ditto.
8232         (avx512f_vpermt2var<mode>3_maskz): Ditto.
8233         (*avx512f_gathersi<mode>): Ditto.
8234         (*avx512f_gathersi<mode>_2): Ditto.
8235         (*avx512f_gatherdi<mode>): Ditto.
8236         (*avx512f_gatherdi<mode>_2): Ditto.
8237         (*avx512f_scattersi<mode>): Ditto.
8238         (*avx512f_scatterdi<mode>): Ditto.
8239         (avx512f_compress<mode>_mask): Ditto.
8240         (avx512f_compressstore<mode>_mask): Ditto.
8241         (avx512f_expand<mode>_mask): Ditto.
8242         * config/i386/subst.md (mask): Change k to Yk.
8243         (mask_scalar_merge): Ditto.
8244         (sd): Ditto.
8245
8246 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
8247
8248         * doc/extend.texi (Vector Extensions): Document ?: in C++.
8249
8250 2014-01-31  Richard Biener  <rguenther@suse.de>
8251
8252         PR middle-end/59990
8253         * builtins.c (fold_builtin_memory_op): Make sure to not
8254         use a floating-point mode or a boolean or enumeral type for
8255         the copy operation.
8256
8257 2014-01-30  DJ Delorie  <dj@redhat.com>
8258
8259         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
8260         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
8261         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
8262         whenever main() has an epilogue.
8263
8264 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8265
8266         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
8267         unused variable "field".
8268         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
8269         (vsx_mergeh_<mode>): Likewise.
8270         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
8271         (altivec_vmrghh): Likewise.
8272         (altivec_vmrghw): Likewise.
8273         (altivec_vmrglb): Likewise.
8274         (altivec_vmrglh): Likewise.
8275         (altivec_vmrglw): Likewise.
8276         (altivec_vspltb): Add missing uses.
8277         (altivec_vsplth): Likewise.
8278         (altivec_vspltw): Likewise.
8279         (altivec_vspltsf): Likewise.
8280
8281 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8282
8283         PR target/59923
8284         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
8285         frame related instructions.
8286
8287 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
8288
8289         PR rtl-optimization/59959
8290         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
8291         any reload of register whose subreg is invalid.
8292
8293 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8294
8295         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
8296         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
8297         Add missing return type - void.
8298
8299 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8300
8301         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
8302         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
8303         remove element index adjustment for endian (now handled in vsx.md
8304         and altivec.md).
8305         (altivec_expand_vec_perm_const): Use
8306         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
8307         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
8308         (vsx_xxspltw_<mode>): Adjust element index for little endian.
8309         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
8310         define_expand and a new define_insn *altivec_vspltb_internal;
8311         adjust for -maltivec=be on a little endian target.
8312         (altivec_vspltb_direct): New.
8313         (altivec_vsplth): Divide into a define_expand and a new
8314         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
8315         little endian target.
8316         (altivec_vsplth_direct): New.
8317         (altivec_vspltw): Divide into a define_expand and a new
8318         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
8319         little endian target.
8320         (altivec_vspltw_direct): New.
8321         (altivec_vspltsf): Divide into a define_expand and a new
8322         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
8323         a little endian target.
8324
8325 2014-01-30  Richard Biener  <rguenther@suse.de>
8326
8327         PR tree-optimization/59993
8328         * tree-ssa-forwprop.c (associate_pointerplus): Check we
8329         can propagate form the earlier stmt and avoid the transform
8330         when the intermediate result is needed.
8331
8332 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
8333
8334         * README.Portability: Fix typo.
8335
8336 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
8337
8338         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
8339         comparison_operator with ordered_comparison_operator.
8340
8341 2014-01-30  Nick Clifton  <nickc@redhat.com>
8342
8343         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
8344         Rename to mn10300_store_multiple_regs.
8345         * config/mn10300/mn10300.c: Likewise.
8346         * config/mn10300/mn10300.md (store_movm): Fix typo: call
8347         store_multiple_regs.
8348         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
8349         Call mn10300_store_multiple_regs.
8350
8351 2014-01-30  Nick Clifton  <nickc@redhat.com>
8352             DJ Delorie  <dj@redhat.com>
8353
8354         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
8355         %fp 2 to keep registers after it properly word-aligned.
8356         (rl78_alloc_physical_registers_umul): Handle the case where both
8357         input operands are the same.
8358
8359 2014-01-30  Richard Biener  <rguenther@suse.de>
8360
8361         PR tree-optimization/59903
8362         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
8363         check properly.
8364
8365 2014-01-30  Jason Merrill  <jason@redhat.com>
8366
8367         PR c++/59633
8368         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
8369
8370         PR c++/59645
8371         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
8372
8373 2014-01-30  Richard Biener  <rguenther@suse.de>
8374
8375         PR tree-optimization/59951
8376         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
8377
8378 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
8379
8380         PR target/59784
8381         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
8382         SFmode to DFmode case.
8383
8384 2014-01-29  DJ Delorie  <dj@redhat.com>
8385
8386         * config/msp430/msp430.opt (-minrt): New.
8387         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
8388         if -minrt given.
8389         (ENDFILE_SPEC): Likewise.
8390
8391 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
8392
8393         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
8394         (estimate_function_body_sizes): Use it.
8395
8396 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
8397
8398         PR c++/58561
8399         * dwarf2out.c (is_cxx_auto): New.
8400         (is_base_type): Use it.
8401         (gen_type_die_with_usage): Likewise.
8402
8403 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8404
8405         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
8406         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
8407         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
8408         -maltivec=be with LE targets.
8409         (vsx_mergeh_<mode>): Likewise.
8410         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
8411         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
8412         (altivec_vmrghb): Replace with define_expand and new
8413         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
8414         (altivec_vmrghb_direct): New define_insn.
8415         (altivec_vmrghh): Replace with define_expand and new
8416         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
8417         (altivec_vmrghh_direct): New define_insn.
8418         (altivec_vmrghw): Replace with define_expand and new
8419         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
8420         (altivec_vmrghw_direct): New define_insn.
8421         (*altivec_vmrghsf): Adjust for endianness.
8422         (altivec_vmrglb): Replace with define_expand and new
8423         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
8424         (altivec_vmrglb_direct): New define_insn.
8425         (altivec_vmrglh): Replace with define_expand and new
8426         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
8427         (altivec_vmrglh_direct): New define_insn.
8428         (altivec_vmrglw): Replace with define_expand and new
8429         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
8430         (altivec_vmrglw_direct): New define_insn.
8431         (*altivec_vmrglsf): Adjust for endianness.
8432         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8433         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8434         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8435         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8436         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8437         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8438         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8439         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8440
8441 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8442
8443         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
8444         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
8445         whitespace.
8446
8447 2014-01-29  Richard Biener  <rguenther@suse.de>
8448
8449         PR tree-optimization/58742
8450         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
8451         associate_pointerplus_align.
8452         (associate_pointerplus_diff): New function.
8453         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
8454         and associate_pointerplus_diff.
8455
8456 2014-01-29  Richard Biener  <rguenther@suse.de>
8457
8458         * lto-streamer.h (LTO_major_version): Bump to 3.
8459         (LTO_minor_version): Reset to 0.
8460
8461 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
8462
8463         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
8464         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
8465         (arm_file_start): Generate correct asm header for armv7ve.
8466         * config/arm/bpabi.h: Add multilib support for armv7ve.
8467         * config/arm/driver-arm.c: Change the architectures of cortex-a7
8468         and cortex-a15 to armv7ve.
8469         * config/arm/t-aprofile: Add multilib support for armv7ve.
8470         * doc/invoke.texi: Document -march=armv7ve.
8471
8472 2014-01-29  Richard Biener  <rguenther@suse.de>
8473
8474         PR tree-optimization/58742
8475         * tree-ssa-forwprop.c (associate_plusminus): Return true
8476         if we changed sth, defer EH cleanup to ...
8477         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
8478         (simplify_mult): New function.
8479
8480 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8481
8482         PR middle-end/59917
8483         PR tree-optimization/59920
8484         * tree.c (build_common_builtin_nodes): Remove
8485         __builtin_setjmp_dispatcher initialization.
8486         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
8487         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
8488         instead of gsi_after_labels + manually skipping debug stmts.
8489         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
8490         ignore bbs with IFN_ABNORMAL_DISPATCHER.
8491         * tree-inline.c (copy_edges_for_bb): Remove
8492         can_make_abnormal_goto argument, instead add abnormal_goto_dest
8493         argument.  Ignore computed_goto_p stmts.  Don't call
8494         make_abnormal_goto_edges.  If a call might need abnormal edges
8495         for non-local gotos, see if it already has an edge to
8496         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
8497         with true argument, don't do anything then, otherwise add
8498         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
8499         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
8500         caller.
8501         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
8502         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
8503         (lower_stmt): Don't set data->calls_builtin_setjmp.
8504         (lower_builtin_setjmp): Adjust comment.
8505         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
8506         * tree-cfg.c (found_computed_goto): Remove.
8507         (factor_computed_gotos): Remove.
8508         (make_goto_expr_edges): Return bool, true for computed gotos.
8509         Don't call make_abnormal_goto_edges.
8510         (build_gimple_cfg): Don't set found_computed_goto, don't call
8511         factor_computed_gotos.
8512         (computed_goto_p): No longer static.
8513         (make_blocks): Don't set found_computed_goto.
8514         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
8515         (make_edges): If make_goto_expr_edges returns true, push bb
8516         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
8517         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
8518         vector.  Record mapping between bbs and OpenMP regions if there
8519         are any, adjust make_gimple_omp_edges caller.  Call
8520         handle_abnormal_edges.
8521         (make_abnormal_goto_edges): Remove.
8522         * tree-cfg.h (make_abnormal_goto_edges): Remove.
8523         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
8524         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
8525         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
8526         * internal-fn.def (ABNORMAL_DISPATCHER): New.
8527         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
8528         filling *region also set *region_idx to (*region)->entry->index.
8529
8530         PR other/58712
8531         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
8532         For REGs set ORIGINAL_REGNO.
8533
8534 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
8535
8536         * doc/md.texi: Mention that a target shouldn't implement
8537         vec_widen_(s|u)mul_even/odd pair if it is less efficient
8538         than hi/lo pair.
8539
8540 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8541
8542         PR tree-optimization/59594
8543         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
8544         a copy of the datarefs vector rather than the vector itself.
8545
8546 2014-01-28  Jason Merrill  <jason@redhat.com>
8547
8548         PR c++/53756
8549         * dwarf2out.c (auto_die): New static.
8550         (gen_type_die_with_usage): Handle C++1y 'auto'.
8551         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
8552         on definition.
8553
8554 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
8555
8556         PR target/59672
8557         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
8558         (SPEC_X32): Likewise.
8559         (SPEC_64): Likewise.
8560         * config/i386/i386.c (ix86_option_override_internal): Turn off
8561         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
8562         for TARGET_16BIT.
8563         (x86_file_start): Output .code16gcc for TARGET_16BIT.
8564         * config/i386/i386.h (TARGET_16BIT): New macro.
8565         (TARGET_16BIT_P): Likewise.
8566         * config/i386/i386.opt: Add m16.
8567         * doc/invoke.texi: Document -m16.
8568
8569 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
8570
8571         PR preprocessor/59935
8572         * input.c (location_get_source_line): Bail out on when line number
8573         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
8574
8575 2014-01-28  Richard Biener  <rguenther@suse.de>
8576
8577         PR tree-optimization/58742
8578         * tree-ssa-forwprop.c (associate_plusminus): Handle
8579         pointer subtraction of the form (T)(P + A) - (T)P.
8580
8581 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8582
8583         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
8584         at const_int_cost.
8585
8586 2014-01-28  Richard Biener  <rguenther@suse.de>
8587
8588         Revert
8589         2014-01-28  Richard Biener  <rguenther@suse.de>
8590
8591         PR rtl-optimization/45364
8592         PR rtl-optimization/59890
8593         * var-tracking.c (local_get_addr_clear_given_value): Handle
8594         already cleared slot.
8595         (val_reset): Handle not allocated local_get_addr_cache.
8596         (vt_find_locations): Use post-order on the inverted CFG.
8597
8598 2014-01-28  Richard Biener  <rguenther@suse.de>
8599
8600         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
8601
8602 2014-01-28  Richard Biener  <rguenther@suse.de>
8603
8604         PR rtl-optimization/45364
8605         PR rtl-optimization/59890
8606         * var-tracking.c (local_get_addr_clear_given_value): Handle
8607         already cleared slot.
8608         (val_reset): Handle not allocated local_get_addr_cache.
8609         (vt_find_locations): Use post-order on the inverted CFG.
8610
8611 2014-01-28  Alan Modra  <amodra@gmail.com>
8612
8613         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
8614         * configure.ac <recursive call for build != host>: Define
8615         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
8616         and LD_FOR_BUILD too.
8617         * configure: Regenerate.
8618
8619 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
8620
8621         * config/i386/i386.c (get_builtin_code_for_version): Separate
8622         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
8623         Broadwell from Haswell.
8624
8625 2014-01-27  Steve Ellcey  <sellcey@mips.com>
8626
8627         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
8628         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
8629         * config/mips/mips.c (mips_option_override): Change setting
8630         of TARGET_DSP.
8631         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
8632         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
8633         Change from Mask to Var.
8634
8635 2014-01-27  Jeff Law  <law@redhat.com>
8636
8637         * ipa-inline.c (inline_small_functions): Fix typo.
8638
8639 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
8640
8641         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
8642         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
8643         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
8644         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
8645         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
8646         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
8647         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
8648         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
8649         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
8650         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
8651         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
8652         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
8653         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
8654         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
8655         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
8656         (_mm512_storeu_epi64): Ditto.
8657         (_mm512_cmpge_epi32_mask): Ditto.
8658         (_mm512_cmpge_epu32_mask): Ditto.
8659         (_mm512_cmpge_epi64_mask): Ditto.
8660         (_mm512_cmpge_epu64_mask): Ditto.
8661         (_mm512_cmple_epi32_mask): Ditto.
8662         (_mm512_cmple_epu32_mask): Ditto.
8663         (_mm512_cmple_epi64_mask): Ditto.
8664         (_mm512_cmple_epu64_mask): Ditto.
8665         (_mm512_cmplt_epi32_mask): Ditto.
8666         (_mm512_cmplt_epu32_mask): Ditto.
8667         (_mm512_cmplt_epi64_mask): Ditto.
8668         (_mm512_cmplt_epu64_mask): Ditto.
8669         (_mm512_cmpneq_epi32_mask): Ditto.
8670         (_mm512_cmpneq_epu32_mask): Ditto.
8671         (_mm512_cmpneq_epi64_mask): Ditto.
8672         (_mm512_cmpneq_epu64_mask): Ditto.
8673         (_mm512_expand_pd): Ditto.
8674         (_mm512_expand_ps): Ditto.
8675         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
8676         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
8677         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
8678         * config/i386/i386.c (ix86_builtins): Add
8679         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
8680         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
8681         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
8682         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
8683         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
8684         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
8685         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
8686         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
8687         IX86_BUILTIN_PMOVUSQW512_MEM.
8688         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
8689         __builtin_ia32_pmovsqd512mem_mask,
8690         __builtin_ia32_pmovqd512mem_mask,
8691         __builtin_ia32_pmovusqw512mem_mask,
8692         __builtin_ia32_pmovsqw512mem_mask,
8693         __builtin_ia32_pmovqw512mem_mask,
8694         __builtin_ia32_pmovusdw512mem_mask,
8695         __builtin_ia32_pmovsdw512mem_mask,
8696         __builtin_ia32_pmovdw512mem_mask,
8697         __builtin_ia32_pmovqb512mem_mask,
8698         __builtin_ia32_pmovusqb512mem_mask,
8699         __builtin_ia32_pmovsqb512mem_mask,
8700         __builtin_ia32_pmovusdb512mem_mask,
8701         __builtin_ia32_pmovsdb512mem_mask,
8702         __builtin_ia32_pmovdb512mem_mask.
8703         (bdesc_args): Add __builtin_ia32_expanddf512,
8704         __builtin_ia32_expandsf512.
8705         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
8706         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
8707         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
8708         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
8709         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
8710         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
8711         (avx512f_<code>v8div16qi2_mask_store): This.
8712         (avx512f_expand<mode>): New.
8713
8714 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
8715
8716         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
8717         New.
8718         (_mm512_mask_prefetch_i64gather_pd): Ditto.
8719         (_mm512_prefetch_i32scatter_pd): Ditto.
8720         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
8721         (_mm512_prefetch_i64scatter_pd): Ditto.
8722         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
8723         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
8724         (_mm512_mask_prefetch_i64gather_ps): Ditto.
8725         (_mm512_prefetch_i32scatter_ps): Ditto.
8726         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
8727         (_mm512_prefetch_i64scatter_ps): Ditto.
8728         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
8729         * config/i386/i386-builtin-types.def: Define
8730         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
8731         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
8732         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
8733         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
8734         IX86_BUILTIN_SCATTERPFQPD.
8735         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
8736         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
8737         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
8738         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
8739         __builtin_ia32_scatterpfqps.
8740         (ix86_expand_builtin): Expand new built-ins.
8741         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
8742         fix memory access data type.
8743         (*avx512pf_gatherpf<mode>_mask): Ditto.
8744         (*avx512pf_gatherpf<mode>): Ditto.
8745         (avx512pf_scatterpf<mode>): Ditto.
8746         (*avx512pf_scatterpf<mode>_mask): Ditto.
8747         (*avx512pf_scatterpf<mode>): Ditto.
8748         (GATHER_SCATTER_SF_MEM_MODE): New.
8749         (avx512pf_gatherpf<mode>df): Ditto.
8750         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8751         (*avx512pf_scatterpf<mode>df): Ditto.
8752
8753 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
8754
8755         PR bootstrap/59934
8756         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
8757         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
8758         reached.
8759
8760 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8761
8762         * common/config/arm/arm-common.c
8763         (arm_rewrite_mcpu): Handle multiple names.
8764         * config/arm/arm.h
8765         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8766
8767 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8768
8769         * gimple-builder.h (create_gimple_tmp): Delete.
8770
8771 2014-01-27  Christian Bruel  <christian.bruel@st.com>
8772
8773         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
8774         words comparisons.
8775
8776 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
8777
8778         * config/pa/pa.md (call): Generate indirect long calls to non-local
8779         functions when outputing 32-bit code.
8780         (call_value): Likewise except for special call to buggy powf function.
8781
8782         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
8783         portable runtime and PIC indirect calls.
8784         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
8785         and PIC call sequences.  Use ldo instead of blr to set return register
8786         in PIC call sequence.
8787
8788 2014-01-25  Walter Lee  <walt@tilera.com>
8789
8790         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
8791         avoid clobbering a live register.
8792
8793 2014-01-25  Walter Lee  <walt@tilera.com>
8794
8795         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
8796         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
8797         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
8798         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
8799
8800 2014-01-25  Walter Lee  <walt@tilera.com>
8801
8802         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
8803         arguments on even registers.
8804         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
8805         STACK_BOUNDARY.
8806         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
8807         (BIGGEST_ALIGNMENT): Ditto.
8808         (BIGGEST_FIELD_ALIGNMENT): Ditto.
8809
8810 2014-01-25  Walter Lee  <walt@tilera.com>
8811
8812         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
8813         insns before bundling.
8814         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
8815
8816 2014-01-25  Walter Lee  <walt@tilera.com>
8817
8818         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
8819         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
8820         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
8821
8822 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8823
8824         * config/mips/constraints.md (kl): Delete.
8825         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
8826         define expands, using...
8827         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
8828         instructions for MIPS16.
8829         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
8830         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
8831
8832 2014-01-25  Walter Lee  <walt@tilera.com>
8833
8834         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
8835         (clzdi2): Ditto.
8836         (ffsdi2): Ditto.
8837
8838 2014-01-25  Walter Lee  <walt@tilera.com>
8839
8840         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
8841         (TARGET_EXPAND_TO_RTL_HOOK): Define.
8842
8843 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8844
8845         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
8846         Handle XOR.
8847
8848 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
8849
8850         * print-rtl.c (in_call_function_usage): New var.
8851         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
8852         EXPR_LIST mode as mode and not as reg note name.
8853
8854         PR middle-end/59561
8855         * cfgloopmanip.c (copy_loop_info): If
8856         loop->warned_aggressive_loop_optimizations, make sure
8857         the flag is set in target loop too.
8858
8859 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8860
8861         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
8862         flag_cilkplus.
8863         * builtins.def: Likewise.
8864         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
8865         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
8866         * ira.c (ira_setup_eliminable_regset): Likewise.
8867         * omp-low.c (gate_expand_omp): Likewise.
8868         (execute_lower_omp): Likewise.
8869         (diagnose_sb_0): Likewise.
8870         (gate_diagnose_omp_blocks): Likewise.
8871         (simd_clone_clauses_extract): Likewise.
8872         (gate): Likewise.
8873
8874 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8875
8876         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
8877         correction for little endian...
8878         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
8879         here.
8880
8881 2014-01-24  Jeff Law  <law@redhat.com>
8882
8883         PR tree-optimization/59919
8884         * tree-vrp.c (find_assert_locations_1): Do not register asserts
8885         for non-returning calls.
8886
8887 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
8888
8889         * common/config/aarch64/aarch64-common.c
8890         (aarch64_rewrite_mcpu): Handle multiple names.
8891         * config/aarch64/aarch64.h
8892         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8893
8894 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
8895
8896         * input.c (add_file_to_cache_tab): Handle the case where fopen
8897         returns NULL.
8898
8899 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
8900
8901         PR target/59929
8902         * config/i386/i386.md (pushsf splitter): Get stack adjustment
8903         from push operand if code of push isn't PRE_DEC.
8904
8905 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
8906
8907         PR target/59909
8908         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
8909         -mquad-memory-atomic.  Update -mquad-memory documentation to say
8910         it is only used for non-atomic loads/stores.
8911
8912         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
8913         -mquad-memory or -mquad-memory-atomic switches.
8914
8915         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
8916         -mquad-memory-atomic to ISA 2.07 support.
8917
8918         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
8919         to separate support of normal quad word memory operations (ldq, stq)
8920         from the atomic quad word memory operations.
8921
8922         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8923         support to separate non-atomic quad word operations from atomic
8924         quad word operations.  Disable non-atomic quad word operations in
8925         little endian mode so that we don't have to swap words after the
8926         load and before the store.
8927         (quad_load_store_p): Add comment about atomic quad word support.
8928         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
8929         options printed with -mdebug=reg.
8930
8931         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
8932         -mquad-memory-atomic as the test for whether we have quad word
8933         atomic instructions.
8934         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
8935         or -mp8-vector are used, allow byte/half-word atomic operations.
8936
8937         * config/rs6000/sync.md (load_lockedti): Insure that the address
8938         is a proper indexed or indirect address for the lqarx instruction.
8939         On little endian systems, swap the hi/lo registers after the lqarx
8940         instruction.
8941         (load_lockedpti): Use indexed_or_indirect_operand predicate to
8942         insure the address is valid for the lqarx instruction.
8943         (store_conditionalti): Insure that the address is a proper indexed
8944         or indirect address for the stqcrx. instruction.  On little endian
8945         systems, swap the hi/lo registers before doing the stqcrx.
8946         instruction.
8947         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
8948         insure the address is valid for the stqcrx. instruction.
8949
8950         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8951         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
8952         type of quad memory support is available.
8953
8954 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
8955
8956         PR regression/59915
8957         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
8958         there is a danger of looping.
8959
8960 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
8961
8962         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8963         force flag_ira_loop_pressure if set via command line.
8964
8965 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8966
8967         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
8968         (ashr_simd): New builtin handling DI mode.
8969         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
8970         (aarch64_sshr_simddi): New match pattern.
8971         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
8972         (vshrd_n_s64): Likewise.
8973         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
8974
8975 2014-01-23  Nick Clifton  <nickc@redhat.com>
8976
8977         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
8978         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
8979         favour of mcu specific scripts.
8980         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
8981         430x multilibs.
8982
8983 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
8984             Alex Velenko  <Alex.Velenko@arm.com>
8985
8986         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
8987         (vaddv_s16): Likewise.
8988         (vaddv_s32): Likewise.
8989         (vaddv_u8): Likewise.
8990         (vaddv_u16): Likewise.
8991         (vaddv_u32): Likewise.
8992         (vaddvq_s8): Likewise.
8993         (vaddvq_s16): Likewise.
8994         (vaddvq_s32): Likewise.
8995         (vaddvq_s64): Likewise.
8996         (vaddvq_u8): Likewise.
8997         (vaddvq_u16): Likewise.
8998         (vaddvq_u32): Likewise.
8999         (vaddvq_u64): Likewise.
9000         (vaddv_f32): Likewise.
9001         (vaddvq_f32): Likewise.
9002         (vaddvq_f64): Likewise.
9003         (vmaxv_f32): Likewise.
9004         (vmaxv_s8): Likewise.
9005         (vmaxv_s16): Likewise.
9006         (vmaxv_s32): Likewise.
9007         (vmaxv_u8): Likewise.
9008         (vmaxv_u16): Likewise.
9009         (vmaxv_u32): Likewise.
9010         (vmaxvq_f32): Likewise.
9011         (vmaxvq_f64): Likewise.
9012         (vmaxvq_s8): Likewise.
9013         (vmaxvq_s16): Likewise.
9014         (vmaxvq_s32): Likewise.
9015         (vmaxvq_u8): Likewise.
9016         (vmaxvq_u16): Likewise.
9017         (vmaxvq_u32): Likewise.
9018         (vmaxnmv_f32): Likewise.
9019         (vmaxnmvq_f32): Likewise.
9020         (vmaxnmvq_f64): Likewise.
9021         (vminv_f32): Likewise.
9022         (vminv_s8): Likewise.
9023         (vminv_s16): Likewise.
9024         (vminv_s32): Likewise.
9025         (vminv_u8): Likewise.
9026         (vminv_u16): Likewise.
9027         (vminv_u32): Likewise.
9028         (vminvq_f32): Likewise.
9029         (vminvq_f64): Likewise.
9030         (vminvq_s8): Likewise.
9031         (vminvq_s16): Likewise.
9032         (vminvq_s32): Likewise.
9033         (vminvq_u8): Likewise.
9034         (vminvq_u16): Likewise.
9035         (vminvq_u32): Likewise.
9036         (vminnmv_f32): Likewise.
9037         (vminnmvq_f32): Likewise.
9038         (vminnmvq_f64): Likewise.
9039
9040 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
9041
9042         * config/aarch64/aarch64-simd.md
9043         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
9044         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
9045         (*aarch64_mul3_elt<mode>): Likewise.
9046         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
9047         (*aarch64_mul3_elt_to_64v2df): Likewise.
9048         (*aarch64_mla_elt<mode>): Likewise.
9049         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
9050         (*aarch64_mls_elt<mode>): Likewise.
9051         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
9052         (*aarch64_fma4_elt<mode>): Likewise.
9053         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
9054         (*aarch64_fma4_elt_to_64v2df): Likewise.
9055         (*aarch64_fnma4_elt<mode>): Likewise.
9056         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
9057         (*aarch64_fnma4_elt_to_64v2df): Likewise.
9058         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
9059         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
9060         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9061         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
9062         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
9063         (aarch64_sqdmull_lane<mode>_internal): Likewise.
9064         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
9065
9066 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9067
9068         * config/aarch64/aarch64-simd.md
9069         (aarch64_be_checked_get_lane<mode>): New define_expand.
9070         * config/aarch64/aarch64-simd-builtins.def
9071         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
9072         New builtin definition.
9073         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
9074         Use new safe be builtin.
9075
9076 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
9077
9078         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
9079         New define_insn.
9080         (aarch64_be_st1<mode>): Likewise.
9081         (aarch_ld1<VALL:mode>): Define_expand modified.
9082         (aarch_st1<VALL:mode>): Likewise.
9083         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
9084         (UNSPEC_ST1): Likewise.
9085
9086 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
9087
9088         * config/microblaze/microblaze.md: Add trap insn and attribute
9089
9090 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
9091
9092         PR preprocessor/58580
9093         * input.h (location_get_source_line): Take an additional line_size
9094         parameter.
9095         (void diagnostics_file_cache_fini): Declare new function.
9096         * input.c (struct fcache): New type.
9097         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
9098         New static constants.
9099         (diagnostic_file_cache_init, total_lines_num)
9100         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
9101         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
9102         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
9103         (get_next_line, read_next_line, goto_next_line, read_line_num):
9104         New static function definitions.
9105         (diagnostic_file_cache_fini): New function.
9106         (location_get_source_line): Take an additional output line_len
9107         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
9108         read_line_num.
9109         * diagnostic.c (diagnostic_finish): Call
9110         diagnostic_file_cache_fini.
9111         (adjust_line): Take an additional input parameter for the length
9112         of the line, rather than calculating it with strlen.
9113         (diagnostic_show_locus): Adjust the use of
9114         location_get_source_line and adjust_line with respect to their new
9115         signature.  While displaying a line now, do not stop at the first
9116         null byte.  Rather, display the zero byte as a space and keep
9117         going until we reach the size of the line.
9118         * Makefile.in: Add vec.o to OBJS-libcommon
9119
9120 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9121             Ilya Tocar     <ilya.tocar@intel.com>
9122
9123         * config/i386/avx512fintrin.h (_mm512_kmov): New.
9124         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
9125         (__builtin_ia32_kmov16): Ditto.
9126         * config/i386/i386.md (UNSPEC_KMOV): New.
9127         (kmovw): Ditto.
9128
9129 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9130
9131         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
9132         (_mm512_storeu_si512): Ditto.
9133
9134 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
9135
9136         PR target/52125
9137         * rtl.h (get_referenced_operands): Declare.
9138         * recog.c (get_referenced_operands): New function.
9139         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
9140         operands have been referenced when recording LO_SUM references.
9141
9142 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
9143
9144         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
9145
9146 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9147
9148         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
9149         Enable for generic and recent AMD targets.
9150
9151 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
9152
9153         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
9154         ARG_SIZE note when adjustment was eliminated.
9155
9156 2014-01-22  Jeff Law  <law@redhat.com>
9157
9158         PR tree-optimization/59597
9159         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
9160         in file.  Accept new argument REGISTERING and use it to modify
9161         dump output appropriately.
9162         (register_jump_thread): Corresponding changes.
9163         (mark_threaded_blocks): Reinstate code to cancel unprofitable
9164         thread paths involving joiner blocks.  Add code to dump cancelled
9165         jump threading paths.
9166
9167 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
9168
9169         PR rtl-optimization/59477
9170         * lra-constraints.c (inherit_in_ebb): Process call for living hard
9171         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
9172
9173 2014-01-22  Tom Tromey  <tromey@redhat.com>
9174
9175         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
9176         PARAMS.
9177         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
9178
9179 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9180
9181         PR rtl-optimization/59896
9182         * lra-constraints.c (process_alt_operands): Check unused note for
9183         matched operands of insn with no output reloads.
9184
9185 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
9186
9187         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
9188         (mips_move_from_gpr_cost): Likewise.
9189
9190 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
9191
9192         PR rtl-optimization/59858
9193         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
9194         ira_class_hard_regs_num.
9195         (process_alt_operands): Increase reject for dying matched operand.
9196
9197 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
9198
9199         PR target/59003
9200         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
9201         smaller than size, perform several stores or loads and stores
9202         at dst + count - size to store or copy all of size bytes, rather
9203         than just last modesize bytes.
9204
9205 2014-01-20  DJ Delorie  <dj@redhat.com>
9206
9207         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
9208         that CLOBBERs are REGs before propogating their values.
9209
9210 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
9211
9212         PR middle-end/59789
9213         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
9214         (cgraph_inline_failed_type): New function.
9215         * cgraph.h (DEFCIFCODE): Add type.
9216         (cgraph_inline_failed_type_t): New enum.
9217         (cgraph_inline_failed_type): New prototype.
9218         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
9219         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
9220         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
9221         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
9222         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
9223         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
9224         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
9225         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
9226         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
9227         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
9228         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
9229         OPTIMIZATION_MISMATCH.
9230         * tree-inline.c (expand_call_inline): Emit errors during
9231         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
9232
9233 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9234
9235         PR target/59685
9236         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
9237         mode attribute in insn output.
9238
9239 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
9240
9241         * output.h (output_constant): Delete.
9242         * varasm.c (output_constant): Make private.
9243
9244 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
9245
9246         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
9247
9248 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9249
9250         PR middle-end/59860
9251         * tree.h (fold_builtin_strcat): New prototype.
9252         * builtins.c (fold_builtin_strcat): No longer static.  Add len
9253         argument, if non-NULL, don't call c_strlen.  Optimize
9254         directly into __builtin_memcpy instead of __builtin_strcpy.
9255         (fold_builtin_2): Adjust fold_builtin_strcat caller.
9256         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
9257
9258 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9259
9260         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9261         for SImode_address_operand operands, having only a REG argument.
9262
9263 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9264
9265         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
9266         loader name using mbig-endian.
9267         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
9268
9269 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9270
9271         * doc/invoke.texi (-march): Clarify documentation for AArch64.
9272         (-mtune): Likewise.
9273         (-mcpu): Likewise.
9274
9275 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
9276
9277         * config/aarch64/aarch64-protos.h
9278         (aarch64_cannot_change_mode_class_ptr): Declare.
9279         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
9280         aarch64_cannot_change_mode_class_ptr): New.
9281         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
9282         backend hook aarch64_cannot_change_mode_class.
9283
9284 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9285
9286         * common/config/aarch64/aarch64-common.c
9287         (aarch64_handle_option): Don't handle any option order logic here.
9288         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
9289         selected_cpu, warn on architecture version mismatch.
9290         (aarch64_override_options): Fix parsing order for option strings.
9291
9292 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9293             Iain Sandoe  <iain@codesourcery.com>
9294
9295         PR bootstrap/59496
9296         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
9297         warning.  Amend comment to reflect current functionality.
9298
9299 2014-01-20  Richard Biener  <rguenther@suse.de>
9300
9301         PR middle-end/59860
9302         * builtins.c (fold_builtin_strcat): Remove case better handled
9303         by tree-ssa-strlen.c.
9304
9305 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
9306
9307         * config/aarch64/aarch64.opt
9308         (mcpu, march, mtune): Make case-insensitive.
9309
9310 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9311
9312         PR target/59880
9313         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9314         if operands[1] is a REG or ZERO_EXTEND of a REG.
9315
9316 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
9317
9318         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
9319
9320 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
9321
9322         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
9323         long non-pic millicode calls.
9324
9325 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9326
9327         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
9328
9329 2014-01-19  Kito Cheng  <kito@0xlab.org>
9330
9331         * builtins.c (expand_movstr): Check movstr expand done or fail.
9332
9333 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9334             H.J. Lu  <hongjiu.lu@intel.com>
9335
9336         PR target/59379
9337         * config/i386/i386.md (*lea<mode>): Zero-extend return register
9338         to DImode for zero-extended addresses.
9339
9340 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
9341
9342         PR rtl-optimization/57763
9343         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
9344         on the new indirect jump_insn and increment LABEL_NUSES (label).
9345
9346 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
9347
9348         PR bootstrap/59580
9349         PR bootstrap/59583
9350         * config.gcc (x86_archs): New variable.
9351         (x86_64_archs): Likewise.
9352         (x86_cpus): Likewise.
9353         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
9354         --with-arch/--with-cpu= options.
9355         Support --with-arch=/--with-cpu={nehalem,westmere,
9356         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
9357
9358 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9359
9360         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
9361         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
9362
9363 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9364
9365         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
9366
9367 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
9368
9369         PR target/58944
9370         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
9371         clear cpp_get_options (parse_in)->warn_unused_macros for
9372         ix86_target_macros_internal with cpp_define.
9373
9374 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
9375
9376         * jump.c (delete_related_insns): Keep (use (insn))s.
9377         * reorg.c (redundant_insn): Check for barriers too.
9378
9379 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9380
9381         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
9382
9383 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
9384
9385         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
9386         call to $$dyncall when TARGET_LONG_CALLS is true.
9387
9388 2014-01-17  Jeff Law  <law@redhat.com>
9389
9390         * ree.c (combine_set_extension): Temporarily disable test for
9391         changing number of hard registers.
9392
9393 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9394
9395         PR middle-end/58125
9396         * ipa-inline-analysis.c (inline_free_summary):
9397         Do not free summary of aliases.
9398
9399 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9400
9401         PR middle-end/59706
9402         * gimplify.c (gimplify_expr): Use create_tmp_var
9403         instead of create_tmp_var_raw.  If cond doesn't have
9404         integral type, don't add the IFN_ANNOTATE builtin at all.
9405
9406 2014-01-17  Martin Jambor  <mjambor@suse.cz>
9407
9408         PR ipa/59736
9409         * ipa-cp.c (prev_edge_clone): New variable.
9410         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
9411         Also resize prev_edge_clone vector.
9412         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
9413         (ipcp_edge_removal_hook): New function.
9414         (ipcp_driver): Register ipcp_edge_removal_hook.
9415
9416 2014-01-17  Andrew Pinski  <apinski@cavium.com>
9417             Steve Ellcey  <sellcey@mips.com>
9418
9419         PR target/59462
9420         * config/mips/mips.c (mips_print_operand): Check operand mode instead
9421         of operator mode.
9422
9423 2014-01-17  Jeff Law  <law@redhat.com>
9424
9425         PR middle-end/57904
9426         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
9427         so that pass_ccp runs first.
9428
9429 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9430
9431         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
9432         (ix86_adjust_cost): Use !TARGET_XXX.
9433         (do_reorder_for_imul): Likewise.
9434         (swap_top_of_ready_list): Likewise.
9435         (ix86_sched_reorder): Likewise.
9436
9437 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9438
9439         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9440         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
9441         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
9442         (intel_memset): New.  Duplicate slm_memset.
9443         (intel_cost): New.  Duplicate slm_cost.
9444         (m_INTEL): New macro.
9445         (processor_target_table): Add "intel".
9446         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
9447         with PROCESSOR_INTEL for "intel".
9448         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
9449         PROCESSOR_SILVERMONT.
9450         (ix86_issue_rate): Likewise.
9451         (ix86_adjust_cost): Likewise.
9452         (ia32_multipass_dfa_lookahead): Likewise.
9453         (swap_top_of_ready_list): Likewise.
9454         (ix86_sched_reorder): Likewise.
9455         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
9456         instead of TARGET_OPT_AGU.
9457         * config/i386/i386.h (TARGET_INTEL): New.
9458         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
9459         (processor_type): Add PROCESSOR_INTEL.
9460         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
9461         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
9462
9463 2014-01-17  Marek Polacek  <polacek@redhat.com>
9464
9465         PR c/58346
9466         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
9467         size is zero.
9468
9469 2014-01-17  Richard Biener  <rguenther@suse.de>
9470
9471         PR tree-optimization/46590
9472         * opts.c (default_options_table): Add entries for
9473         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
9474         all enabled at -O1 but not for -Og.
9475         * common.opt (fbranch-count-reg): Remove Init(1).
9476         (fmove-loop-invariants): Likewise.
9477         (ftree-pta): Likewise.
9478
9479 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9480
9481         * config/i386/i386.c (ix86_data_alignment): For compatibility with
9482         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
9483         decls to at least the GCC 4.8 used alignments.
9484
9485         PR fortran/59440
9486         * tree-nested.c (convert_nonlocal_reference_stmt,
9487         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
9488         of GIMPLE_BIND stmts, adjust associated decls.
9489
9490 2014-01-17  Richard Biener  <rguenther@suse.de>
9491
9492         PR tree-optimization/46590
9493         * vec.h (vec<>::bseach): New member function implementing
9494         binary search according to C89 bsearch.
9495         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
9496         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
9497         bitmap pointer again.  Make accesses_in_loop a flat array.
9498         (mem_ref_obstack): New global.
9499         (outermost_indep_loop): Adjust for mem_ref->stored changes.
9500         (mark_ref_stored): Likewise.
9501         (ref_indep_loop_p_2): Likewise.
9502         (set_ref_stored_in_loop): New helper function.
9503         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
9504         (memref_free): Adjust.
9505         (record_mem_ref_loc): Simplify.
9506         (gather_mem_refs_stmt): Adjust.
9507         (sort_locs_in_loop_postorder_cmp): New function.
9508         (analyze_memory_references): Sort accesses_in_loop after
9509         loop postorder number.
9510         (find_ref_loc_in_loop_cmp): New function.
9511         (for_all_locs_in_loop): Find relevant cluster of locs in
9512         accesses_in_loop and iterate without recursion.
9513         (execute_sm): Avoid uninit warning.
9514         (struct ref_always_accessed): Simplify.
9515         (ref_always_accessed::operator ()): Likewise.
9516         (ref_always_accessed_p): Likewise.
9517         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
9518         loop postorder numbers here.
9519         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
9520         numbers.
9521
9522 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9523
9524         PR c++/57945
9525         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
9526         on decls for which assemble_alias has been called.
9527
9528 2014-01-17  Nick Clifton  <nickc@redhat.com>
9529
9530         * config/msp430/msp430.opt: (mcpu): New option.
9531         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
9532         (msp430_option_override): Parse target_cpu.  If the MCU name
9533         matches a generic string, clear target_mcu.
9534         (msp430_attr): Allow numeric interrupt values up to 63.
9535         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
9536         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
9537         option.
9538         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
9539         Add mcpu matches.
9540         * config/msp430/msp430.md (popm): Use %J rather than %I.
9541         (addsi3): Use msp430_nonimmediate_operand for operand 2.
9542         (addhi_cy_i): Use immediate_operand for operand 2.
9543         * doc/invoke.texi: Document -mcpu option.
9544
9545 2014-01-17  Richard Biener  <rguenther@suse.de>
9546
9547         PR rtl-optimization/38518
9548         * df.h (df_analyze_loop): Declare.
9549         * df-core.c: Include cfgloop.h.
9550         (df_analyze_1): Split out main part of df_analyze.
9551         (df_analyze): Adjust.
9552         (loop_inverted_post_order_compute): New function.
9553         (loop_post_order_compute): Likewise.
9554         (df_analyze_loop): New function avoiding whole-function
9555         postorder computes.
9556         * loop-invariant.c (find_defs): Use df_analyze_loop.
9557         (find_invariants): Adjust.
9558         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
9559
9560 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
9561
9562         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
9563         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
9564
9565 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
9566
9567         * ipa-ref.c (ipa_remove_stmt_references): Fix references
9568         traversal when removing references.
9569
9570 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
9571
9572         PR ipa/59775
9573         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
9574
9575 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
9576
9577         PR middle-end/56791
9578         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
9579         pushing a reload for an autoinc when we had previously reloaded an
9580         inner part of the address.
9581
9582 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
9583
9584         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
9585         field.
9586         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
9587         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
9588         when not giving up or versioning for alias only because of
9589         loop->safelen.
9590         (vect_analyze_data_ref_dependences): Set to true.
9591         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
9592         is a GIMPLE_PHI.
9593         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
9594         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
9595         to the condition.
9596
9597         PR middle-end/58344
9598         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
9599
9600         PR target/59839
9601         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
9602         operand 0 predicate for gathers, use a new pseudo as subtarget.
9603
9604 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
9605
9606         PR middle-end/59609
9607         * lra-constraints.c (process_alt_operands): Add printing debug info.
9608         Check absence of input/output reloads for matched operands too.
9609
9610 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
9611
9612         PR rtl-optimization/59835
9613         * ira.c (ira_init_register_move_cost): Increase cost for
9614         impossible modes.
9615
9616 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
9617
9618         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
9619
9620 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
9621
9622         PR target/59780
9623         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
9624         non-register objects.  Use gen_(high/low)part more consistently.
9625         Fix assertions.
9626
9627 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9628
9629         PR target/59844
9630         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
9631         endian support, remove tests for WORDS_BIG_ENDIAN.
9632         (p8_mfvsrd_3_<mode>): Likewise.
9633         (reload_gpr_from_vsx<mode>): Likewise.
9634         (reload_gpr_from_vsxsf): Likewise.
9635         (p8_mfvsrd_4_disf): Likewise.
9636
9637 2014-01-16  Richard Biener  <rguenther@suse.de>
9638
9639         PR rtl-optimization/46590
9640         * lcm.c (compute_antinout_edge): Use postorder iteration.
9641         (compute_laterin): Use inverted postorder iteration.
9642
9643 2014-01-16  Nick Clifton  <nickc@redhat.com>
9644
9645         PR middle-end/28865
9646         * varasm.c (output_constant): Return the number of bytes actually
9647         emitted.
9648         (output_constructor_array_range): Update the field size with the
9649         number of bytes emitted by output_constant.
9650         (output_constructor_regular_field): Likewise.  Also do not
9651         complain if the total number of bytes emitted is now greater
9652         than the expected fieldpos.
9653         * output.h (output_constant): Update prototype and descriptive comment.
9654
9655 2014-01-16  Marek Polacek  <polacek@redhat.com>
9656
9657         PR middle-end/59827
9658         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
9659         it is error_mark_node.
9660
9661 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
9662
9663         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
9664         VALID_AVX256_REG_OR_OI_MODE.
9665
9666 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
9667
9668         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
9669         current procedure should be profiled.
9670
9671 2014-01-15  Andrew Pinski  <apinski@cavium.com>
9672
9673         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
9674         of moving from/to the STACK_REG register class.
9675
9676 2014-01-15  Richard Henderson  <rth@redhat.com>
9677
9678         PR debug/54694
9679         * reginfo.c (global_regs_decl): Globalize.
9680         * rtl.h (global_regs_decl): Declare.
9681         * ira.c (do_reload): Diagnose frame_pointer_needed and it
9682         reserved via global_regs.
9683
9684 2014-01-15  Teresa Johnson  <tejohnson@google.com>
9685
9686         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
9687
9688 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
9689
9690         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
9691         and vmulosh rather than call gen_vec_widen_smult_*.
9692         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
9693         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
9694         (vec_widen_smult_even_v16qi): Likewise.
9695         (vec_widen_umult_even_v8hi): Likewise.
9696         (vec_widen_smult_even_v8hi): Likewise.
9697         (vec_widen_umult_odd_v16qi): Likewise.
9698         (vec_widen_smult_odd_v16qi): Likewise.
9699         (vec_widen_umult_odd_v8hi): Likewise.
9700         (vec_widen_smult_odd_v8hi): Likewise.
9701         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
9702         vmuloub rather than call gen_vec_widen_umult_*.
9703         (vec_widen_umult_lo_v16qi): Likewise.
9704         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
9705         vmulosb rather than call gen_vec_widen_smult_*.
9706         (vec_widen_smult_lo_v16qi): Likewise.
9707         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
9708         rather than call gen_vec_widen_umult_*.
9709         (vec_widen_umult_lo_v8hi): Likewise.
9710         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
9711         rather than call gen_vec_widen_smult_*.
9712         (vec_widen_smult_lo_v8hi): Likewise.
9713
9714 2014-01-15  Jeff Law  <law@redhat.com>
9715
9716         PR tree-optimization/59747
9717         * ree.c (find_and_remove_re): Properly handle case where a second
9718         eliminated extension requires widening a copy created for elimination
9719         of a prior extension.
9720         (combine_set_extension): Ensure that the number of hard regs needed
9721         for a destination register does not change when we widen it.
9722
9723 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9724
9725         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
9726         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
9727         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
9728         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
9729         (avr-*-rtems*): Likewise.
9730         (bfin*-rtems*): Likewise.
9731         (moxie-*-rtems*): Likewise.
9732         (h8300-*-rtems*): Likewise.
9733         (i[34567]86-*-rtems*): Likewise.
9734         (lm32-*-rtems*): Likewise.
9735         (m32r-*-rtems*): Likewise.
9736         (m68k-*-rtems*): Likewise.
9737         (microblaze*-*-rtems*): Likewise.
9738         (mips*-*-rtems*): Likewise.
9739         (powerpc-*-rtems*): Likewise.
9740         (sh-*-rtems*): Likewise.
9741         (sparc-*-rtems*): Likewise.
9742         (sparc64-*-rtems*): Likewise.
9743         (v850-*-rtems*): Likewise.
9744         (m32c-*-rtems*): Likewise.
9745
9746 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
9747
9748         PR rtl-optimization/59511
9749         * ira.c (ira_init_register_move_cost): Use memory costs for some
9750         cases of register move cost calculations.
9751         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
9752         instead of BB frequency.
9753         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
9754         * lra-assigns.c (find_hard_regno_for): Ditto.
9755
9756 2014-01-15  Richard Biener  <rguenther@suse.de>
9757
9758         PR tree-optimization/59822
9759         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
9760         (vectorizable_load): Use it to hoist defs of uses of invariant
9761         loads out of the loop.
9762
9763 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
9764             Kugan Vivekanandarajah  <kuganv@linaro.org>
9765
9766         PR target/59695
9767         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
9768         truncation.
9769
9770 2014-01-15  Richard Biener  <rguenther@suse.de>
9771
9772         PR rtl-optimization/59802
9773         * lcm.c (compute_available): Use inverted postorder to seed
9774         the initial worklist.
9775
9776 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9777
9778         PR target/59803
9779         * config/s390/s390.c (s390_preferred_reload_class): Don't return
9780         ADDR_REGS for invalid symrefs in non-PIC code.
9781
9782 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
9783
9784         PR other/58712
9785         * builtins.c (determine_block_size): Initialize *probable_max_size
9786         even if len_rtx is CONST_INT.
9787
9788 2014-01-14  Andrew Pinski  <apinski@cavium.com>
9789
9790         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
9791         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
9792         (cortexa53_tunings): Likewise.
9793         (aarch64_sched_issue_rate): New function.
9794         (TARGET_SCHED_ISSUE_RATE): Define.
9795
9796 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9797
9798         * ira-costs.c (find_costs_and_classes): Add missed
9799         ira_init_register_move_cost_if_necessary.
9800
9801 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9802
9803         PR target/59787
9804         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
9805
9806 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
9807
9808         PR target/59794
9809         * config/i386/i386.c (type_natural_mode): Add a bool parameter
9810         to indicate if type is used for function return value.  Warn ABI
9811         change if the vector mode isn't available for function return value.
9812         (ix86_function_arg_advance): Pass false to type_natural_mode.
9813         (ix86_function_arg): Likewise.
9814         (ix86_gimplify_va_arg): Likewise.
9815         (function_arg_32): Don't warn ABI change.
9816         (ix86_function_value): Pass true to type_natural_mode.
9817         (ix86_return_in_memory): Likewise.
9818         (ix86_struct_value_rtx): Removed.
9819         (TARGET_STRUCT_VALUE_RTX): Likewise.
9820
9821 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9822
9823         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
9824         converting a conditional jump into a conditional return.
9825
9826 2014-01-14  Richard Biener  <rguenther@suse.de>
9827
9828         PR tree-optimization/58921
9829         PR tree-optimization/59006
9830         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
9831         hoisting invariant stmts.
9832         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
9833         invariant loads on the preheader edge if possible.
9834
9835 2014-01-14  Joey Ye  <joey.ye@arm.com>
9836
9837         * doc/plugin.texi (Building GCC plugins): Update to C++.
9838
9839 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
9840
9841         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
9842         (_mm_rcp28_round_ss): Ditto.
9843         (_mm_rsqrt28_round_sd): Ditto.
9844         (_mm_rsqrt28_round_ss): Ditto.
9845         (_mm_rcp28_sd): Ditto.
9846         (_mm_rcp28_ss): Ditto.
9847         (_mm_rsqrt28_sd): Ditto.
9848         (_mm_rsqrt28_ss): Ditto.
9849         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
9850         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
9851         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
9852         (IX86_BUILTIN_RCP28SD): Ditto.
9853         (IX86_BUILTIN_RCP28SS): Ditto.
9854         (IX86_BUILTIN_RSQRT28SD): Ditto.
9855         (IX86_BUILTIN_RSQRT28SS): Ditto.
9856         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
9857         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
9858         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
9859         (ix86_expand_special_args_builtin): Expand new FTYPE.
9860         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
9861         (srcp14<mode>): Make insn unary.
9862         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
9863         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
9864         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
9865         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
9866         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
9867         Fix rounding: make it SAE only.
9868         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
9869         Ditto.
9870         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
9871         Ditto.
9872         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
9873         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
9874         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
9875         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
9876         (round_saeonly_mask_scalar_operand4): Ditto.
9877         (round_saeonly_mask_scalar_op3): Ditto.
9878         (round_saeonly_mask_scalar_op4): Ditto.
9879
9880 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9881
9882         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9883         Implement -maltivec=be for vec_insert and vec_extract.
9884
9885 2014-01-10  DJ Delorie  <dj@redhat.com>
9886
9887         * config/msp430/msp430.md (call_internal): Don't allow memory
9888         references with SP as the base register.
9889         (call_value_internal): Likewise.
9890         * config/msp430/constraints.md (Yc): New.  For memory references
9891         that don't use SP as a base register.
9892
9893         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
9894         "an integer without a # prefix"
9895         * config/msp430/msp430.md (epilogue_helper): Use it.
9896
9897 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9898
9899         PR target/59617
9900         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
9901         AVX512F gather builtins.
9902         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
9903         on gather decls with INTEGER_TYPE masktype.
9904         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
9905         directly into the builtin rather than hoisting it before loop.
9906
9907         PR tree-optimization/59387
9908         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
9909         (scev_const_prop): If folded_casts and type has undefined overflow,
9910         use force_gimple_operand instead of force_gimple_operand_gsi and
9911         for each added stmt if it is assign with
9912         arith_code_with_undefined_signed_overflow, call
9913         rewrite_to_defined_overflow.
9914         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
9915         gimple-fold.h instead.
9916         (arith_code_with_undefined_signed_overflow,
9917         rewrite_to_defined_overflow): Moved to ...
9918         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
9919         rewrite_to_defined_overflow): ... here.  No longer static.
9920         Include gimplify-me.h.
9921         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
9922         rewrite_to_defined_overflow): New prototypes.
9923
9924 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9925
9926         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
9927
9928 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
9929
9930         * builtins.c (get_object_alignment_2): Minor tweak.
9931         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
9932
9933 2014-01-13  Christian Bruel  <christian.bruel@st.com>
9934
9935         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
9936         optimized non constant lengths.
9937
9938 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9939
9940         PR libgomp/59194
9941         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
9942         load as __atomic_load_N if possible.
9943
9944 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9945
9946         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
9947         target parameter.
9948         (rs6000_expand_builtin): Adjust call.
9949
9950 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9951
9952         PR target/58115
9953         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
9954         * config/rs6000/rs6000.c: Include target-globals.h.
9955         (rs6000_set_current_function): Instead of doing target_reinit
9956         unconditionally, use save_target_globals_default_opts and
9957         restore_target_globals.
9958
9959         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
9960         FPSCR.
9961         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
9962         (rs6000_expand_builtin): Handle mffs and mtfsf.
9963         (rs6000_init_builtins): Define mffs and mtfsf.
9964         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
9965         (rs6000_mffs): New pattern.
9966         (rs6000_mtfsf): New pattern.
9967
9968 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
9969
9970         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
9971         Start narrowing with START.  Apply candidate-use pair
9972         and check overall cost in narrowing.
9973         (iv_ca_prune): Pass new argument.
9974
9975 2014-01-10  Jeff Law  <law@redhat.com>
9976
9977         PR middle-end/59743
9978         * ree.c (combine_reaching_defs): Ensure the defining statement
9979         occurs before the extension when optimizing extensions with
9980         different source and destination hard registers.
9981
9982 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
9983
9984         PR ipa/58585
9985         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
9986         vtables into the type inheritance graph.
9987
9988 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9989
9990         PR rtl-optimization/59754
9991         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
9992         modes in the REGNO != REGNO case.
9993
9994 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9995
9996         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
9997
9998 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9999
10000         PR tree-optimization/59745
10001         * tree-predcom.c (tree_predictive_commoning_loop): Call
10002         free_affine_expand_cache if giving up because components is NULL.
10003
10004         * target-globals.c (save_target_globals): Allocate < 4KB structs using
10005         GC in payload of target_globals struct instead of allocating them on
10006         the heap and the larger structs separately using GC.
10007         * target-globals.h (struct target_globals): Make regs, hard_regs,
10008         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
10009         of GTY((skip)) and change type to void *.
10010         (reset_target_globals): Cast loads from those fields to corresponding
10011         types.
10012
10013 2014-01-10  Steve Ellcey  <sellcey@mips.com>
10014
10015         PR plugins/59335
10016         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
10017         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
10018         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
10019
10020 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
10021
10022         PR target/59744
10023         * aarch64-modes.def (CC_Zmode): New flags mode.
10024         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
10025         represents an equality.
10026         (aarch64_get_condition_code): Handle CC_Zmode.
10027         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
10028
10029 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10030
10031         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
10032         extraction in good case.
10033
10034 2014-01-10  Richard Biener  <rguenther@suse.de>
10035
10036         PR tree-optimization/59374
10037         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
10038         checking after SLP discovery.  Mark stmts not participating
10039         in any SLP instance properly.
10040
10041 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10042
10043         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
10044         when handling a SET rtx.
10045
10046 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10047
10048         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
10049         (cortex-a57): Likewise.
10050         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
10051
10052 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10053
10054         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
10055         non-iwmmxt builtins.
10056
10057 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
10058
10059         PR ipa/58252
10060         PR ipa/59226
10061         * ipa-devirt.c record_target_from_binfo): Take as argument
10062         stack of binfos and lookup matching one for virtual inheritance.
10063         (possible_polymorphic_call_targets_1): Update.
10064
10065 2014-01-10  Huacai Chen  <chenhc@lemote.com>
10066
10067         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
10068         kernel strings for Loongson-2E/2F/3A.
10069
10070 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
10071
10072         PR middle-end/59670
10073         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
10074         is_gimple_call before calling gimple_call_internal_p.
10075
10076 2014-01-09  Steve Ellcey  <sellcey@mips.com>
10077
10078         * Makefile.in (TREE_FLOW_H): Remove.
10079         (TREE_SSA_H): Add file names from tree-flow.h.
10080         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
10081         * tree.h: Remove tree-flow.h reference.
10082         * hash-table.h: Remove tree-flow.h reference.
10083         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
10084         reference with tree-ssa-loop.h.
10085
10086 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10087
10088         * doc/invoke.texi: Add -maltivec={be,le} options, and document
10089         default element-order behavior for -maltivec.
10090         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
10091         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
10092         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
10093         when targeting big endian, at least for now.
10094         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
10095
10096 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10097
10098         PR middle-end/47735
10099         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
10100         var satisfies use_register_for_decl, just take into account type
10101         alignment, rather than decl alignment.
10102
10103         PR tree-optimization/59622
10104         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
10105         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
10106         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
10107         Don't devirtualize for inplace at all.  For targets.length () == 1,
10108         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
10109
10110 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10111
10112         * config/i386/i386.md (cpu): Remove the unused btver1.
10113
10114 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
10115
10116         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
10117
10118 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10119
10120         PR target/58115
10121         * tree-core.h (struct target_globals): New forward declaration.
10122         (struct tree_target_option): Add globals field.
10123         * tree.h (TREE_TARGET_GLOBALS): Define.
10124         (prepare_target_option_nodes_for_pch): New prototype.
10125         * target-globals.h (struct target_globals): Define even if
10126         !SWITCHABLE_TARGET.
10127         * tree.c (prepare_target_option_node_for_pch,
10128         prepare_target_option_nodes_for_pch): New functions.
10129         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
10130         * config/i386/i386.c: Include target-globals.h.
10131         (ix86_set_current_function): Instead of doing target_reinit
10132         unconditionally, use save_target_globals_default_opts and
10133         restore_target_globals.
10134
10135 2014-01-09  Richard Biener  <rguenther@suse.de>
10136
10137         PR tree-optimization/59715
10138         * tree-cfg.h (split_critical_edges): Declare.
10139         * tree-cfg.c (split_critical_edges): Export.
10140         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
10141
10142 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
10143
10144         * cfgexpand.c (expand_stack_vars): Optionally disable
10145         asan stack protection.
10146         (expand_used_vars): Likewise.
10147         (partition_stack_vars): Likewise.
10148         * asan.c (asan_emit_stack_protection): Optionally disable
10149         after return stack usage.
10150         (instrument_derefs): Optionally disable memory access instrumentation.
10151         (instrument_builtin_call): Likewise.
10152         (instrument_strlen_call): Likewise.
10153         (asan_protect_global): Optionally disable global variables protection.
10154         * doc/invoke.texi: Added doc for new options.
10155         * params.def: Added new options.
10156         * params.h: Likewise.
10157
10158 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
10159
10160         PR rtl-optimization/59724
10161         * ifcvt.c (cond_exec_process_if_block): Don't call
10162         flow_find_head_matching_sequence with 0 longest_match.
10163         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
10164         non-active insns if !stop_after.
10165         (try_head_merge_bb): Revert 2014-01-07 changes.
10166
10167 2014-01-08  Jeff Law  <law@redhat.com>
10168
10169         * ree.c (get_sub_rtx): New function, extracted from...
10170         (merge_def_and_ext): Here.
10171         (combine_reaching_defs): Use get_sub_rtx.
10172
10173 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
10174
10175         * cgraph.h (varpool_variable_node): Do not choke on null node.
10176
10177 2014-01-08  Catherine Moore  <clm@codesourcery.com>
10178
10179         * config/mips/mips.md (simple_return): Attempt to use JRC
10180         for microMIPS.
10181         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
10182
10183 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10184
10185         PR rtl-optimization/59137
10186         * reorg.c (steal_delay_list_from_target): Call update_block for
10187         elided insns.
10188         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
10189
10190 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10191
10192         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10193         two duplicate entries.
10194
10195 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
10196
10197         Revert:
10198         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
10199
10200         * config/mips/mips.c (mips_truncated_op_cost): New function.
10201         (mips_rtx_costs): Adjust test for BADDU.
10202         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
10203
10204         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
10205
10206         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
10207         (*baddu_si): ...this new pattern.
10208
10209 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10210
10211         PR ipa/59722
10212         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
10213
10214 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10215
10216         PR middle-end/57748
10217         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
10218         inner_reference_p.
10219         (expand_expr, expand_normal): Adjust.
10220         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
10221         inner_reference_p. Use inner_reference_p to expand inner references.
10222         (store_expr): Adjust.
10223         * cfgexpand.c (expand_call_stmt): Adjust.
10224
10225 2014-01-08  Rong Xu  <xur@google.com>
10226
10227         * gcov-io.c (gcov_var): Move from gcov-io.h.
10228         (gcov_position): Ditto.
10229         (gcov_is_error): Ditto.
10230         (gcov_rewrite): Ditto.
10231         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
10232         only part to libgcc/libgcov.h.
10233
10234 2014-01-08  Marek Polacek  <polacek@redhat.com>
10235
10236         PR middle-end/59669
10237         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
10238
10239 2014-01-08  Marek Polacek  <polacek@redhat.com>
10240
10241         PR sanitizer/59667
10242         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
10243
10244 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10245
10246         PR rtl-optimization/59649
10247         * stor-layout.c (get_mode_bounds): For BImode return
10248         0 and STORE_FLAG_VALUE.
10249
10250 2014-01-08  Richard Biener  <rguenther@suse.de>
10251
10252         PR middle-end/59630
10253         * gimple.h (is_gimple_builtin_call): Remove.
10254         (gimple_builtin_call_types_compatible_p): New.
10255         (gimple_call_builtin_p): New overload.
10256         * gimple.c (is_gimple_builtin_call): Remove.
10257         (validate_call): Rename to ...
10258         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
10259         check return types.
10260         (validate_type): New static function.
10261         (gimple_call_builtin_p): New overload and adjust.
10262         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
10263         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
10264         (gimple_fold_stmt_to_constant_1): Likewise.
10265         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
10266
10267 2014-01-08  Richard Biener  <rguenther@suse.de>
10268
10269         PR middle-end/59471
10270         * gimplify.c (gimplify_expr): Gimplify register-register type
10271         VIEW_CONVERT_EXPRs to separate stmts.
10272
10273 2014-01-07  Jeff Law  <law@redhat.com>
10274
10275         PR middle-end/53623
10276         * ree.c (combine_set_extension): Handle case where source
10277         and destination registers in an extension insn are different.
10278         (combine_reaching_defs): Allow source and destination registers
10279         in extension to be different under limited circumstances.
10280         (add_removable_extension): Remove restriction that the
10281         source and destination registers in the extension are the same.
10282         (find_and_remove_re): Emit a copy from the extension's
10283         destination to its source after the defining insn if
10284         the source and destination registers are different.
10285
10286         PR middle-end/59285
10287         * ifcvt.c (merge_if_block): If we are merging a block with more than
10288         one successor with a block with no successors, remove any BARRIER
10289         after the second block.
10290
10291 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
10292
10293         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
10294
10295 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
10296
10297         PR target/59652
10298         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
10299         for 14-bit register offsets when INT14_OK_STRICT is false.
10300
10301 2014-01-07  Roland Stigge  <stigge@antcom.de>
10302             Michael Meissner  <meissner@linux.vnet.ibm.com>
10303
10304         PR 57386/target
10305         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
10306         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
10307
10308 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
10309
10310         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
10311         -mcpu.
10312
10313 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
10314
10315         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
10316         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
10317         rtx is const0_rtx or not.
10318
10319 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
10320
10321         PR target/58115
10322         * target-globals.c (save_target_globals): Remove this_fn_optab
10323         handling.
10324         * toplev.c: Include optabs.h.
10325         (target_reinit): Temporarily restore the global options if another
10326         set of options are in force.
10327
10328 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
10329
10330         PR rtl-optimization/58668
10331         * cfgcleanup.c (flow_find_cross_jump): Don't count
10332         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
10333         to determine what is counted.
10334         (flow_find_head_matching_sequence): Use active_insn_p to determine
10335         what is counted.
10336         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
10337         counting change.
10338         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
10339         determine what is counted.
10340
10341         PR tree-optimization/59643
10342         * tree-predcom.c (split_data_refs_to_components): If one dr is
10343         read and one write, determine_offset fails and the write isn't
10344         in the bad component, just put the read into the bad component.
10345
10346 2014-01-07  Mike Stump  <mikestump@comcast.net>
10347             Jakub Jelinek  <jakub@redhat.com>
10348
10349         PR pch/59436
10350         * tree-core.h (struct tree_optimization_option): Change optabs
10351         type from unsigned char * to void *.
10352         * optabs.c (init_tree_optimization_optabs): Adjust
10353         TREE_OPTIMIZATION_OPTABS initialization.
10354
10355 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
10356
10357         PR target/59644
10358         * config/i386/i386.h (struct machine_function): Add
10359         no_drap_save_restore field.
10360         * config/i386/i386.c (ix86_save_reg): Use
10361         !cfun->machine->no_drap_save_restore instead of
10362         crtl->stack_realign_needed.
10363         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
10364         this function clears frame_pointer_needed.  Set
10365         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
10366         and DRAP reg is needed.
10367
10368 2014-01-06  Marek Polacek  <polacek@redhat.com>
10369
10370         PR c/57773
10371         * doc/implement-c.texi: Mention that other integer types are
10372         permitted as bit-field types in strictly conforming mode.
10373
10374 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
10375
10376         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
10377         is newly allocated.
10378
10379 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
10380
10381         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
10382
10383 2014-01-06  Martin Jambor  <mjambor@suse.cz>
10384
10385         PR ipa/59008
10386         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
10387         to int.
10388         * ipa-prop.c (ipa_print_node_params): Fix indentation.
10389
10390 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
10391
10392         PR debug/59350
10393         PR debug/59510
10394         * var-tracking.c (add_stores): Preserve the value of the source even if
10395         we don't record the store.
10396
10397 2014-01-06  Terry Guo  <terry.guo@arm.com>
10398
10399         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
10400
10401 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
10402
10403         PR bootstrap/59541
10404         * config/darwin.c (darwin_function_section): Adjust return values to
10405         correspond to optimisation changes made in r206070.
10406
10407 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
10408
10409         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
10410         from prefetch_block tune setting.
10411         (nocona_cost): Correct size of prefetch block to 64.
10412
10413 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
10414
10415         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
10416         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
10417         used to save the static chain register in the computation of the offset
10418         from which the FP registers need to be restored.
10419
10420 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
10421
10422         PR tree-optimization/59519
10423         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
10424         ICE if get_current_def (current_new_name) is already non-NULL, as long
10425         as it is a phi result of some other phi in *new_exit_bb that has
10426         the same argument.
10427
10428         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
10429         or vmovdqu* for misaligned_operand.
10430         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
10431         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
10432         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
10433         aligned_mem for AVX512F masked aligned load and store builtins and for
10434         non-temporal moves.
10435
10436 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
10437
10438         PR tree-optimization/59651
10439         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
10440         Address range for negative step should be added by TYPE_SIZE_UNIT.
10441
10442 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
10443
10444         * config/m68k/m68k.c (handle_move_double): Handle pushes with
10445         overlapping registers also for registers other than the stack pointer.
10446
10447 2014-01-03  Marek Polacek  <polacek@redhat.com>
10448
10449         PR other/59661
10450         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
10451         __builtin_FILE.
10452
10453 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
10454
10455         PR target/59625
10456         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
10457         asm goto as jump.
10458
10459         * config/i386/i386.md (MODE_SIZE): New mode attribute.
10460         (push splitter): Use <P:MODE_SIZE> instead of
10461         GET_MODE_SIZE (<P:MODE>mode).
10462         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
10463         (mov -1, reg peephole2): Likewise.
10464         * config/i386/sse.md (*mov<mode>_internal,
10465         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
10466         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
10467         *<code><mode>3, *andnot<mode>3<mask_name>,
10468         <mask_codefor><code><mode>3<mask_name>): Likewise.
10469         * config/i386/subst.md (mask_mode512bit_condition,
10470         sd_mask_mode512bit_condition): Likewise.
10471
10472 2014-01-02  Xinliang David Li  <davidxl@google.com>
10473
10474         PR tree-optimization/59303
10475         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
10476         (dump_predicates): Better output format.
10477         (pred_equal_p): New function.
10478         (is_neq_relop_p): Ditto.
10479         (is_neq_zero_form_p): Ditto.
10480         (pred_expr_equal_p): Ditto.
10481         (pred_neg_p): Ditto.
10482         (simplify_pred): Ditto.
10483         (simplify_preds_2): Ditto.
10484         (simplify_preds_3): Ditto.
10485         (simplify_preds_4): Ditto.
10486         (simplify_preds): Ditto.
10487         (push_pred): Ditto.
10488         (push_to_worklist): Ditto.
10489         (get_pred_info_from_cmp): Ditto.
10490         (is_degenerated_phi): Ditto.
10491         (normalize_one_pred_1): Ditto.
10492         (normalize_one_pred): Ditto.
10493         (normalize_one_pred_chain): Ditto.
10494         (normalize_preds): Ditto.
10495         (normalize_cond_1): Remove function.
10496         (normalize_cond): Ditto.
10497         (is_gcond_subset_of): Ditto.
10498         (is_subset_of_any): Ditto.
10499         (is_or_set_subset_of): Ditto.
10500         (is_and_set_subset_of): Ditto.
10501         (is_norm_cond_subset_of): Ditto.
10502         (pred_chain_length_cmp): Ditto.
10503         (convert_control_dep_chain_into_preds): Type change.
10504         (find_predicates): Ditto.
10505         (find_def_preds): Ditto.
10506         (destroy_predicates_vecs): Ditto.
10507         (find_matching_predicates_in_rest_chains): Ditto.
10508         (use_pred_not_overlap_with_undef_path_pred): Ditto.
10509         (is_pred_expr_subset): Ditto.
10510         (is_pred_chain_subset_of): Ditto.
10511         (is_included_in): Ditto.
10512         (is_superset_of): Ditto.
10513
10514 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10515
10516         Update copyright years.
10517
10518 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10519
10520         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
10521         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
10522         config/arc/arc.md, config/arc/arc.opt,
10523         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
10524         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
10525         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
10526         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
10527         config/linux-protos.h, config/linux.c, config/winnt-c.c,
10528         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
10529         vtable-verify.c, vtable-verify.h: Use the standard form for the
10530         copyright notice.
10531
10532 2014-01-02  Tobias Burnus  <burnus@net-b.de>
10533
10534         * gcc.c (process_command): Update copyright notice dates.
10535         * gcov-dump.c: Ditto.
10536         * gcov.c: Ditto.
10537         * doc/cpp.texi: Bump @copying's copyright year.
10538         * doc/cppinternals.texi: Ditto.
10539         * doc/gcc.texi: Ditto.
10540         * doc/gccint.texi: Ditto.
10541         * doc/gcov.texi: Ditto.
10542         * doc/install.texi: Ditto.
10543         * doc/invoke.texi: Ditto.
10544
10545 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
10546
10547         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
10548
10549 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
10550
10551         * config/i386/sse.md (*mov<mode>_internal): Guard
10552         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
10553
10554         PR rtl-optimization/59647
10555         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
10556         new_rtx into UNSIGNED_FLOAT rtxes.
10557 \f
10558 Copyright (C) 2014 Free Software Foundation, Inc.
10559
10560 Copying and distribution of this file, with or without modification,
10561 are permitted in any medium without royalty provided the copyright
10562 notice and this notice are preserved.