sysdeps/x86_64/multiarch/mempcpy.S

   1 /* Multiple versions of mempcpy
   2    All versions must be listed in ifunc-impl-list.c.
   3    Copyright (C) 2010-2015 Free Software Foundation, Inc.
   4    Contributed by Intel Corporation.
   5    This file is part of the GNU C Library.
   6
   7    The GNU C Library is free software; you can redistribute it and/or
   8    modify it under the terms of the GNU Lesser General Public
   9    License as published by the Free Software Foundation; either
  10    version 2.1 of the License, or (at your option) any later version.
  11
  12    The GNU C Library is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15    Lesser General Public License for more details.
  16
  17    You should have received a copy of the GNU Lesser General Public
  18    License along with the GNU C Library; if not, see
  19    <http://www.gnu.org/licenses/>.  */
  20
  21 #include <sysdep.h>
  22 #include <init-arch.h>
  23
  24 /* Define multiple versions only for the definition in lib and for
  25    DSO.  In static binaries we need mempcpy before the initialization
  26    happened.  */
  27 #if defined SHARED && IS_IN (libc)
  28 ENTRY(__mempcpy)
  29         .type   __mempcpy, @gnu_indirect_function
  30         cmpl    $0, KIND_OFFSET+__cpu_features(%rip)
  31         jne     1f
  32         call    __init_cpu_features
  33 1:      leaq    __mempcpy_sse2(%rip), %rax
  34         testl   $bit_SSSE3, __cpu_features+CPUID_OFFSET+index_SSSE3(%rip)
  35         jz      2f
  36         leaq    __mempcpy_ssse3(%rip), %rax
  37         testl   $bit_Fast_Copy_Backward, __cpu_features+FEATURE_OFFSET+index_Fast_Copy_Backward(%rip)
  38         jz      2f
  39         leaq    __mempcpy_ssse3_back(%rip), %rax
  40         testl   $bit_AVX_Usable, __cpu_features+FEATURE_OFFSET+index_AVX_Usable(%rip)
  41         jz      2f
  42         leaq    __mempcpy_avx_unaligned(%rip), %rax
  43 2:      ret
  44 END(__mempcpy)
  45
  46 # undef ENTRY
  47 # define ENTRY(name) \
  48         .type __mempcpy_sse2, @function; \
  49         .p2align 4; \
  50         .globl __mempcpy_sse2; \
  51         .hidden __mempcpy_sse2; \
  52         __mempcpy_sse2: cfi_startproc; \
  53         CALL_MCOUNT
  54 # undef END
  55 # define END(name) \
  56         cfi_endproc; .size __mempcpy_sse2, .-__mempcpy_sse2
  57
  58 # undef ENTRY_CHK
  59 # define ENTRY_CHK(name) \
  60         .type __mempcpy_chk_sse2, @function; \
  61         .globl __mempcpy_chk_sse2; \
  62         .p2align 4; \
  63         __mempcpy_chk_sse2: cfi_startproc; \
  64         CALL_MCOUNT
  65 # undef END_CHK
  66 # define END_CHK(name) \
  67         cfi_endproc; .size __mempcpy_chk_sse2, .-__mempcpy_chk_sse2
  68
  69 # undef libc_hidden_def
  70 # undef libc_hidden_builtin_def
  71 /* It doesn't make sense to send libc-internal mempcpy calls through a PLT.
  72    The speedup we get from using SSSE3 instruction is likely eaten away
  73    by the indirect call in the PLT.  */
  74 # define libc_hidden_def(name) \
  75         .globl __GI_mempcpy; __GI_mempcpy = __mempcpy_sse2
  76 # define libc_hidden_builtin_def(name) \
  77         .globl __GI___mempcpy; __GI___mempcpy = __mempcpy_sse2
  78 #endif
  79
  80 #include "../mempcpy.S"