-#!/usr/bin/env perl
+#! /usr/bin/env perl
+# Copyright 2011-2016 The OpenSSL Project Authors. All Rights Reserved.
+#
+# Licensed under the Apache License 2.0 (the "License"). You may not use
+# this file except in compliance with the License. You can obtain a copy
+# in the file LICENSE in the source distribution or at
+# https://www.openssl.org/source/license.html
+
#
# ====================================================================
# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
# Sandy Bridge 5.05[+5.0(6.1)] 10.06(11.15) 5.98(7.05) +68%(+58%)
# Ivy Bridge 5.05[+4.6] 9.65 5.54 +74%
# Haswell 4.43[+3.6(4.2)] 8.00(8.58) 4.55(5.21) +75%(+65%)
+# Skylake 2.63[+3.5(4.1)] 6.17(6.69) 4.23(4.44) +46%(+51%)
# Bulldozer 5.77[+6.0] 11.72 6.37 +84%
+# Ryzen(**) 2.71[+1.93] 4.64 2.74 +69%
+# Goldmont(**) 3.82[+1.70] 5.52 4.20 +31%
#
# AES-192-CBC
# Westmere 4.51 9.81 6.80 +44%
# Sandy Bridge 7.05 12.06(13.15) 7.12(7.72) +69%(+70%)
# Ivy Bridge 7.05 11.65 7.12 +64%
# Haswell 6.19 9.76(10.34) 6.21(6.25) +57%(+65%)
+# Skylake 3.62 7.16(7.68) 4.56(4.76) +57%(+61%)
# Bulldozer 8.00 13.95 8.25 +69%
+# Ryzen(**) 3.71 5.64 3.72 +52%
+# Goldmont(**) 5.35 7.05 5.76 +22%
#
# (*) There are two code paths: SSSE3 and AVX. See sha1-568.pl for
# background information. Above numbers in parentheses are SSSE3
# results collected on AVX-capable CPU, i.e. apply on OSes that
# don't support AVX.
+# (**) SHAEXT results.
#
# Needless to mention that it makes no sense to implement "stitched"
# *decrypt* subroutine. Because *both* AESNI-CBC decrypt and SHA1
# (**) Execution is fully dominated by integer code sequence and
# SIMD still hardly shows [in single-process benchmark;-]
-$flavour = shift;
-$output = shift;
-if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
$avx=1 if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
`ml64 2>&1` =~ /Version ([0-9]+)\./ &&
$1>=10);
+$avx=1 if (!$avx && `$ENV{CC} -v 2>&1` =~ /((?:^clang|LLVM) version|.*based on LLVM) ([3-9]\.[0-9]+)/ && $2>=3.0);
+
+$shaext=1; ### set to zero if compiling for 1.0.1
$stitched_decrypt=0;
-open OUT,"| \"$^X\" $xlate $flavour $output";
+open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\""
+ or die "can't call $xlate: $!";
*STDOUT=*OUT;
# void aesni_cbc_sha1_enc(const void *inp,
.type aesni_cbc_sha1_enc,\@abi-omnipotent
.align 32
aesni_cbc_sha1_enc:
+.cfi_startproc
# caller should check for SSSE3 and AES-NI bits
mov OPENSSL_ia32cap_P+0(%rip),%r10d
mov OPENSSL_ia32cap_P+4(%rip),%r11
+___
+$code.=<<___ if ($shaext);
bt \$61,%r11 # check SHA bit
jc aesni_cbc_sha1_enc_shaext
___
$code.=<<___;
jmp aesni_cbc_sha1_enc_ssse3
ret
+.cfi_endproc
.size aesni_cbc_sha1_enc,.-aesni_cbc_sha1_enc
___
.type aesni_cbc_sha1_enc_ssse3,\@function,6
.align 32
aesni_cbc_sha1_enc_ssse3:
+.cfi_startproc
mov `($win64?56:8)`(%rsp),$inp # load 7th argument
#shr \$6,$len # debugging artefact
#jz .Lepilogue_ssse3 # debugging artefact
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
lea `-104-($win64?10*16:0)`(%rsp),%rsp
+.cfi_adjust_cfa_offset `104+($win64?10*16:0)`
#mov $in0,$inp # debugging artefact
#lea 64(%rsp),$ctx # debugging artefact
___
$r++; unshift(@rndkey,pop(@rndkey));
};
-sub Xupdate_ssse3_16_31() # recall that $Xi starts wtih 4
+sub Xupdate_ssse3_16_31() # recall that $Xi starts with 4
{ use integer;
my $body = shift;
my @insns = (&$body,&$body,&$body,&$body); # 40 instructions
___
$code.=<<___;
lea `104+($win64?10*16:0)`(%rsp),%rsi
+.cfi_def_cfa %rsi,56
mov 0(%rsi),%r15
+.cfi_restore %r15
mov 8(%rsi),%r14
+.cfi_restore %r14
mov 16(%rsi),%r13
+.cfi_restore %r13
mov 24(%rsi),%r12
+.cfi_restore %r12
mov 32(%rsi),%rbp
+.cfi_restore %rbp
mov 40(%rsi),%rbx
+.cfi_restore %rbx
lea 48(%rsi),%rsp
+.cfi_def_cfa %rsp,8
.Lepilogue_ssse3:
ret
+.cfi_endproc
.size aesni_cbc_sha1_enc_ssse3,.-aesni_cbc_sha1_enc_ssse3
___
if ($stitched_decrypt) {{{
# reset
($in0,$out,$len,$key,$ivp,$ctx,$inp)=("%rdi","%rsi","%rdx","%rcx","%r8","%r9","%r10");
-$j=$jj=$r=$sn=$rx=0;
+$j=$jj=$r=$rx=0;
$Xi=4;
# reassign for Atom Silvermont (see above)
sub body_20_39_dec () { # b^d^c
# on entry @T[0]=b^d
return &body_40_59_dec() if ($rx==39);
-
+
my @r=@body_20_39;
unshift (@r,@aes256_dec[$rx]) if (@aes256_dec[$rx]);
.type aesni256_cbc_sha1_dec,\@abi-omnipotent
.align 32
aesni256_cbc_sha1_dec:
+.cfi_startproc
# caller should check for SSSE3 and AES-NI bits
mov OPENSSL_ia32cap_P+0(%rip),%r10d
mov OPENSSL_ia32cap_P+4(%rip),%r11d
$code.=<<___;
jmp aesni256_cbc_sha1_dec_ssse3
ret
+.cfi_endproc
.size aesni256_cbc_sha1_dec,.-aesni256_cbc_sha1_dec
.type aesni256_cbc_sha1_dec_ssse3,\@function,6
.align 32
aesni256_cbc_sha1_dec_ssse3:
+.cfi_startproc
mov `($win64?56:8)`(%rsp),$inp # load 7th argument
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
lea `-104-($win64?10*16:0)`(%rsp),%rsp
+.cfi_adjust_cfa_offset `104+($win64?10*16:0)`
___
$code.=<<___ if ($win64);
movaps %xmm6,96+0(%rsp)
___
$code.=<<___;
lea `104+($win64?10*16:0)`(%rsp),%rsi
+.cfi_cfa_def %rsi,56
mov 0(%rsi),%r15
+.cfi_restore %r15
mov 8(%rsi),%r14
+.cfi_restore %r14
mov 16(%rsi),%r13
+.cfi_restore %r13
mov 24(%rsi),%r12
+.cfi_restore %r12
mov 32(%rsi),%rbp
+.cfi_restore %rbp
mov 40(%rsi),%rbx
+.cfi_restore %rbx
lea 48(%rsi),%rsp
+.cfi_cfa_def %rsp,8
.Lepilogue_dec_ssse3:
ret
+.cfi_endproc
.size aesni256_cbc_sha1_dec_ssse3,.-aesni256_cbc_sha1_dec_ssse3
___
}}}
-$j=$jj=$r=$sn=$rx=0;
+$j=$jj=$r=$rx=0;
if ($avx) {
my ($in0,$out,$len,$key,$ivp,$ctx,$inp)=("%rdi","%rsi","%rdx","%rcx","%r8","%r9","%r10");
.type aesni_cbc_sha1_enc_avx,\@function,6
.align 32
aesni_cbc_sha1_enc_avx:
+.cfi_startproc
mov `($win64?56:8)`(%rsp),$inp # load 7th argument
#shr \$6,$len # debugging artefact
#jz .Lepilogue_avx # debugging artefact
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
lea `-104-($win64?10*16:0)`(%rsp),%rsp
+.cfi_adjust_cfa_offset `104+($win64?10*16:0)`
#mov $in0,$inp # debugging artefact
#lea 64(%rsp),$ctx # debugging artefact
___
$r++; unshift(@rndkey,pop(@rndkey));
};
-sub Xupdate_avx_16_31() # recall that $Xi starts wtih 4
+sub Xupdate_avx_16_31() # recall that $Xi starts with 4
{ use integer;
my $body = shift;
my @insns = (&$body,&$body,&$body,&$body); # 40 instructions
___
$code.=<<___;
lea `104+($win64?10*16:0)`(%rsp),%rsi
+.cfi_def_cfa %rsi,56
mov 0(%rsi),%r15
+.cfi_restore %r15
mov 8(%rsi),%r14
+.cfi_restore %r14
mov 16(%rsi),%r13
+.cfi_restore %r13
mov 24(%rsi),%r12
+.cfi_restore %r12
mov 32(%rsi),%rbp
+.cfi_restore %rbp
mov 40(%rsi),%rbx
+.cfi_restore %rbx
lea 48(%rsi),%rsp
+.cfi_def_cfa %rsp,8
.Lepilogue_avx:
ret
+.cfi_endproc
.size aesni_cbc_sha1_enc_avx,.-aesni_cbc_sha1_enc_avx
___
# reset
($in0,$out,$len,$key,$ivp,$ctx,$inp)=("%rdi","%rsi","%rdx","%rcx","%r8","%r9","%r10");
-$j=$jj=$r=$sn=$rx=0;
+$j=$jj=$r=$rx=0;
$Xi=4;
@aes256_dec = (
.type aesni256_cbc_sha1_dec_avx,\@function,6
.align 32
aesni256_cbc_sha1_dec_avx:
+.cfi_startproc
mov `($win64?56:8)`(%rsp),$inp # load 7th argument
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
lea `-104-($win64?10*16:0)`(%rsp),%rsp
+.cfi_adjust_cfa_offset `104+($win64?10*16:0)`
___
$code.=<<___ if ($win64);
movaps %xmm6,96+0(%rsp)
___
$code.=<<___;
lea `104+($win64?10*16:0)`(%rsp),%rsi
+.cfi_def_cfa %rsi,56
mov 0(%rsi),%r15
+.cfi_restore %r15
mov 8(%rsi),%r14
+.cfi_restore %r14
mov 16(%rsi),%r13
+.cfi_restore %r13
mov 24(%rsi),%r12
+.cfi_restore %r12
mov 32(%rsi),%rbp
+.cfi_restore %rbp
mov 40(%rsi),%rbx
+.cfi_restore %rbx
lea 48(%rsi),%rsp
+.cfi_def_cfa %rsp,8
.Lepilogue_dec_avx:
ret
+.cfi_endproc
.size aesni256_cbc_sha1_dec_avx,.-aesni256_cbc_sha1_dec_avx
___
}}}
.asciz "AESNI-CBC+SHA1 stitch for x86_64, CRYPTOGAMS by <appro\@openssl.org>"
.align 64
___
- {{{
+ if ($shaext) {{{
($in0,$out,$len,$key,$ivp,$ctx,$inp)=("%rdi","%rsi","%rdx","%rcx","%r8","%r9","%r10");
$rounds="%r11d";
.type aesni_cbc_sha1_enc_shaext,\@function,6
.align 32
aesni_cbc_sha1_enc_shaext:
+.cfi_startproc
mov `($win64?56:8)`(%rsp),$inp # load 7th argument
___
$code.=<<___ if ($win64);
- lea `-8-4*16`(%rsp),%rsp
+ lea `-8-10*16`(%rsp),%rsp
movaps %xmm6,-8-10*16(%rax)
movaps %xmm7,-8-9*16(%rax)
movaps %xmm8,-8-8*16(%rax)
mov 240($key),$rounds
sub $in0,$out
movups ($key),$rndkey0 # $key[0]
+ movups ($ivp),$iv # load IV
movups 16($key),$rndkey[0] # forward reference
lea 112($key),$key # size optimization
___
$code.=<<___;
ret
+.cfi_endproc
.size aesni_cbc_sha1_enc_shaext,.-aesni_cbc_sha1_enc_shaext
___
}}}
lea (%rsi,%r10),%r10 # epilogue label
cmp %r10,%rbx # context->Rip>=epilogue label
jae .Lcommon_seh_tail
+___
+$code.=<<___ if ($shaext);
+ lea aesni_cbc_sha1_enc_shaext(%rip),%r10
+ cmp %r10,%rbx
+ jb .Lseh_no_shaext
+ lea (%rax),%rsi
+ lea 512($context),%rdi # &context.Xmm6
+ mov \$20,%ecx
+ .long 0xa548f3fc # cld; rep movsq
+ lea 168(%rax),%rax # adjust stack pointer
+ jmp .Lcommon_seh_tail
+.Lseh_no_shaext:
+___
+$code.=<<___;
lea 96(%rax),%rsi
lea 512($context),%rdi # &context.Xmm6
mov \$20,%ecx
.rva .LSEH_end_aesni_cbc_sha1_enc_avx
.rva .LSEH_info_aesni_cbc_sha1_enc_avx
___
+$code.=<<___ if ($shaext);
+ .rva .LSEH_begin_aesni_cbc_sha1_enc_shaext
+ .rva .LSEH_end_aesni_cbc_sha1_enc_shaext
+ .rva .LSEH_info_aesni_cbc_sha1_enc_shaext
+___
$code.=<<___;
.section .xdata
.align 8
.rva ssse3_handler
.rva .Lprologue_avx,.Lepilogue_avx # HandlerData[]
___
+$code.=<<___ if ($shaext);
+.LSEH_info_aesni_cbc_sha1_enc_shaext:
+ .byte 9,0,0,0
+ .rva ssse3_handler
+ .rva .Lprologue_shaext,.Lepilogue_shaext # HandlerData[]
+___
}
####################################################################
projects / openssl.git / blobdiff