generic-library/openssl
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

sha1-586.pl: update from master.

Browse Source
This commit is contained in:
Andy Polyakov 2013-05-19 21:32:29 +02:00
parent e72f7677bf
commit 5c8ae6d93e
1 changed files with 84 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

131
crypto/sha/asm/sha1-586.pl
View File

@ -1,7 +1,7 @@
#!/usr/bin/env perl #!/usr/bin/env perl
# ==================================================================== # ====================================================================
# [Re]written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL # [Re]written by Andy Polyakov <appro@openssl.org> for the OpenSSL
# project. The module is, however, dual licensed under OpenSSL and # project. The module is, however, dual licensed under OpenSSL and
# CRYPTOGAMS licenses depending on where you obtain it. For further # CRYPTOGAMS licenses depending on where you obtain it. For further
# details see http://www.openssl.org/~appro/cryptogams/. # details see http://www.openssl.org/~appro/cryptogams/.
@ -88,10 +88,13 @@
# PIII 11.5 - # PIII 11.5 -
# P4 10.6 - # P4 10.6 -
# AMD K8 7.1 - # AMD K8 7.1 -
# Core2 7.3 6.1/+20% - # Core2 7.3 6.0/+22% -
# Atom 12.5 9.5(*)/+32% - # Atom 12.5 9.3(*)/+35% -
# Westmere 7.3 5.6/+30% - # Westmere 7.3 5.5/+33% -
# Sandy Bridge 8.8 6.2/+40% 5.1(**)/+70% # Sandy Bridge 8.8 6.2/+40% 5.1(**)/+73%
# Ivy Bridge 7.2 4.8/+51% 4.7(**)/+53%
# Bulldozer 11.6 6.0/+92%
# VIA Nano 10.6 7.4/+43%
# #
# (*) Loop is 1056 instructions long and expected result is ~8.25. # (*) Loop is 1056 instructions long and expected result is ~8.25.
# It remains mystery [to me] why ILP is limited to 1.7. # It remains mystery [to me] why ILP is limited to 1.7.
@ -116,6 +119,10 @@ $ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32n" &&
`nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/ && `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/ &&
$1>=2.03); # first version supporting AVX $1>=2.03); # first version supporting AVX
$ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32" &&
`ml 2>&1` =~ /Version ([0-9]+)\./ &&
$1>=10); # first version supporting AVX
&external_label("OPENSSL_ia32cap_P") if ($xmm); &external_label("OPENSSL_ia32cap_P") if ($xmm);
@ -416,6 +423,7 @@ my $Xi=4; # 4xSIMD Xupdate round, start pre-seeded
my @X=map("xmm$_",(4..7,0..3)); # pre-seeded for $Xi=4 my @X=map("xmm$_",(4..7,0..3)); # pre-seeded for $Xi=4
my @V=($A,$B,$C,$D,$E); my @V=($A,$B,$C,$D,$E);
my $j=0; # hash round my $j=0; # hash round
my $rx=0;
my @T=($T,$tmp1); my @T=($T,$tmp1);
my $inp; my $inp;
@ -501,8 +509,11 @@ my $_ror=sub { &ror(@_) };
&movdqa (&QWP(0+16,"esp"),@X[-3&7]); &movdqa (&QWP(0+16,"esp"),@X[-3&7]);
&psubd (@X[-3&7],@X[3]); &psubd (@X[-3&7],@X[3]);
&movdqa (&QWP(0+32,"esp"),@X[-2&7]); &movdqa (&QWP(0+32,"esp"),@X[-2&7]);
&mov (@T[1],$C);
&psubd (@X[-2&7],@X[3]); &psubd (@X[-2&7],@X[3]);
&xor (@T[1],$D);
&movdqa (@X[0],@X[-3&7]); &movdqa (@X[0],@X[-3&7]);
&and (@T[0],@T[1]);
&jmp (&label("loop")); &jmp (&label("loop"));
###################################################################### ######################################################################
@ -609,7 +620,7 @@ sub Xupdate_ssse3_16_31() # recall that $Xi starts wtih 4
sub Xupdate_ssse3_32_79() sub Xupdate_ssse3_32_79()
{ use integer; { use integer;
my $body = shift; my $body = shift;
my @insns = (&$body,&$body,&$body,&$body); # 32 to 48 instructions my @insns = (&$body,&$body,&$body,&$body); # 32 to 44 instructions
my ($a,$b,$c,$d,$e); my ($a,$b,$c,$d,$e);
&movdqa (@X[2],@X[-1&7]) if ($Xi==8); &movdqa (@X[2],@X[-1&7]) if ($Xi==8);
@ -743,51 +754,66 @@ sub Xtail_ssse3()
foreach (@insns) { eval; } foreach (@insns) { eval; }
} }
sub body_00_19 () { sub body_00_19 () { # ((c^d)&b)^d
# on start @T[0]=(c^d)&b
return &body_20_39() if ($rx==19); $rx++;
(
'($a,$b,$c,$d,$e)=@V;'.
'&$_ror ($b,$j?7:2);', # $b>>>2
'&xor (@T[0],$d);',
'&mov (@T[1],$a);', # $b in next round
'&add ($e,&DWP(4*($j&15),"esp"));', # X[]+K xfer
'&xor ($b,$c);', # $c^$d for next round
'&$_rol ($a,5);',
'&add ($e,@T[0]);',
'&and (@T[1],$b);', # ($b&($c^$d)) for next round
'&xor ($b,$c);', # restore $b
'&add ($e,$a);' .'$j++; unshift(@V,pop(@V)); unshift(@T,pop(@T));'
);
}
sub body_20_39 () { # b^d^c
# on entry @T[0]=b^d
return &body_40_59() if ($rx==39); $rx++;
( (
'($a,$b,$c,$d,$e)=@V;'. '($a,$b,$c,$d,$e)=@V;'.
'&add ($e,&DWP(4*($j&15),"esp"));', # X[]+K xfer '&add ($e,&DWP(4*($j&15),"esp"));', # X[]+K xfer
'&xor ($c,$d);', '&xor (@T[0],$d) if($j==19);'.
'&xor (@T[0],$c) if($j> 19);', # ($b^$d^$c)
'&mov (@T[1],$a);', # $b in next round '&mov (@T[1],$a);', # $b in next round
'&$_rol ($a,5);',
'&and (@T[0],$c);', # ($b&($c^$d))
'&xor ($c,$d);', # restore $c
'&xor (@T[0],$d);',
'&add ($e,$a);',
'&$_ror ($b,$j?7:2);', # $b>>>2
'&add ($e,@T[0]);' .'$j++; unshift(@V,pop(@V)); unshift(@T,pop(@T));'
);
}
sub body_20_39 () {
(
'($a,$b,$c,$d,$e)=@V;'.
'&add ($e,&DWP(4*($j++&15),"esp"));', # X[]+K xfer
'&xor (@T[0],$d);', # ($b^$d)
'&mov (@T[1],$a);', # $b in next round
'&$_rol ($a,5);',
'&xor (@T[0],$c);', # ($b^$d^$c)
'&add ($e,$a);',
'&$_ror ($b,7);', # $b>>>2
'&add ($e,@T[0]);' .'unshift(@V,pop(@V)); unshift(@T,pop(@T));'
);
}
sub body_40_59 () {
(
'($a,$b,$c,$d,$e)=@V;'.
'&mov (@T[1],$c);',
'&xor ($c,$d);',
'&add ($e,&DWP(4*($j++&15),"esp"));', # X[]+K xfer
'&and (@T[1],$d);',
'&and (@T[0],$c);', # ($b&($c^$d))
'&$_ror ($b,7);', # $b>>>2
'&add ($e,@T[1]);',
'&mov (@T[1],$a);', # $b in next round
'&$_rol ($a,5);', '&$_rol ($a,5);',
'&add ($e,@T[0]);', '&add ($e,@T[0]);',
'&xor ($c,$d);', # restore $c '&xor (@T[1],$c) if ($j< 79);', # $b^$d for next round
'&add ($e,$a);' .'unshift(@V,pop(@V)); unshift(@T,pop(@T));'
'&$_ror ($b,7);', # $b>>>2
'&add ($e,$a);' .'$j++; unshift(@V,pop(@V)); unshift(@T,pop(@T));'
);
}
sub body_40_59 () { # ((b^c)&(c^d))^c
# on entry @T[0]=(b^c), (c^=d)
$rx++;
(
'($a,$b,$c,$d,$e)=@V;'.
'&add ($e,&DWP(4*($j&15),"esp"));', # X[]+K xfer
'&and (@T[0],$c) if ($j>=40);', # (b^c)&(c^d)
'&xor ($c,$d) if ($j>=40);', # restore $c
'&$_ror ($b,7);', # $b>>>2
'&mov (@T[1],$a);', # $b for next round
'&xor (@T[0],$c);',
'&$_rol ($a,5);',
'&add ($e,@T[0]);',
'&xor (@T[1],$c) if ($j==59);'.
'&xor (@T[1],$b) if ($j< 59);', # b^c for next round
'&xor ($b,$c) if ($j< 59);', # c^d for next round
'&add ($e,$a);' .'$j++; unshift(@V,pop(@V)); unshift(@T,pop(@T));'
); );
} }
@ -825,10 +851,13 @@ sub body_40_59 () {
&mov (&DWP(4,@T[1]),@T[0]); &mov (&DWP(4,@T[1]),@T[0]);
&add ($E,&DWP(16,@T[1])); &add ($E,&DWP(16,@T[1]));
&mov (&DWP(8,@T[1]),$C); &mov (&DWP(8,@T[1]),$C);
&mov ($B,@T[0]); &mov ($B,$C);
&mov (&DWP(12,@T[1]),$D); &mov (&DWP(12,@T[1]),$D);
&xor ($B,$D);
&mov (&DWP(16,@T[1]),$E); &mov (&DWP(16,@T[1]),$E);
&and ($B,@T[0]);
&movdqa (@X[0],@X[-3&7]); &movdqa (@X[0],@X[-3&7]);
&xchg ($B,@T[0]);
&jmp (&label("loop")); &jmp (&label("loop"));
@ -853,6 +882,8 @@ sub body_40_59 () {
&function_end("_sha1_block_data_order_ssse3"); &function_end("_sha1_block_data_order_ssse3");
$rx=0; # reset
if ($ymm) { if ($ymm) {
my $Xi=4; # 4xSIMD Xupdate round, start pre-seeded my $Xi=4; # 4xSIMD Xupdate round, start pre-seeded
my @X=map("xmm$_",(4..7,0..3)); # pre-seeded for $Xi=4 my @X=map("xmm$_",(4..7,0..3)); # pre-seeded for $Xi=4
@ -940,8 +971,11 @@ my $_ror=sub { &shrd(@_[0],@_) };
&vpaddd (@X[1],@X[-3&7],@X[3]); &vpaddd (@X[1],@X[-3&7],@X[3]);
&vpaddd (@X[2],@X[-2&7],@X[3]); &vpaddd (@X[2],@X[-2&7],@X[3]);
&vmovdqa(&QWP(0,"esp"),@X[0]); # X[]+K xfer to IALU &vmovdqa(&QWP(0,"esp"),@X[0]); # X[]+K xfer to IALU
&mov (@T[1],$C);
&vmovdqa(&QWP(0+16,"esp"),@X[1]); &vmovdqa(&QWP(0+16,"esp"),@X[1]);
&xor (@T[1],$D);
&vmovdqa(&QWP(0+32,"esp"),@X[2]); &vmovdqa(&QWP(0+32,"esp"),@X[2]);
&and (@T[0],@T[1]);
&jmp (&label("loop")); &jmp (&label("loop"));
sub Xupdate_avx_16_31() # recall that $Xi starts wtih 4 sub Xupdate_avx_16_31() # recall that $Xi starts wtih 4
@ -1025,7 +1059,7 @@ sub Xupdate_avx_16_31() # recall that $Xi starts wtih 4
sub Xupdate_avx_32_79() sub Xupdate_avx_32_79()
{ use integer; { use integer;
my $body = shift; my $body = shift;
my @insns = (&$body,&$body,&$body,&$body); # 32 to 48 instructions my @insns = (&$body,&$body,&$body,&$body); # 32 to 44 instructions
my ($a,$b,$c,$d,$e); my ($a,$b,$c,$d,$e);
&vpalignr(@X[2],@X[-1&7],@X[-2&7],8); # compose "X[-6]" &vpalignr(@X[2],@X[-1&7],@X[-2&7],8); # compose "X[-6]"
@ -1188,10 +1222,13 @@ sub Xtail_avx()
&add ($D,&DWP(12,@T[1])); &add ($D,&DWP(12,@T[1]));
&mov (&DWP(4,@T[1]),@T[0]); &mov (&DWP(4,@T[1]),@T[0]);
&add ($E,&DWP(16,@T[1])); &add ($E,&DWP(16,@T[1]));
&mov ($B,$C);
&mov (&DWP(8,@T[1]),$C); &mov (&DWP(8,@T[1]),$C);
&mov ($B,@T[0]); &xor ($B,$D);
&mov (&DWP(12,@T[1]),$D); &mov (&DWP(12,@T[1]),$D);
&and ($B,@T[0]);
&mov (&DWP(16,@T[1]),$E); &mov (&DWP(16,@T[1]),$E);
&xchg ($B,@T[0]);
&jmp (&label("loop")); &jmp (&label("loop"));