mirror of
https://github.com/openssl/openssl.git
synced 2024-11-27 05:21:51 +08:00
Add VIS3 Montgomery multiplication.
This commit is contained in:
Andy Polyakov
parent
d35c0ff30b
commit
947d78275b
@ -130,7 +130,7 @@ my $x86_elf_asm="$x86_asm:elf";
|
||||
|
||||
my $x86_64_asm="x86_64cpuid.o:x86_64-gcc.o x86_64-mont.o x86_64-mont5.o x86_64-gf2m.o modexp512-x86_64.o::aes-x86_64.o vpaes-x86_64.o bsaes-x86_64.o aesni-x86_64.o aesni-sha1-x86_64.o::md5-x86_64.o:sha1-x86_64.o sha256-x86_64.o sha512-x86_64.o::rc4-x86_64.o rc4-md5-x86_64.o:::wp-x86_64.o:cmll-x86_64.o cmll_misc.o:ghash-x86_64.o:e_padlock-x86_64.o";
|
||||
my $ia64_asm="ia64cpuid.o:bn-ia64.o ia64-mont.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o::rc4-ia64.o rc4_skey.o:::::ghash-ia64.o::void";
|
||||
my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o:des_enc-sparc.o fcrypt_b.o:aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o::md5-sparcv9.o:sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o::::::camellia.o cmll_misc.o cmll_cbc.o cmllt4-sparcv9.o:ghash-sparcv9.o::void";
|
||||
my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o:des_enc-sparc.o fcrypt_b.o:aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o::md5-sparcv9.o:sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o::::::camellia.o cmll_misc.o cmll_cbc.o cmllt4-sparcv9.o:ghash-sparcv9.o::void";
|
||||
my $sparcv8_asm=":sparcv8.o:des_enc-sparc.o fcrypt_b.o:::::::::::::void";
|
||||
my $alpha_asm="alphacpuid.o:bn_asm.o alpha-mont.o:::::sha1-alpha.o:::::::ghash-alpha.o::void";
|
||||
my $mips64_asm=":bn-mips.o mips-mont.o::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o sha512-mips.o::::::::";
|
||||
|
||||
18
TABLE
18
TABLE
@ -174,7 +174,7 @@ $sys_id =
|
||||
$lflags =
|
||||
$bn_ops = BN_LLONG RC2_CHAR RC4_CHUNK DES_INT DES_PTR DES_RISC2 BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -2616,7 +2616,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK_LL DES_PTR DES_RISC1 DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -2649,7 +2649,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -4398,7 +4398,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -4596,7 +4596,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_INT DES_PTR DES_RISC1 DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -5454,7 +5454,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK_LL DES_PTR DES_RISC1 DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -5487,7 +5487,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -5586,7 +5586,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_INT DES_PTR DES_RISC1 DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
@ -5619,7 +5619,7 @@ $sys_id = ULTRASPARC
|
||||
$lflags = -lsocket -lnsl -ldl
|
||||
$bn_ops = BN_LLONG RC4_CHAR RC4_CHUNK DES_INT DES_PTR DES_RISC1 DES_UNROLL BF_PTR
|
||||
$cpuid_obj = sparcv9cap.o sparccpuid.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o
|
||||
$bn_obj = bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o
|
||||
$des_obj = des_enc-sparc.o fcrypt_b.o
|
||||
$aes_obj = aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o
|
||||
$bf_obj =
|
||||
|
||||
@ -77,6 +77,8 @@ sparcv9a-mont.s: asm/sparcv9a-mont.pl
|
||||
$(PERL) asm/sparcv9a-mont.pl $(CFLAGS) > $@
|
||||
sparcv9-mont.s: asm/sparcv9-mont.pl
|
||||
$(PERL) asm/sparcv9-mont.pl $(CFLAGS) > $@
|
||||
vis3-mont.s: asm/vis3-mont.pl
|
||||
$(PERL) asm/vis3-mont.pl $(CFLAGS) > $@
|
||||
|
||||
bn-mips3.o: asm/mips3.s
|
||||
@if [ "$(CC)" = "gcc" ]; then \
|
||||
|
||||
373
crypto/bn/asm/vis3-mont.pl
Normal file
373
crypto/bn/asm/vis3-mont.pl
Normal file
@ -0,0 +1,373 @@
|
||||
#!/usr/bin/env perl
|
||||
|
||||
# ====================================================================
|
||||
# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
|
||||
# project. The module is, however, dual licensed under OpenSSL and
|
||||
# CRYPTOGAMS licenses depending on where you obtain it. For further
|
||||
# details see http://www.openssl.org/~appro/cryptogams/.
|
||||
# ====================================================================
|
||||
|
||||
# October 2012.
|
||||
#
|
||||
# SPARCv9 VIS3 Montgomery multiplicaion procedure suitable for T3 and
|
||||
# onward. There are three new instructions used here: umulxhi,
|
||||
# addxc[cc] and initializing store. On T3 RSA private key operations
|
||||
# are 1.54/1.87/2.11/2.26 times faster for 512/1024/2048/4096-bit key
|
||||
# lengths. This is without dedicated squaring procedure. On T4
|
||||
# corresponding coefficients are 1.47/2.10/2.80/2.90x, which is mostly
|
||||
# for reference purposes, because T4 has dedicated Montgomery
|
||||
# multiplication and squaring *instructions* that deliver even more.
|
||||
|
||||
$bits=32;
|
||||
for (@ARGV) { $bits=64 if (/\-m64/ || /\-xarch\=v9/); }
|
||||
if ($bits==64) { $bias=2047; $frame=192; }
|
||||
else { $bias=0; $frame=112; }
|
||||
|
||||
$code.=<<___ if ($bits==64);
|
||||
.register %g2,#scratch
|
||||
.register %g3,#scratch
|
||||
___
|
||||
$code.=<<___;
|
||||
.section ".text",#alloc,#execinstr
|
||||
___
|
||||
|
||||
($n0,$m0,$m1,$lo0,$hi0, $lo1,$hi1,$aj,$alo,$nj,$nlo,$tj)=
|
||||
(map("%g$_",(1..5)),map("%o$_",(0..5,7)));
|
||||
|
||||
# int bn_mul_mont(
|
||||
$rp="%o0"; # BN_ULONG *rp,
|
||||
$ap="%o1"; # const BN_ULONG *ap,
|
||||
$bp="%o2"; # const BN_ULONG *bp,
|
||||
$np="%o3"; # const BN_ULONG *np,
|
||||
$n0p="%o4"; # const BN_ULONG *n0,
|
||||
$num="%o5"; # int num); # caller ensures that num is even
|
||||
# and >=6
|
||||
$code.=<<___;
|
||||
.globl bn_mul_mont_vis3
|
||||
.align 32
|
||||
bn_mul_mont_vis3:
|
||||
add %sp, $bias, %g4 ! real top of stack
|
||||
sll $num, 2, $num ! size in bytes
|
||||
add $num, 63, %g5
|
||||
andn %g5, 63, %g5 ! buffer size rounded up to 64 bytes
|
||||
add %g5, %g5, %g1
|
||||
add %g5, %g1, %g1 ! 3*buffer size
|
||||
sub %g4, %g1, %g1
|
||||
andn %g1, 63, %g1 ! align at 64 byte
|
||||
sub %g1, $frame, %g1 ! new top of stack
|
||||
sub %g1, %g4, %g1
|
||||
|
||||
save %sp, %g1, %sp
|
||||
___
|
||||
|
||||
# +-------------------------------+<----- %sp
|
||||
# . .
|
||||
# +-------------------------------+<----- aligned at 64 bytes
|
||||
# | __int64 tmp[0] |
|
||||
# +-------------------------------+
|
||||
# . .
|
||||
# . .
|
||||
# +-------------------------------+<----- aligned at 64 bytes
|
||||
# | __int64 ap[1..0] | converted ap[]
|
||||
# +-------------------------------+
|
||||
# | __int64 np[1..0] | converted np[]
|
||||
# +-------------------------------+
|
||||
# | __int64 ap[3..2] |
|
||||
# . .
|
||||
# . .
|
||||
# +-------------------------------+
|
||||
($rp,$ap,$bp,$np,$n0p,$num)=map("%i$_",(0..5));
|
||||
($t0,$t1,$t2,$t3,$cnt,$tp,$bufsz,$anp)=map("%l$_",(0..7));
|
||||
($ovf,$i)=($t0,$t1);
|
||||
$code.=<<___;
|
||||
ld [$n0p+0], $t0 ! pull n0[0..1] value
|
||||
add %sp, $bias+$frame, $tp
|
||||
ld [$n0p+4], $t1
|
||||
add $tp, %g5, $anp
|
||||
ld [$bp+0], $t2 ! m0=bp[0]
|
||||
sllx $t1, 32, $n0
|
||||
ld [$bp+4], $t3
|
||||
or $t0, $n0, $n0
|
||||
add $bp, 8, $bp
|
||||
|
||||
ld [$ap+0], $t0 ! ap[0]
|
||||
sllx $t3, 32, $m0
|
||||
ld [$ap+4], $t1
|
||||
or $t2, $m0, $m0
|
||||
|
||||
ld [$ap+8], $t2 ! ap[1]
|
||||
sllx $t1, 32, $aj
|
||||
ld [$ap+12], $t3
|
||||
or $t0, $aj, $aj
|
||||
add $ap, 16, $ap
|
||||
stxa $aj, [$anp]0xe2 ! converted ap[0]
|
||||
|
||||
mulx $aj, $m0, $lo0 ! ap[0]*bp[0]
|
||||
umulxhi $aj, $m0, $hi0
|
||||
|
||||
ld [$np+0], $t0 ! np[0]
|
||||
sllx $t3, 32, $aj
|
||||
ld [$np+4], $t1
|
||||
or $t2, $aj, $aj
|
||||
|
||||
ld [$np+8], $t2 ! np[1]
|
||||
sllx $t1, 32, $nj
|
||||
ld [$np+12], $t3
|
||||
or $t0, $nj, $nj
|
||||
add $np, 16, $np
|
||||
stx $nj, [$anp+8] ! converted np[0]
|
||||
|
||||
mulx $lo0, $n0, $m1 ! "tp[0]"*n0
|
||||
stx $aj, [$anp+16] ! converted ap[1]
|
||||
|
||||
mulx $aj, $m0, $alo ! ap[1]*bp[0]
|
||||
umulxhi $aj, $m0, $aj ! ahi=aj
|
||||
|
||||
mulx $nj, $m1, $lo1 ! np[0]*m1
|
||||
umulxhi $nj, $m1, $hi1
|
||||
|
||||
sllx $t3, 32, $nj
|
||||
or $t2, $nj, $nj
|
||||
stx $nj, [$anp+24] ! converted np[1]
|
||||
add $anp, 32, $anp
|
||||
|
||||
addcc $lo0, $lo1, $lo1
|
||||
addxc %g0, $hi1, $hi1
|
||||
|
||||
mulx $nj, $m1, $nlo ! np[1]*m1
|
||||
umulxhi $nj, $m1, $nj ! nhi=nj
|
||||
|
||||
ba .L1st
|
||||
sub $num, 24, $cnt ! cnt=num-3
|
||||
|
||||
.align 16
|
||||
.L1st:
|
||||
ld [$ap+0], $t0 ! ap[j]
|
||||
addcc $alo, $hi0, $lo0
|
||||
ld [$ap+4], $t1
|
||||
addxc $aj, %g0, $hi0
|
||||
|
||||
sllx $t1, 32, $aj
|
||||
add $ap, 8, $ap
|
||||
or $t0, $aj, $aj
|
||||
stxa $aj, [$anp]0xe2 ! converted ap[j]
|
||||
|
||||
ld [$np+0], $t2 ! np[j]
|
||||
addcc $nlo, $hi1, $lo1
|
||||
ld [$np+4], $t3
|
||||
addxc $nj, %g0, $hi1 ! nhi=nj
|
||||
|
||||
sllx $t3, 32, $nj
|
||||
add $np, 8, $np
|
||||
mulx $aj, $m0, $alo ! ap[j]*bp[0]
|
||||
or $t2, $nj, $nj
|
||||
umulxhi $aj, $m0, $aj ! ahi=aj
|
||||
stx $nj, [$anp+8] ! converted np[j]
|
||||
add $anp, 16, $anp ! anp++
|
||||
|
||||
mulx $nj, $m1, $nlo ! np[j]*m1
|
||||
addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0]
|
||||
umulxhi $nj, $m1, $nj ! nhi=nj
|
||||
addxc %g0, $hi1, $hi1
|
||||
stxa $lo1, [$tp]0xe2 ! tp[j-1]
|
||||
add $tp, 8, $tp ! tp++
|
||||
|
||||
brnz,pt $cnt, .L1st
|
||||
sub $cnt, 8, $cnt ! j--
|
||||
!.L1st
|
||||
addcc $alo, $hi0, $lo0
|
||||
addxc $aj, %g0, $hi0 ! ahi=aj
|
||||
|
||||
addcc $nlo, $hi1, $lo1
|
||||
addxc $nj, %g0, $hi1
|
||||
addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0]
|
||||
addxc %g0, $hi1, $hi1
|
||||
stxa $lo1, [$tp]0xe2 ! tp[j-1]
|
||||
add $tp, 8, $tp
|
||||
|
||||
addcc $hi0, $hi1, $hi1
|
||||
addxc %g0, %g0, $ovf ! upmost overflow bit
|
||||
stxa $hi1, [$tp]0xe2
|
||||
add $tp, 8, $tp
|
||||
|
||||
ba .Louter
|
||||
sub $num, 16, $i ! i=num-2
|
||||
|
||||
.align 16
|
||||
.Louter:
|
||||
ld [$bp+0], $t2 ! m0=bp[i]
|
||||
ld [$bp+4], $t3
|
||||
|
||||
sub $anp, $num, $anp ! rewind
|
||||
sub $tp, $num, $tp
|
||||
sub $anp, $num, $anp
|
||||
|
||||
add $bp, 8, $bp
|
||||
sllx $t3, 32, $m0
|
||||
ldx [$anp+0], $aj ! ap[0]
|
||||
or $t2, $m0, $m0
|
||||
ldx [$anp+8], $nj ! np[0]
|
||||
|
||||
mulx $aj, $m0, $lo0 ! ap[0]*bp[i]
|
||||
ldx [$tp], $tj ! tp[0]
|
||||
umulxhi $aj, $m0, $hi0
|
||||
ldx [$anp+16], $aj ! ap[1]
|
||||
addcc $lo0, $tj, $lo0 ! ap[0]*bp[i]+tp[0]
|
||||
mulx $aj, $m0, $alo ! ap[1]*bp[i]
|
||||
addxc %g0, $hi0, $hi0
|
||||
mulx $lo0, $n0, $m1 ! tp[0]*n0
|
||||
umulxhi $aj, $m0, $aj ! ahi=aj
|
||||
mulx $nj, $m1, $lo1 ! np[0]*m1
|
||||
umulxhi $nj, $m1, $hi1
|
||||
ldx [$anp+24], $nj ! np[1]
|
||||
add $anp, 32, $anp
|
||||
addcc $lo1, $lo0, $lo1
|
||||
mulx $nj, $m1, $nlo ! np[1]*m1
|
||||
addxc %g0, $hi1, $hi1
|
||||
umulxhi $nj, $m1, $nj ! nhi=nj
|
||||
|
||||
ba .Linner
|
||||
sub $num, 24, $cnt ! cnt=num-3
|
||||
.align 16
|
||||
.Linner:
|
||||
addcc $alo, $hi0, $lo0
|
||||
ldx [$tp+8], $tj ! tp[j]
|
||||
addxc $aj, %g0, $hi0 ! ahi=aj
|
||||
ldx [$anp+0], $aj ! ap[j]
|
||||
addcc $nlo, $hi1, $lo1
|
||||
mulx $aj, $m0, $alo ! ap[j]*bp[i]
|
||||
addxc $nj, %g0, $hi1 ! nhi=nj
|
||||
ldx [$anp+8], $nj ! np[j]
|
||||
add $anp, 16, $anp
|
||||
umulxhi $aj, $m0, $aj ! ahi=aj
|
||||
addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j]
|
||||
mulx $nj, $m1, $nlo ! np[j]*m1
|
||||
addxc %g0, $hi0, $hi0
|
||||
umulxhi $nj, $m1, $nj ! nhi=nj
|
||||
addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j]
|
||||
addxc %g0, $hi1, $hi1
|
||||
stx $lo1, [$tp] ! tp[j-1]
|
||||
add $tp, 8, $tp
|
||||
brnz,pt $cnt, .Linner
|
||||
sub $cnt, 8, $cnt
|
||||
!.Linner
|
||||
ldx [$tp+8], $tj ! tp[j]
|
||||
addcc $alo, $hi0, $lo0
|
||||
addxc $aj, %g0, $hi0 ! ahi=aj
|
||||
addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j]
|
||||
addxc %g0, $hi0, $hi0
|
||||
|
||||
addcc $nlo, $hi1, $lo1
|
||||
addxc $nj, %g0, $hi1 ! nhi=nj
|
||||
addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j]
|
||||
addxc %g0, $hi1, $hi1
|
||||
stx $lo1, [$tp] ! tp[j-1]
|
||||
|
||||
subcc %g0, $ovf, %g0 ! move upmost overflow to CCR.xcc
|
||||
addxccc $hi1, $hi0, $hi1
|
||||
addxc %g0, %g0, $ovf
|
||||
stx $hi1, [$tp+8]
|
||||
add $tp, 16, $tp
|
||||
|
||||
brnz,pt $i, .Louter
|
||||
sub $i, 8, $i
|
||||
|
||||
sub $anp, $num, $anp ! rewind
|
||||
sub $tp, $num, $tp
|
||||
sub $anp, $num, $anp
|
||||
ba .Lsub
|
||||
subcc $num, 8, $cnt ! cnt=num-1 and clear CCR.xcc
|
||||
|
||||
.align 16
|
||||
.Lsub:
|
||||
ldx [$tp], $tj
|
||||
add $tp, 8, $tp
|
||||
ldx [$anp+8], $nj
|
||||
add $anp, 16, $anp
|
||||
subccc $tj, $nj, $t2 ! tp[j]-np[j]
|
||||
srlx $tj, 32, $tj
|
||||
srlx $nj, 32, $nj
|
||||
subccc $tj, $nj, $t3
|
||||
add $rp, 8, $rp
|
||||
st $t2, [$rp-4] ! reverse order
|
||||
st $t3, [$rp-8]
|
||||
brnz,pt $cnt, .Lsub
|
||||
sub $cnt, 8, $cnt
|
||||
|
||||
sub $anp, $num, $anp ! rewind
|
||||
sub $tp, $num, $tp
|
||||
sub $anp, $num, $anp
|
||||
sub $rp, $num, $rp
|
||||
|
||||
subc $ovf, %g0, $ovf ! handle upmost overflow bit
|
||||
and $tp, $ovf, $ap
|
||||
andn $rp, $ovf, $np
|
||||
or $np, $ap, $ap ! ap=borrow?tp:rp
|
||||
ba .Lcopy
|
||||
sub $num, 8, $cnt
|
||||
|
||||
.align 16
|
||||
.Lcopy: ! copy or in-place refresh
|
||||
ld [$ap+0], $t2
|
||||
ld [$ap+4], $t3
|
||||
add $ap, 8, $ap
|
||||
stx %g0, [$tp] ! zap
|
||||
add $tp, 8, $tp
|
||||
stx %g0, [$anp] ! zap
|
||||
stx %g0, [$anp+8]
|
||||
add $anp, 16, $anp
|
||||
st $t3, [$rp+0] ! flip order
|
||||
st $t2, [$rp+4]
|
||||
add $rp, 8, $rp
|
||||
brnz $cnt, .Lcopy
|
||||
sub $cnt, 8, $cnt
|
||||
|
||||
mov 1, %o0
|
||||
ret
|
||||
restore
|
||||
.type bn_mul_mont_vis3, #function
|
||||
.size bn_mul_mont_vis3, .-bn_mul_mont_vis3
|
||||
.asciz "Montgomery Multiplication for SPARCv9 VIS3, CRYPTOGAMS by <appro\@openssl.org>"
|
||||
.align 4
|
||||
___
|
||||
|
||||
# Purpose of these subroutines is to explicitly encode VIS instructions,
|
||||
# so that one can compile the module without having to specify VIS
|
||||
# extentions on compiler command line, e.g. -xarch=v9 vs. -xarch=v9a.
|
||||
# Idea is to reserve for option to produce "universal" binary and let
|
||||
# programmer detect if current CPU is VIS capable at run-time.
|
||||
sub unvis3 {
|
||||
my ($mnemonic,$rs1,$rs2,$rd)=@_;
|
||||
my %bias = ( "g" => 0, "o" => 8, "l" => 16, "i" => 24 );
|
||||
my ($ref,$opf);
|
||||
my %visopf = ( "addxc" => 0x011,
|
||||
"addxccc" => 0x013,
|
||||
"umulxhi" => 0x016 );
|
||||
|
||||
$ref = "$mnemonic\t$rs1,$rs2,$rd";
|
||||
|
||||
if ($opf=$visopf{$mnemonic}) {
|
||||
foreach ($rs1,$rs2,$rd) {
|
||||
return $ref if (!/%([goli])([0-9])/);
|
||||
$_=$bias{$1}+$2;
|
||||
}
|
||||
|
||||
return sprintf ".word\t0x%08x !%s",
|
||||
0x81b00000|$rd<<25|$rs1<<14|$opf<<5|$rs2,
|
||||
$ref;
|
||||
} else {
|
||||
return $ref;
|
||||
}
|
||||
}
|
||||
|
||||
foreach (split("\n",$code)) {
|
||||
s/\`([^\`]*)\`/eval $1/ge;
|
||||
|
||||
s/\b(umulxhi|addxc[c]{0,2})\s+(%[goli][0-7]),\s*(%[goli][0-7]),\s*(%[goli][0-7])/
|
||||
&unvis3($1,$2,$3,$4)
|
||||
/ge;
|
||||
|
||||
print $_,"\n";
|
||||
}
|
||||
|
||||
close STDOUT;
|
||||
@ -15,15 +15,20 @@ unsigned int OPENSSL_sparcv9cap_P[2]={SPARCV9_TICK_PRIVILEGED,0};
|
||||
|
||||
int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num)
|
||||
{
|
||||
int bn_mul_mont_vis3(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num);
|
||||
int bn_mul_mont_fpu(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num);
|
||||
int bn_mul_mont_int(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num);
|
||||
|
||||
if (num>=8 && !(num&1) &&
|
||||
(OPENSSL_sparcv9cap_P[0]&(SPARCV9_PREFER_FPU|SPARCV9_VIS1)) ==
|
||||
(SPARCV9_PREFER_FPU|SPARCV9_VIS1))
|
||||
return bn_mul_mont_fpu(rp,ap,bp,np,n0,num);
|
||||
else
|
||||
return bn_mul_mont_int(rp,ap,bp,np,n0,num);
|
||||
if (!(num&1) && num>=6)
|
||||
{
|
||||
if ((OPENSSL_sparcv9cap_P[0]&SPARCV9_VIS3))
|
||||
return bn_mul_mont_vis3(rp,ap,bp,np,n0,num);
|
||||
else if (num>=8 &&
|
||||
(OPENSSL_sparcv9cap_P[0]&(SPARCV9_PREFER_FPU|SPARCV9_VIS1)) ==
|
||||
(SPARCV9_PREFER_FPU|SPARCV9_VIS1))
|
||||
return bn_mul_mont_fpu(rp,ap,bp,np,n0,num);
|
||||
}
|
||||
return bn_mul_mont_int(rp,ap,bp,np,n0,num);
|
||||
}
|
||||
|
||||
unsigned long _sparcv9_rdtick(void);
|
||||
@ -35,7 +40,7 @@ unsigned long _sparcv9_rdcfr(void);
|
||||
void _sparcv9_vis3_probe(void);
|
||||
unsigned long _sparcv9_random(void);
|
||||
size_t _sparcv9_vis1_instrument_bus(unsigned int *,size_t);
|
||||
size_t _sparcv8_vis1_instrument_bus2(unsigned int *,size_t,size_t);
|
||||
size_t _sparcv9_vis1_instrument_bus2(unsigned int *,size_t,size_t);
|
||||
|
||||
unsigned long OPENSSL_rdtsc(void)
|
||||
{
|
||||
@ -51,7 +56,7 @@ unsigned long OPENSSL_rdtsc(void)
|
||||
|
||||
size_t OPENSSL_instrument_bus(unsigned int *out,size_t cnt)
|
||||
{
|
||||
if (OPENSSL_sparcv9cap_P[0]&(SPARCV9_TICK_PRIVILEGED|SPARCV9_BLK) ==
|
||||
if ((OPENSSL_sparcv9cap_P[0]&(SPARCV9_TICK_PRIVILEGED|SPARCV9_BLK)) ==
|
||||
SPARCV9_BLK)
|
||||
return _sparcv9_vis1_instrument_bus(out,cnt);
|
||||
else
|
||||
@ -60,7 +65,7 @@ size_t OPENSSL_instrument_bus(unsigned int *out,size_t cnt)
|
||||
|
||||
size_t OPENSSL_instrument_bus2(unsigned int *out,size_t cnt,size_t max)
|
||||
{
|
||||
if (OPENSSL_sparcv9cap_P[0]&(SPARCV9_TICK_PRIVILEGED|SPARCV9_BLK) ==
|
||||
if ((OPENSSL_sparcv9cap_P[0]&(SPARCV9_TICK_PRIVILEGED|SPARCV9_BLK)) ==
|
||||
SPARCV9_BLK)
|
||||
return _sparcv9_vis1_instrument_bus2(out,cnt,max);
|
||||
else
|
||||
|
||||
Loading…
Reference in New Issue
Block a user