aboutsummaryrefslogtreecommitdiff
path: root/openssl/crypto/bn
diff options
context:
space:
mode:
authorMike DePaulo <mikedep333@gmail.com>2015-01-10 12:26:41 -0500
committerMike DePaulo <mikedep333@gmail.com>2015-01-10 15:20:57 -0500
commitee914bf036b78dcbde9bf694794c15482d721028 (patch)
tree13793fdc02dbf47bd4dd5e93861bccac71d58887 /openssl/crypto/bn
parentfaa5026e540d03f858265b2796054d685f687383 (diff)
downloadvcxsrv-ee914bf036b78dcbde9bf694794c15482d721028.tar.gz
vcxsrv-ee914bf036b78dcbde9bf694794c15482d721028.tar.bz2
vcxsrv-ee914bf036b78dcbde9bf694794c15482d721028.zip
Update openssl to version openssl-1.0.1k
Conflicts: openssl/Makefile
Diffstat (limited to 'openssl/crypto/bn')
-rw-r--r--openssl/crypto/bn/asm/mips.pl611
-rw-r--r--openssl/crypto/bn/asm/mips3.s2201
-rw-r--r--openssl/crypto/bn/asm/x86_64-gcc.c34
-rw-r--r--openssl/crypto/bn/bn.h4
-rw-r--r--openssl/crypto/bn/bn_asm.c16
-rw-r--r--openssl/crypto/bn/bn_ctx.c2
-rw-r--r--openssl/crypto/bn/bn_div.c8
-rw-r--r--openssl/crypto/bn/bntest.c178
8 files changed, 319 insertions, 2735 deletions
diff --git a/openssl/crypto/bn/asm/mips.pl b/openssl/crypto/bn/asm/mips.pl
index d2f3ef7bb..215c9a748 100644
--- a/openssl/crypto/bn/asm/mips.pl
+++ b/openssl/crypto/bn/asm/mips.pl
@@ -1872,6 +1872,41 @@ ___
($a_4,$a_5,$a_6,$a_7)=($b_0,$b_1,$b_2,$b_3);
+sub add_c2 () {
+my ($hi,$lo,$c0,$c1,$c2,
+ $warm, # !$warm denotes first call with specific sequence of
+ # $c_[XYZ] when there is no Z-carry to accumulate yet;
+ $an,$bn # these two are arguments for multiplication which
+ # result is used in *next* step [which is why it's
+ # commented as "forward multiplication" below];
+ )=@_;
+$code.=<<___;
+ mflo $lo
+ mfhi $hi
+ $ADDU $c0,$lo
+ sltu $at,$c0,$lo
+ $MULTU $an,$bn # forward multiplication
+ $ADDU $c0,$lo
+ $ADDU $at,$hi
+ sltu $lo,$c0,$lo
+ $ADDU $c1,$at
+ $ADDU $hi,$lo
+___
+$code.=<<___ if (!$warm);
+ sltu $c2,$c1,$at
+ $ADDU $c1,$hi
+ sltu $hi,$c1,$hi
+ $ADDU $c2,$hi
+___
+$code.=<<___ if ($warm);
+ sltu $at,$c1,$at
+ $ADDU $c1,$hi
+ $ADDU $c2,$at
+ sltu $hi,$c1,$hi
+ $ADDU $c2,$hi
+___
+}
+
$code.=<<___;
.align 5
@@ -1920,21 +1955,10 @@ $code.=<<___;
sltu $at,$c_2,$t_1
$ADDU $c_3,$t_2,$at
$ST $c_2,$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_1 # mul_add_c(a[1],b[1],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_1,$a_1); # mul_add_c(a[1],b[1],c3,c1,c2);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_3,$t_1
@@ -1945,67 +1969,19 @@ $code.=<<___;
sltu $at,$c_1,$t_2
$ADDU $c_2,$at
$ST $c_3,2*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_3,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_2 # mul_add_c2(a[1],b[2],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_4,$a_0 # mul_add_c2(a[4],b[0],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
+___
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0,
+ $a_1,$a_2); # mul_add_c2(a[1],b[2],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_4,$a_0); # mul_add_c2(a[4],b[0],c2,c3,c1);
+$code.=<<___;
$ST $c_1,3*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_1,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_3,$a_1 # mul_add_c2(a[3],b[1],c2,c3,c1);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_1,$at
- $MULTU $a_2,$a_2 # mul_add_c(a[2],b[2],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
+___
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0,
+ $a_3,$a_1); # mul_add_c2(a[3],b[1],c2,c3,c1);
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1,
+ $a_2,$a_2); # mul_add_c(a[2],b[2],c2,c3,c1);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_2,$t_1
@@ -2016,97 +1992,23 @@ $code.=<<___;
sltu $at,$c_3,$t_2
$ADDU $c_1,$at
$ST $c_2,4*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_4 # mul_add_c2(a[1],b[4],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_2,$at
- $MULTU $a_2,$a_3 # mul_add_c2(a[2],b[3],c3,c1,c2);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $MULTU $a_6,$a_0 # mul_add_c2(a[6],b[0],c1,c2,c3);
- $ADDU $c_2,$at
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_1,$a_4); # mul_add_c2(a[1],b[4],c3,c1,c2);
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1,
+ $a_2,$a_3); # mul_add_c2(a[2],b[3],c3,c1,c2);
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1,
+ $a_6,$a_0); # mul_add_c2(a[6],b[0],c1,c2,c3);
+$code.=<<___;
$ST $c_3,5*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_3,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_5,$a_1 # mul_add_c2(a[5],b[1],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_4,$a_2 # mul_add_c2(a[4],b[2],c1,c2,c3);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_3,$a_3 # mul_add_c(a[3],b[3],c1,c2,c3);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
+___
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0,
+ $a_5,$a_1); # mul_add_c2(a[5],b[1],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_4,$a_2); # mul_add_c2(a[4],b[2],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_3,$a_3); # mul_add_c(a[3],b[3],c1,c2,c3);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_1,$t_1
@@ -2117,112 +2019,25 @@ $code.=<<___;
sltu $at,$c_2,$t_2
$ADDU $c_3,$at
$ST $c_1,6*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_1,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_6 # mul_add_c2(a[1],b[6],c2,c3,c1);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_1,$at
- $MULTU $a_2,$a_5 # mul_add_c2(a[2],b[5],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_1,$at
- $MULTU $a_3,$a_4 # mul_add_c2(a[3],b[4],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_1,$at
- $MULTU $a_7,$a_1 # mul_add_c2(a[7],b[1],c3,c1,c2);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
+___
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0,
+ $a_1,$a_6); # mul_add_c2(a[1],b[6],c2,c3,c1);
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1,
+ $a_2,$a_5); # mul_add_c2(a[2],b[5],c2,c3,c1);
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1,
+ $a_3,$a_4); # mul_add_c2(a[3],b[4],c2,c3,c1);
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1,
+ $a_7,$a_1); # mul_add_c2(a[7],b[1],c3,c1,c2);
+$code.=<<___;
$ST $c_2,7*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_6,$a_2 # mul_add_c2(a[6],b[2],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_2,$at
- $MULTU $a_5,$a_3 # mul_add_c2(a[5],b[3],c3,c1,c2);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_2,$at
- $MULTU $a_4,$a_4 # mul_add_c(a[4],b[4],c3,c1,c2);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_6,$a_2); # mul_add_c2(a[6],b[2],c3,c1,c2);
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1,
+ $a_5,$a_3); # mul_add_c2(a[5],b[3],c3,c1,c2);
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1,
+ $a_4,$a_4); # mul_add_c(a[4],b[4],c3,c1,c2);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_3,$t_1
@@ -2233,82 +2048,21 @@ $code.=<<___;
sltu $at,$c_1,$t_2
$ADDU $c_2,$at
$ST $c_3,8*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_3,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_3,$a_6 # mul_add_c2(a[3],b[6],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_4,$a_5 # mul_add_c2(a[4],b[5],c1,c2,c3);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_7,$a_3 # mul_add_c2(a[7],b[3],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
+___
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0,
+ $a_3,$a_6); # mul_add_c2(a[3],b[6],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_4,$a_5); # mul_add_c2(a[4],b[5],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_7,$a_3); # mul_add_c2(a[7],b[3],c2,c3,c1);
+$code.=<<___;
$ST $c_1,9*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_1,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_6,$a_4 # mul_add_c2(a[6],b[4],c2,c3,c1);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_1,$at
- $MULTU $a_5,$a_5 # mul_add_c(a[5],b[5],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
+___
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0,
+ $a_6,$a_4); # mul_add_c2(a[6],b[4],c2,c3,c1);
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1,
+ $a_5,$a_5); # mul_add_c(a[5],b[5],c2,c3,c1);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_2,$t_1
@@ -2319,52 +2073,17 @@ $code.=<<___;
sltu $at,$c_3,$t_2
$ADDU $c_1,$at
$ST $c_2,10*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_5,$a_6 # mul_add_c2(a[5],b[6],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_2,$at
- $MULTU $a_7,$a_5 # mul_add_c2(a[7],b[5],c1,c2,c3);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_5,$a_6); # mul_add_c2(a[5],b[6],c3,c1,c2);
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1,
+ $a_7,$a_5); # mul_add_c2(a[7],b[5],c1,c2,c3);
+$code.=<<___;
$ST $c_3,11*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_3,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_6,$a_6 # mul_add_c(a[6],b[6],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
+___
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0,
+ $a_6,$a_6); # mul_add_c(a[6],b[6],c1,c2,c3);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_1,$t_1
@@ -2375,21 +2094,10 @@ $code.=<<___;
sltu $at,$c_2,$t_2
$ADDU $c_3,$at
$ST $c_1,12*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_1,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_7,$a_7 # mul_add_c(a[7],b[7],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
+___
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0,
+ $a_7,$a_7); # mul_add_c(a[7],b[7],c3,c1,c2);
+$code.=<<___;
$ST $c_2,13*$BNSZ($a0)
mflo $t_1
@@ -2457,21 +2165,10 @@ $code.=<<___;
sltu $at,$c_2,$t_1
$ADDU $c_3,$t_2,$at
$ST $c_2,$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_1 # mul_add_c(a[1],b[1],c3,c1,c2);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_1,$a_1); # mul_add_c(a[1],b[1],c3,c1,c2);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_3,$t_1
@@ -2482,52 +2179,17 @@ $code.=<<___;
sltu $at,$c_1,$t_2
$ADDU $c_2,$at
$ST $c_3,2*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_3,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_1,$a_2 # mul_add_c(a2[1],b[2],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
- mflo $t_1
- mfhi $t_2
- slt $at,$t_2,$zero
- $ADDU $c_3,$at
- $MULTU $a_3,$a_1 # mul_add_c2(a[3],b[1],c2,c3,c1);
- $SLL $t_2,1
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_1,$t_1
- sltu $at,$c_1,$t_1
- $ADDU $t_2,$at
- $ADDU $c_2,$t_2
- sltu $at,$c_2,$t_2
- $ADDU $c_3,$at
+___
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0,
+ $a_1,$a_2); # mul_add_c2(a2[1],b[2],c1,c2,c3);
+ &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1,
+ $a_3,$a_1); # mul_add_c2(a[3],b[1],c2,c3,c1);
+$code.=<<___;
$ST $c_1,3*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_1,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_2,$a_2 # mul_add_c(a[2],b[2],c2,c3,c1);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_2,$t_1
- sltu $at,$c_2,$t_1
- $ADDU $t_2,$at
- $ADDU $c_3,$t_2
- sltu $at,$c_3,$t_2
- $ADDU $c_1,$at
+___
+ &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0,
+ $a_2,$a_2); # mul_add_c(a[2],b[2],c2,c3,c1);
+$code.=<<___;
mflo $t_1
mfhi $t_2
$ADDU $c_2,$t_1
@@ -2538,21 +2200,10 @@ $code.=<<___;
sltu $at,$c_3,$t_2
$ADDU $c_1,$at
$ST $c_2,4*$BNSZ($a0)
-
- mflo $t_1
- mfhi $t_2
- slt $c_2,$t_2,$zero
- $SLL $t_2,1
- $MULTU $a_3,$a_3 # mul_add_c(a[3],b[3],c1,c2,c3);
- slt $a2,$t_1,$zero
- $ADDU $t_2,$a2
- $SLL $t_1,1
- $ADDU $c_3,$t_1
- sltu $at,$c_3,$t_1
- $ADDU $t_2,$at
- $ADDU $c_1,$t_2
- sltu $at,$c_1,$t_2
- $ADDU $c_2,$at
+___
+ &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0,
+ $a_3,$a_3); # mul_add_c(a[3],b[3],c1,c2,c3);
+$code.=<<___;
$ST $c_3,5*$BNSZ($a0)
mflo $t_1
diff --git a/openssl/crypto/bn/asm/mips3.s b/openssl/crypto/bn/asm/mips3.s
deleted file mode 100644
index dca4105c7..000000000
--- a/openssl/crypto/bn/asm/mips3.s
+++ /dev/null
@@ -1,2201 +0,0 @@
-.rdata
-.asciiz "mips3.s, Version 1.1"
-.asciiz "MIPS III/IV ISA artwork by Andy Polyakov <appro@fy.chalmers.se>"
-
-/*
- * ====================================================================
- * Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
- * project.
- *
- * Rights for redistribution and usage in source and binary forms are
- * granted according to the OpenSSL license. Warranty of any kind is
- * disclaimed.
- * ====================================================================
- */
-
-/*
- * This is my modest contributon to the OpenSSL project (see
- * http://www.openssl.org/ for more information about it) and is
- * a drop-in MIPS III/IV ISA replacement for crypto/bn/bn_asm.c
- * module. For updates see http://fy.chalmers.se/~appro/hpe/.
- *
- * The module is designed to work with either of the "new" MIPS ABI(5),
- * namely N32 or N64, offered by IRIX 6.x. It's not ment to work under
- * IRIX 5.x not only because it doesn't support new ABIs but also
- * because 5.x kernels put R4x00 CPU into 32-bit mode and all those
- * 64-bit instructions (daddu, dmultu, etc.) found below gonna only
- * cause illegal instruction exception:-(
- *
- * In addition the code depends on preprocessor flags set up by MIPSpro
- * compiler driver (either as or cc) and therefore (probably?) can't be
- * compiled by the GNU assembler. GNU C driver manages fine though...
- * I mean as long as -mmips-as is specified or is the default option,
- * because then it simply invokes /usr/bin/as which in turn takes
- * perfect care of the preprocessor definitions. Another neat feature
- * offered by the MIPSpro assembler is an optimization pass. This gave
- * me the opportunity to have the code looking more regular as all those
- * architecture dependent instruction rescheduling details were left to
- * the assembler. Cool, huh?
- *
- * Performance improvement is astonishing! 'apps/openssl speed rsa dsa'
- * goes way over 3 times faster!
- *
- * <appro@fy.chalmers.se>
- */
-#include <asm.h>
-#include <regdef.h>
-
-#if _MIPS_ISA>=4
-#define MOVNZ(cond,dst,src) \
- movn dst,src,cond
-#else
-#define MOVNZ(cond,dst,src) \
- .set noreorder; \
- bnezl cond,.+8; \
- move dst,src; \
- .set reorder
-#endif
-
-.text
-
-.set noat
-.set reorder
-
-#define MINUS4 v1
-
-.align 5
-LEAF(bn_mul_add_words)
- .set noreorder
- bgtzl a2,.L_bn_mul_add_words_proceed
- ld t0,0(a1)
- jr ra
- move v0,zero
- .set reorder
-
-.L_bn_mul_add_words_proceed:
- li MINUS4,-4
- and ta0,a2,MINUS4
- move v0,zero
- beqz ta0,.L_bn_mul_add_words_tail
-
-.L_bn_mul_add_words_loop:
- dmultu t0,a3
- ld t1,0(a0)
- ld t2,8(a1)
- ld t3,8(a0)
- ld ta0,16(a1)
- ld ta1,16(a0)
- daddu t1,v0
- sltu v0,t1,v0 /* All manuals say it "compares 32-bit
- * values", but it seems to work fine
- * even on 64-bit registers. */
- mflo AT
- mfhi t0
- daddu t1,AT
- daddu v0,t0
- sltu AT,t1,AT
- sd t1,0(a0)
- daddu v0,AT
-
- dmultu t2,a3
- ld ta2,24(a1)
- ld ta3,24(a0)
- daddu t3,v0
- sltu v0,t3,v0
- mflo AT
- mfhi t2
- daddu t3,AT
- daddu v0,t2
- sltu AT,t3,AT
- sd t3,8(a0)
- daddu v0,AT
-
- dmultu ta0,a3
- subu a2,4
- PTR_ADD a0,32
- PTR_ADD a1,32
- daddu ta1,v0
- sltu v0,ta1,v0
- mflo AT
- mfhi ta0
- daddu ta1,AT
- daddu v0,ta0
- sltu AT,ta1,AT
- sd ta1,-16(a0)
- daddu v0,AT
-
-
- dmultu ta2,a3
- and ta0,a2,MINUS4
- daddu ta3,v0
- sltu v0,ta3,v0
- mflo AT
- mfhi ta2
- daddu ta3,AT
- daddu v0,ta2
- sltu AT,ta3,AT
- sd ta3,-8(a0)
- daddu v0,AT
- .set noreorder
- bgtzl ta0,.L_bn_mul_add_words_loop
- ld t0,0(a1)
-
- bnezl a2,.L_bn_mul_add_words_tail
- ld t0,0(a1)
- .set reorder
-
-.L_bn_mul_add_words_return:
- jr ra
-
-.L_bn_mul_add_words_tail:
- dmultu t0,a3
- ld t1,0(a0)
- subu a2,1
- daddu t1,v0
- sltu v0,t1,v0
- mflo AT
- mfhi t0
- daddu t1,AT
- daddu v0,t0
- sltu AT,t1,AT
- sd t1,0(a0)
- daddu v0,AT
- beqz a2,.L_bn_mul_add_words_return
-
- ld t0,8(a1)
- dmultu t0,a3
- ld t1,8(a0)
- subu a2,1
- daddu t1,v0
- sltu v0,t1,v0
- mflo AT
- mfhi t0
- daddu t1,AT
- daddu v0,t0
- sltu AT,t1,AT
- sd t1,8(a0)
- daddu v0,AT
- beqz a2,.L_bn_mul_add_words_return
-
- ld t0,16(a1)
- dmultu t0,a3
- ld t1,16(a0)
- daddu t1,v0
- sltu v0,t1,v0
- mflo AT
- mfhi t0
- daddu t1,AT
- daddu v0,t0
- sltu AT,t1,AT
- sd t1,16(a0)
- daddu v0,AT
- jr ra
-END(bn_mul_add_words)
-
-.align 5
-LEAF(bn_mul_words)
- .set noreorder
- bgtzl a2,.L_bn_mul_words_proceed
- ld t0,0(a1)
- jr ra
- move v0,zero
- .set reorder
-
-.L_bn_mul_words_proceed:
- li MINUS4,-4
- and ta0,a2,MINUS4
- move v0,zero
- beqz ta0,.L_bn_mul_words_tail
-
-.L_bn_mul_words_loop:
- dmultu t0,a3
- ld t2,8(a1)
- ld ta0,16(a1)
- ld ta2,24(a1)
- mflo AT
- mfhi t0
- daddu v0,AT
- sltu t1,v0,AT
- sd v0,0(a0)
- daddu v0,t1,t0
-
- dmultu t2,a3
- subu a2,4
- PTR_ADD a0,32
- PTR_ADD a1,32
- mflo AT
- mfhi t2
- daddu v0,AT
- sltu t3,v0,AT
- sd v0,-24(a0)
- daddu v0,t3,t2
-
- dmultu ta0,a3
- mflo AT
- mfhi ta0
- daddu v0,AT
- sltu ta1,v0,AT
- sd v0,-16(a0)
- daddu v0,ta1,ta0
-
-
- dmultu ta2,a3
- and ta0,a2,MINUS4
- mflo AT
- mfhi ta2
- daddu v0,AT
- sltu ta3,v0,AT
- sd v0,-8(a0)
- daddu v0,ta3,ta2
- .set noreorder
- bgtzl ta0,.L_bn_mul_words_loop
- ld t0,0(a1)
-
- bnezl a2,.L_bn_mul_words_tail
- ld t0,0(a1)
- .set reorder
-
-.L_bn_mul_words_return:
- jr ra
-
-.L_bn_mul_words_tail:
- dmultu t0,a3
- subu a2,1
- mflo AT
- mfhi t0
- daddu v0,AT
- sltu t1,v0,AT
- sd v0,0(a0)
- daddu v0,t1,t0
- beqz a2,.L_bn_mul_words_return
-
- ld t0,8(a1)
- dmultu t0,a3
- subu a2,1
- mflo AT
- mfhi t0
- daddu v0,AT
- sltu t1,v0,AT
- sd v0,8(a0)
- daddu v0,t1,t0
- beqz a2,.L_bn_mul_words_return
-
- ld t0,16(a1)
- dmultu t0,a3
- mflo AT
- mfhi t0
- daddu v0,AT
- sltu t1,v0,AT
- sd v0,16(a0)
- daddu v0,t1,t0
- jr ra
-END(bn_mul_words)
-
-.align 5
-LEAF(bn_sqr_words)
- .set noreorder
- bgtzl a2,.L_bn_sqr_words_proceed
- ld t0,0(a1)
- jr ra
- move v0,zero
- .set reorder
-
-.L_bn_sqr_words_proceed:
- li MINUS4,-4
- and ta0,a2,MINUS4
- move v0,zero
- beqz ta0,.L_bn_sqr_words_tail
-
-.L_bn_sqr_words_loop:
- dmultu t0,t0
- ld t2,8(a1)
- ld ta0,16(a1)
- ld ta2,24(a1)
- mflo t1
- mfhi t0
- sd t1,0(a0)
- sd t0,8(a0)
-
- dmultu t2,t2
- subu a2,4
- PTR_ADD a0,64
- PTR_ADD a1,32
- mflo t3
- mfhi t2
- sd t3,-48(a0)
- sd t2,-40(a0)
-
- dmultu ta0,ta0
- mflo ta1
- mfhi ta0
- sd ta1,-32(a0)
- sd ta0,-24(a0)
-
-
- dmultu ta2,ta2
- and ta0,a2,MINUS4
- mflo ta3
- mfhi ta2
- sd ta3,-16(a0)
- sd ta2,-8(a0)
-
- .set noreorder
- bgtzl ta0,.L_bn_sqr_words_loop
- ld t0,0(a1)
-
- bnezl a2,.L_bn_sqr_words_tail
- ld t0,0(a1)
- .set reorder
-
-.L_bn_sqr_words_return:
- move v0,zero
- jr ra
-
-.L_bn_sqr_words_tail:
- dmultu t0,t0
- subu a2,1
- mflo t1
- mfhi t0
- sd t1,0(a0)
- sd t0,8(a0)
- beqz a2,.L_bn_sqr_words_return
-
- ld t0,8(a1)
- dmultu t0,t0
- subu a2,1
- mflo t1
- mfhi t0
- sd t1,16(a0)
- sd t0,24(a0)
- beqz a2,.L_bn_sqr_words_return
-
- ld t0,16(a1)
- dmultu t0,t0
- mflo t1
- mfhi t0
- sd t1,32(a0)
- sd t0,40(a0)
- jr ra
-END(bn_sqr_words)
-
-.align 5
-LEAF(bn_add_words)
- .set noreorder
- bgtzl a3,.L_bn_add_words_proceed
- ld t0,0(a1)
- jr ra
- move v0,zero
- .set reorder
-
-.L_bn_add_words_proceed:
- li MINUS4,-4
- and AT,a3,MINUS4
- move v0,zero
- beqz AT,.L_bn_add_words_tail
-
-.L_bn_add_words_loop:
- ld ta0,0(a2)
- subu a3,4
- ld t1,8(a1)
- and AT,a3,MINUS4
- ld t2,16(a1)
- PTR_ADD a2,32
- ld t3,24(a1)
- PTR_ADD a0,32
- ld ta1,-24(a2)
- PTR_ADD a1,32
- ld ta2,-16(a2)
- ld ta3,-8(a2)
- daddu ta0,t0
- sltu t8,ta0,t0
- daddu t0,ta0,v0
- sltu v0,t0,ta0
- sd t0,-32(a0)
- daddu v0,t8
-
- daddu ta1,t1
- sltu t9,ta1,t1
- daddu t1,ta1,v0
- sltu v0,t1,ta1
- sd t1,-24(a0)
- daddu v0,t9
-
- daddu ta2,t2
- sltu t8,ta2,t2
- daddu t2,ta2,v0
- sltu v0,t2,ta2
- sd t2,-16(a0)
- daddu v0,t8
-
- daddu ta3,t3
- sltu t9,ta3,t3
- daddu t3,ta3,v0
- sltu v0,t3,ta3
- sd t3,-8(a0)
- daddu v0,t9
-
- .set noreorder
- bgtzl AT,.L_bn_add_words_loop
- ld t0,0(a1)
-
- bnezl a3,.L_bn_add_words_tail
- ld t0,0(a1)
- .set reorder
-
-.L_bn_add_words_return:
- jr ra
-
-.L_bn_add_words_tail:
- ld ta0,0(a2)
- daddu ta0,t0
- subu a3,1
- sltu t8,ta0,t0
- daddu t0,ta0,v0
- sltu v0,t0,ta0
- sd t0,0(a0)
- daddu v0,t8
- beqz a3,.L_bn_add_words_return
-
- ld t1,8(a1)
- ld ta1,8(a2)
- daddu ta1,t1
- subu a3,1
- sltu t9,ta1,t1
- daddu t1,ta1,v0
- sltu v0,t1,ta1
- sd t1,8(a0)
- daddu v0,t9
- beqz a3,.L_bn_add_words_return
-
- ld t2,16(a1)
- ld ta2,16(a2)
- daddu ta2,t2
- sltu t8,ta2,t2
- daddu t2,ta2,v0
- sltu v0,t2,ta2
- sd t2,16(a0)
- daddu v0,t8
- jr ra
-END(bn_add_words)
-
-.align 5
-LEAF(bn_sub_words)
- .set noreorder
- bgtzl a3,.L_bn_sub_words_proceed
- ld t0,0(a1)
- jr ra
- move v0,zero
- .set reorder
-
-.L_bn_sub_words_proceed:
- li MINUS4,-4
- and AT,a3,MINUS4
- move v0,zero
- beqz AT,.L_bn_sub_words_tail
-
-.L_bn_sub_words_loop:
- ld ta0,0(a2)
- subu a3,4
- ld t1,8(a1)
- and AT,a3,MINUS4
- ld t2,16(a1)
- PTR_ADD a2,32
- ld t3,24(a1)
- PTR_ADD a0,32
- ld ta1,-24(a2)
- PTR_ADD a1,32
- ld ta2,-16(a2)
- ld ta3,-8(a2)
- sltu t8,t0,ta0
- dsubu t0,ta0
- dsubu ta0,t0,v0
- sd ta0,-32(a0)
- MOVNZ (t0,v0,t8)
-
- sltu t9,t1,ta1
- dsubu t1,ta1
- dsubu ta1,t1,v0
- sd ta1,-24(a0)
- MOVNZ (t1,v0,t9)
-
-
- sltu t8,t2,ta2
- dsubu t2,ta2
- dsubu ta2,t2,v0
- sd ta2,-16(a0)
- MOVNZ (t2,v0,t8)
-
- sltu t9,t3,ta3
- dsubu t3,ta3
- dsubu ta3,t3,v0
- sd ta3,-8(a0)
- MOVNZ (t3,v0,t9)
-
- .set noreorder
- bgtzl AT,.L_bn_sub_words_loop
- ld t0,0(a1)
-
- bnezl a3,.L_bn_sub_words_tail
- ld t0,0(a1)
- .set reorder
-
-.L_bn_sub_words_return:
- jr ra
-
-.L_bn_sub_words_tail:
- ld ta0,0(a2)
- subu a3,1
- sltu t8,t0,ta0
- dsubu t0,ta0
- dsubu ta0,t0,v0
- MOVNZ (t0,v0,t8)
- sd ta0,0(a0)
- beqz a3,.L_bn_sub_words_return
-
- ld t1,8(a1)
- subu a3,1
- ld ta1,8(a2)
- sltu t9,t1,ta1
- dsubu t1,ta1
- dsubu ta1,t1,v0
- MOVNZ (t1,v0,t9)
- sd ta1,8(a0)
- beqz a3,.L_bn_sub_words_return
-
- ld t2,16(a1)
- ld ta2,16(a2)
- sltu t8,t2,ta2
- dsubu t2,ta2
- dsubu ta2,t2,v0
- MOVNZ (t2,v0,t8)
- sd ta2,16(a0)
- jr ra
-END(bn_sub_words)
-
-#undef MINUS4
-
-.align 5
-LEAF(bn_div_3_words)
- .set reorder
- move a3,a0 /* we know that bn_div_words doesn't
- * touch a3, ta2, ta3 and preserves a2
- * so that we can save two arguments
- * and return address in registers
- * instead of stack:-)
- */
- ld a0,(a3)
- move ta2,a1
- ld a1,-8(a3)
- bne a0,a2,.L_bn_div_3_words_proceed
- li v0,-1
- jr ra
-.L_bn_div_3_words_proceed:
- move ta3,ra
- bal bn_div_words
- move ra,ta3
- dmultu ta2,v0
- ld t2,-16(a3)
- move ta0,zero
- mfhi t1
- mflo t0
- sltu t8,t1,v1
-.L_bn_div_3_words_inner_loop:
- bnez t8,.L_bn_div_3_words_inner_loop_done
- sgeu AT,t2,t0
- seq t9,t1,v1
- and AT,t9
- sltu t3,t0,ta2
- daddu v1,a2
- dsubu t1,t3
- dsubu t0,ta2
- sltu t8,t1,v1
- sltu ta0,v1,a2
- or t8,ta0
- .set noreorder
- beqzl AT,.L_bn_div_3_words_inner_loop
- dsubu v0,1
- .set reorder
-.L_bn_div_3_words_inner_loop_done:
- jr ra
-END(bn_div_3_words)
-
-.align 5
-LEAF(bn_div_words)
- .set noreorder
- bnezl a2,.L_bn_div_words_proceed
- move v1,zero
- jr ra
- li v0,-1 /* I'd rather signal div-by-zero
- * which can be done with 'break 7' */
-
-.L_bn_div_words_proceed:
- bltz a2,.L_bn_div_words_body
- move t9,v1
- dsll a2,1
- bgtz a2,.-4
- addu t9,1
-
- .set reorder
- negu t1,t9
- li t2,-1
- dsll t2,t1
- and t2,a0
- dsrl AT,a1,t1
- .set noreorder
- bnezl t2,.+8
- break 6 /* signal overflow */
- .set reorder
- dsll a0,t9
- dsll a1,t9
- or a0,AT
-
-#define QT ta0
-#define HH ta1
-#define DH v1
-.L_bn_div_words_body:
- dsrl DH,a2,32
- sgeu AT,a0,a2
- .set noreorder
- bnezl AT,.+8
- dsubu a0,a2
- .set reorder
-
- li QT,-1
- dsrl HH,a0,32
- dsrl QT,32 /* q=0xffffffff */
- beq DH,HH,.L_bn_div_words_skip_div1
- ddivu zero,a0,DH
- mflo QT
-.L_bn_div_words_skip_div1:
- dmultu a2,QT
- dsll t3,a0,32
- dsrl AT,a1,32
- or t3,AT
- mflo t0
- mfhi t1
-.L_bn_div_words_inner_loop1:
- sltu t2,t3,t0
- seq t8,HH,t1
- sltu AT,HH,t1
- and t2,t8
- sltu v0,t0,a2
- or AT,t2
- .set noreorder
- beqz AT,.L_bn_div_words_inner_loop1_done
- dsubu t1,v0
- dsubu t0,a2
- b .L_bn_div_words_inner_loop1
- dsubu QT,1
- .set reorder
-.L_bn_div_words_inner_loop1_done:
-
- dsll a1,32
- dsubu a0,t3,t0
- dsll v0,QT,32
-
- li QT,-1
- dsrl HH,a0,32
- dsrl QT,32 /* q=0xffffffff */
- beq DH,HH,.L_bn_div_words_skip_div2
- ddivu zero,a0,DH
- mflo QT
-.L_bn_div_words_skip_div2:
-#undef DH
- dmultu a2,QT
- dsll t3,a0,32
- dsrl AT,a1,32
- or t3,AT
- mflo t0
- mfhi t1
-.L_bn_div_words_inner_loop2:
- sltu t2,t3,t0
- seq t8,HH,t1
- sltu AT,HH,t1
- and t2,t8
- sltu v1,t0,a2
- or AT,t2
- .set noreorder
- beqz AT,.L_bn_div_words_inner_loop2_done
- dsubu t1,v1
- dsubu t0,a2
- b .L_bn_div_words_inner_loop2
- dsubu QT,1
- .set reorder
-.L_bn_div_words_inner_loop2_done:
-#undef HH
-
- dsubu a0,t3,t0
- or v0,QT
- dsrl v1,a0,t9 /* v1 contains remainder if anybody wants it */
- dsrl a2,t9 /* restore a2 */
- jr ra
-#undef QT
-END(bn_div_words)
-
-#define a_0 t0
-#define a_1 t1
-#define a_2 t2
-#define a_3 t3
-#define b_0 ta0
-#define b_1 ta1
-#define b_2 ta2
-#define b_3 ta3
-
-#define a_4 s0
-#define a_5 s2
-#define a_6 s4
-#define a_7 a1 /* once we load a[7] we don't need a anymore */
-#define b_4 s1
-#define b_5 s3
-#define b_6 s5
-#define b_7 a2 /* once we load b[7] we don't need b anymore */
-
-#define t_1 t8
-#define t_2 t9
-
-#define c_1 v0
-#define c_2 v1
-#define c_3 a3
-
-#define FRAME_SIZE 48
-
-.align 5
-LEAF(bn_mul_comba8)
- .set noreorder
- PTR_SUB sp,FRAME_SIZE
- .frame sp,64,ra
- .set reorder
- ld a_0,0(a1) /* If compiled with -mips3 option on
- * R5000 box assembler barks on this
- * line with "shouldn't have mult/div
- * as last instruction in bb (R10K
- * bug)" warning. If anybody out there
- * has a clue about how to circumvent
- * this do send me a note.
- * <appro@fy.chalmers.se>
- */
- ld b_0,0(a2)
- ld a_1,8(a1)
- ld a_2,16(a1)
- ld a_3,24(a1)
- ld b_1,8(a2)
- ld b_2,16(a2)
- ld b_3,24(a2)
- dmultu a_0,b_0 /* mul_add_c(a[0],b[0],c1,c2,c3); */
- sd s0,0(sp)
- sd s1,8(sp)
- sd s2,16(sp)
- sd s3,24(sp)
- sd s4,32(sp)
- sd s5,40(sp)
- mflo c_1
- mfhi c_2
-
- dmultu a_0,b_1 /* mul_add_c(a[0],b[1],c2,c3,c1); */
- ld a_4,32(a1)
- ld a_5,40(a1)
- ld a_6,48(a1)
- ld a_7,56(a1)
- ld b_4,32(a2)
- ld b_5,40(a2)
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu c_3,t_2,AT
- dmultu a_1,b_0 /* mul_add_c(a[1],b[0],c2,c3,c1); */
- ld b_6,48(a2)
- ld b_7,56(a2)
- sd c_1,0(a0) /* r[0]=c1; */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- sd c_2,8(a0) /* r[1]=c2; */
-
- dmultu a_2,b_0 /* mul_add_c(a[2],b[0],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- dmultu a_1,b_1 /* mul_add_c(a[1],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_0,b_2 /* mul_add_c(a[0],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,16(a0) /* r[2]=c3; */
-
- dmultu a_0,b_3 /* mul_add_c(a[0],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu c_3,c_2,t_2
- dmultu a_1,b_2 /* mul_add_c(a[1],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_2,b_1 /* mul_add_c(a[2],b[1],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_3,b_0 /* mul_add_c(a[3],b[0],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,24(a0) /* r[3]=c1; */
-
- dmultu a_4,b_0 /* mul_add_c(a[4],b[0],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- dmultu a_3,b_1 /* mul_add_c(a[3],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_2,b_2 /* mul_add_c(a[2],b[2],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_1,b_3 /* mul_add_c(a[1],b[3],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_0,b_4 /* mul_add_c(a[0],b[4],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,32(a0) /* r[4]=c2; */
-
- dmultu a_0,b_5 /* mul_add_c(a[0],b[5],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_1,b_4 /* mul_add_c(a[1],b[4],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_2,b_3 /* mul_add_c(a[2],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_3,b_2 /* mul_add_c(a[3],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_4,b_1 /* mul_add_c(a[4],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_5,b_0 /* mul_add_c(a[5],b[0],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,40(a0) /* r[5]=c3; */
-
- dmultu a_6,b_0 /* mul_add_c(a[6],b[0],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu c_3,c_2,t_2
- dmultu a_5,b_1 /* mul_add_c(a[5],b[1],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_4,b_2 /* mul_add_c(a[4],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_3,b_3 /* mul_add_c(a[3],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_2,b_4 /* mul_add_c(a[2],b[4],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_1,b_5 /* mul_add_c(a[1],b[5],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_0,b_6 /* mul_add_c(a[0],b[6],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,48(a0) /* r[6]=c1; */
-
- dmultu a_0,b_7 /* mul_add_c(a[0],b[7],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- dmultu a_1,b_6 /* mul_add_c(a[1],b[6],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_2,b_5 /* mul_add_c(a[2],b[5],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_3,b_4 /* mul_add_c(a[3],b[4],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_4,b_3 /* mul_add_c(a[4],b[3],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_5,b_2 /* mul_add_c(a[5],b[2],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_6,b_1 /* mul_add_c(a[6],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_7,b_0 /* mul_add_c(a[7],b[0],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,56(a0) /* r[7]=c2; */
-
- dmultu a_7,b_1 /* mul_add_c(a[7],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_6,b_2 /* mul_add_c(a[6],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_5,b_3 /* mul_add_c(a[5],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_4,b_4 /* mul_add_c(a[4],b[4],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_3,b_5 /* mul_add_c(a[3],b[5],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_2,b_6 /* mul_add_c(a[2],b[6],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_1,b_7 /* mul_add_c(a[1],b[7],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,64(a0) /* r[8]=c3; */
-
- dmultu a_2,b_7 /* mul_add_c(a[2],b[7],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu c_3,c_2,t_2
- dmultu a_3,b_6 /* mul_add_c(a[3],b[6],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_4,b_5 /* mul_add_c(a[4],b[5],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_5,b_4 /* mul_add_c(a[5],b[4],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_6,b_3 /* mul_add_c(a[6],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_7,b_2 /* mul_add_c(a[7],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,72(a0) /* r[9]=c1; */
-
- dmultu a_7,b_3 /* mul_add_c(a[7],b[3],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- dmultu a_6,b_4 /* mul_add_c(a[6],b[4],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_5,b_5 /* mul_add_c(a[5],b[5],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_4,b_6 /* mul_add_c(a[4],b[6],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_3,b_7 /* mul_add_c(a[3],b[7],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,80(a0) /* r[10]=c2; */
-
- dmultu a_4,b_7 /* mul_add_c(a[4],b[7],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_5,b_6 /* mul_add_c(a[5],b[6],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_6,b_5 /* mul_add_c(a[6],b[5],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_7,b_4 /* mul_add_c(a[7],b[4],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,88(a0) /* r[11]=c3; */
-
- dmultu a_7,b_5 /* mul_add_c(a[7],b[5],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu c_3,c_2,t_2
- dmultu a_6,b_6 /* mul_add_c(a[6],b[6],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_5,b_7 /* mul_add_c(a[5],b[7],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,96(a0) /* r[12]=c1; */
-
- dmultu a_6,b_7 /* mul_add_c(a[6],b[7],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- dmultu a_7,b_6 /* mul_add_c(a[7],b[6],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,104(a0) /* r[13]=c2; */
-
- dmultu a_7,b_7 /* mul_add_c(a[7],b[7],c3,c1,c2); */
- ld s0,0(sp)
- ld s1,8(sp)
- ld s2,16(sp)
- ld s3,24(sp)
- ld s4,32(sp)
- ld s5,40(sp)
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sd c_3,112(a0) /* r[14]=c3; */
- sd c_1,120(a0) /* r[15]=c1; */
-
- PTR_ADD sp,FRAME_SIZE
-
- jr ra
-END(bn_mul_comba8)
-
-.align 5
-LEAF(bn_mul_comba4)
- .set reorder
- ld a_0,0(a1)
- ld b_0,0(a2)
- ld a_1,8(a1)
- ld a_2,16(a1)
- dmultu a_0,b_0 /* mul_add_c(a[0],b[0],c1,c2,c3); */
- ld a_3,24(a1)
- ld b_1,8(a2)
- ld b_2,16(a2)
- ld b_3,24(a2)
- mflo c_1
- mfhi c_2
- sd c_1,0(a0)
-
- dmultu a_0,b_1 /* mul_add_c(a[0],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu c_3,t_2,AT
- dmultu a_1,b_0 /* mul_add_c(a[1],b[0],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- sd c_2,8(a0)
-
- dmultu a_2,b_0 /* mul_add_c(a[2],b[0],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- dmultu a_1,b_1 /* mul_add_c(a[1],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_0,b_2 /* mul_add_c(a[0],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,16(a0)
-
- dmultu a_0,b_3 /* mul_add_c(a[0],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu c_3,c_2,t_2
- dmultu a_1,b_2 /* mul_add_c(a[1],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_2,b_1 /* mul_add_c(a[2],b[1],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_3,b_0 /* mul_add_c(a[3],b[0],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,24(a0)
-
- dmultu a_3,b_1 /* mul_add_c(a[3],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu c_1,c_3,t_2
- dmultu a_2,b_2 /* mul_add_c(a[2],b[2],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_1,b_3 /* mul_add_c(a[1],b[3],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,32(a0)
-
- dmultu a_2,b_3 /* mul_add_c(a[2],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu c_2,c_1,t_2
- dmultu a_3,b_2 /* mul_add_c(a[3],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,40(a0)
-
- dmultu a_3,b_3 /* mul_add_c(a[3],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sd c_1,48(a0)
- sd c_2,56(a0)
-
- jr ra
-END(bn_mul_comba4)
-
-#undef a_4
-#undef a_5
-#undef a_6
-#undef a_7
-#define a_4 b_0
-#define a_5 b_1
-#define a_6 b_2
-#define a_7 b_3
-
-.align 5
-LEAF(bn_sqr_comba8)
- .set reorder
- ld a_0,0(a1)
- ld a_1,8(a1)
- ld a_2,16(a1)
- ld a_3,24(a1)
-
- dmultu a_0,a_0 /* mul_add_c(a[0],b[0],c1,c2,c3); */
- ld a_4,32(a1)
- ld a_5,40(a1)
- ld a_6,48(a1)
- ld a_7,56(a1)
- mflo c_1
- mfhi c_2
- sd c_1,0(a0)
-
- dmultu a_0,a_1 /* mul_add_c2(a[0],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu c_3,t_2,AT
- sd c_2,8(a0)
-
- dmultu a_2,a_0 /* mul_add_c2(a[2],b[0],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_1,a_1 /* mul_add_c(a[1],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,16(a0)
-
- dmultu a_0,a_3 /* mul_add_c2(a[0],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt c_3,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_1,a_2 /* mul_add_c2(a[1],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,24(a0)
-
- dmultu a_4,a_0 /* mul_add_c2(a[4],b[0],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_3,a_1 /* mul_add_c2(a[3],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_1,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_2,a_2 /* mul_add_c(a[2],b[2],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,32(a0)
-
- dmultu a_0,a_5 /* mul_add_c2(a[0],b[5],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_1,a_4 /* mul_add_c2(a[1],b[4],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_2,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_2,a_3 /* mul_add_c2(a[2],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_2,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,40(a0)
-
- dmultu a_6,a_0 /* mul_add_c2(a[6],b[0],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt c_3,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_5,a_1 /* mul_add_c2(a[5],b[1],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_4,a_2 /* mul_add_c2(a[4],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_3,a_3 /* mul_add_c(a[3],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,48(a0)
-
- dmultu a_0,a_7 /* mul_add_c2(a[0],b[7],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_1,a_6 /* mul_add_c2(a[1],b[6],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_1,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_2,a_5 /* mul_add_c2(a[2],b[5],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_1,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_3,a_4 /* mul_add_c2(a[3],b[4],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_1,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,56(a0)
-
- dmultu a_7,a_1 /* mul_add_c2(a[7],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_6,a_2 /* mul_add_c2(a[6],b[2],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_2,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_5,a_3 /* mul_add_c2(a[5],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_2,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_4,a_4 /* mul_add_c(a[4],b[4],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,64(a0)
-
- dmultu a_2,a_7 /* mul_add_c2(a[2],b[7],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt c_3,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_3,a_6 /* mul_add_c2(a[3],b[6],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_4,a_5 /* mul_add_c2(a[4],b[5],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,72(a0)
-
- dmultu a_7,a_3 /* mul_add_c2(a[7],b[3],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_6,a_4 /* mul_add_c2(a[6],b[4],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_1,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_5,a_5 /* mul_add_c(a[5],b[5],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,80(a0)
-
- dmultu a_4,a_7 /* mul_add_c2(a[4],b[7],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_5,a_6 /* mul_add_c2(a[5],b[6],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_2,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,88(a0)
-
- dmultu a_7,a_5 /* mul_add_c2(a[7],b[5],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt c_3,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_6,a_6 /* mul_add_c(a[6],b[6],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,96(a0)
-
- dmultu a_6,a_7 /* mul_add_c2(a[6],b[7],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,104(a0)
-
- dmultu a_7,a_7 /* mul_add_c(a[7],b[7],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sd c_3,112(a0)
- sd c_1,120(a0)
-
- jr ra
-END(bn_sqr_comba8)
-
-.align 5
-LEAF(bn_sqr_comba4)
- .set reorder
- ld a_0,0(a1)
- ld a_1,8(a1)
- ld a_2,16(a1)
- ld a_3,24(a1)
- dmultu a_0,a_0 /* mul_add_c(a[0],b[0],c1,c2,c3); */
- mflo c_1
- mfhi c_2
- sd c_1,0(a0)
-
- dmultu a_0,a_1 /* mul_add_c2(a[0],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu c_3,t_2,AT
- sd c_2,8(a0)
-
- dmultu a_2,a_0 /* mul_add_c2(a[2],b[0],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- dmultu a_1,a_1 /* mul_add_c(a[1],b[1],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,16(a0)
-
- dmultu a_0,a_3 /* mul_add_c2(a[0],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt c_3,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- dmultu a_1,a_2 /* mul_add_c(a2[1],b[2],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- slt AT,t_2,zero
- daddu c_3,AT
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sltu AT,c_2,t_2
- daddu c_3,AT
- sd c_1,24(a0)
-
- dmultu a_3,a_1 /* mul_add_c2(a[3],b[1],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- slt c_1,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- dmultu a_2,a_2 /* mul_add_c(a[2],b[2],c2,c3,c1); */
- mflo t_1
- mfhi t_2
- daddu c_2,t_1
- sltu AT,c_2,t_1
- daddu t_2,AT
- daddu c_3,t_2
- sltu AT,c_3,t_2
- daddu c_1,AT
- sd c_2,32(a0)
-
- dmultu a_2,a_3 /* mul_add_c2(a[2],b[3],c3,c1,c2); */
- mflo t_1
- mfhi t_2
- slt c_2,t_2,zero
- dsll t_2,1
- slt a2,t_1,zero
- daddu t_2,a2
- dsll t_1,1
- daddu c_3,t_1
- sltu AT,c_3,t_1
- daddu t_2,AT
- daddu c_1,t_2
- sltu AT,c_1,t_2
- daddu c_2,AT
- sd c_3,40(a0)
-
- dmultu a_3,a_3 /* mul_add_c(a[3],b[3],c1,c2,c3); */
- mflo t_1
- mfhi t_2
- daddu c_1,t_1
- sltu AT,c_1,t_1
- daddu t_2,AT
- daddu c_2,t_2
- sd c_1,48(a0)
- sd c_2,56(a0)
-
- jr ra
-END(bn_sqr_comba4)
diff --git a/openssl/crypto/bn/asm/x86_64-gcc.c b/openssl/crypto/bn/asm/x86_64-gcc.c
index 31476abeb..2d39407f4 100644
--- a/openssl/crypto/bn/asm/x86_64-gcc.c
+++ b/openssl/crypto/bn/asm/x86_64-gcc.c
@@ -273,6 +273,10 @@ BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
/* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */
+/*
+ * Keep in mind that carrying into high part of multiplication result
+ * can not overflow, because it cannot be all-ones.
+ */
#if 0
/* original macros are kept for reference purposes */
#define mul_add_c(a,b,c0,c1,c2) { \
@@ -287,10 +291,10 @@ BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
BN_ULONG ta=(a),tb=(b),t0; \
t1 = BN_UMULT_HIGH(ta,tb); \
t0 = ta * tb; \
- t2 = t1+t1; c2 += (t2<t1)?1:0; \
- t1 = t0+t0; t2 += (t1<t0)?1:0; \
- c0 += t1; t2 += (c0<t1)?1:0; \
+ c0 += t0; t2 = t1+((c0<t0)?1:0);\
c1 += t2; c2 += (c1<t2)?1:0; \
+ c0 += t0; t1 += (c0<t0)?1:0; \
+ c1 += t1; c2 += (c1<t1)?1:0; \
}
#else
#define mul_add_c(a,b,c0,c1,c2) do { \
@@ -328,22 +332,14 @@ BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
: "=a"(t1),"=d"(t2) \
: "a"(a),"m"(b) \
: "cc"); \
- asm ("addq %0,%0; adcq %2,%1" \
- : "+d"(t2),"+r"(c2) \
- : "g"(0) \
- : "cc"); \
- asm ("addq %0,%0; adcq %2,%1" \
- : "+a"(t1),"+d"(t2) \
- : "g"(0) \
- : "cc"); \
- asm ("addq %2,%0; adcq %3,%1" \
- : "+r"(c0),"+d"(t2) \
- : "a"(t1),"g"(0) \
- : "cc"); \
- asm ("addq %2,%0; adcq %3,%1" \
- : "+r"(c1),"+r"(c2) \
- : "d"(t2),"g"(0) \
- : "cc"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
} while (0)
#endif
diff --git a/openssl/crypto/bn/bn.h b/openssl/crypto/bn/bn.h
index 21a1a3fe3..c4d618522 100644
--- a/openssl/crypto/bn/bn.h
+++ b/openssl/crypto/bn/bn.h
@@ -780,7 +780,9 @@ int RAND_pseudo_bytes(unsigned char *buf,int num);
#define bn_wcheck_size(bn, words) \
do { \
const BIGNUM *_bnum2 = (bn); \
- assert(words <= (_bnum2)->dmax && words >= (_bnum2)->top); \
+ assert((words) <= (_bnum2)->dmax && (words) >= (_bnum2)->top); \
+ /* avoid unused variable warning with NDEBUG */ \
+ (void)(_bnum2); \
} while(0)
#else /* !BN_DEBUG */
diff --git a/openssl/crypto/bn/bn_asm.c b/openssl/crypto/bn/bn_asm.c
index c43c91cc0..a33b63411 100644
--- a/openssl/crypto/bn/bn_asm.c
+++ b/openssl/crypto/bn/bn_asm.c
@@ -438,6 +438,10 @@ BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
/* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */
+/*
+ * Keep in mind that carrying into high part of multiplication result
+ * can not overflow, because it cannot be all-ones.
+ */
#ifdef BN_LLONG
#define mul_add_c(a,b,c0,c1,c2) \
t=(BN_ULLONG)a*b; \
@@ -478,10 +482,10 @@ BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
#define mul_add_c2(a,b,c0,c1,c2) { \
BN_ULONG ta=(a),tb=(b),t0; \
BN_UMULT_LOHI(t0,t1,ta,tb); \
- t2 = t1+t1; c2 += (t2<t1)?1:0; \
- t1 = t0+t0; t2 += (t1<t0)?1:0; \
- c0 += t1; t2 += (c0<t1)?1:0; \
+ c0 += t0; t2 = t1+((c0<t0)?1:0);\
c1 += t2; c2 += (c1<t2)?1:0; \
+ c0 += t0; t1 += (c0<t0)?1:0; \
+ c1 += t1; c2 += (c1<t1)?1:0; \
}
#define sqr_add_c(a,i,c0,c1,c2) { \
@@ -508,10 +512,10 @@ BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
BN_ULONG ta=(a),tb=(b),t0; \
t1 = BN_UMULT_HIGH(ta,tb); \
t0 = ta * tb; \
- t2 = t1+t1; c2 += (t2<t1)?1:0; \
- t1 = t0+t0; t2 += (t1<t0)?1:0; \
- c0 += t1; t2 += (c0<t1)?1:0; \
+ c0 += t0; t2 = t1+((c0<t0)?1:0);\
c1 += t2; c2 += (c1<t2)?1:0; \
+ c0 += t0; t1 += (c0<t0)?1:0; \
+ c1 += t1; c2 += (c1<t1)?1:0; \
}
#define sqr_add_c(a,i,c0,c1,c2) { \
diff --git a/openssl/crypto/bn/bn_ctx.c b/openssl/crypto/bn/bn_ctx.c
index 3f2256f67..90aa3aeb9 100644
--- a/openssl/crypto/bn/bn_ctx.c
+++ b/openssl/crypto/bn/bn_ctx.c
@@ -158,7 +158,7 @@ static void ctxdbg(BN_CTX *ctx)
unsigned int bnidx = 0, fpidx = 0;
BN_POOL_ITEM *item = ctx->pool.head;
BN_STACK *stack = &ctx->stack;
- fprintf(stderr,"(%08x): ", (unsigned int)ctx);
+ fprintf(stderr,"(%16p): ", ctx);
while(bnidx < ctx->used)
{
fprintf(stderr,"%03x ", item->vals[bnidx++ % BN_CTX_POOL_SIZE].dmax);
diff --git a/openssl/crypto/bn/bn_div.c b/openssl/crypto/bn/bn_div.c
index 7b2403185..0ec90e805 100644
--- a/openssl/crypto/bn/bn_div.c
+++ b/openssl/crypto/bn/bn_div.c
@@ -189,15 +189,17 @@ int BN_div(BIGNUM *dv, BIGNUM *rm, const BIGNUM *num, const BIGNUM *divisor,
int no_branch=0;
/* Invalid zero-padding would have particularly bad consequences
- * in the case of 'num', so don't just rely on bn_check_top() for this one
+ * so don't just rely on bn_check_top() here
* (bn_check_top() works only for BN_DEBUG builds) */
- if (num->top > 0 && num->d[num->top - 1] == 0)
+ if ((num->top > 0 && num->d[num->top - 1] == 0) ||
+ (divisor->top > 0 && divisor->d[divisor->top - 1] == 0))
{
BNerr(BN_F_BN_DIV,BN_R_NOT_INITIALIZED);
return 0;
}
bn_check_top(num);
+ bn_check_top(divisor);
if ((BN_get_flags(num, BN_FLG_CONSTTIME) != 0) || (BN_get_flags(divisor, BN_FLG_CONSTTIME) != 0))
{
@@ -207,7 +209,7 @@ int BN_div(BIGNUM *dv, BIGNUM *rm, const BIGNUM *num, const BIGNUM *divisor,
bn_check_top(dv);
bn_check_top(rm);
/* bn_check_top(num); */ /* 'num' has been checked already */
- bn_check_top(divisor);
+ /* bn_check_top(divisor); */ /* 'divisor' has been checked already */
if (BN_is_zero(divisor))
{
diff --git a/openssl/crypto/bn/bntest.c b/openssl/crypto/bn/bntest.c
index 06f5954ac..48bc63384 100644
--- a/openssl/crypto/bn/bntest.c
+++ b/openssl/crypto/bn/bntest.c
@@ -107,6 +107,7 @@ int test_mod(BIO *bp,BN_CTX *ctx);
int test_mod_mul(BIO *bp,BN_CTX *ctx);
int test_mod_exp(BIO *bp,BN_CTX *ctx);
int test_mod_exp_mont_consttime(BIO *bp,BN_CTX *ctx);
+int test_mod_exp_mont5(BIO *bp, BN_CTX *ctx);
int test_exp(BIO *bp,BN_CTX *ctx);
int test_gf2m_add(BIO *bp);
int test_gf2m_mod(BIO *bp);
@@ -249,6 +250,7 @@ int main(int argc, char *argv[])
message(out,"BN_mod_exp_mont_consttime");
if (!test_mod_exp_mont_consttime(out,ctx)) goto err;
+ if (!test_mod_exp_mont5(out,ctx)) goto err;
(void)BIO_flush(out);
message(out,"BN_exp");
@@ -676,44 +678,98 @@ int test_mul(BIO *bp)
int test_sqr(BIO *bp, BN_CTX *ctx)
{
- BIGNUM a,c,d,e;
- int i;
+ BIGNUM *a,*c,*d,*e;
+ int i, ret = 0;
- BN_init(&a);
- BN_init(&c);
- BN_init(&d);
- BN_init(&e);
+ a = BN_new();
+ c = BN_new();
+ d = BN_new();
+ e = BN_new();
+ if (a == NULL || c == NULL || d == NULL || e == NULL)
+ {
+ goto err;
+ }
for (i=0; i<num0; i++)
{
- BN_bntest_rand(&a,40+i*10,0,0);
- a.neg=rand_neg();
- BN_sqr(&c,&a,ctx);
+ BN_bntest_rand(a,40+i*10,0,0);
+ a->neg=rand_neg();
+ BN_sqr(c,a,ctx);
if (bp != NULL)
{
if (!results)
{
- BN_print(bp,&a);
+ BN_print(bp,a);
BIO_puts(bp," * ");
- BN_print(bp,&a);
+ BN_print(bp,a);
BIO_puts(bp," - ");
}
- BN_print(bp,&c);
+ BN_print(bp,c);
BIO_puts(bp,"\n");
}
- BN_div(&d,&e,&c,&a,ctx);
- BN_sub(&d,&d,&a);
- if(!BN_is_zero(&d) || !BN_is_zero(&e))
- {
- fprintf(stderr,"Square test failed!\n");
- return 0;
- }
+ BN_div(d,e,c,a,ctx);
+ BN_sub(d,d,a);
+ if(!BN_is_zero(d) || !BN_is_zero(e))
+ {
+ fprintf(stderr,"Square test failed!\n");
+ goto err;
+ }
}
- BN_free(&a);
- BN_free(&c);
- BN_free(&d);
- BN_free(&e);
- return(1);
+
+ /* Regression test for a BN_sqr overflow bug. */
+ BN_hex2bn(&a,
+ "80000000000000008000000000000001FFFFFFFFFFFFFFFE0000000000000000");
+ BN_sqr(c, a, ctx);
+ if (bp != NULL)
+ {
+ if (!results)
+ {
+ BN_print(bp,a);
+ BIO_puts(bp," * ");
+ BN_print(bp,a);
+ BIO_puts(bp," - ");
+ }
+ BN_print(bp,c);
+ BIO_puts(bp,"\n");
+ }
+ BN_mul(d, a, a, ctx);
+ if (BN_cmp(c, d))
+ {
+ fprintf(stderr, "Square test failed: BN_sqr and BN_mul produce "
+ "different results!\n");
+ goto err;
+ }
+
+ /* Regression test for a BN_sqr overflow bug. */
+ BN_hex2bn(&a,
+ "80000000000000000000000080000001FFFFFFFE000000000000000000000000");
+ BN_sqr(c, a, ctx);
+ if (bp != NULL)
+ {
+ if (!results)
+ {
+ BN_print(bp,a);
+ BIO_puts(bp," * ");
+ BN_print(bp,a);
+ BIO_puts(bp," - ");
+ }
+ BN_print(bp,c);
+ BIO_puts(bp,"\n");
+ }
+ BN_mul(d, a, a, ctx);
+ if (BN_cmp(c, d))
+ {
+ fprintf(stderr, "Square test failed: BN_sqr and BN_mul produce "
+ "different results!\n");
+ goto err;
+ }
+ ret = 1;
+err:
+ if (a != NULL) BN_free(a);
+ if (c != NULL) BN_free(c);
+ if (d != NULL) BN_free(d);
+ if (e != NULL) BN_free(e);
+ return ret;
}
int test_mont(BIO *bp, BN_CTX *ctx)
@@ -1012,6 +1068,80 @@ int test_mod_exp_mont_consttime(BIO *bp, BN_CTX *ctx)
return(1);
}
+/* Test constant-time modular exponentiation with 1024-bit inputs,
+ * which on x86_64 cause a different code branch to be taken.
+ */
+int test_mod_exp_mont5(BIO *bp, BN_CTX *ctx)
+ {
+ BIGNUM *a,*p,*m,*d,*e;
+
+ BN_MONT_CTX *mont;
+
+ a=BN_new();
+ p=BN_new();
+ m=BN_new();
+ d=BN_new();
+ e=BN_new();
+
+ mont = BN_MONT_CTX_new();
+
+ BN_bntest_rand(m,1024,0,1); /* must be odd for montgomery */
+ /* Zero exponent */
+ BN_bntest_rand(a,1024,0,0);
+ BN_zero(p);
+ if(!BN_mod_exp_mont_consttime(d,a,p,m,ctx,NULL))
+ return 0;
+ if(!BN_is_one(d))
+ {
+ fprintf(stderr, "Modular exponentiation test failed!\n");
+ return 0;
+ }
+ /* Zero input */
+ BN_bntest_rand(p,1024,0,0);
+ BN_zero(a);
+ if(!BN_mod_exp_mont_consttime(d,a,p,m,ctx,NULL))
+ return 0;
+ if(!BN_is_zero(d))
+ {
+ fprintf(stderr, "Modular exponentiation test failed!\n");
+ return 0;
+ }
+ /* Craft an input whose Montgomery representation is 1,
+ * i.e., shorter than the modulus m, in order to test
+ * the const time precomputation scattering/gathering.
+ */
+ BN_one(a);
+ BN_MONT_CTX_set(mont,m,ctx);
+ if(!BN_from_montgomery(e,a,mont,ctx))
+ return 0;
+ if(!BN_mod_exp_mont_consttime(d,e,p,m,ctx,NULL))
+ return 0;
+ if(!BN_mod_exp_simple(a,e,p,m,ctx))
+ return 0;
+ if(BN_cmp(a,d) != 0)
+ {
+ fprintf(stderr,"Modular exponentiation test failed!\n");
+ return 0;
+ }
+ /* Finally, some regular test vectors. */
+ BN_bntest_rand(e,1024,0,0);
+ if(!BN_mod_exp_mont_consttime(d,e,p,m,ctx,NULL))
+ return 0;
+ if(!BN_mod_exp_simple(a,e,p,m,ctx))
+ return 0;
+ if(BN_cmp(a,d) != 0)
+ {
+ fprintf(stderr,"Modular exponentiation test failed!\n");
+ return 0;
+ }
+ BN_free(a);
+ BN_free(p);
+ BN_free(m);
+ BN_free(d);
+ BN_free(e);
+ return(1);
+ }
+
int test_exp(BIO *bp, BN_CTX *ctx)
{
BIGNUM *a,*b,*d,*e,*one;