Fix side channel in ecp_nistz256-armv8.pl
authorFangming.Fang <fangming.fang@arm.com>
Mon, 30 Dec 2019 12:15:37 +0000 (12:15 +0000)
committerNicola Tuveri <nic.tuv@gmail.com>
Sun, 5 Jan 2020 08:20:58 +0000 (10:20 +0200)
This change addresses a potential side-channel vulnerability in
the internals of nistz256 low level operations for armv8.

Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9239)

crypto/ec/asm/ecp_nistz256-armv8.pl

index 2bcf130bb6c1f03624ab2a03c7ff3f4b9fa72c7b..3be0c1c02665ba283adb6aef0abe530cf37362fc 100644 (file)
@@ -725,7 +725,7 @@ $code.=<<___;
 .align 5
 ecp_nistz256_point_double:
        .inst   0xd503233f              // paciasp
-       stp     x29,x30,[sp,#-80]!
+       stp     x29,x30,[sp,#-96]!
        add     x29,sp,#0
        stp     x19,x20,[sp,#16]
        stp     x21,x22,[sp,#32]
@@ -858,7 +858,7 @@ ecp_nistz256_point_double:
        add     sp,x29,#0               // destroy frame
        ldp     x19,x20,[x29,#16]
        ldp     x21,x22,[x29,#32]
-       ldp     x29,x30,[sp],#80
+       ldp     x29,x30,[sp],#96
        .inst   0xd50323bf              // autiasp
        ret
 .size  ecp_nistz256_point_double,.-ecp_nistz256_point_double
@@ -875,7 +875,7 @@ my ($res_x,$res_y,$res_z,
 my ($Z1sqr, $Z2sqr) = ($Hsqr, $Rsqr);
 # above map() describes stack layout with 12 temporary
 # 256-bit vectors on top.
-my ($rp_real,$ap_real,$bp_real,$in1infty,$in2infty,$temp)=map("x$_",(21..26));
+my ($rp_real,$ap_real,$bp_real,$in1infty,$in2infty,$temp0,$temp1,$temp2)=map("x$_",(21..28));
 
 $code.=<<___;
 .globl ecp_nistz256_point_add
@@ -883,12 +883,13 @@ $code.=<<___;
 .align 5
 ecp_nistz256_point_add:
        .inst   0xd503233f              // paciasp
-       stp     x29,x30,[sp,#-80]!
+       stp     x29,x30,[sp,#-96]!
        add     x29,sp,#0
        stp     x19,x20,[sp,#16]
        stp     x21,x22,[sp,#32]
        stp     x23,x24,[sp,#48]
        stp     x25,x26,[sp,#64]
+       stp     x27,x28,[sp,#80]
        sub     sp,sp,#32*12
 
        ldp     $a0,$a1,[$bp,#64]       // in2_z
@@ -902,7 +903,7 @@ ecp_nistz256_point_add:
        orr     $t2,$a2,$a3
        orr     $in2infty,$t0,$t2
        cmp     $in2infty,#0
-       csetm   $in2infty,ne            // !in2infty
+       csetm   $in2infty,ne            // ~in2infty
        add     $rp,sp,#$Z2sqr
        bl      __ecp_nistz256_sqr_mont // p256_sqr_mont(Z2sqr, in2_z);
 
@@ -912,7 +913,7 @@ ecp_nistz256_point_add:
        orr     $t2,$a2,$a3
        orr     $in1infty,$t0,$t2
        cmp     $in1infty,#0
-       csetm   $in1infty,ne            // !in1infty
+       csetm   $in1infty,ne            // ~in1infty
        add     $rp,sp,#$Z1sqr
        bl      __ecp_nistz256_sqr_mont // p256_sqr_mont(Z1sqr, in1_z);
 
@@ -953,7 +954,7 @@ ecp_nistz256_point_add:
 
        orr     $acc0,$acc0,$acc1       // see if result is zero
        orr     $acc2,$acc2,$acc3
-       orr     $temp,$acc0,$acc2
+       orr     $temp0,$acc0,$acc2      // ~is_equal(S1,S2)
 
        add     $bp,sp,#$Z2sqr
        add     $rp,sp,#$U1
@@ -974,32 +975,21 @@ ecp_nistz256_point_add:
 
        orr     $acc0,$acc0,$acc1       // see if result is zero
        orr     $acc2,$acc2,$acc3
-       orr     $acc0,$acc0,$acc2
-       tst     $acc0,$acc0
-       b.ne    .Ladd_proceed           // is_equal(U1,U2)?
+       orr     $acc0,$acc0,$acc2       // ~is_equal(U1,U2)
 
-       tst     $in1infty,$in2infty
-       b.eq    .Ladd_proceed           // (in1infty || in2infty)?
+       mvn     $temp1,$in1infty        // -1/0 -> 0/-1
+       mvn     $temp2,$in2infty        // -1/0 -> 0/-1
+       orr     $acc0,$acc0,$temp1
+       orr     $acc0,$acc0,$temp2
+       orr     $acc0,$acc0,$temp0
+       cbnz    $acc0,.Ladd_proceed     // if(~is_equal(U1,U2) | in1infty | in2infty | ~is_equal(S1,S2))
 
-       tst     $temp,$temp
-       b.eq    .Ladd_double            // is_equal(S1,S2)?
-
-       eor     $a0,$a0,$a0
-       eor     $a1,$a1,$a1
-       stp     $a0,$a1,[$rp_real]
-       stp     $a0,$a1,[$rp_real,#16]
-       stp     $a0,$a1,[$rp_real,#32]
-       stp     $a0,$a1,[$rp_real,#48]
-       stp     $a0,$a1,[$rp_real,#64]
-       stp     $a0,$a1,[$rp_real,#80]
-       b       .Ladd_done
-
-.align 4
 .Ladd_double:
        mov     $ap,$ap_real
        mov     $rp,$rp_real
        ldp     x23,x24,[x29,#48]
        ldp     x25,x26,[x29,#64]
+       ldp     x27,x28,[x29,#80]
        add     sp,sp,#32*(12-4)        // difference in stack frames
        b       .Ldouble_shortcut
 
@@ -1084,14 +1074,14 @@ ___
 for($i=0;$i<64;$i+=32) {               # conditional moves
 $code.=<<___;
        ldp     $acc0,$acc1,[$ap_real,#$i]      // in1
-       cmp     $in1infty,#0                    // !$in1intfy, remember?
+       cmp     $in1infty,#0                    // ~$in1intfy, remember?
        ldp     $acc2,$acc3,[$ap_real,#$i+16]
        csel    $t0,$a0,$t0,ne
        csel    $t1,$a1,$t1,ne
        ldp     $a0,$a1,[sp,#$res_x+$i+32]      // res
        csel    $t2,$a2,$t2,ne
        csel    $t3,$a3,$t3,ne
-       cmp     $in2infty,#0                    // !$in2intfy, remember?
+       cmp     $in2infty,#0                    // ~$in2intfy, remember?
        ldp     $a2,$a3,[sp,#$res_x+$i+48]
        csel    $acc0,$t0,$acc0,ne
        csel    $acc1,$t1,$acc1,ne
@@ -1105,13 +1095,13 @@ ___
 }
 $code.=<<___;
        ldp     $acc0,$acc1,[$ap_real,#$i]      // in1
-       cmp     $in1infty,#0                    // !$in1intfy, remember?
+       cmp     $in1infty,#0                    // ~$in1intfy, remember?
        ldp     $acc2,$acc3,[$ap_real,#$i+16]
        csel    $t0,$a0,$t0,ne
        csel    $t1,$a1,$t1,ne
        csel    $t2,$a2,$t2,ne
        csel    $t3,$a3,$t3,ne
-       cmp     $in2infty,#0                    // !$in2intfy, remember?
+       cmp     $in2infty,#0                    // ~$in2intfy, remember?
        csel    $acc0,$t0,$acc0,ne
        csel    $acc1,$t1,$acc1,ne
        csel    $acc2,$t2,$acc2,ne
@@ -1125,7 +1115,8 @@ $code.=<<___;
        ldp     x21,x22,[x29,#32]
        ldp     x23,x24,[x29,#48]
        ldp     x25,x26,[x29,#64]
-       ldp     x29,x30,[sp],#80
+       ldp     x27,x28,[x29,#80]
+       ldp     x29,x30,[sp],#96
        .inst   0xd50323bf              // autiasp
        ret
 .size  ecp_nistz256_point_add,.-ecp_nistz256_point_add
@@ -1169,7 +1160,7 @@ ecp_nistz256_point_add_affine:
        orr     $t2,$a2,$a3
        orr     $in1infty,$t0,$t2
        cmp     $in1infty,#0
-       csetm   $in1infty,ne            // !in1infty
+       csetm   $in1infty,ne            // ~in1infty
 
        ldp     $acc0,$acc1,[$bp]       // in2_x
        ldp     $acc2,$acc3,[$bp,#16]
@@ -1183,7 +1174,7 @@ ecp_nistz256_point_add_affine:
        orr     $t0,$t0,$t2
        orr     $in2infty,$acc0,$t0
        cmp     $in2infty,#0
-       csetm   $in2infty,ne            // !in2infty
+       csetm   $in2infty,ne            // ~in2infty
 
        add     $rp,sp,#$Z1sqr
        bl      __ecp_nistz256_sqr_mont // p256_sqr_mont(Z1sqr, in1_z);
@@ -1293,14 +1284,14 @@ ___
 for($i=0;$i<64;$i+=32) {               # conditional moves
 $code.=<<___;
        ldp     $acc0,$acc1,[$ap_real,#$i]      // in1
-       cmp     $in1infty,#0                    // !$in1intfy, remember?
+       cmp     $in1infty,#0                    // ~$in1intfy, remember?
        ldp     $acc2,$acc3,[$ap_real,#$i+16]
        csel    $t0,$a0,$t0,ne
        csel    $t1,$a1,$t1,ne
        ldp     $a0,$a1,[sp,#$res_x+$i+32]      // res
        csel    $t2,$a2,$t2,ne
        csel    $t3,$a3,$t3,ne
-       cmp     $in2infty,#0                    // !$in2intfy, remember?
+       cmp     $in2infty,#0                    // ~$in2intfy, remember?
        ldp     $a2,$a3,[sp,#$res_x+$i+48]
        csel    $acc0,$t0,$acc0,ne
        csel    $acc1,$t1,$acc1,ne
@@ -1317,13 +1308,13 @@ ___
 }
 $code.=<<___;
        ldp     $acc0,$acc1,[$ap_real,#$i]      // in1
-       cmp     $in1infty,#0                    // !$in1intfy, remember?
+       cmp     $in1infty,#0                    // ~$in1intfy, remember?
        ldp     $acc2,$acc3,[$ap_real,#$i+16]
        csel    $t0,$a0,$t0,ne
        csel    $t1,$a1,$t1,ne
        csel    $t2,$a2,$t2,ne
        csel    $t3,$a3,$t3,ne
-       cmp     $in2infty,#0                    // !$in2intfy, remember?
+       cmp     $in2infty,#0                    // ~$in2intfy, remember?
        csel    $acc0,$t0,$acc0,ne
        csel    $acc1,$t1,$acc1,ne
        csel    $acc2,$t2,$acc2,ne