The elliptic curve arithmetic library used by the EC-DH KPP implementation
assumes big endian byte order, and unconditionally reverses the byte
and word order of multi-limb quantities. On big endian systems, the byte
reordering is not necessary, while the word ordering needs to be retained.

So replace the __swab64() invocation with a call to be64_to_cpu() which
should do the right thing for both little and big endian builds.

Cc: <sta...@vger.kernel.org> # v4.9+
Signed-off-by: Ard Biesheuvel <ard.biesheu...@linaro.org>
---
 crypto/ecc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/crypto/ecc.c b/crypto/ecc.c
index dfe114bc0c4a..8ee787723c5c 100644
--- a/crypto/ecc.c
+++ b/crypto/ecc.c
@@ -1284,10 +1284,11 @@ EXPORT_SYMBOL(ecc_point_mult_shamir);
 static inline void ecc_swap_digits(const u64 *in, u64 *out,
                                   unsigned int ndigits)
 {
+       const __be64 *src = (__force __be64 *)in;
        int i;
 
        for (i = 0; i < ndigits; i++)
-               out[i] = __swab64(in[ndigits - 1 - i]);
+               out[i] = be64_to_cpu(src[ndigits - 1 - i]);
 }
 
 static int __ecc_is_key_valid(const struct ecc_curve *curve,
-- 
2.20.1

Reply via email to