Chuck's recently-added RPCSEC GSS krb5 KUnit test
(net/sunrpc/auth_gss/gss_krb5_test.c) is failing on arm64, specifically
the RFC 3962 test cases (I'm just pasting the output of 1 case, but all
6 cases fail):
---8<---
[ 237.255197] # Subtest: RFC 3962 encryption
[ 237.255588] # RFC 3962 encryption: EXPECTATION FAILED at net/sunrpc/auth_gss/gss_krb5_test.c:772
Expected memcmp(param->next_iv->data, iv, param->next_iv->len) == 0, but
memcmp(param->next_iv->data, iv, param->next_iv->len) == 1 (0x1)
IV mismatch
---8<---
If I disable the hardware accelerated ciphers
(CONFIG_CRYPTO_AES_ARM64_CE_BLK and CONFIG_CRYPTO_AES_ARM64_NEON_BLK),
then the test works.
Likewise, if I modify Chuck's test to explicitly request
"cts(cbc(aes-generic))", then the test works.
The problem is that the asm helper aes_cbc_cts_encrypt in
arch/arm64/crypto/aes-modes.S doesn't return the next IV.
If I make the following change, then the test works:
diff --git a/arch/arm64/crypto/aes-modes.S b/arch/arm64/crypto/aes-modes.S
index 0e834a2c062c..477605fad76b 100644
--- a/arch/arm64/crypto/aes-modes.S
+++ b/arch/arm64/crypto/aes-modes.S
@@ -268,6 +268,7 @@ AES_FUNC_START(aes_cbc_cts_encrypt)
add x4, x0, x4
st1 {v0.16b}, [x4] /* overlapping stores */
st1 {v1.16b}, [x0]
+ st1 {v1.16b}, [x5]
ret
AES_FUNC_END(aes_cbc_cts_encrypt)
But I don't know if that change is at all correct! (I've never even
looked at arm64 asm before). If someone who's knowledgeable about this
code could chime in, I'd appreciate it.
-Scott
