[Bug 247986] geli: authentication panics on Via Nano

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=247986

--- Comment #3 from John Baldwin <jhb at FreeBSD.org> ---
Ok, it may have to do with the size of the request.  The '-z' flag will try
various sizes.  However, it may be good to figure out what algorithms the kyua
test is using and then test that specifically, e.g. 'cryptocheck -a
aes-xts+sha256hmac -d padlock0 -z'.

Hmm, however, I do see a bug, but it's a bug in 12 as well.  The padlock()
function returns the crp_error value from its process() callback.  Probably the
cipher and hash functions just never returned errors before until gaining
verify mode in head which can return EBADMSG.

Try this:

Index: sys/crypto/via/padlock.c
===================================================================
--- padlock.c   (revision 363276)
+++ padlock.c   (working copy)
@@ -275,7 +275,7 @@ out:
 #endif
        crp->crp_etype = error;
        crypto_done(crp);
-       return (error);
+       return (0);
 }

 static device_method_t padlock_methods[] = {

-- 
You are receiving this mail because:
You are the assignee for the bug.