> Your patch seems to fix the EIO prob, but I got (till now) one:
>
> Feb 23 09:08:13 kernel: NFS: server hp, readdir reply truncated
> Feb 23 09:08:13 kernel: NFS: nr=170, slots=3, len=7
Yes, this is the expected behavior. The first detected overrun sets a flag to
make subsequent readdir calls pass the buffer size in longwords. If this problem
turns out to be prevalent, we could make a mount option to enable it from the
start.
> The "RPC garbage" problem is repeatable and I remember, that I had it
> also happen on copying files. The last time I tried the 2.1 kernel I
> promptly restarted my box to 2.0 since nearly everything is
> nfs-mounted here. So I had only the xemacs case.
OK, if the RPC garbage is repeatable, try uncommenting the #define
NFS_PAD_WRITES and that will probably fix it. Some servers are picky about
getting RPC messages that aren't lonword multiples.
Regards,
Bill
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu