FreeBSD6.0-BETA1 panics

[Kris Kennaway]

On Tue, Aug 02, 2005 at 11:27:35AM -0400, rick at snowhite.cis.uoguelph.ca wrote:
> > I suspect the server is returning bogus data, perhaps because of the
> > locking problems.
> 
> My server doesn't support the lockd/statd protocol, so v3 mounts won't
> have any advisory locking support. Is that likely to be the cause of this?
> (I don't know anything about the inner workings of cvs.)

I meant vnode locking in the kernel (i.e. race between the two nfsd
processes reading files on the server).  Everything is read-only here
(and static on the server), so there is no write collision from the
client.

> > OK, I triggered another deadlock while running a simultaneous cvs
> > checkout on the server (via ufs) and over nfs3 from a remote machine.
> 
> Ok, thanks for the info. I'll try and figure this one out.
> (Given your original panic, I assume you're running a kernel with
>  DEBUG_LOCKS and DEBUG_VFS_LOCKS options? That would have caught the
>  obvious "forgot to unlock the vnode" type problems, which would suggest
>  a race between the local ufs syscall and newnfsd for vnode locking.
>  Could be a fun one to find:-)

Only DEBUG_LOCKS, I think..I'll add the other.

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-fs/attachments/20050802/bb645800/attachment.bin