Device departure processing

[Pawel Jakub Dawidek]

On Mon, Jan 31, 2011 at 02:39:08PM -0700, Justin T. Gibbs wrote:
> On both -current and -stable (v16), I've been able to reliably panic the
> system by pulling a drive (leaf vdev in a RAIDZ2) while I/O is
> in progress.  The source of the panic is either from GEOM complaining
> that vdev_geom's consumer isn't idle, or "use after free" bugs caused
> by in-flight I/O (often due to a re-probe).  The following patch resolves
> the issue for me.

I was able to reproduce the panic on v28 and I integrated your patch
there.

> >From a brief review of the v28 code, it seems appropriate there too.
> The code in vdev_geom_orphan is identical save for an added call to
> zfs_post_remove().  In our port, this call appears to do nothing more
> than to post the devctl notice event a little earlier than via the
> scheduled removal of the device by the spa code.  Is there some benefit
> I'm missing that makes the zfs_post_remove() necessary?

Without it I see no info about vdev removal in devd.

-- 
Pawel Jakub Dawidek                       http://www.wheelsystems.com
pjd at FreeBSD.org                           http://www.FreeBSD.org
FreeBSD committer                         Am I Evil? Yes, I Am!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/zfs-devel/attachments/20110204/3489cfc6/attachment.pgp