Re: loop device auto release patch

[Ayan George <ayan@ayan.net>]

Phillip Susi <psusi@cfl.rr.com> wrote:
> On 07/15/2011 03:19 PM, Andrew Morton wrote:
> > The patch appears to do two unrelated things.  That's generally frowned
> > upon, but doesn't bother me much if the patch is small.
> > 
> > Still, splitting it into two patches (in which the bugfix is staged
> > first) would be advantageous for people who might wish to backport the
> > fix into earlier kernels.
>
> It looks like this got stalled and someone emailed me asking what
> happened to it.  I'm not sure that splitting the patch in two makes
> sense.  I don't see how it does two unrelated things. 

I agree -- I can't think of a meaningful way to split it up. I am
stalled in testing it however.

I watched the values of lo->lo_device and bdev using systemtap and
noticed cases where they could be out of sync which *might* make
this patch invalid.  I'd like a little more time to investigate.

-ayan