Re: bug in 2.6.22-rc2: loop mount limited to one single iso image

[Kay Sievers <kay.sievers@vrfy.org>]

On Sun, 2007-05-20 at 09:10 -0700, Ray Lee wrote:
> On 5/20/07, Kay Sievers <kay.sievers@vrfy.org> wrote:
> > On 5/20/07, Ray Lee <ray-lk@madrabbit.org> wrote:
> > > On 5/19/07, Al Viro <viro@ftp.linux.org.uk> wrote:
> > > > On Sat, May 19, 2007 at 11:16:59PM -0700, Ray Lee wrote:
> > > > > Ken? Ball's in your court. As the patch isn't providing a killer
> > > > > feature for 2.6.22, I'd suggest just reverting it for now until the
> > > > > issues are ironed out.
> > > >
> > > > Hold it.  The real question here is which logics do we want there.
> > > > IOW, and how many device nodes do we want to appear and _when_ do
> > > > we want them to appear?
> > >
> > > The when part is what looks to make it racy. I'm guessing that we're
> > > relying on udev to create those loop nodes. If so, I think any scheme
> > > that creates more on demand would give transient mount errors while
> > > it's waiting on udev to create more nodes.
> > >
> > > Perhaps if we were to start with 8 loop nodes at init (as we have in
> > > 2.6.21), and then always maintain a margin of 8 (or 4, or...) when
> > > they start being used or detached?
> >
> > Until the tools can request dynamic loop device allocation from the
> > kernel before they want to use the device, you can create as many as
> > needed "static" loop* nodes in /lib/udev/devices/, which will be
> > copied to /dev/ early on every bootup.
> 
> Except that's different than current behavior presented to userspace.
> IOW, we broke userspace for anyone using udev. Which is, y'know, a lot
> of us.
> 
> We're at -rc2 right now. Given that, it looks like we have two
> options. First is to revert all this for now and try again when the
> patch has had more testing and agreement (as this isn't a major
> feature we're talking about here; it's effectively just a cleanup that
> happened to have unfortunate side-effects).
> 
> The second option is that we could have the loop device start with 8
> nodes populated, which would match current behavior.
> 
> A third option of requiring new userspace for 2.6.22 is a non-starter.

Right, providing "preallocated" devices, 8 or the number given in
max_loop, sounds like the best option until the tools can handle that.

Thanks,
Kay

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/