Re: [PATCH v4 4/5] virtiofs: Skip submounts in sget_fc()

From: Greg Kurz
Date: Fri May 21 2021 - 06:07:02 EST


On Fri, 21 May 2021 10:50:34 +0200
Miklos Szeredi <miklos@xxxxxxxxxx> wrote:

> On Fri, 21 May 2021 at 10:39, Greg Kurz <groug@xxxxxxxx> wrote:
> >
> > On Fri, 21 May 2021 10:26:27 +0200
> > Miklos Szeredi <miklos@xxxxxxxxxx> wrote:
> >
> > > On Thu, 20 May 2021 at 17:47, Greg Kurz <groug@xxxxxxxx> wrote:
> > > >
> > > > All submounts share the same virtio-fs device instance as the root
> > > > mount. If the same virtiofs filesystem is mounted again, sget_fc()
> > > > is likely to pick up any of these submounts and reuse it instead of
> > > > the root mount.
> > > >
> > > > On the server side:
> > > >
> > > > # mkdir ${some_dir}
> > > > # mkdir ${some_dir}/mnt1
> > > > # mount -t tmpfs none ${some_dir}/mnt1
> > > > # touch ${some_dir}/mnt1/THIS_IS_MNT1
> > > > # mkdir ${some_dir}/mnt2
> > > > # mount -t tmpfs none ${some_dir}/mnt2
> > > > # touch ${some_dir}/mnt2/THIS_IS_MNT2
> > > >
> > > > On the client side:
> > > >
> > > > # mkdir /mnt/virtiofs1
> > > > # mount -t virtiofs myfs /mnt/virtiofs1
> > > > # ls /mnt/virtiofs1
> > > > mnt1 mnt2
> > > > # grep virtiofs /proc/mounts
> > > > myfs /mnt/virtiofs1 virtiofs rw,seclabel,relatime 0 0
> > > > none on /mnt/mnt1 type virtiofs (rw,relatime,seclabel)
> > > > none on /mnt/mnt2 type virtiofs (rw,relatime,seclabel)
> > > >
> > > > And now remount it again:
> > > >
> > > > # mount -t virtiofs myfs /mnt/virtiofs2
> > > > # grep virtiofs /proc/mounts
> > > > myfs /mnt/virtiofs1 virtiofs rw,seclabel,relatime 0 0
> > > > none on /mnt/mnt1 type virtiofs (rw,relatime,seclabel)
> > > > none on /mnt/mnt2 type virtiofs (rw,relatime,seclabel)
> > > > myfs /mnt/virtiofs2 virtiofs rw,seclabel,relatime 0 0
> > > > # ls /mnt/virtiofs2
> > > > THIS_IS_MNT2
> > > >
> > > > Submount mnt2 was picked-up instead of the root mount.
> > >
> >
> > > Why is this a problem?
> > >
> >
> > It seems very weird to mount the same filesystem again
> > and to end up in one of its submounts. We should have:
> >
> > # ls /mnt/virtiofs2
> > mnt1 mnt2
>
> Okay, sorry, I understand the problem. The solution is wrong,
> however: the position of the submount on that list is no indication
> that it's the right one (it's possible that the root sb will go away
> and only a sub-sb will remain).
>

Ah... I had myself convinced this could not happen, i.e. you can't
unmount a parent sb with a sub-sb still mounted.

How can this happen ?

> Even just setting a flag in the root, indicating that it's the root
> isn't fully going to solve the problem.
>
> Here's issue in full:
>
> case 1: no connection for "myfs" exists
> - need to create fuse_conn, sb
>
> case 2: connection for "myfs" exists but only sb for submount

How would we know this sb isn't a root sb ?

> - only create sb for root, reuse fuse_conn
>
> case 3: connection for "myfs" as well as root sb exists
> - reuse sb
>
> I'll think about how to fix this properly, it's probably going to be
> rather more involved...
>

Sure. BTW I'm wondering why we never reuse sbs for submounts ?

> Thanks,
> Miklos