Re: [PATCH] fs: Remove implicit nodev for new mounts in non-root userns

[Serge Hallyn <serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org>]

Quoting Andy Lutomirski (luto-kltTT9wpgjJwATOyAt5JVQ@public.gmane.org):
> On Fri, Aug 15, 2014 at 12:37 PM, Serge Hallyn <serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org> wrote:
> > Quoting Andy Lutomirski (luto-kltTT9wpgjJwATOyAt5JVQ@public.gmane.org):
> >> On Fri, Aug 15, 2014 at 12:05 PM, Serge Hallyn <serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org> wrote:
> >> > Quoting Andy Lutomirski (luto-kltTT9wpgjJwATOyAt5JVQ@public.gmane.org):
> >> >> Currently, creating a new mount (as opposed to bindmount) in a
> >> >> non-root userns will implicitly set nodev unless the fs is devpts.
> >> >> Something like this will be necessary for file systems that allow
> >> >> the mounter to create device nodes without using mknod (e.g. FUSE
> >> >> if/when that is allowed), but none of the currently allowed
> >> >> filesystems do this.
> >> >
> >> > Hi,
> >> >
> >> > Sorry, I'm probably thinking stupidly, but I don't see this restriction
> >> > being the case
> >> >
> >> > serge@sl:~$ mount | grep tmp
> >> > [...]
> >> > tmpfs on /run type tmpfs (rw,noexec,nosuid,size=10%,mode=0755)
> >> > serge@sl:~$ sudo mknod /run/kvm c 10 232
> >> > [sudo] password for serge:
> >> > serge@sl:~$ echo $?
> >> > 0
> >> > serge@sl:~$ ls -l /run/kvm
> >> > crw-r--r-- 1 root root 10, 232 Aug 15 14:04 /run/kvm
> >> >
> >> > But you seem to be saying I shouldn't be allowed to create a device inside
> >> > a tmpfs.  What am I overlooking?
> >>
> >> I assume you're in the root userns.  This patch is unnecessary, and
> >> has no effect, if you're in the root userns.
> >
> > Right, but I thought you were justifying adding FS_USERNS_DEV_MOUNT by saying
> > that you cannot mknod in those filesystems.  But I see you actually said
> > "without using mknod".  I guess I don't understand that caveat.
> 
> IIUC, there are two ways that a user could put a device node into
> their filesystem.
> 
> The obvious way is using mknod.  But mknod has its own perfectly valid
> permission checks, and it doesn't need any special handling at mount
> time.
> 
> The less obvious way is to mount a filesystem that already contains a
> device node or to mount a filesystem that gives some other means of
> inserting a device node (e.g. a network filesystem or FUSE).  Those
> might allow inserting device nodes without passing a global capability
> check, so unprivileged users in a userns must not be allowed to mount
> such a filesystem without MNT_NODEV | MNT_LOCK_NODEV.
> 
> Fortunately, none of the existing FS_USERNS_MOUNT filesystems have
> that property.  FUSE will, but we don't support FUSE in a userns yet
> (unfortunately -- it would be a *very* useful feature.)
> 
> I think that, if we ever allow FUSE in a userns, we should return

Which, btw, I'm hoping we'll be allowing soon.

> -EPERM when trying to mount it unless the user specifies MS_NODEV,

In either case we can think that through when the time comes.

> which is what this patch does.  I don't think there's any reason to
> play complicated games to allow programs to get away with omitting
> MS_NODEV.

Thanks, Andy.