Re: do_change_type(): refuse to operate on unmounted/not ours mounts

[Al Viro <viro@zeniv.linux.org.uk>]

On Wed, Aug 13, 2025 at 01:09:27PM -0600, Tycho Andersen wrote:
> On Wed, Aug 13, 2025 at 07:56:01PM +0100, Al Viro wrote:
> > @@ -3347,18 +3360,11 @@ static int do_set_group(struct path *from_path, struct path *to_path)
> >  
> >  	namespace_lock();
> >  
> > -	err = -EINVAL;
> > -	/* To and From must be mounted */
> > -	if (!is_mounted(&from->mnt))
> > -		goto out;
> > -	if (!is_mounted(&to->mnt))
> > -		goto out;
> > -
> > -	err = -EPERM;
> > -	/* We should be allowed to modify mount namespaces of both mounts */
> > -	if (!ns_capable(from->mnt_ns->user_ns, CAP_SYS_ADMIN))
> > +	err = may_change_propagation(from);
> > +	if (err)
> >  		goto out;
> > -	if (!ns_capable(to->mnt_ns->user_ns, CAP_SYS_ADMIN))
> > +	err = may_change_propagation(from);
> 
> Just driving by, but I guess you mean "to" here.

D'oh...  Yes, of course.  Fun question: would our selftests have caught
that?
[checks]
move_mount_set_group_test.c doesn't have anything in that area, nothing in
LTP or xfstests either, AFAICS...  And I don't see anything in
https://github.com/checkpoint-restore/criu
either - there are uses of MOVE_MOUNT_SET_GROUP, but they are well-buried
and I don't see anything in their tests that would even try to poke into
that thing...

Before we go and try to cobble something up, does anybody know of a place
where regression tests for MOVE_MOUNT_SET_GROUP could be picked from?