Re: [PATCH RESEND v2 16/18] fuse: Support fuse filesystems outside of init_user_ns

[Seth Forshee <seth.forshee@canonical.com>]

On Wed, Mar 09, 2016 at 03:48:22PM +0100, Miklos Szeredi wrote:
> On Wed, Mar 9, 2016 at 3:18 PM, Seth Forshee <seth.forshee@canonical.com> wrote:
> > On Wed, Mar 09, 2016 at 12:29:23PM +0100, Miklos Szeredi wrote:
> >> On Mon, Jan 04, 2016 at 12:03:55PM -0600, Seth Forshee wrote:
> 
> >> > -static int parse_fuse_opt(char *opt, struct fuse_mount_data *d, int is_bdev)
> >> > +static int parse_fuse_opt(char *opt, struct fuse_mount_data *d, int is_bdev,
> >> > +                     struct user_namespace *user_ns)
> >> >  {
> >> >     char *p;
> >> >     memset(d, 0, sizeof(struct fuse_mount_data));
> >> >     d->max_read = ~0;
> >> >     d->blksize = FUSE_DEFAULT_BLKSIZE;
> >> > +   d->user_id = make_kuid(user_ns, 0);
> >> > +   d->group_id = make_kgid(user_ns, 0);
> >>
> >> It is true that if "user_id=" or "group_id" options were omitted we used the
> >> zero uid/gid values.  However, this isn't actually used by anybody AFAIK, and
> >> generalizing it for userns doesn't seem to make much sense.
> >>
> >> So I suggest we that we instead return an error if mounting from a userns AND
> >> neither "allow_other" nor both "user_id" and "group_id" are specified.
> >
> > But those are also used for ownership of the connection files in
> > fusectl. In an allow_other mount shouldn't those files by owned by
> > namespace root and not global root?
> 
> Yes.
> 
> Can't we use current_cred()->uid/gid? Or fsuid/fsgid maybe?

That would be a departure from the current behavior in the !allow_other
case for unprivileged users. Since those mounts are done by an suid
helper all of those ids would be root in the userns, wouldn't they?

> When we have true unprivileged mounts, the user_id/group_id options
> become redundant anyway and we can just use the current credentials.

True, but we don't yet have that.

Thanks,
Seth