Re: [RFC][PATCH 2/2] fanotify: support limited functionality for unprivileged users

[Amir Goldstein <amir73il@gmail.com>]

On Tue, Feb 16, 2021 at 8:12 PM Amir Goldstein <amir73il@gmail.com> wrote:
>
> On Tue, Feb 16, 2021 at 7:01 PM Jan Kara <jack@suse.cz> wrote:
> >
> > On Sun 24-01-21 20:42:04, Amir Goldstein wrote:
> > > Add limited support for unprivileged fanotify event listener.
> > > An unprivileged event listener does not get an open file descriptor in
> > > the event nor the process pid of another process.  An unprivileged event
> > > listener cannot request permission events, cannot set mount/filesystem
> > > marks and cannot request unlimited queue/marks.
> > >
> > > This enables the limited functionality similar to inotify when watching a
> > > set of files and directories for OPEN/ACCESS/MODIFY/CLOSE events, without
> > > requiring SYS_CAP_ADMIN privileges.
> > >
> > > The FAN_REPORT_DFID_NAME init flag, provide a method for an unprivileged
> > > event listener watching a set of directories (with FAN_EVENT_ON_CHILD)
> > > to monitor all changes inside those directories.
> > >
> > > This typically requires that the listener keeps a map of watched directory
> > > fid to dirfd (O_PATH), where fid is obtained with name_to_handle_at()
> > > before starting to watch for changes.
> > >
> > > When getting an event, the reported fid of the parent should be resolved
> > > to dirfd and fstatsat(2) with dirfd and name should be used to query the
> > > state of the filesystem entry.
> > >
> > > Note that even though events do not report the event creator pid,
> > > fanotify does not merge similar events on the same object that were
> > > generated by different processes. This is aligned with exiting behavior
> > > when generating processes are outside of the listener pidns (which
> > > results in reporting 0 pid to listener).
> > >
> > > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> >
> > The patch looks mostly good to me. Just two questions:
> >
> > a) Remind me please, why did we decide pid isn't safe to report to
> > unpriviledged listeners?
>
> Just because the information that process X modified file Y is not an
> information that user can generally obtain without extra capabilities(?)
> I can add a flag FAN_REPORT_OWN_PID to make that behavior
> explicit and then we can relax reporting pids later.
>

FYI a patch for flag FAN_REPORT_SELF_PID is pushed to branch
fanotify_unpriv.

The UAPI feels a bit awkward with this flag, but that is the easiest way
to start without worrying about disclosing pids.

I guess we can require that unprivileged listener has pid 1 in its own
pid ns. The outcome is similar to FAN_REPORT_SELF_PID, except
it can also get pids of its children which is probably fine.

I am not sure if this is a reasonable option from users POV.

> >
> > b) Why did we decide returning open file descriptors isn't safe for
> > unpriviledged listeners? Is it about FMODE_NONOTIFY?
> >
>
> Don't remember something in particular. I feels risky.
>
> > I'm not opposed to either but I'm wondering. Also with b) old style
> > fanotify events are not very useful so maybe we could just disallow all
> > notification groups without FID/DFID reporting? In the future if we ever
> > decide returning open fds is safe or how to do it, we can enable that group
> > type for unpriviledged users. However just starting to return open fds
> > later won't fly because listener has to close these fds when receiving
> > events.
> >
>
> I like this option better.
>

This is also pushed to branch fanotify_unpriv.
With all the behavior specified explicitly in fanotify_init() and
fanotify_mark() flags, there is no need for the internal
FANOTIFY_UNPRIV group flag, which looks better IMO.

Thanks,
Amir.