fs/ufs: three reviewed fixes looking for a tree

Hi Christian,

I have three UFS fixes that Jan has reviewed, but they haven't been
picked up yet. I'm not quite sure which tree they should go through.

There isn't a separate MAINTAINERS entry for fs/ufs, and
get_maintainer.pl doesn't turn up much for files under it. I'm pinging
you since you've taken most of the recent UFS changes, but please let me
know if I should send these elsewhere.

ufs: create the root dentry after loading cylinder metadata
Cc: stable, Reviewed-by: Jan Kara
https://lore.kernel.org/all/[email protected]/

ufs: validate cylinder group metadata before caching it
Cc: stable, Reviewed-by: Jan Kara
https://lore.kernel.org/all/[email protected]/

ufs: do not treat unreadable directory blocks as empty
Reviewed-by: Jan Kara
https://lore.kernel.org/all/[email protected]/

The first two fix a mount teardown NULL dereference and an out-of-bounds
access caused by on-disk cylinder group fields; both can be triggered by
a crafted filesystem image. The third fixes an rmdir case where an
unreadable directory block could cause a non-empty directory to be
treated as empty.

If useful, I can resend these as a small series with Jan's review tags
collected, or rebase them onto whichever tree you'd prefer.

I also have a larger series from Aug 1 that hasn't had any replies yet:

[PATCH 00/10] ufs: correctness fixes for valid on-disk filesystems
https://lore.kernel.org/all/[email protected]/

That series is a bit different from the three above: the issues show up
on valid FreeBSD filesystems rather than crafted images. It includes a
symlink case that can read past the end of the device and panic on
unlink, a tail relocation bug that can corrupt data on 64 KiB pages,
and missing enforcement of on-disk immutable and append-only flags.

It's a larger review job, which is why I kept it separate. No rush on
the series as a whole, but there are a few fairly serious bugs in there.

Thanks,
Ali