On Mon, 2021-02-01 at 17:18 +0000, Luis Henriques wrote:
> Jeff Layton <jlayton@xxxxxxxxxx> writes:
>
> > If we have a dentry which represents a no-key name, then we need to test
> > whether the parent directory's encryption key has since been added. Do
> > that before we test anything else about the dentry.
> >
> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > ---
> > fs/ceph/dir.c | 4 ++++
> > 1 file changed, 4 insertions(+)
> >
> > diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
> > index 236c381ab6bd..cb7ff91a243a 100644
> > --- a/fs/ceph/dir.c
> > +++ b/fs/ceph/dir.c
> > @@ -1726,6 +1726,10 @@ static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags)
> > dout("d_revalidate %p '%pd' inode %p offset 0x%llx\n", dentry,
> > dentry, inode, ceph_dentry(dentry)->offset);
> >
> >
> > + valid = fscrypt_d_revalidate(dentry, flags);
> > + if (valid <= 0)
> > + return valid;
> > +
>
> This one took me a while to figure out, but eventually got there.
> Initially I was seeing this error:
>
> crypt: ceph: 1 inode(s) still busy after removing key with identifier f019f4a1c5d5665675218f89fccfa3c7, including ino 1099511627791
>
> and, when umounting the filesystem I would get the warning in
> fs/dcache.c:1623.
>
> Anyway, the patch below should fix it.
>
> Unfortunately I didn't had a lot of time to look into the -experimental
> branch yet. On my TODO list for the next few days.
>
> Cheers,
Well spotted! I think the better fix though is to just move the
fscrypt_d_revalidate call up before the point where we take the parent
reference. I'll fix that up in my tree. Thanks for tracking that down!
--
Jeff Layton <jlayton@xxxxxxxxxx>
