Re: nfsfh.c:nfsd_findparent lookup("..") failure fix in 2.4.4 - xfs related?

Trond Myklebust (trond.myklebust@fys.uio.no)
18 Oct 2001 12:42:01 +0200


>>>>> " " == Peter Wächtler <pwaechtler@loewe-komp.de> writes:

> The following diff was made in 2.4.4. diff -u --recursive
> --new-file v2.4.4/linux/fs/nfsd/nfsfh.c linux/fs/nfsd/nfsfh.c
> --- v2.4.4/linux/fs/nfsd/nfsfh.c Fri Feb 9 11:29:44 2001
> +++ linux/fs/nfsd/nfsfh.c Sat May 19 17:47:55 2001
> @@ -244,6 +245,11 @@
> */
> pdentry = child->d_inode->i_op->lookup(child->d_inode,
> tdentry); d_drop(tdentry); /* we never want ".." hashed
> */
> + if (!pdentry && tdentry->d_inode == NULL) {
> + /* File system cannot find ".." ... sad but possible */
> + dput(tdentry);
> + pdentry = ERR_PTR(-EINVAL);
> + }

> Umh. How is it possible to have a valid dentry which has no
> parent? Even "/.." is linked to "/."

Who said these are *valid* dentries?

There's no way you can fit full path information into an NFS
filehandle, so when the client passes such an object to nfsd, the
latter sometimes has to make so-called 'disconnected' dentries. These
are dentries which contain no path information, and are basically just
providing a dentry alias for the inode.

The code you are looking at attempts to call the filesystem in order
to lookup a file named '..' from just such a disconnected dentry. The
purpose being to actually recreate the path, and hence convert the
dentry into a valid one...

Cheers,
Trond
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/