[PATCH] knfsd: nfsd: call nfsd_setuser() on fh_compose(), fix nfsd4 permissions problem

In the typical v2/v3 case the only new filehandles used as arguments to
operations are filehandles taken directly off the wire, which don't get
dentries until fh_verify() is called.

But in v4 the filehandles that are arguments to operations were often created
by previous operations (putrootfh, lookup, etc.) using fh_compose, which sets
the dentry in the filehandle without calling nfsd_setuser().

This also means that, for example, if filesystem B is mounted on filesystem A,
and filesystem A is exported without root-squashing, then a client can bypass
the rootsquashing on B using a compound that starts at a filehandle in A,
crosses into B using lookups, and then does stuff in B.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

fs/nfsd/nfsfh.c

@@ -187,13 +187,6 @@ fh_verify(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, int access)
 			goto out;
 		}
 
-		/* Set user creds for this exportpoint */
-		error = nfsd_setuser(rqstp, exp);
-		if (error) {
-			error = nfserrno(error);
-			goto out;
-		}
-
 		/*
 		 * Look up the dentry using the NFS file handle.
 		 */
@@ -251,6 +244,14 @@ fh_verify(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, int access)
 	}
 	cache_get(&exp->h);
 
+	/* Set user creds for this exportpoint; necessary even in the "just
+	 * checking" case because this may be a filehandle that was created by
+	 * fh_compose, and that is about to be used in another nfsv4 compound
+	 * operation */
+	error = nfserrno(nfsd_setuser(rqstp, exp));
+	if (error)
+		goto out;
+
 	error = nfsd_mode_check(rqstp, dentry->d_inode->i_mode, type);
 	if (error)
 		goto out;