Fixes for the dentry refcounting leak I introduced in 4.8-rc1, and for

races in the LOCK code which appear to go back to the big nfsd state
 lock removal from 3.17.
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1
 
 iQIcBAABAgAGBQJXrkCqAAoJECebzXlCjuG+vJQP/RiBMW04XXV3Pe61S7URqVIr
 USTpsXJApxGQOhU6XrEMYWGz9Ya70aMpS9MmKqXtYtlIQN6gVSmC+t8YNPIPs2oF
 1n9a9tQhX/hI1Ipe8vjWmarLH31GOhQqbAd6RdUwHWGrMyWeajBKCms9UZ1bdG42
 dXBvlH7A8aoFJUY9GXerf2b2hyz34KFJmNxSx5e70XjF3Wq4HaQCCTKU8RFJmDxd
 PVYTz/0CR3bbtRKJkDHs6jRo1Qr9PZJXVxiRhvG113XrbmVZcBqTZ4Ee/2vwjRvr
 obxzQGMO7cb/GT6Iqly1tkdMfp/miVS/gPRXXcLQRJXNDfZwoRZF/2LMgABiDn62
 WXxgd6uqnexb3AAuCSpIW1HTgWLX+YekVYHdlZBs+YsTY2Q/jDNsy3yYDzA257yT
 HaHh0oyWmiiJQ+SgOc/KI1ony3aRtF+WclKsr2vQtmC/DmRkOBXpeFTcPN9K51aa
 BhNGVnC/YgZojgJsECZ+9VaYcDZ13UGFIoSPOA0zRWZRbDLw5TIt0mgnes63s3mp
 9pTGlEe0hhjG21JNbU3vxXrWcXY5o5fXaU3oCWlQJmk/dNehbBkY4x3fj/a1S2sK
 nxd8mrUruYcXEYvdjYGef38zUcJmhY/26wq0DGczm3jcxHfCDv4/ydhHa6TTuI8A
 HUQjymCrr3sWrWHmYATR
 =d1R8
 -----END PGP SIGNATURE-----

Merge tag 'nfsd-4.8-1' of git://linux-nfs.org/~bfields/linux

Pull nfsd fixes from Bruce Fields:
 "Fixes for the dentry refcounting leak I introduced in 4.8-rc1, and for
  races in the LOCK code which appear to go back to the big nfsd state
  lock removal from 3.17"

* tag 'nfsd-4.8-1' of git://linux-nfs.org/~bfields/linux:
  nfsd: don't return an unhashed lock stateid after taking mutex
  nfsd: Fix race between FREE_STATEID and LOCK
  nfsd: fix dentry refcounting on create

fs/nfsd/nfs4state.c

@@ -4903,6 +4903,32 @@ nfsd4_test_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	return nfs_ok;
 }
 
+static __be32
+nfsd4_free_lock_stateid(stateid_t *stateid, struct nfs4_stid *s)
+{
+	struct nfs4_ol_stateid *stp = openlockstateid(s);
+	__be32 ret;
+
+	mutex_lock(&stp->st_mutex);
+
+	ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
+	if (ret)
+		goto out;
+
+	ret = nfserr_locks_held;
+	if (check_for_locks(stp->st_stid.sc_file,
+			    lockowner(stp->st_stateowner)))
+		goto out;
+
+	release_lock_stateid(stp);
+	ret = nfs_ok;
+
+out:
+	mutex_unlock(&stp->st_mutex);
+	nfs4_put_stid(s);
+	return ret;
+}
+
 __be32
 nfsd4_free_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		   struct nfsd4_free_stateid *free_stateid)
@@ -4910,7 +4936,6 @@ nfsd4_free_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	stateid_t *stateid = &free_stateid->fr_stateid;
 	struct nfs4_stid *s;
 	struct nfs4_delegation *dp;
-	struct nfs4_ol_stateid *stp;
 	struct nfs4_client *cl = cstate->session->se_client;
 	__be32 ret = nfserr_bad_stateid;
 
@@ -4929,18 +4954,9 @@ nfsd4_free_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		ret = nfserr_locks_held;
 		break;
 	case NFS4_LOCK_STID:
-		ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
-		if (ret)
-			break;
-		stp = openlockstateid(s);
-		ret = nfserr_locks_held;
-		if (check_for_locks(stp->st_stid.sc_file,
-				    lockowner(stp->st_stateowner)))
-			break;
-		WARN_ON(!unhash_lock_stateid(stp));
+		atomic_inc(&s->sc_count);
 		spin_unlock(&cl->cl_lock);
-		nfs4_put_stid(s);
-		ret = nfs_ok;
+		ret = nfsd4_free_lock_stateid(stateid, s);
 		goto out;
 	case NFS4_REVOKED_DELEG_STID:
 		dp = delegstateid(s);
@@ -5507,7 +5523,7 @@ static __be32
 lookup_or_create_lock_state(struct nfsd4_compound_state *cstate,
 			    struct nfs4_ol_stateid *ost,
 			    struct nfsd4_lock *lock,
-			    struct nfs4_ol_stateid **lst, bool *new)
+			    struct nfs4_ol_stateid **plst, bool *new)
 {
 	__be32 status;
 	struct nfs4_file *fi = ost->st_stid.sc_file;
@@ -5515,7 +5531,9 @@ lookup_or_create_lock_state(struct nfsd4_compound_state *cstate,
 	struct nfs4_client *cl = oo->oo_owner.so_client;
 	struct inode *inode = d_inode(cstate->current_fh.fh_dentry);
 	struct nfs4_lockowner *lo;
+	struct nfs4_ol_stateid *lst;
 	unsigned int strhashval;
+	bool hashed;
 
 	lo = find_lockowner_str(cl, &lock->lk_new_owner);
 	if (!lo) {
@@ -5531,12 +5549,27 @@ lookup_or_create_lock_state(struct nfsd4_compound_state *cstate,
 			goto out;
 	}
 
-	*lst = find_or_create_lock_stateid(lo, fi, inode, ost, new);
-	if (*lst == NULL) {
+retry:
+	lst = find_or_create_lock_stateid(lo, fi, inode, ost, new);
+	if (lst == NULL) {
 		status = nfserr_jukebox;
 		goto out;
 	}
+
+	mutex_lock(&lst->st_mutex);
+
+	/* See if it's still hashed to avoid race with FREE_STATEID */
+	spin_lock(&cl->cl_lock);
+	hashed = !list_empty(&lst->st_perfile);
+	spin_unlock(&cl->cl_lock);
+
+	if (!hashed) {
+		mutex_unlock(&lst->st_mutex);
+		nfs4_put_stid(&lst->st_stid);
+		goto retry;
+	}
 	status = nfs_ok;
+	*plst = lst;
 out:
 	nfs4_put_stateowner(&lo->lo_owner);
 	return status;
@@ -5603,8 +5636,6 @@ nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 			goto out;
 		status = lookup_or_create_lock_state(cstate, open_stp, lock,
 							&lock_stp, &new);
-		if (status == nfs_ok)
-			mutex_lock(&lock_stp->st_mutex);
 	} else {
 		status = nfs4_preprocess_seqid_op(cstate,
 				       lock->lk_old_lock_seqid,

fs/nfsd/vfs.c

@@ -1252,10 +1252,13 @@ nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	if (IS_ERR(dchild))
 		return nfserrno(host_err);
 	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
-	if (err) {
-		dput(dchild);
+	/*
+	 * We unconditionally drop our ref to dchild as fh_compose will have
+	 * already grabbed its own ref for it.
+	 */
+	dput(dchild);
+	if (err)
 		return err;
-	}
 	return nfsd_create_locked(rqstp, fhp, fname, flen, iap, type,
 					rdev, resfhp);
 }