NFS: Discover NFSv4 server trunking when mounting

"Server trunking" is a fancy named for a multi-homed NFS server.
Trunking might occur if a client sends NFS requests for a single
workload to multiple network interfaces on the same server.  There
are some implications for NFSv4 state management that make it useful
for a client to know if a single NFSv4 server instance is
multi-homed.  (Note this is only a consideration for NFSv4, not for
legacy versions of NFS, which are stateless).

If a client cares about server trunking, no NFSv4 operations can
proceed until that client determines who it is talking to.  Thus
server IP trunking discovery must be done when the client first
encounters an unfamiliar server IP address.

The nfs_get_client() function walks the nfs_client_list and matches
on server IP address.  The outcome of that walk tells us immediately
if we have an unfamiliar server IP address.  It invokes
nfs_init_client() in this case.  Thus, nfs4_init_client() is a good
spot to perform trunking discovery.

Discovery requires a client to establish a fresh client ID, so our
client will now send SETCLIENTID or EXCHANGE_ID as the first NFS
operation after a successful ping, rather than waiting for an
application to perform an operation that requires NFSv4 state.

The exact process for detecting trunking is different for NFSv4.0 and
NFSv4.1, so a minorversion-specific init_client callout method is
introduced.

CLID_INUSE recovery is important for the trunking discovery process.
CLID_INUSE is a sign the server recognizes the client's nfs_client_id4
id string, but the client is using the wrong principal this time for
the SETCLIENTID operation.  The SETCLIENTID must be retried with a
series of different principals until one works, and then the rest of
trunking discovery can proceed.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 253 insertions, 0 deletions

diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 612f5ebaabac..14ddd4d30966 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -185,6 +185,7 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 				    rpc_authflavor_t authflavour)
 {
 	char buf[INET6_ADDRSTRLEN + 1];
+	struct nfs_client *old;
 	int error;
 
 	if (clp->cl_cons_state == NFS_CS_READY) {
@@ -230,6 +231,17 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 
 	if (!nfs4_has_session(clp))
 		nfs_mark_client_ready(clp, NFS_CS_READY);
+
+	error = nfs4_discover_server_trunking(clp, &old);
+	if (error < 0)
+		goto error;
+	if (clp != old) {
+		clp->cl_preserve_clid = true;
+		nfs_put_client(clp);
+		clp = old;
+		atomic_inc(&clp->cl_count);
+	}
+
 	return clp;
 
 error:
@@ -239,6 +251,247 @@ error:
 	return ERR_PTR(error);
 }
 
+/*
+ * Returns true if the client IDs match
+ */
+static bool nfs4_match_clientids(struct nfs_client *a, struct nfs_client *b)
+{
+	if (a->cl_clientid != b->cl_clientid) {
+		dprintk("NFS: --> %s client ID %llx does not match %llx\n",
+			__func__, a->cl_clientid, b->cl_clientid);
+		return false;
+	}
+	dprintk("NFS: --> %s client ID %llx matches %llx\n",
+		__func__, a->cl_clientid, b->cl_clientid);
+	return true;
+}
+
+/*
+ * SETCLIENTID just did a callback update with the callback ident in
+ * "drop," but server trunking discovery claims "drop" and "keep" are
+ * actually the same server.  Swap the callback IDs so that "keep"
+ * will continue to use the callback ident the server now knows about,
+ * and so that "keep"'s original callback ident is destroyed when
+ * "drop" is freed.
+ */
+static void nfs4_swap_callback_idents(struct nfs_client *keep,
+				      struct nfs_client *drop)
+{
+	struct nfs_net *nn = net_generic(keep->cl_net, nfs_net_id);
+	unsigned int save = keep->cl_cb_ident;
+
+	if (keep->cl_cb_ident == drop->cl_cb_ident)
+		return;
+
+	dprintk("%s: keeping callback ident %u and dropping ident %u\n",
+		__func__, keep->cl_cb_ident, drop->cl_cb_ident);
+
+	spin_lock(&nn->nfs_client_lock);
+
+	idr_replace(&nn->cb_ident_idr, keep, drop->cl_cb_ident);
+	keep->cl_cb_ident = drop->cl_cb_ident;
+
+	idr_replace(&nn->cb_ident_idr, drop, save);
+	drop->cl_cb_ident = save;
+
+	spin_unlock(&nn->nfs_client_lock);
+}
+
+/**
+ * nfs40_walk_client_list - Find server that recognizes a client ID
+ *
+ * @new: nfs_client with client ID to test
+ * @result: OUT: found nfs_client, or new
+ * @cred: credential to use for trunking test
+ *
+ * Returns zero, a negative errno, or a negative NFS4ERR status.
+ * If zero is returned, an nfs_client pointer is planted in "result."
+ *
+ * NB: nfs40_walk_client_list() relies on the new nfs_client being
+ *     the last nfs_client on the list.
+ */
+int nfs40_walk_client_list(struct nfs_client *new,
+			   struct nfs_client **result,
+			   struct rpc_cred *cred)
+{
+	struct nfs_net *nn = net_generic(new->cl_net, nfs_net_id);
+	struct nfs_client *pos, *n, *prev = NULL;
+	struct nfs4_setclientid_res clid = {
+		.clientid	= new->cl_clientid,
+		.confirm	= new->cl_confirm,
+	};
+	int status;
+
+	spin_lock(&nn->nfs_client_lock);
+	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
+		/* If "pos" isn't marked ready, we can't trust the
+		 * remaining fields in "pos" */
+		if (pos->cl_cons_state < NFS_CS_READY)
+			continue;
+
+		if (pos->rpc_ops != new->rpc_ops)
+			continue;
+
+		if (pos->cl_proto != new->cl_proto)
+			continue;
+
+		if (pos->cl_minorversion != new->cl_minorversion)
+			continue;
+
+		if (pos->cl_clientid != new->cl_clientid)
+			continue;
+
+		atomic_inc(&pos->cl_count);
+		spin_unlock(&nn->nfs_client_lock);
+
+		if (prev)
+			nfs_put_client(prev);
+
+		status = nfs4_proc_setclientid_confirm(pos, &clid, cred);
+		if (status == 0) {
+			nfs4_swap_callback_idents(pos, new);
+
+			nfs_put_client(pos);
+			*result = pos;
+			dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
+				__func__, pos, atomic_read(&pos->cl_count));
+			return 0;
+		}
+		if (status != -NFS4ERR_STALE_CLIENTID) {
+			nfs_put_client(pos);
+			dprintk("NFS: <-- %s status = %d, no result\n",
+				__func__, status);
+			return status;
+		}
+
+		spin_lock(&nn->nfs_client_lock);
+		prev = pos;
+	}
+
+	/*
+	 * No matching nfs_client found.  This should be impossible,
+	 * because the new nfs_client has already been added to
+	 * nfs_client_list by nfs_get_client().
+	 *
+	 * Don't BUG(), since the caller is holding a mutex.
+	 */
+	if (prev)
+		nfs_put_client(prev);
+	spin_unlock(&nn->nfs_client_lock);
+	pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
+	return -NFS4ERR_STALE_CLIENTID;
+}
+
+#ifdef CONFIG_NFS_V4_1
+/*
+ * Returns true if the server owners match
+ */
+static bool
+nfs4_match_serverowners(struct nfs_client *a, struct nfs_client *b)
+{
+	struct nfs41_server_owner *o1 = a->cl_serverowner;
+	struct nfs41_server_owner *o2 = b->cl_serverowner;
+
+	if (o1->minor_id != o2->minor_id) {
+		dprintk("NFS: --> %s server owner minor IDs do not match\n",
+			__func__);
+		return false;
+	}
+
+	if (o1->major_id_sz != o2->major_id_sz)
+		goto out_major_mismatch;
+	if (memcmp(o1->major_id, o2->major_id, o1->major_id_sz) != 0)
+		goto out_major_mismatch;
+
+	dprintk("NFS: --> %s server owners match\n", __func__);
+	return true;
+
+out_major_mismatch:
+	dprintk("NFS: --> %s server owner major IDs do not match\n",
+		__func__);
+	return false;
+}
+
+/**
+ * nfs41_walk_client_list - Find nfs_client that matches a client/server owner
+ *
+ * @new: nfs_client with client ID to test
+ * @result: OUT: found nfs_client, or new
+ * @cred: credential to use for trunking test
+ *
+ * Returns zero, a negative errno, or a negative NFS4ERR status.
+ * If zero is returned, an nfs_client pointer is planted in "result."
+ *
+ * NB: nfs41_walk_client_list() relies on the new nfs_client being
+ *     the last nfs_client on the list.
+ */
+int nfs41_walk_client_list(struct nfs_client *new,
+			   struct nfs_client **result,
+			   struct rpc_cred *cred)
+{
+	struct nfs_net *nn = net_generic(new->cl_net, nfs_net_id);
+	struct nfs_client *pos, *n, *prev = NULL;
+	int error;
+
+	spin_lock(&nn->nfs_client_lock);
+	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
+		/* If "pos" isn't marked ready, we can't trust the
+		 * remaining fields in "pos", especially the client
+		 * ID and serverowner fields.  Wait for CREATE_SESSION
+		 * to finish. */
+		if (pos->cl_cons_state < NFS_CS_READY) {
+			atomic_inc(&pos->cl_count);
+			spin_unlock(&nn->nfs_client_lock);
+
+			if (prev)
+				nfs_put_client(prev);
+			prev = pos;
+
+			error = nfs_wait_client_init_complete(pos);
+			if (error < 0) {
+				nfs_put_client(pos);
+				continue;
+			}
+
+			spin_lock(&nn->nfs_client_lock);
+		}
+
+		if (pos->rpc_ops != new->rpc_ops)
+			continue;
+
+		if (pos->cl_proto != new->cl_proto)
+			continue;
+
+		if (pos->cl_minorversion != new->cl_minorversion)
+			continue;
+
+		if (!nfs4_match_clientids(pos, new))
+			continue;
+
+		if (!nfs4_match_serverowners(pos, new))
+			continue;
+
+		spin_unlock(&nn->nfs_client_lock);
+		dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
+			__func__, pos, atomic_read(&pos->cl_count));
+
+		*result = pos;
+		return 0;
+	}
+
+	/*
+	 * No matching nfs_client found.  This should be impossible,
+	 * because the new nfs_client has already been added to
+	 * nfs_client_list by nfs_get_client().
+	 *
+	 * Don't BUG(), since the caller is holding a mutex.
+	 */
+	spin_unlock(&nn->nfs_client_lock);
+	pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
+	return -NFS4ERR_STALE_CLIENTID;
+}
+#endif	/* CONFIG_NFS_V4_1 */
+
 static void nfs4_destroy_server(struct nfs_server *server)
 {
 	nfs_server_return_all_delegations(server);