cifs: Check for timeout on Negotiate stage

Some servers seem to accept connections while booting but never send
the SMBNegotiate response neither close the connection, causing all
processes accessing the share hang on uninterruptible sleep state.

This happens when the cifs_demultiplex_thread detects the server is
unresponsive so releases the socket and start trying to reconnect.
At some point, the faulty server will accept the socket and the TCP
status will be set to NeedNegotiate. The first issued command accessing
the share will start the negotiation (pid 5828 below), but the response
will never arrive so other commands will be blocked waiting on the mutex
(pid 55352).

This patch checks for unresponsive servers also on the negotiate stage
releasing the socket and reconnecting if the response is not received
and checking again the tcp state when the mutex is acquired.

PID: 55352  TASK: ffff880fd6cc02c0  CPU: 0   COMMAND: "ls"
 #0 [ffff880fd9add9f0] schedule at ffffffff81467eb9
 #1 [ffff880fd9addb38] __mutex_lock_slowpath at ffffffff81468fe0
 #2 [ffff880fd9addba8] mutex_lock at ffffffff81468b1a
 #3 [ffff880fd9addbc0] cifs_reconnect_tcon at ffffffffa042f905 [cifs]
 #4 [ffff880fd9addc60] smb_init at ffffffffa042faeb [cifs]
 #5 [ffff880fd9addca0] CIFSSMBQPathInfo at ffffffffa04360b5 [cifs]
 ....

Which is waiting a mutex owned by:

PID: 5828   TASK: ffff880fcc55e400  CPU: 0   COMMAND: "xxxx"
 #0 [ffff880fbfdc19b8] schedule at ffffffff81467eb9
 #1 [ffff880fbfdc1b00] wait_for_response at ffffffffa044f96d [cifs]
 #2 [ffff880fbfdc1b60] SendReceive at ffffffffa04505ce [cifs]
 #3 [ffff880fbfdc1bb0] CIFSSMBNegotiate at ffffffffa0438d79 [cifs]
 #4 [ffff880fbfdc1c50] cifs_negotiate_protocol at ffffffffa043b383 [cifs]
 #5 [ffff880fbfdc1c80] cifs_reconnect_tcon at ffffffffa042f911 [cifs]
 #6 [ffff880fbfdc1d20] smb_init at ffffffffa042faeb [cifs]
 #7 [ffff880fbfdc1d60] CIFSSMBQFSInfo at ffffffffa0434eb0 [cifs]
 ....

Signed-off-by: Samuel Cabrero <scabrero@suse.de>
Reviewed-by: Aurélien Aptel <aaptel@suse.de>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>

3 files changed, 26 insertions, 1 deletions

diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c
index 48455afefec8..7cbe28315971 100644
--- a/fs/cifs/cifssmb.c
+++ b/fs/cifs/cifssmb.c
@@ -178,6 +178,18 @@ cifs_reconnect_tcon(struct cifs_tcon *tcon, int smb_command)
 	 * reconnect the same SMB session
 	 */
 	mutex_lock(&ses->session_mutex);
+
+	/*
+	 * Recheck after acquire mutex. If another thread is negotiating
+	 * and the server never sends an answer the socket will be closed
+	 * and tcpStatus set to reconnect.
+	 */
+	if (server->tcpStatus == CifsNeedReconnect) {
+		rc = -EHOSTDOWN;
+		mutex_unlock(&ses->session_mutex);
+		goto out;
+	}
+
 	rc = cifs_negotiate_protocol(0, ses);
 	if (rc == 0 && ses->need_reconnect)
 		rc = cifs_setup_session(0, ses, nls_codepage);
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 83a8f52cd879..5aa2d278ca84 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -509,7 +509,8 @@ server_unresponsive(struct TCP_Server_Info *server)
 	 * 65s kernel_recvmsg times out, and we see that we haven't gotten
 	 *     a response in >60s.
 	 */
-	if (server->tcpStatus == CifsGood &&
+	if ((server->tcpStatus == CifsGood ||
+	    server->tcpStatus == CifsNeedNegotiate) &&
 	    time_after(jiffies, server->lstrp + 2 * server->echo_interval)) {
 		cifs_dbg(VFS, "Server %s has not responded in %lu seconds. Reconnecting...\n",
 			 server->hostname, (2 * server->echo_interval) / HZ);
diff --git a/fs/cifs/smb2pdu.c b/fs/cifs/smb2pdu.c
index d7595e735304..5531e7ee1210 100644
--- a/fs/cifs/smb2pdu.c
+++ b/fs/cifs/smb2pdu.c
@@ -238,6 +238,18 @@ smb2_reconnect(__le16 smb2_command, struct cifs_tcon *tcon)
 	 * the same SMB session
 	 */
 	mutex_lock(&tcon->ses->session_mutex);
+
+	/*
+	 * Recheck after acquire mutex. If another thread is negotiating
+	 * and the server never sends an answer the socket will be closed
+	 * and tcpStatus set to reconnect.
+	 */
+	if (server->tcpStatus == CifsNeedReconnect) {
+		rc = -EHOSTDOWN;
+		mutex_unlock(&tcon->ses->session_mutex);
+		goto out;
+	}
+
 	rc = cifs_negotiate_protocol(0, tcon->ses);
 	if (!rc && tcon->ses->need_reconnect)
 		rc = cifs_setup_session(0, tcon->ses, nls_codepage);