diff mbox

[[PATCH,v1] 23/37] [CIFS] SMBD: Implement API for upper layer to reconnect transport

Message ID 1501704648-20159-24-git-send-email-longli@exchange.microsoft.com (mailing list archive)
State New, archived
Headers show

Commit Message

Long Li Aug. 2, 2017, 8:10 p.m. UTC 
From: Long Li <longli@microsoft.com>

Upper layer may request to reconnect to server for multiple reason. Disconnect can happen at SMB layer or at SMBD layer. Add function to support reconnect when transport is disconnected.

Signed-off-by: Long Li <longli@microsoft.com>
---
 fs/cifs/cifsrdma.c | 24 ++++++++++++++++++++++++
 fs/cifs/cifsrdma.h |  3 +++
 2 files changed, 27 insertions(+)

Comments

Tom Talpey Aug. 14, 2017, 9:02 p.m. UTC | #1 
> -----Original Message-----
> From: linux-cifs-owner@vger.kernel.org [mailto:linux-cifs-
> owner@vger.kernel.org] On Behalf Of Long Li
> Sent: Wednesday, August 2, 2017 4:11 PM
> To: Steve French <sfrench@samba.org>; linux-cifs@vger.kernel.org; samba-
> technical@lists.samba.org; linux-kernel@vger.kernel.org
> Cc: Long Li <longli@microsoft.com>
> Subject: [[PATCH v1] 23/37] [CIFS] SMBD: Implement API for upper layer to
> reconnect transport
> 
> +int cifs_reconnect_rdma_session(struct TCP_Server_Info *server)
> +{
> +       log_rdma_event("reconnecting rdma session\n");
> +
> +       // why reconnect while it is still connected?
> +       if (server->rdma_ses->transport_status == CIFS_RDMA_CONNECTED) {
> +               log_rdma_event("still connected, not reconnecting\n");
> +               return -EINVAL;
> +       }

Why is this check needed?

> +
> +       // wait until the transport is destroyed
> +       while (server->rdma_ses->transport_status != CIFS_RDMA_DESTROYED)
> +               msleep(1);

Polling!? Please plan to implement a proper handshake for connection logic.

--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Long Li Aug. 14, 2017, 11:37 p.m. UTC | #2 
> -----Original Message-----
> From: Tom Talpey
> Sent: Monday, August 14, 2017 2:03 PM
> To: Long Li <longli@microsoft.com>; Steve French <sfrench@samba.org>;
> linux-cifs@vger.kernel.org; samba-technical@lists.samba.org; linux-
> kernel@vger.kernel.org
> Subject: RE: [[PATCH v1] 23/37] [CIFS] SMBD: Implement API for upper layer
> to reconnect transport
> 
> > -----Original Message-----
> > From: linux-cifs-owner@vger.kernel.org [mailto:linux-cifs-
> > owner@vger.kernel.org] On Behalf Of Long Li
> > Sent: Wednesday, August 2, 2017 4:11 PM
> > To: Steve French <sfrench@samba.org>; linux-cifs@vger.kernel.org;
> > samba- technical@lists.samba.org; linux-kernel@vger.kernel.org
> > Cc: Long Li <longli@microsoft.com>
> > Subject: [[PATCH v1] 23/37] [CIFS] SMBD: Implement API for upper layer
> > to reconnect transport
> >
> > +int cifs_reconnect_rdma_session(struct TCP_Server_Info *server) {
> > +       log_rdma_event("reconnecting rdma session\n");
> > +
> > +       // why reconnect while it is still connected?
> > +       if (server->rdma_ses->transport_status == CIFS_RDMA_CONNECTED)
> {
> > +               log_rdma_event("still connected, not reconnecting\n");
> > +               return -EINVAL;
> > +       }
> 
> Why is this check needed?

This was used in early stage of development. It's probably not needed anymore. Will look into this.

> 
> > +
> > +       // wait until the transport is destroyed
> > +       while (server->rdma_ses->transport_status !=
> CIFS_RDMA_DESTROYED)
> > +               msleep(1);
> 
> Polling!? Please plan to implement a proper handshake for connection logic.

Will look into using wait queue.
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/cifs/cifsrdma.c b/fs/cifs/cifsrdma.c
index 67a11d9..4681cda 100644
--- a/fs/cifs/cifsrdma.c
+++ b/fs/cifs/cifsrdma.c
@@ -1170,6 +1170,30 @@  static void destroy_receive_buffers(struct cifs_rdma_info *info)
 		mempool_free(response, info->response_mempool);
 }
 
+int cifs_reconnect_rdma_session(struct TCP_Server_Info *server)
+{
+	log_rdma_event("reconnecting rdma session\n");
+
+	// why reconnect while it is still connected?
+	if (server->rdma_ses->transport_status == CIFS_RDMA_CONNECTED) {
+		log_rdma_event("still connected, not reconnecting\n");
+		return -EINVAL;
+	}
+
+	// wait until the transport is destroyed
+	while (server->rdma_ses->transport_status != CIFS_RDMA_DESTROYED)
+		msleep(1);
+
+	if (server->rdma_ses)
+		kfree(server->rdma_ses);
+
+	log_rdma_event("creating rdma session\n");
+	server->rdma_ses = cifs_create_rdma_session(
+		server, (struct sockaddr *) &server->dstaddr);
+
+	return server->rdma_ses ? 0 : -ENOENT;
+}
+
 struct cifs_rdma_info* cifs_create_rdma_session(
 	struct TCP_Server_Info *server, struct sockaddr *dstaddr)
 {
diff --git a/fs/cifs/cifsrdma.h b/fs/cifs/cifsrdma.h
index 36f3e4c..c27db6f 100644
--- a/fs/cifs/cifsrdma.h
+++ b/fs/cifs/cifsrdma.h
@@ -212,6 +212,9 @@  struct cifs_rdma_response {
 struct cifs_rdma_info* cifs_create_rdma_session(
 	struct TCP_Server_Info *server, struct sockaddr *dstaddr);
 
+// Reconnect SMBDirect session
+int cifs_reconnect_rdma_session(struct TCP_Server_Info *server);
+
 // SMBDirect interface for carrying upper layer CIFS I/O
 int cifs_rdma_read(
 	struct cifs_rdma_info *rdma, char *buf, unsigned int to_read);