A bunch of updates and cleanup within the transport layer,

particularly with a focus on RDMA.
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1.4.14 (GNU/Linux)
 Comment: GPGTools - http://gpgtools.org
 
 iQIcBAABAgAGBQJTRXV1AAoJEDZk62b0Tg6xYWAP/0Kfp9/8Z05SQ1fnJveySw1O
 nKK//1/GBmquntqFAVHI6yJRLzPJ+Z/Y4u4a0qriwfgpPQvUJQrL77tY/VfEBUPB
 VoJG1tj7lpdLrO3p4YyPkPjymyC7YOoFNjGEstWFg7HetwnnqqZL2LB+5yJzyqjx
 y9nv1HzsrbAE7j8C4hQ1Nmds5muUb5VhnTtPhjrx4tP1sWWh8XTVJbsVDiEqx6cu
 uJXFFTbkONr9jKfv+Ki3H2pZej2yD7w4tU4lkdcGNyij/Q4Xn1iERqroW2/GT6Cl
 AXxlIKN24ASjWo4VqW0Wf8gO8vbUtHRChoiZ69DvzNTkbWmAIWSFHyQJ4cinwuyr
 UbOQZuccO59QtpNVpBvG/vjnbI54rg+VGLy+xE0vcrBDlyoptc56IAFSg8zJY5UN
 ysbyHCGME//9VZ1zeeZvkMjm8z5Enp6x4zmtnUHmufO7DVMTFUePED6U1u9WIyP5
 FFy5EboXMSh97yB8REvbIlY2MgBJWYdnyzLKFMeRpzC8fOXJqBoCX8i3Z5SkMYWJ
 1FS/pGr7ec/VX1iHXSYi9hhzTJ6o9mEmOIhaO4UcqAuK8Rk2jbCp1Lx0iDhmJtdT
 zjofGDe57ro7nOZf8A/TI5z+6BZq8KYVfZrXtYaPqC4rXOzJAox/yHlz9FmAJYgv
 ssOIKXX0ujLwqrMBVatj
 =yzy/
 -----END PGP SIGNATURE-----

Merge tag 'for-linus-3.15' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs

Pull 9p changes from Eric Van Hensbergen:
 "A bunch of updates and cleanup within the transport layer,
  particularly with a focus on RDMA"

* tag 'for-linus-3.15' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  9pnet_rdma: check token type before int conversion
  9pnet: trans_fd : allocate struct p9_trans_fd and struct p9_conn together.
  9pnet: p9_client->conn field is unused. Remove it.
  9P: Get rid of REQ_STATUS_FLSH
  9pnet_rdma: add cancelled()
  9pnet_rdma: update request status during send
  9P: Add cancelled() to the transport functions.
  net: Mark function as static in 9p/client.c
  9P: Add memory barriers to protect request fields over cb/rpc threads handoff

include/net/9p/client.h

@@ -67,7 +67,6 @@ enum p9_trans_status {
  * @REQ_STATUS_ALLOC: request has been allocated but not sent
  * @REQ_STATUS_UNSENT: request waiting to be sent
  * @REQ_STATUS_SENT: request sent to server
- * @REQ_STATUS_FLSH: a flush has been sent for this request
  * @REQ_STATUS_RCVD: response received from server
  * @REQ_STATUS_FLSHD: request has been flushed
  * @REQ_STATUS_ERROR: request encountered an error on the client side
@@ -83,7 +82,6 @@ enum p9_req_status_t {
 	REQ_STATUS_ALLOC,
 	REQ_STATUS_UNSENT,
 	REQ_STATUS_SENT,
-	REQ_STATUS_FLSH,
 	REQ_STATUS_RCVD,
 	REQ_STATUS_FLSHD,
 	REQ_STATUS_ERROR,
@@ -130,7 +128,6 @@ struct p9_req_t {
  * @proto_version: 9P protocol version to use
  * @trans_mod: module API instantiated with this client
  * @trans: tranport instance state and API
- * @conn: connection state information used by trans_fd
  * @fidpool: fid handle accounting for session
  * @fidlist: List of active fid handles
  * @tagpool - transaction id accounting for session
@@ -159,7 +156,6 @@ struct p9_client {
 	struct p9_trans_module *trans_mod;
 	enum p9_trans_status status;
 	void *trans;
-	struct p9_conn *conn;
 
 	struct p9_idpool *fidpool;
 	struct list_head fidlist;
@@ -261,7 +257,7 @@ int p9_client_mkdir_dotl(struct p9_fid *fid, char *name, int mode,
 int p9_client_lock_dotl(struct p9_fid *fid, struct p9_flock *flock, u8 *status);
 int p9_client_getlock_dotl(struct p9_fid *fid, struct p9_getlock *fl);
 struct p9_req_t *p9_tag_lookup(struct p9_client *, u16);
-void p9_client_cb(struct p9_client *c, struct p9_req_t *req);
+void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status);
 
 int p9_parse_header(struct p9_fcall *, int32_t *, int8_t *, int16_t *, int);
 int p9stat_read(struct p9_client *, char *, int, struct p9_wstat *);

include/net/9p/transport.h

@@ -40,6 +40,8 @@
  * @close: member function to discard a connection on this transport
  * @request: member function to issue a request to the transport
  * @cancel: member function to cancel a request (if it hasn't been sent)
+ * @cancelled: member function to notify that a cancelled request will not
+ *             not receive a reply
  *
  * This is the basic API for a transport module which is registered by the
  * transport module with the 9P core network module and used by the client
@@ -58,6 +60,7 @@ struct p9_trans_module {
 	void (*close) (struct p9_client *);
 	int (*request) (struct p9_client *, struct p9_req_t *req);
 	int (*cancel) (struct p9_client *, struct p9_req_t *req);
+	int (*cancelled)(struct p9_client *, struct p9_req_t *req);
 	int (*zc_request)(struct p9_client *, struct p9_req_t *,
 			  char *, char *, int , int, int, int);
 };

net/9p/client.c

@@ -415,9 +415,17 @@ static void p9_free_req(struct p9_client *c, struct p9_req_t *r)
  * req: request received
  *
  */
-void p9_client_cb(struct p9_client *c, struct p9_req_t *req)
+void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status)
 {
 	p9_debug(P9_DEBUG_MUX, " tag %d\n", req->tc->tag);
+
+	/*
+	 * This barrier is needed to make sure any change made to req before
+	 * the other thread wakes up will indeed be seen by the waiting side.
+	 */
+	smp_wmb();
+	req->status = status;
+
 	wake_up(req->wq);
 	p9_debug(P9_DEBUG_MUX, "wakeup: %d\n", req->tc->tag);
 }
@@ -655,16 +663,13 @@ static int p9_client_flush(struct p9_client *c, struct p9_req_t *oldreq)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-
 	/*
 	 * if we haven't received a response for oldreq,
 	 * remove it from the list
 	 */
-	if (oldreq->status == REQ_STATUS_FLSH) {
+	if (oldreq->status == REQ_STATUS_SENT)
-		spin_lock(&c->lock);
+		if (c->trans_mod->cancelled)
-		list_del(&oldreq->req_list);
+			c->trans_mod->cancelled(c, oldreq);
-		spin_unlock(&c->lock);
-	}
 
 	p9_free_req(c, req);
 	return 0;
@@ -751,6 +756,12 @@ again:
 	err = wait_event_interruptible(*req->wq,
 				       req->status >= REQ_STATUS_RCVD);
 
+	/*
+	 * Make sure our req is coherent with regard to updates in other
+	 * threads - echoes to wmb() in the callback
+	 */
+	smp_rmb();
+
 	if ((err == -ERESTARTSYS) && (c->status == Connected)
 				  && (type == P9_TFLUSH)) {
 		sigpending = 1;

net/9p/trans_fd.c

@@ -66,20 +66,6 @@ struct p9_fd_opts {
 	int privport;
 };
 
-/**
- * struct p9_trans_fd - transport state
- * @rd: reference to file to read from
- * @wr: reference of file to write to
- * @conn: connection state reference
- *
- */
-
-struct p9_trans_fd {
-	struct file *rd;
-	struct file *wr;
-	struct p9_conn *conn;
-};
-
 /*
   * Option Parsing (code inspired by NFS code)
   *  - a little lazy - parse all fd-transport options
@@ -159,6 +145,20 @@ struct p9_conn {
 	unsigned long wsched;
 };
 
+/**
+ * struct p9_trans_fd - transport state
+ * @rd: reference to file to read from
+ * @wr: reference of file to write to
+ * @conn: connection state reference
+ *
+ */
+
+struct p9_trans_fd {
+	struct file *rd;
+	struct file *wr;
+	struct p9_conn conn;
+};
+
 static void p9_poll_workfn(struct work_struct *work);
 
 static DEFINE_SPINLOCK(p9_poll_lock);
@@ -212,15 +212,9 @@ static void p9_conn_cancel(struct p9_conn *m, int err)
 	m->err = err;
 
 	list_for_each_entry_safe(req, rtmp, &m->req_list, req_list) {
-		req->status = REQ_STATUS_ERROR;
-		if (!req->t_err)
-			req->t_err = err;
 		list_move(&req->req_list, &cancel_list);
 	}
 	list_for_each_entry_safe(req, rtmp, &m->unsent_req_list, req_list) {
-		req->status = REQ_STATUS_ERROR;
-		if (!req->t_err)
-			req->t_err = err;
 		list_move(&req->req_list, &cancel_list);
 	}
 	spin_unlock_irqrestore(&m->client->lock, flags);
@@ -228,7 +222,9 @@ static void p9_conn_cancel(struct p9_conn *m, int err)
 	list_for_each_entry_safe(req, rtmp, &cancel_list, req_list) {
 		p9_debug(P9_DEBUG_ERROR, "call back req %p\n", req);
 		list_del(&req->req_list);
-		p9_client_cb(m->client, req);
+		if (!req->t_err)
+			req->t_err = err;
+		p9_client_cb(m->client, req, REQ_STATUS_ERROR);
 	}
 }
 
@@ -302,6 +298,7 @@ static void p9_read_work(struct work_struct *work)
 {
 	int n, err;
 	struct p9_conn *m;
+	int status = REQ_STATUS_ERROR;
 
 	m = container_of(work, struct p9_conn, rq);
 
@@ -348,8 +345,7 @@ static void p9_read_work(struct work_struct *work)
 			 "mux %p pkt: size: %d bytes tag: %d\n", m, n, tag);
 
 		m->req = p9_tag_lookup(m->client, tag);
-		if (!m->req || (m->req->status != REQ_STATUS_SENT &&
+		if (!m->req || (m->req->status != REQ_STATUS_SENT)) {
-					m->req->status != REQ_STATUS_FLSH)) {
 			p9_debug(P9_DEBUG_ERROR, "Unexpected packet tag %d\n",
 				 tag);
 			err = -EIO;
@@ -375,10 +371,10 @@ static void p9_read_work(struct work_struct *work)
 		p9_debug(P9_DEBUG_TRANS, "got new packet\n");
 		spin_lock(&m->client->lock);
 		if (m->req->status != REQ_STATUS_ERROR)
-			m->req->status = REQ_STATUS_RCVD;
+			status = REQ_STATUS_RCVD;
 		list_del(&m->req->req_list);
 		spin_unlock(&m->client->lock);
-		p9_client_cb(m->client, m->req);
+		p9_client_cb(m->client, m->req, status);
 		m->rbuf = NULL;
 		m->rpos = 0;
 		m->rsize = 0;
@@ -573,21 +569,19 @@ p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p)
 }
 
 /**
- * p9_conn_create - allocate and initialize the per-session mux data
+ * p9_conn_create - initialize the per-session mux data
  * @client: client instance
  *
  * Note: Creates the polling task if this is the first session.
  */
 
-static struct p9_conn *p9_conn_create(struct p9_client *client)
+static void p9_conn_create(struct p9_client *client)
 {
 	int n;
-	struct p9_conn *m;
+	struct p9_trans_fd *ts = client->trans;
+	struct p9_conn *m = &ts->conn;
 
 	p9_debug(P9_DEBUG_TRANS, "client %p msize %d\n", client, client->msize);
-	m = kzalloc(sizeof(struct p9_conn), GFP_KERNEL);
-	if (!m)
-		return ERR_PTR(-ENOMEM);
 
 	INIT_LIST_HEAD(&m->mux_list);
 	m->client = client;
@@ -609,8 +603,6 @@ static struct p9_conn *p9_conn_create(struct p9_client *client)
 		p9_debug(P9_DEBUG_TRANS, "mux %p can write\n", m);
 		set_bit(Wpending, &m->wsched);
 	}
-
-	return m;
 }
 
 /**
@@ -669,7 +661,7 @@ static int p9_fd_request(struct p9_client *client, struct p9_req_t *req)
 {
 	int n;
 	struct p9_trans_fd *ts = client->trans;
-	struct p9_conn *m = ts->conn;
+	struct p9_conn *m = &ts->conn;
 
 	p9_debug(P9_DEBUG_TRANS, "mux %p task %p tcall %p id %d\n",
 		 m, current, req->tc, req->tc->id);
@@ -704,14 +696,26 @@ static int p9_fd_cancel(struct p9_client *client, struct p9_req_t *req)
 		list_del(&req->req_list);
 		req->status = REQ_STATUS_FLSHD;
 		ret = 0;
-	} else if (req->status == REQ_STATUS_SENT)
+	}
-		req->status = REQ_STATUS_FLSH;
-
 	spin_unlock(&client->lock);
 
 	return ret;
 }
 
+static int p9_fd_cancelled(struct p9_client *client, struct p9_req_t *req)
+{
+	p9_debug(P9_DEBUG_TRANS, "client %p req %p\n", client, req);
+
+	/* we haven't received a response for oldreq,
+	 * remove it from the list.
+	 */
+	spin_lock(&client->lock);
+	list_del(&req->req_list);
+	spin_unlock(&client->lock);
+
+	return 0;
+}
+
 /**
  * parse_opts - parse mount options into p9_fd_opts structure
  * @params: options string passed from mount
@@ -780,7 +784,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 
 static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 {
-	struct p9_trans_fd *ts = kmalloc(sizeof(struct p9_trans_fd),
+	struct p9_trans_fd *ts = kzalloc(sizeof(struct p9_trans_fd),
 					   GFP_KERNEL);
 	if (!ts)
 		return -ENOMEM;
@@ -806,9 +810,8 @@ static int p9_socket_open(struct p9_client *client, struct socket *csocket)
 {
 	struct p9_trans_fd *p;
 	struct file *file;
-	int ret;
 
-	p = kmalloc(sizeof(struct p9_trans_fd), GFP_KERNEL);
+	p = kzalloc(sizeof(struct p9_trans_fd), GFP_KERNEL);
 	if (!p)
 		return -ENOMEM;
 
@@ -829,20 +832,12 @@ static int p9_socket_open(struct p9_client *client, struct socket *csocket)
 
 	p->rd->f_flags |= O_NONBLOCK;
 
-	p->conn = p9_conn_create(client);
+	p9_conn_create(client);
-	if (IS_ERR(p->conn)) {
-		ret = PTR_ERR(p->conn);
-		p->conn = NULL;
-		kfree(p);
-		sockfd_put(csocket);
-		sockfd_put(csocket);
-		return ret;
-	}
 	return 0;
 }
 
 /**
- * p9_mux_destroy - cancels all pending requests and frees mux resources
+ * p9_mux_destroy - cancels all pending requests of mux
  * @m: mux to destroy
  *
  */
@@ -859,7 +854,6 @@ static void p9_conn_destroy(struct p9_conn *m)
 	p9_conn_cancel(m, -ECONNRESET);
 
 	m->client = NULL;
-	kfree(m);
 }
 
 /**
@@ -881,7 +875,7 @@ static void p9_fd_close(struct p9_client *client)
 
 	client->status = Disconnected;
 
-	p9_conn_destroy(ts->conn);
+	p9_conn_destroy(&ts->conn);
 
 	if (ts->rd)
 		fput(ts->rd);
@@ -1033,14 +1027,7 @@ p9_fd_create(struct p9_client *client, const char *addr, char *args)
 		return err;
 
 	p = (struct p9_trans_fd *) client->trans;
-	p->conn = p9_conn_create(client);
+	p9_conn_create(client);
-	if (IS_ERR(p->conn)) {
-		err = PTR_ERR(p->conn);
-		p->conn = NULL;
-		fput(p->rd);
-		fput(p->wr);
-		return err;
-	}
 
 	return 0;
 }
@@ -1053,6 +1040,7 @@ static struct p9_trans_module p9_tcp_trans = {
 	.close = p9_fd_close,
 	.request = p9_fd_request,
 	.cancel = p9_fd_cancel,
+	.cancelled = p9_fd_cancelled,
 	.owner = THIS_MODULE,
 };
 
@@ -1064,6 +1052,7 @@ static struct p9_trans_module p9_unix_trans = {
 	.close = p9_fd_close,
 	.request = p9_fd_request,
 	.cancel = p9_fd_cancel,
+	.cancelled = p9_fd_cancelled,
 	.owner = THIS_MODULE,
 };
 
@@ -1075,6 +1064,7 @@ static struct p9_trans_module p9_fd_trans = {
 	.close = p9_fd_close,
 	.request = p9_fd_request,
 	.cancel = p9_fd_cancel,
+	.cancelled = p9_fd_cancelled,
 	.owner = THIS_MODULE,
 };
 

net/9p/trans_rdma.c

@@ -193,6 +193,8 @@ static int parse_opts(char *params, struct p9_rdma_opts *opts)
 		if (!*p)
 			continue;
 		token = match_token(p, tokens, args);
+		if (token == Opt_err)
+			continue;
 		r = match_int(&args[0], &option);
 		if (r < 0) {
 			p9_debug(P9_DEBUG_ERROR,
@@ -305,8 +307,7 @@ handle_recv(struct p9_client *client, struct p9_trans_rdma *rdma,
 	}
 
 	req->rc = c->rc;
-	req->status = REQ_STATUS_RCVD;
+	p9_client_cb(client, req, REQ_STATUS_RCVD);
-	p9_client_cb(client, req);
 
 	return;
 
@@ -511,6 +512,11 @@ dont_need_post_recv:
 		goto send_error;
 	}
 
+	/* Mark request as `sent' *before* we actually send it,
+	 * because doing if after could erase the REQ_STATUS_RCVD
+	 * status in case of a very fast reply.
+	 */
+	req->status = REQ_STATUS_SENT;
 	err = ib_post_send(rdma->qp, &wr, &bad_wr);
 	if (err)
 		goto send_error;
@@ -520,6 +526,7 @@ dont_need_post_recv:
 
  /* Handle errors that happened during or while preparing the send: */
  send_error:
+	req->status = REQ_STATUS_ERROR;
 	kfree(c);
 	p9_debug(P9_DEBUG_ERROR, "Error %d in rdma_request()\n", err);
 
@@ -582,12 +589,24 @@ static struct p9_trans_rdma *alloc_rdma(struct p9_rdma_opts *opts)
 	return rdma;
 }
 
-/* its not clear to me we can do anything after send has been posted */
 static int rdma_cancel(struct p9_client *client, struct p9_req_t *req)
 {
+	/* Nothing to do here.
+	 * We will take care of it (if we have to) in rdma_cancelled()
+	 */
 	return 1;
 }
 
+/* A request has been fully flushed without a reply.
+ * That means we have posted one buffer in excess.
+ */
+static int rdma_cancelled(struct p9_client *client, struct p9_req_t *req)
+{
+	struct p9_trans_rdma *rdma = client->trans;
+	atomic_inc(&rdma->excess_rc);
+	return 0;
+}
+
 /**
  * trans_create_rdma - Transport method for creating atransport instance
  * @client: client instance
@@ -721,6 +740,7 @@ static struct p9_trans_module p9_rdma_trans = {
 	.close = rdma_close,
 	.request = rdma_request,
 	.cancel = rdma_cancel,
+	.cancelled = rdma_cancelled,
 };
 
 /**

net/9p/trans_virtio.c

@@ -164,8 +164,7 @@ static void req_done(struct virtqueue *vq)
 		p9_debug(P9_DEBUG_TRANS, ": rc %p\n", rc);
 		p9_debug(P9_DEBUG_TRANS, ": lookup tag %d\n", rc->tag);
 		req = p9_tag_lookup(chan->client, rc->tag);
-		req->status = REQ_STATUS_RCVD;
+		p9_client_cb(chan->client, req, REQ_STATUS_RCVD);
-		p9_client_cb(chan->client, req);
 	}
 }