5 files changed, 268 insertions, 161 deletions

diff --git a/net/9p/client.c b/net/9p/client.c
index addc116cecf..0004cbaac4a 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -127,7 +127,7 @@ static int parse_opts(char *opts, struct p9_client *clnt)
 	char *s;
 	int ret = 0;
 
-	clnt->proto_version = p9_proto_2000u;
+	clnt->proto_version = p9_proto_2000L;
 	clnt->msize = 8192;
 
 	if (!opts)
@@ -204,6 +204,17 @@ free_and_return:
 	return ret;
 }
 
+static struct p9_fcall *p9_fcall_alloc(int alloc_msize)
+{
+	struct p9_fcall *fc;
+	fc = kmalloc(sizeof(struct p9_fcall) + alloc_msize, GFP_NOFS);
+	if (!fc)
+		return NULL;
+	fc->capacity = alloc_msize;
+	fc->sdata = (char *) fc + sizeof(struct p9_fcall);
+	return fc;
+}
+
 /**
  * p9_tag_alloc - lookup/allocate a request by tag
  * @c: client session to lookup tag within
@@ -256,39 +267,36 @@ p9_tag_alloc(struct p9_client *c, u16 tag, unsigned int max_size)
 	col = tag % P9_ROW_MAXTAG;
 
 	req = &c->reqs[row][col];
-	if (!req->tc) {
+	if (!req->wq) {
 		req->wq = kmalloc(sizeof(wait_queue_head_t), GFP_NOFS);
-		if (!req->wq) {
-			pr_err("Couldn't grow tag array\n");
-			return ERR_PTR(-ENOMEM);
-		}
+		if (!req->wq)
+			goto grow_failed;
 		init_waitqueue_head(req->wq);
-		req->tc = kmalloc(sizeof(struct p9_fcall) + alloc_msize,
-				  GFP_NOFS);
-		req->rc = kmalloc(sizeof(struct p9_fcall) + alloc_msize,
-				  GFP_NOFS);
-		if ((!req->tc) || (!req->rc)) {
-			pr_err("Couldn't grow tag array\n");
-			kfree(req->tc);
-			kfree(req->rc);
-			kfree(req->wq);
-			req->tc = req->rc = NULL;
-			req->wq = NULL;
-			return ERR_PTR(-ENOMEM);
-		}
-		req->tc->capacity = alloc_msize;
-		req->rc->capacity = alloc_msize;
-		req->tc->sdata = (char *) req->tc + sizeof(struct p9_fcall);
-		req->rc->sdata = (char *) req->rc + sizeof(struct p9_fcall);
 	}
 
+	if (!req->tc)
+		req->tc = p9_fcall_alloc(alloc_msize);
+	if (!req->rc)
+		req->rc = p9_fcall_alloc(alloc_msize);
+	if (!req->tc || !req->rc)
+		goto grow_failed;
+
 	p9pdu_reset(req->tc);
 	p9pdu_reset(req->rc);
 
 	req->tc->tag = tag-1;
 	req->status = REQ_STATUS_ALLOC;
 
-	return &c->reqs[row][col];
+	return req;
+
+grow_failed:
+	pr_err("Couldn't grow tag array\n");
+	kfree(req->tc);
+	kfree(req->rc);
+	kfree(req->wq);
+	req->tc = req->rc = NULL;
+	req->wq = NULL;
+	return ERR_PTR(-ENOMEM);
 }
 
 /**
@@ -407,9 +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);
 }
@@ -647,13 +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. */
-	spin_lock(&c->lock);
-	if (oldreq->status == REQ_STATUS_FLSH)
-		list_del(&oldreq->req_list);
-	spin_unlock(&c->lock);
+	/*
+	 * if we haven't received a response for oldreq,
+	 * remove it from the list
+	 */
+	if (oldreq->status == REQ_STATUS_SENT)
+		if (c->trans_mod->cancelled)
+			c->trans_mod->cancelled(c, oldreq);
 
 	p9_free_req(c, req);
 	return 0;
@@ -740,6 +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;
@@ -976,6 +998,7 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 {
 	int err;
 	struct p9_client *clnt;
+	char *client_id;
 
 	err = 0;
 	clnt = kmalloc(sizeof(struct p9_client), GFP_KERNEL);
@@ -984,6 +1007,10 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 
 	clnt->trans_mod = NULL;
 	clnt->trans = NULL;
+
+	client_id = utsname()->nodename;
+	memcpy(clnt->name, client_id, strlen(client_id) + 1);
+
 	spin_lock_init(&clnt->lock);
 	INIT_LIST_HEAD(&clnt->fidlist);
 
diff --git a/net/9p/trans_common.c b/net/9p/trans_common.c
index de8df957867..2ee3879161b 100644
--- a/net/9p/trans_common.c
+++ b/net/9p/trans_common.c
@@ -24,11 +24,11 @@
  */
 void p9_release_pages(struct page **pages, int nr_pages)
 {
-	int i = 0;
-	while (pages[i] && nr_pages--) {
-		put_page(pages[i]);
-		i++;
-	}
+	int i;
+
+	for (i = 0; i < nr_pages; i++)
+		if (pages[i])
+			put_page(pages[i]);
 }
 EXPORT_SYMBOL(p9_release_pages);
 
diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c
index 02efb25c295..80d08f6664c 100644
--- a/net/9p/trans_fd.c
+++ b/net/9p/trans_fd.c
@@ -63,20 +63,7 @@ struct p9_fd_opts {
 	int rfd;
 	int wfd;
 	u16 port;
-};
-
-/**
- * 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;
+	int privport;
 };
 
 /*
@@ -87,12 +74,15 @@ struct p9_trans_fd {
 enum {
 	/* Options that take integer arguments */
 	Opt_port, Opt_rfdno, Opt_wfdno, Opt_err,
+	/* Options that take no arguments */
+	Opt_privport,
 };
 
 static const match_table_t tokens = {
 	{Opt_port, "port=%u"},
 	{Opt_rfdno, "rfdno=%u"},
 	{Opt_wfdno, "wfdno=%u"},
+	{Opt_privport, "privport"},
 	{Opt_err, NULL},
 };
 
@@ -155,12 +145,29 @@ 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);
 static LIST_HEAD(p9_poll_pending_list);
 static DECLARE_WORK(p9_poll_work, p9_poll_workfn);
 
+static unsigned int p9_ipport_resv_min = P9_DEF_MIN_RESVPORT;
+static unsigned int p9_ipport_resv_max = P9_DEF_MAX_RESVPORT;
+
 static void p9_mux_poll_stop(struct p9_conn *m)
 {
 	unsigned long flags;
@@ -205,15 +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);
@@ -221,7 +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);
 	}
 }
 
@@ -237,10 +240,10 @@ p9_fd_poll(struct p9_client *client, struct poll_table_struct *pt)
 	if (!ts)
 		return -EREMOTEIO;
 
-	if (!ts->rd->f_op || !ts->rd->f_op->poll)
+	if (!ts->rd->f_op->poll)
 		return -EIO;
 
-	if (!ts->wr->f_op || !ts->wr->f_op->poll)
+	if (!ts->wr->f_op->poll)
 		return -EIO;
 
 	ret = ts->rd->f_op->poll(ts->rd, pt);
@@ -295,6 +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);
 
@@ -341,8 +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 &&
-					m->req->status != REQ_STATUS_FLSH)) {
+		if (!m->req || (m->req->status != REQ_STATUS_SENT)) {
 			p9_debug(P9_DEBUG_ERROR, "Unexpected packet tag %d\n",
 				 tag);
 			err = -EIO;
@@ -368,10 +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;
@@ -566,21 +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;
@@ -602,8 +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;
 }
 
 /**
@@ -662,7 +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);
@@ -697,14 +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
@@ -741,7 +752,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 		if (!*p)
 			continue;
 		token = match_token(p, tokens, args);
-		if (token != Opt_err) {
+		if ((token != Opt_err) && (token != Opt_privport)) {
 			r = match_int(&args[0], &option);
 			if (r < 0) {
 				p9_debug(P9_DEBUG_ERROR,
@@ -759,6 +770,9 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 		case Opt_wfdno:
 			opts->wfd = option;
 			break;
+		case Opt_privport:
+			opts->privport = 1;
+			break;
 		default:
 			continue;
 		}
@@ -770,7 +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;
@@ -796,9 +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;
 
@@ -819,20 +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);
-	if (IS_ERR(p->conn)) {
-		ret = PTR_ERR(p->conn);
-		p->conn = NULL;
-		kfree(p);
-		sockfd_put(csocket);
-		sockfd_put(csocket);
-		return ret;
-	}
+	p9_conn_create(client);
 	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
  *
  */
@@ -849,7 +854,6 @@ static void p9_conn_destroy(struct p9_conn *m)
 	p9_conn_cancel(m, -ECONNRESET);
 
 	m->client = NULL;
-	kfree(m);
 }
 
 /**
@@ -871,7 +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);
@@ -898,6 +902,24 @@ static inline int valid_ipaddr4(const char *buf)
 	return 0;
 }
 
+static int p9_bind_privport(struct socket *sock)
+{
+	struct sockaddr_in cl;
+	int port, err = -EINVAL;
+
+	memset(&cl, 0, sizeof(cl));
+	cl.sin_family = AF_INET;
+	cl.sin_addr.s_addr = INADDR_ANY;
+	for (port = p9_ipport_resv_max; port >= p9_ipport_resv_min; port--) {
+		cl.sin_port = htons((ushort)port);
+		err = kernel_bind(sock, (struct sockaddr *)&cl, sizeof(cl));
+		if (err != -EADDRINUSE)
+			break;
+	}
+	return err;
+}
+
+
 static int
 p9_fd_create_tcp(struct p9_client *client, const char *addr, char *args)
 {
@@ -926,6 +948,16 @@ p9_fd_create_tcp(struct p9_client *client, const char *addr, char *args)
 		return err;
 	}
 
+	if (opts.privport) {
+		err = p9_bind_privport(csocket);
+		if (err < 0) {
+			pr_err("%s (%d): problem binding to privport\n",
+			       __func__, task_pid_nr(current));
+			sock_release(csocket);
+			return err;
+		}
+	}
+
 	err = csocket->ops->connect(csocket,
 				    (struct sockaddr *)&sin_server,
 				    sizeof(struct sockaddr_in), 0);
@@ -995,14 +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);
-	if (IS_ERR(p->conn)) {
-		err = PTR_ERR(p->conn);
-		p->conn = NULL;
-		fput(p->rd);
-		fput(p->wr);
-		return err;
-	}
+	p9_conn_create(client);
 
 	return 0;
 }
@@ -1010,11 +1035,12 @@ p9_fd_create(struct p9_client *client, const char *addr, char *args)
 static struct p9_trans_module p9_tcp_trans = {
 	.name = "tcp",
 	.maxsize = MAX_SOCK_BUF,
-	.def = 1,
+	.def = 0,
 	.create = p9_fd_create_tcp,
 	.close = p9_fd_close,
 	.request = p9_fd_request,
 	.cancel = p9_fd_cancel,
+	.cancelled = p9_fd_cancelled,
 	.owner = THIS_MODULE,
 };
 
@@ -1026,6 +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,
 };
 
@@ -1037,6 +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,
 };
 
diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
index 2c69ddd691a..14ad43b5cf8 100644
--- a/net/9p/trans_rdma.c
+++ b/net/9p/trans_rdma.c
@@ -57,9 +57,7 @@
 #define P9_RDMA_IRD		0
 #define P9_RDMA_ORD		0
 #define P9_RDMA_TIMEOUT		30000		/* 30 seconds */
-#define P9_RDMA_MAXSIZE		(4*4096)	/* Min SGE is 4, so we can
-						 * safely advertise a maxsize
-						 * of 64k */
+#define P9_RDMA_MAXSIZE		(1024*1024)	/* 1MB */
 
 /**
  * struct p9_trans_rdma - RDMA transport instance
@@ -75,7 +73,9 @@
  * @sq_depth: The depth of the Send Queue
  * @sq_sem: Semaphore for the SQ
  * @rq_depth: The depth of the Receive Queue.
- * @rq_count: Count of requests in the Receive Queue.
+ * @rq_sem: Semaphore for the RQ
+ * @excess_rc : Amount of posted Receive Contexts without a pending request.
+ *		See rdma_request()
  * @addr: The remote peer's address
  * @req_lock: Protects the active request list
  * @cm_done: Completion event for connection management tracking
@@ -100,7 +100,8 @@ struct p9_trans_rdma {
 	int sq_depth;
 	struct semaphore sq_sem;
 	int rq_depth;
-	atomic_t rq_count;
+	struct semaphore rq_sem;
+	atomic_t excess_rc;
 	struct sockaddr_in addr;
 	spinlock_t req_lock;
 
@@ -192,6 +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,
@@ -296,9 +299,15 @@ handle_recv(struct p9_client *client, struct p9_trans_rdma *rdma,
 	if (!req)
 		goto err_out;
 
+	/* Check that we have not yet received a reply for this request.
+	 */
+	if (unlikely(req->rc)) {
+		pr_err("Duplicate reply for request %d", tag);
+		goto err_out;
+	}
+
 	req->rc = c->rc;
-	req->status = REQ_STATUS_RCVD;
-	p9_client_cb(client, req);
+	p9_client_cb(client, req, REQ_STATUS_RCVD);
 
 	return;
 
@@ -336,8 +345,8 @@ static void cq_comp_handler(struct ib_cq *cq, void *cq_context)
 
 		switch (c->wc_op) {
 		case IB_WC_RECV:
-			atomic_dec(&rdma->rq_count);
 			handle_recv(client, rdma, c, wc.status, wc.byte_len);
+			up(&rdma->rq_sem);
 			break;
 
 		case IB_WC_SEND:
@@ -421,32 +430,33 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req)
 	struct p9_rdma_context *c = NULL;
 	struct p9_rdma_context *rpl_context = NULL;
 
+	/* When an error occurs between posting the recv and the send,
+	 * there will be a receive context posted without a pending request.
+	 * Since there is no way to "un-post" it, we remember it and skip
+	 * post_recv() for the next request.
+	 * So here,
+	 * see if we are this `next request' and need to absorb an excess rc.
+	 * If yes, then drop and free our own, and do not recv_post().
+	 **/
+	if (unlikely(atomic_read(&rdma->excess_rc) > 0)) {
+		if ((atomic_sub_return(1, &rdma->excess_rc) >= 0)) {
+			/* Got one ! */
+			kfree(req->rc);
+			req->rc = NULL;
+			goto dont_need_post_recv;
+		} else {
+			/* We raced and lost. */
+			atomic_inc(&rdma->excess_rc);
+		}
+	}
+
 	/* Allocate an fcall for the reply */
 	rpl_context = kmalloc(sizeof *rpl_context, GFP_NOFS);
 	if (!rpl_context) {
 		err = -ENOMEM;
-		goto err_close;
-	}
-
-	/*
-	 * If the request has a buffer, steal it, otherwise
-	 * allocate a new one.  Typically, requests should already
-	 * have receive buffers allocated and just swap them around
-	 */
-	if (!req->rc) {
-		req->rc = kmalloc(sizeof(struct p9_fcall)+client->msize,
-				  GFP_NOFS);
-		if (req->rc) {
-			req->rc->sdata = (char *) req->rc +
-						sizeof(struct p9_fcall);
-			req->rc->capacity = client->msize;
-		}
+		goto recv_error;
 	}
 	rpl_context->rc = req->rc;
-	if (!rpl_context->rc) {
-		err = -ENOMEM;
-		goto err_free2;
-	}
 
 	/*
 	 * Post a receive buffer for this request. We need to ensure
@@ -455,29 +465,35 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req)
 	 * outstanding request, so we must keep a count to avoid
 	 * overflowing the RQ.
 	 */
-	if (atomic_inc_return(&rdma->rq_count) <= rdma->rq_depth) {
-		err = post_recv(client, rpl_context);
-		if (err)
-			goto err_free1;
-	} else
-		atomic_dec(&rdma->rq_count);
+	if (down_interruptible(&rdma->rq_sem)) {
+		err = -EINTR;
+		goto recv_error;
+	}
 
+	err = post_recv(client, rpl_context);
+	if (err) {
+		p9_debug(P9_DEBUG_FCALL, "POST RECV failed\n");
+		goto recv_error;
+	}
 	/* remove posted receive buffer from request structure */
 	req->rc = NULL;
 
+dont_need_post_recv:
 	/* Post the request */
 	c = kmalloc(sizeof *c, GFP_NOFS);
 	if (!c) {
 		err = -ENOMEM;
-		goto err_free1;
+		goto send_error;
 	}
 	c->req = req;
 
 	c->busa = ib_dma_map_single(rdma->cm_id->device,
 				    c->req->tc->sdata, c->req->tc->size,
 				    DMA_TO_DEVICE);
-	if (ib_dma_mapping_error(rdma->cm_id->device, c->busa))
-		goto error;
+	if (ib_dma_mapping_error(rdma->cm_id->device, c->busa)) {
+		err = -EIO;
+		goto send_error;
+	}
 
 	sge.addr = c->busa;
 	sge.length = c->req->tc->size;
@@ -491,22 +507,38 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req)
 	wr.sg_list = &sge;
 	wr.num_sge = 1;
 
-	if (down_interruptible(&rdma->sq_sem))
-		goto error;
+	if (down_interruptible(&rdma->sq_sem)) {
+		err = -EINTR;
+		goto send_error;
+	}
 
-	return ib_post_send(rdma->qp, &wr, &bad_wr);
+	/* 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;
 
- error:
+	/* Success */
+	return 0;
+
+ /* Handle errors that happened during or while preparing the send: */
+ send_error:
+	req->status = REQ_STATUS_ERROR;
 	kfree(c);
-	kfree(rpl_context->rc);
-	kfree(rpl_context);
-	p9_debug(P9_DEBUG_ERROR, "EIO\n");
-	return -EIO;
- err_free1:
-	kfree(rpl_context->rc);
- err_free2:
+	p9_debug(P9_DEBUG_ERROR, "Error %d in rdma_request()\n", err);
+
+	/* Ach.
+	 *  We did recv_post(), but not send. We have one recv_post in excess.
+	 */
+	atomic_inc(&rdma->excess_rc);
+	return err;
+
+ /* Handle errors that happened during or while preparing post_recv(): */
+ recv_error:
 	kfree(rpl_context);
- err_close:
 	spin_lock_irqsave(&rdma->req_lock, flags);
 	if (rdma->state < P9_RDMA_CLOSING) {
 		rdma->state = P9_RDMA_CLOSING;
@@ -551,17 +583,30 @@ static struct p9_trans_rdma *alloc_rdma(struct p9_rdma_opts *opts)
 	spin_lock_init(&rdma->req_lock);
 	init_completion(&rdma->cm_done);
 	sema_init(&rdma->sq_sem, rdma->sq_depth);
-	atomic_set(&rdma->rq_count, 0);
+	sema_init(&rdma->rq_sem, rdma->rq_depth);
+	atomic_set(&rdma->excess_rc, 0);
 
 	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
@@ -695,6 +740,7 @@ static struct p9_trans_module p9_rdma_trans = {
 	.close = rdma_close,
 	.request = rdma_request,
 	.cancel = rdma_cancel,
+	.cancelled = rdma_cancelled,
 };
 
 /**
diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index e1c26b10183..6940d8fe897 100644
--- a/net/9p/trans_virtio.c
+++ b/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);
+		p9_client_cb(chan->client, req, REQ_STATUS_RCVD);
 	}
 }
 
@@ -340,7 +339,10 @@ static int p9_get_mapped_pages(struct virtio_chan *chan,
 		int count = nr_pages;
 		while (nr_pages) {
 			s = rest_of_page(data);
-			pages[index++] = kmap_to_page(data);
+			if (is_vmalloc_addr(data))
+				pages[index++] = vmalloc_to_page(data);
+			else
+				pages[index++] = kmap_to_page(data);
 			data += s;
 			nr_pages--;
 		}
@@ -544,9 +546,7 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 
 	chan->inuse = false;
 	if (virtio_has_feature(vdev, VIRTIO_9P_MOUNT_TAG)) {
-		vdev->config->get(vdev,
-				offsetof(struct virtio_9p_config, tag_len),
-				&tag_len, sizeof(tag_len));
+		virtio_cread(vdev, struct virtio_9p_config, tag_len, &tag_len);
 	} else {
 		err = -EINVAL;
 		goto out_free_vq;
@@ -556,8 +556,9 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 		err = -ENOMEM;
 		goto out_free_vq;
 	}
-	vdev->config->get(vdev, offsetof(struct virtio_9p_config, tag),
-			tag, tag_len);
+
+	virtio_cread_bytes(vdev, offsetof(struct virtio_9p_config, tag),
+			   tag, tag_len);
 	chan->tag = tag;
 	chan->tag_len = tag_len;
 	err = sysfs_create_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
@@ -577,6 +578,10 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 	mutex_lock(&virtio_9p_lock);
 	list_add_tail(&chan->chan_list, &virtio_chan_list);
 	mutex_unlock(&virtio_9p_lock);
+
+	/* Let udev rules use the new mount_tag attribute. */
+	kobject_uevent(&(vdev->dev.kobj), KOBJ_CHANGE);
+
 	return 0;
 
 out_free_tag:
@@ -654,6 +659,7 @@ static void p9_virtio_remove(struct virtio_device *vdev)
 	list_del(&chan->chan_list);
 	mutex_unlock(&virtio_9p_lock);
 	sysfs_remove_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
+	kobject_uevent(&(vdev->dev.kobj), KOBJ_CHANGE);
 	kfree(chan->tag);
 	kfree(chan->vc_wq);
 	kfree(chan);
@@ -694,7 +700,7 @@ static struct p9_trans_module p9_virtio_trans = {
 	 * page in zero copy.
 	 */
 	.maxsize = PAGE_SIZE * (VIRTQUEUE_NUM - 3),
-	.def = 0,
+	.def = 1,
 	.owner = THIS_MODULE,
 };