1 files changed, 524 insertions, 217 deletions

diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index de7a9f532ed..6940d8fe897 100644
--- a/net/9p/trans_virtio.c
+++ b/net/9p/trans_virtio.c
@@ -1,12 +1,10 @@
 /*
- * The Guest 9p transport driver
+ * The Virtio 9p transport driver
  *
  * This is a block based transport driver based on the lguest block driver
  * code.
  *
- */
-/*
- *  Copyright (C) 2007 Eric Van Hensbergen, IBM Corporation
+ *  Copyright (C) 2007, 2008 Eric Van Hensbergen, IBM Corporation
  *
  *  Based on virtio console driver
  *  Copyright (C) 2006, 2007 Rusty Russell, IBM Corporation
@@ -28,6 +26,8 @@
  *
  */
 
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
 #include <linux/in.h>
 #include <linux/module.h>
 #include <linux/net.h>
@@ -39,90 +39,68 @@
 #include <linux/inet.h>
 #include <linux/idr.h>
 #include <linux/file.h>
+#include <linux/highmem.h>
+#include <linux/slab.h>
 #include <net/9p/9p.h>
 #include <linux/parser.h>
+#include <net/9p/client.h>
 #include <net/9p/transport.h>
 #include <linux/scatterlist.h>
+#include <linux/swap.h>
 #include <linux/virtio.h>
 #include <linux/virtio_9p.h>
+#include "trans_common.h"
 
 #define VIRTQUEUE_NUM	128
 
 /* a single mutex to manage channel initialization and attachment */
-static DECLARE_MUTEX(virtio_9p_lock);
-/* global which tracks highest initialized channel */
-static int chan_index;
-
-#define P9_INIT_MAXTAG	16
-
-#define REQ_STATUS_IDLE	0
-#define REQ_STATUS_SENT 1
-#define REQ_STATUS_RCVD 2
-#define REQ_STATUS_FLSH 3
-
-struct p9_req_t {
-	int status;
-	wait_queue_head_t *wq;
-};
-
-/* We keep all per-channel information in a structure.
+static DEFINE_MUTEX(virtio_9p_lock);
+static DECLARE_WAIT_QUEUE_HEAD(vp_wq);
+static atomic_t vp_pinned = ATOMIC_INIT(0);
+
+/**
+ * struct virtio_chan - per-instance transport information
+ * @initialized: whether the channel is initialized
+ * @inuse: whether the channel is in use
+ * @lock: protects multiple elements within this structure
+ * @client: client instance
+ * @vdev: virtio dev associated with this channel
+ * @vq: virtio queue associated with this channel
+ * @sg: scatter gather list which is used to pack a request (protected?)
+ *
+ * We keep all per-channel information in a structure.
  * This structure is allocated within the devices dev->mem space.
  * A pointer to the structure will get put in the transport private.
+ *
  */
-static struct virtio_chan {
-	bool initialized;		/* channel is initialized */
-	bool inuse;			/* channel is in use */
+
+struct virtio_chan {
+	bool inuse;
 
 	spinlock_t lock;
 
+	struct p9_client *client;
 	struct virtio_device *vdev;
 	struct virtqueue *vq;
-
-	struct p9_idpool *tagpool;
-	struct p9_req_t *reqs;
-	int max_tag;
-
+	int ring_bufs_avail;
+	wait_queue_head_t *vc_wq;
+	/* This is global limit. Since we don't have a global structure,
+	 * will be placing it in each channel.
+	 */
+	unsigned long p9_max_pages;
 	/* Scatterlist: can be too big for stack. */
 	struct scatterlist sg[VIRTQUEUE_NUM];
-} channels[MAX_9P_CHAN];
 
-/* Lookup requests by tag */
-static struct p9_req_t *p9_lookup_tag(struct virtio_chan *c, u16 tag)
-{
-	/* This looks up the original request by tag so we know which
-	 * buffer to read the data into */
-	tag++;
-
-	while (tag >= c->max_tag) {
-		int old_max = c->max_tag;
-		int count;
-
-		if (c->max_tag)
-			c->max_tag *= 2;
-		else
-			c->max_tag = P9_INIT_MAXTAG;
-
-		c->reqs = krealloc(c->reqs, sizeof(struct p9_req_t)*c->max_tag,
-								GFP_ATOMIC);
-		if (!c->reqs) {
-			printk(KERN_ERR "Couldn't grow tag array\n");
-			BUG();
-		}
-		for (count = old_max; count < c->max_tag; count++) {
-			c->reqs[count].status = REQ_STATUS_IDLE;
-			c->reqs[count].wq = kmalloc(sizeof(wait_queue_head_t),
-								GFP_ATOMIC);
-			if (!c->reqs[count].wq) {
-				printk(KERN_ERR "Couldn't grow tag array\n");
-				BUG();
-			}
-			init_waitqueue_head(c->reqs[count].wq);
-		}
-	}
+	int tag_len;
+	/*
+	 * tag name to identify a mount Non-null terminated
+	 */
+	char *tag;
 
-	return &c->reqs[tag];
-}
+	struct list_head chan_list;
+};
 
+static struct list_head virtio_chan_list;
 
 /* How many bytes left in this page. */
 static unsigned int rest_of_page(void *data)
@@ -130,48 +108,82 @@ static unsigned int rest_of_page(void *data)
 	return PAGE_SIZE - ((unsigned long)data % PAGE_SIZE);
 }
 
-static void p9_virtio_close(struct p9_trans *trans)
-{
-	struct virtio_chan *chan = trans->priv;
-	int count;
-	unsigned int flags;
-
-	spin_lock_irqsave(&chan->lock, flags);
-	p9_idpool_destroy(chan->tagpool);
-	for (count = 0; count < chan->max_tag; count++)
-		kfree(chan->reqs[count].wq);
-	kfree(chan->reqs);
-	chan->max_tag = 0;
-	spin_unlock_irqrestore(&chan->lock, flags);
+/**
+ * p9_virtio_close - reclaim resources of a channel
+ * @client: client instance
+ *
+ * This reclaims a channel by freeing its resources and
+ * reseting its inuse flag.
+ *
+ */
 
-	down(&virtio_9p_lock);
-	chan->inuse = false;
-	up(&virtio_9p_lock);
+static void p9_virtio_close(struct p9_client *client)
+{
+	struct virtio_chan *chan = client->trans;
 
-	kfree(trans);
+	mutex_lock(&virtio_9p_lock);
+	if (chan)
+		chan->inuse = false;
+	mutex_unlock(&virtio_9p_lock);
 }
 
+/**
+ * req_done - callback which signals activity from the server
+ * @vq: virtio queue activity was received on
+ *
+ * This notifies us that the server has triggered some activity
+ * on the virtio channel - most likely a response to request we
+ * sent.  Figure out which requests now have responses and wake up
+ * those threads.
+ *
+ * Bugs: could do with some additional sanity checking, but appears to work.
+ *
+ */
+
 static void req_done(struct virtqueue *vq)
 {
 	struct virtio_chan *chan = vq->vdev->priv;
 	struct p9_fcall *rc;
 	unsigned int len;
-	unsigned long flags;
 	struct p9_req_t *req;
+	unsigned long flags;
 
-	spin_lock_irqsave(&chan->lock, flags);
-	while ((rc = chan->vq->vq_ops->get_buf(chan->vq, &len)) != NULL) {
-		req = p9_lookup_tag(chan, rc->tag);
-		req->status = REQ_STATUS_RCVD;
-		wake_up(req->wq);
+	p9_debug(P9_DEBUG_TRANS, ": request done\n");
+
+	while (1) {
+		spin_lock_irqsave(&chan->lock, flags);
+		rc = virtqueue_get_buf(chan->vq, &len);
+		if (rc == NULL) {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			break;
+		}
+		chan->ring_bufs_avail = 1;
+		spin_unlock_irqrestore(&chan->lock, flags);
+		/* Wakeup if anyone waiting for VirtIO ring space. */
+		wake_up(chan->vc_wq);
+		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);
+		p9_client_cb(chan->client, req, REQ_STATUS_RCVD);
 	}
-	/* In case queue is stopped waiting for more buffers. */
-	spin_unlock_irqrestore(&chan->lock, flags);
 }
 
-static int
-pack_sg_list(struct scatterlist *sg, int start, int limit, char *data,
-								int count)
+/**
+ * pack_sg_list - pack a scatter gather list from a linear buffer
+ * @sg: scatter/gather list to pack into
+ * @start: which segment of the sg_list to start at
+ * @limit: maximum segment to pack data to
+ * @data: data to pack into scatter/gather list
+ * @count: amount of data to pack into the scatter/gather list
+ *
+ * sg_lists have multiple segments of various sizes.  This will pack
+ * arbitrary data into an existing scatter gather list, segmenting the
+ * data as necessary within constraints.
+ *
+ */
+
+static int pack_sg_list(struct scatterlist *sg, int start,
+			int limit, char *data, int count)
 {
 	int s;
 	int index = start;
@@ -180,103 +192,341 @@ pack_sg_list(struct scatterlist *sg, int start, int limit, char *data,
 		s = rest_of_page(data);
 		if (s > count)
 			s = count;
+		BUG_ON(index > limit);
+		/* Make sure we don't terminate early. */
+		sg_unmark_end(&sg[index]);
 		sg_set_buf(&sg[index++], data, s);
 		count -= s;
 		data += s;
-		BUG_ON(index > limit);
 	}
-
+	if (index-start)
+		sg_mark_end(&sg[index - 1]);
 	return index-start;
 }
 
+/* We don't currently allow canceling of virtio requests */
+static int p9_virtio_cancel(struct p9_client *client, struct p9_req_t *req)
+{
+	return 1;
+}
+
+/**
+ * pack_sg_list_p - Just like pack_sg_list. Instead of taking a buffer,
+ * this takes a list of pages.
+ * @sg: scatter/gather list to pack into
+ * @start: which segment of the sg_list to start at
+ * @pdata: a list of pages to add into sg.
+ * @nr_pages: number of pages to pack into the scatter/gather list
+ * @data: data to pack into scatter/gather list
+ * @count: amount of data to pack into the scatter/gather list
+ */
 static int
-p9_virtio_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc)
+pack_sg_list_p(struct scatterlist *sg, int start, int limit,
+	       struct page **pdata, int nr_pages, char *data, int count)
 {
-	int in, out;
-	int n, err, size;
-	struct virtio_chan *chan = t->priv;
-	char *rdata;
-	struct p9_req_t *req;
-	unsigned long flags;
+	int i = 0, s;
+	int data_off;
+	int index = start;
 
-	if (*rc == NULL) {
-		*rc = kmalloc(sizeof(struct p9_fcall) + t->msize, GFP_KERNEL);
-		if (!*rc)
-			return -ENOMEM;
+	BUG_ON(nr_pages > (limit - start));
+	/*
+	 * if the first page doesn't start at
+	 * page boundary find the offset
+	 */
+	data_off = offset_in_page(data);
+	while (nr_pages) {
+		s = rest_of_page(data);
+		if (s > count)
+			s = count;
+		/* Make sure we don't terminate early. */
+		sg_unmark_end(&sg[index]);
+		sg_set_page(&sg[index++], pdata[i++], s, data_off);
+		data_off = 0;
+		data += s;
+		count -= s;
+		nr_pages--;
 	}
 
-	rdata = (char *)*rc+sizeof(struct p9_fcall);
+	if (index-start)
+		sg_mark_end(&sg[index - 1]);
+	return index - start;
+}
 
-	n = P9_NOTAG;
-	if (tc->id != P9_TVERSION) {
-		n = p9_idpool_get(chan->tagpool);
-		if (n < 0)
-			return -ENOMEM;
-	}
+/**
+ * p9_virtio_request - issue a request
+ * @client: client instance issuing the request
+ * @req: request to be issued
+ *
+ */
 
+static int
+p9_virtio_request(struct p9_client *client, struct p9_req_t *req)
+{
+	int err;
+	int in, out, out_sgs, in_sgs;
+	unsigned long flags;
+	struct virtio_chan *chan = client->trans;
+	struct scatterlist *sgs[2];
+
+	p9_debug(P9_DEBUG_TRANS, "9p debug: virtio request\n");
+
+	req->status = REQ_STATUS_SENT;
+req_retry:
 	spin_lock_irqsave(&chan->lock, flags);
-	req = p9_lookup_tag(chan, n);
-	spin_unlock_irqrestore(&chan->lock, flags);
 
-	p9_set_tag(tc, n);
+	out_sgs = in_sgs = 0;
+	/* Handle out VirtIO ring buffers */
+	out = pack_sg_list(chan->sg, 0,
+			   VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
+	if (out)
+		sgs[out_sgs++] = chan->sg;
+
+	in = pack_sg_list(chan->sg, out,
+			  VIRTQUEUE_NUM, req->rc->sdata, req->rc->capacity);
+	if (in)
+		sgs[out_sgs + in_sgs++] = chan->sg + out;
+
+	err = virtqueue_add_sgs(chan->vq, sgs, out_sgs, in_sgs, req->tc,
+				GFP_ATOMIC);
+	if (err < 0) {
+		if (err == -ENOSPC) {
+			chan->ring_bufs_avail = 0;
+			spin_unlock_irqrestore(&chan->lock, flags);
+			err = wait_event_interruptible(*chan->vc_wq,
+							chan->ring_bufs_avail);
+			if (err  == -ERESTARTSYS)
+				return err;
+
+			p9_debug(P9_DEBUG_TRANS, "Retry virtio request\n");
+			goto req_retry;
+		} else {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			p9_debug(P9_DEBUG_TRANS,
+				 "virtio rpc add_sgs returned failure\n");
+			return -EIO;
+		}
+	}
+	virtqueue_kick(chan->vq);
+	spin_unlock_irqrestore(&chan->lock, flags);
 
-	P9_DPRINTK(P9_DEBUG_TRANS, "9p debug: virtio rpc tag %d\n", n);
+	p9_debug(P9_DEBUG_TRANS, "virtio request kicked\n");
+	return 0;
+}
 
-	out = pack_sg_list(chan->sg, 0, VIRTQUEUE_NUM, tc->sdata, tc->size);
-	in = pack_sg_list(chan->sg, out, VIRTQUEUE_NUM-out, rdata, t->msize);
+static int p9_get_mapped_pages(struct virtio_chan *chan,
+			       struct page **pages, char *data,
+			       int nr_pages, int write, int kern_buf)
+{
+	int err;
+	if (!kern_buf) {
+		/*
+		 * We allow only p9_max_pages pinned. We wait for the
+		 * Other zc request to finish here
+		 */
+		if (atomic_read(&vp_pinned) >= chan->p9_max_pages) {
+			err = wait_event_interruptible(vp_wq,
+			      (atomic_read(&vp_pinned) < chan->p9_max_pages));
+			if (err == -ERESTARTSYS)
+				return err;
+		}
+		err = p9_payload_gup(data, &nr_pages, pages, write);
+		if (err < 0)
+			return err;
+		atomic_add(nr_pages, &vp_pinned);
+	} else {
+		/* kernel buffer, no need to pin pages */
+		int s, index = 0;
+		int count = nr_pages;
+		while (nr_pages) {
+			s = rest_of_page(data);
+			if (is_vmalloc_addr(data))
+				pages[index++] = vmalloc_to_page(data);
+			else
+				pages[index++] = kmap_to_page(data);
+			data += s;
+			nr_pages--;
+		}
+		nr_pages = count;
+	}
+	return nr_pages;
+}
 
+/**
+ * p9_virtio_zc_request - issue a zero copy request
+ * @client: client instance issuing the request
+ * @req: request to be issued
+ * @uidata: user bffer that should be ued for zero copy read
+ * @uodata: user buffer that shoud be user for zero copy write
+ * @inlen: read buffer size
+ * @olen: write buffer size
+ * @hdrlen: reader header size, This is the size of response protocol data
+ *
+ */
+static int
+p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req,
+		     char *uidata, char *uodata, int inlen,
+		     int outlen, int in_hdr_len, int kern_buf)
+{
+	int in, out, err, out_sgs, in_sgs;
+	unsigned long flags;
+	int in_nr_pages = 0, out_nr_pages = 0;
+	struct page **in_pages = NULL, **out_pages = NULL;
+	struct virtio_chan *chan = client->trans;
+	struct scatterlist *sgs[4];
+
+	p9_debug(P9_DEBUG_TRANS, "virtio request\n");
+
+	if (uodata) {
+		out_nr_pages = p9_nr_pages(uodata, outlen);
+		out_pages = kmalloc(sizeof(struct page *) * out_nr_pages,
+				    GFP_NOFS);
+		if (!out_pages) {
+			err = -ENOMEM;
+			goto err_out;
+		}
+		out_nr_pages = p9_get_mapped_pages(chan, out_pages, uodata,
+						   out_nr_pages, 0, kern_buf);
+		if (out_nr_pages < 0) {
+			err = out_nr_pages;
+			kfree(out_pages);
+			out_pages = NULL;
+			goto err_out;
+		}
+	}
+	if (uidata) {
+		in_nr_pages = p9_nr_pages(uidata, inlen);
+		in_pages = kmalloc(sizeof(struct page *) * in_nr_pages,
+				   GFP_NOFS);
+		if (!in_pages) {
+			err = -ENOMEM;
+			goto err_out;
+		}
+		in_nr_pages = p9_get_mapped_pages(chan, in_pages, uidata,
+						  in_nr_pages, 1, kern_buf);
+		if (in_nr_pages < 0) {
+			err = in_nr_pages;
+			kfree(in_pages);
+			in_pages = NULL;
+			goto err_out;
+		}
+	}
 	req->status = REQ_STATUS_SENT;
+req_retry_pinned:
+	spin_lock_irqsave(&chan->lock, flags);
 
-	if (chan->vq->vq_ops->add_buf(chan->vq, chan->sg, out, in, tc)) {
-		P9_DPRINTK(P9_DEBUG_TRANS,
-			"9p debug: virtio rpc add_buf returned failure");
-		return -EIO;
-	}
+	out_sgs = in_sgs = 0;
 
-	chan->vq->vq_ops->kick(chan->vq);
+	/* out data */
+	out = pack_sg_list(chan->sg, 0,
+			   VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
 
-	wait_event(*req->wq, req->status == REQ_STATUS_RCVD);
+	if (out)
+		sgs[out_sgs++] = chan->sg;
 
-	size = le32_to_cpu(*(__le32 *) rdata);
+	if (out_pages) {
+		sgs[out_sgs++] = chan->sg + out;
+		out += pack_sg_list_p(chan->sg, out, VIRTQUEUE_NUM,
+				      out_pages, out_nr_pages, uodata, outlen);
+	}
+		
+	/*
+	 * Take care of in data
+	 * For example TREAD have 11.
+	 * 11 is the read/write header = PDU Header(7) + IO Size (4).
+	 * Arrange in such a way that server places header in the
+	 * alloced memory and payload onto the user buffer.
+	 */
+	in = pack_sg_list(chan->sg, out,
+			  VIRTQUEUE_NUM, req->rc->sdata, in_hdr_len);
+	if (in)
+		sgs[out_sgs + in_sgs++] = chan->sg + out;
+
+	if (in_pages) {
+		sgs[out_sgs + in_sgs++] = chan->sg + out + in;
+		in += pack_sg_list_p(chan->sg, out + in, VIRTQUEUE_NUM,
+				     in_pages, in_nr_pages, uidata, inlen);
+	}
 
-	err = p9_deserialize_fcall(rdata, size, *rc, t->extended);
+	BUG_ON(out_sgs + in_sgs > ARRAY_SIZE(sgs));
+	err = virtqueue_add_sgs(chan->vq, sgs, out_sgs, in_sgs, req->tc,
+				GFP_ATOMIC);
 	if (err < 0) {
-		P9_DPRINTK(P9_DEBUG_TRANS,
-			"9p debug: virtio rpc deserialize returned %d\n", err);
-		return err;
+		if (err == -ENOSPC) {
+			chan->ring_bufs_avail = 0;
+			spin_unlock_irqrestore(&chan->lock, flags);
+			err = wait_event_interruptible(*chan->vc_wq,
+						       chan->ring_bufs_avail);
+			if (err  == -ERESTARTSYS)
+				goto err_out;
+
+			p9_debug(P9_DEBUG_TRANS, "Retry virtio request\n");
+			goto req_retry_pinned;
+		} else {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			p9_debug(P9_DEBUG_TRANS,
+				 "virtio rpc add_sgs returned failure\n");
+			err = -EIO;
+			goto err_out;
+		}
 	}
-
-#ifdef CONFIG_NET_9P_DEBUG
-	if ((p9_debug_level&P9_DEBUG_FCALL) == P9_DEBUG_FCALL) {
-		char buf[150];
-
-		p9_printfcall(buf, sizeof(buf), *rc, t->extended);
-		printk(KERN_NOTICE ">>> %p %s\n", t, buf);
+	virtqueue_kick(chan->vq);
+	spin_unlock_irqrestore(&chan->lock, flags);
+	p9_debug(P9_DEBUG_TRANS, "virtio request kicked\n");
+	err = wait_event_interruptible(*req->wq,
+				       req->status >= REQ_STATUS_RCVD);
+	/*
+	 * Non kernel buffers are pinned, unpin them
+	 */
+err_out:
+	if (!kern_buf) {
+		if (in_pages) {
+			p9_release_pages(in_pages, in_nr_pages);
+			atomic_sub(in_nr_pages, &vp_pinned);
+		}
+		if (out_pages) {
+			p9_release_pages(out_pages, out_nr_pages);
+			atomic_sub(out_nr_pages, &vp_pinned);
+		}
+		/* wakeup anybody waiting for slots to pin pages */
+		wake_up(&vp_wq);
 	}
-#endif
+	kfree(in_pages);
+	kfree(out_pages);
+	return err;
+}
 
-	if (n != P9_NOTAG && p9_idpool_check(n, chan->tagpool))
-		p9_idpool_put(n, chan->tagpool);
+static ssize_t p9_mount_tag_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct virtio_chan *chan;
+	struct virtio_device *vdev;
 
-	req->status = REQ_STATUS_IDLE;
+	vdev = dev_to_virtio(dev);
+	chan = vdev->priv;
 
-	return 0;
+	return snprintf(buf, chan->tag_len + 1, "%s", chan->tag);
 }
 
+static DEVICE_ATTR(mount_tag, 0444, p9_mount_tag_show, NULL);
+
+/**
+ * p9_virtio_probe - probe for existence of 9P virtio channels
+ * @vdev: virtio device to probe
+ *
+ * This probes for existing virtio channels.
+ *
+ */
+
 static int p9_virtio_probe(struct virtio_device *vdev)
 {
+	__u16 tag_len;
+	char *tag;
 	int err;
 	struct virtio_chan *chan;
-	int index;
-
-	down(&virtio_9p_lock);
-	index = chan_index++;
-	chan = &channels[index];
-	up(&virtio_9p_lock);
 
-	if (chan_index > MAX_9P_CHAN) {
-		printk(KERN_ERR "9p: virtio: Maximum channels exceeded\n");
-		BUG();
+	chan = kmalloc(sizeof(struct virtio_chan), GFP_KERNEL);
+	if (!chan) {
+		pr_err("Failed to allocate virtio 9P channel\n");
 		err = -ENOMEM;
 		goto fail;
 	}
@@ -284,7 +534,7 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 	chan->vdev = vdev;
 
 	/* We expect one virtqueue, for requests. */
-	chan->vq = vdev->config->find_vq(vdev, 0, req_done);
+	chan->vq = virtio_find_single_vq(vdev, req_done, "requests");
 	if (IS_ERR(chan->vq)) {
 		err = PTR_ERR(chan->vq);
 		goto out_free_vq;
@@ -295,113 +545,169 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 	sg_init_table(chan->sg, VIRTQUEUE_NUM);
 
 	chan->inuse = false;
-	chan->initialized = true;
+	if (virtio_has_feature(vdev, VIRTIO_9P_MOUNT_TAG)) {
+		virtio_cread(vdev, struct virtio_9p_config, tag_len, &tag_len);
+	} else {
+		err = -EINVAL;
+		goto out_free_vq;
+	}
+	tag = kmalloc(tag_len, GFP_KERNEL);
+	if (!tag) {
+		err = -ENOMEM;
+		goto out_free_vq;
+	}
+
+	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);
+	if (err) {
+		goto out_free_tag;
+	}
+	chan->vc_wq = kmalloc(sizeof(wait_queue_head_t), GFP_KERNEL);
+	if (!chan->vc_wq) {
+		err = -ENOMEM;
+		goto out_free_tag;
+	}
+	init_waitqueue_head(chan->vc_wq);
+	chan->ring_bufs_avail = 1;
+	/* Ceiling limit to avoid denial of service attacks */
+	chan->p9_max_pages = nr_free_buffer_pages()/4;
+
+	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:
+	kfree(tag);
 out_free_vq:
-	vdev->config->del_vq(chan->vq);
+	vdev->config->del_vqs(vdev);
+	kfree(chan);
 fail:
-	down(&virtio_9p_lock);
-	chan_index--;
-	up(&virtio_9p_lock);
 	return err;
 }
 
-/* This sets up a transport channel for 9p communication.  Right now
+
+/**
+ * p9_virtio_create - allocate a new virtio channel
+ * @client: client instance invoking this transport
+ * @devname: string identifying the channel to connect to (unused)
+ * @args: args passed from sys_mount() for per-transport options (unused)
+ *
+ * This sets up a transport channel for 9p communication.  Right now
  * we only match the first available channel, but eventually we couldlook up
  * alternate channels by matching devname versus a virtio_config entry.
  * We use a simple reference count mechanism to ensure that only a single
- * mount has a channel open at a time. */
-static struct p9_trans *
-p9_virtio_create(const char *devname, char *args, int msize,
-							unsigned char extended)
+ * mount has a channel open at a time.
+ *
+ */
+
+static int
+p9_virtio_create(struct p9_client *client, const char *devname, char *args)
 {
-	struct p9_trans *trans;
-	struct virtio_chan *chan = channels;
-	int index = 0;
-
-	down(&virtio_9p_lock);
-	while (index < MAX_9P_CHAN) {
-		if (chan->initialized && !chan->inuse) {
-			chan->inuse = true;
-			break;
-		} else {
-			index++;
-			chan = &channels[index];
+	struct virtio_chan *chan;
+	int ret = -ENOENT;
+	int found = 0;
+
+	mutex_lock(&virtio_9p_lock);
+	list_for_each_entry(chan, &virtio_chan_list, chan_list) {
+		if (!strncmp(devname, chan->tag, chan->tag_len) &&
+		    strlen(devname) == chan->tag_len) {
+			if (!chan->inuse) {
+				chan->inuse = true;
+				found = 1;
+				break;
+			}
+			ret = -EBUSY;
 		}
 	}
-	up(&virtio_9p_lock);
+	mutex_unlock(&virtio_9p_lock);
 
-	if (index >= MAX_9P_CHAN) {
-		printk(KERN_ERR "9p: no channels available\n");
-		return ERR_PTR(-ENODEV);
+	if (!found) {
+		pr_err("no channels available\n");
+		return ret;
 	}
 
-	chan->tagpool = p9_idpool_create();
-	if (IS_ERR(chan->tagpool)) {
-		printk(KERN_ERR "9p: couldn't allocate tagpool\n");
-		return ERR_PTR(-ENOMEM);
-	}
-	p9_idpool_get(chan->tagpool); /* reserve tag 0 */
-	chan->max_tag = 0;
-	chan->reqs = NULL;
-
-	trans = kmalloc(sizeof(struct p9_trans), GFP_KERNEL);
-	if (!trans) {
-		printk(KERN_ERR "9p: couldn't allocate transport\n");
-		return ERR_PTR(-ENOMEM);
-	}
-	trans->extended = extended;
-	trans->msize = msize;
-	trans->close = p9_virtio_close;
-	trans->rpc = p9_virtio_rpc;
-	trans->priv = chan;
+	client->trans = (void *)chan;
+	client->status = Connected;
+	chan->client = client;
 
-	return trans;
+	return 0;
 }
 
+/**
+ * p9_virtio_remove - clean up resources associated with a virtio device
+ * @vdev: virtio device to remove
+ *
+ */
+
 static void p9_virtio_remove(struct virtio_device *vdev)
 {
 	struct virtio_chan *chan = vdev->priv;
 
-	BUG_ON(chan->inuse);
+	if (chan->inuse)
+		p9_virtio_close(chan->client);
+	vdev->config->del_vqs(vdev);
 
-	if (chan->initialized) {
-		vdev->config->del_vq(chan->vq);
-		chan->initialized = false;
-	}
-}
+	mutex_lock(&virtio_9p_lock);
+	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);
 
-#define VIRTIO_ID_9P 9
+}
 
 static struct virtio_device_id id_table[] = {
 	{ VIRTIO_ID_9P, VIRTIO_DEV_ANY_ID },
 	{ 0 },
 };
 
+static unsigned int features[] = {
+	VIRTIO_9P_MOUNT_TAG,
+};
+
 /* The standard "struct lguest_driver": */
 static struct virtio_driver p9_virtio_drv = {
-	.driver.name = 	KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
-	.id_table =	id_table,
-	.probe = 	p9_virtio_probe,
-	.remove =	p9_virtio_remove,
+	.feature_table  = features,
+	.feature_table_size = ARRAY_SIZE(features),
+	.driver.name    = KBUILD_MODNAME,
+	.driver.owner	= THIS_MODULE,
+	.id_table	= id_table,
+	.probe		= p9_virtio_probe,
+	.remove		= p9_virtio_remove,
 };
 
 static struct p9_trans_module p9_virtio_trans = {
 	.name = "virtio",
 	.create = p9_virtio_create,
-	.maxsize = PAGE_SIZE*16,
-	.def = 0,
+	.close = p9_virtio_close,
+	.request = p9_virtio_request,
+	.zc_request = p9_virtio_zc_request,
+	.cancel = p9_virtio_cancel,
+	/*
+	 * We leave one entry for input and one entry for response
+	 * headers. We also skip one more entry to accomodate, address
+	 * that are not at page boundary, that can result in an extra
+	 * page in zero copy.
+	 */
+	.maxsize = PAGE_SIZE * (VIRTQUEUE_NUM - 3),
+	.def = 1,
+	.owner = THIS_MODULE,
 };
 
 /* The standard init function */
 static int __init p9_virtio_init(void)
 {
-	int count;
-
-	for (count = 0; count < MAX_9P_CHAN; count++)
-		channels[count].initialized = false;
+	INIT_LIST_HEAD(&virtio_chan_list);
 
 	v9fs_register_trans(&p9_virtio_trans);
 	return register_virtio_driver(&p9_virtio_drv);
@@ -410,6 +716,7 @@ static int __init p9_virtio_init(void)
 static void __exit p9_virtio_cleanup(void)
 {
 	unregister_virtio_driver(&p9_virtio_drv);
+	v9fs_unregister_trans(&p9_virtio_trans);
 }
 
 module_init(p9_virtio_init);