10 files changed, 716 insertions, 8 deletions

diff --git a/drivers/xen/Kconfig b/drivers/xen/Kconfig
index 4b75a16de009..526187c8a12d 100644
--- a/drivers/xen/Kconfig
+++ b/drivers/xen/Kconfig
@@ -17,3 +17,27 @@ config XEN_SCRUB_PAGES
           is not accidentally visible to other domains.  Is it more
           secure, but slightly less efficient.
           If in doubt, say yes.
+
+config XENFS
+        tristate "Xen filesystem"
+        depends on XEN
+        default y
+        help
+          The xen filesystem provides a way for domains to share
+          information with each other and with the hypervisor.
+          For example, by reading and writing the "xenbus" file, guests
+          may pass arbitrary information to the initial domain.
+          If in doubt, say yes.
+
+config XEN_COMPAT_XENFS
+       bool "Create compatibility mount point /proc/xen"
+       depends on XENFS
+       default y
+       help
+         The old xenstore userspace tools expect to find "xenbus"
+         under /proc/xen, but "xenbus" is now found at the root of the
+         xenfs filesystem.  Selecting this causes the kernel to create
+         the compatibilty mount point /proc/xen if it is running on
+         a xen platform.
+         If in doubt, say yes.
+
diff --git a/drivers/xen/Makefile b/drivers/xen/Makefile
index d2a8fdf0e191..ff8accc9e103 100644
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -1,5 +1,7 @@
 obj-y   += grant-table.o features.o events.o manage.o
 obj-y   += xenbus/
+
 obj-$(CONFIG_HOTPLUG_CPU)       += cpu_hotplug.o
 obj-$(CONFIG_XEN_XENCOMM)       += xencomm.o
 obj-$(CONFIG_XEN_BALLOON)       += balloon.o
+obj-$(CONFIG_XENFS)             += xenfs/
\ No newline at end of file
diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
index b2a03184a246..773d1cf23283 100644
--- a/drivers/xen/xenbus/xenbus_probe.c
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -40,6 +40,7 @@
 #include <linux/ctype.h>
 #include <linux/fcntl.h>
 #include <linux/mm.h>
+#include <linux/proc_fs.h>
 #include <linux/notifier.h>
 #include <linux/kthread.h>
 #include <linux/mutex.h>
@@ -55,7 +56,10 @@
 #include "xenbus_comms.h"
 #include "xenbus_probe.h"
 
+
 int xen_store_evtchn;
+EXPORT_SYMBOL(xen_store_evtchn);
+
 struct xenstore_domain_interface *xen_store_interface;
 static unsigned long xen_store_mfn;
 
@@ -166,6 +170,9 @@ static int read_backend_details(struct xenbus_device *xendev)
         return read_otherend_details(xendev, "backend-id", "backend");
 }
 
+static struct device_attribute xenbus_dev_attrs[] = {
+        __ATTR_NULL
+};
 
 /* Bus type for frontend drivers. */
 static struct xen_bus_type xenbus_frontend = {
@@ -174,12 +181,13 @@ static struct xen_bus_type xenbus_frontend = {
         .get_bus_id = frontend_bus_id,
         .probe = xenbus_probe_frontend,
         .bus = {
-                .name     = "xen",
-                .match    = xenbus_match,
-                .uevent   = xenbus_uevent,
-                .probe    = xenbus_dev_probe,
-                .remove   = xenbus_dev_remove,
-                .shutdown = xenbus_dev_shutdown,
+                .name      = "xen",
+                .match     = xenbus_match,
+                .uevent    = xenbus_uevent,
+                .probe     = xenbus_dev_probe,
+                .remove    = xenbus_dev_remove,
+                .shutdown  = xenbus_dev_shutdown,
+                .dev_attrs = xenbus_dev_attrs,
         },
 };
 
@@ -852,6 +860,14 @@ static int __init xenbus_probe_init(void)
         if (!xen_initial_domain())
                 xenbus_probe(NULL);
 
+#ifdef CONFIG_XEN_COMPAT_XENFS
+        /*
+         * Create xenfs mountpoint in /proc for compatibility with
+         * utilities that expect to find "xenbus" under "/proc/xen".
+         */
+        proc_mkdir("xen", NULL);
+#endif
+
         return 0;
 
   out_unreg_back:
diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c
index 7f2f91c0e11d..e325eab4724d 100644
--- a/drivers/xen/xenbus/xenbus_xs.c
+++ b/drivers/xen/xenbus/xenbus_xs.c
@@ -184,6 +184,7 @@ void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
 
         return ret;
 }
+EXPORT_SYMBOL(xenbus_dev_request_and_reply);
 
 /* Send message to xs, get kmalloc'ed reply.  ERR_PTR() on error. */
 static void *xs_talkv(struct xenbus_transaction t,
diff --git a/drivers/xen/xenfs/Makefile b/drivers/xen/xenfs/Makefile
new file mode 100644
index 000000000000..25275c3bbdff
--- /dev/null
+++ b/drivers/xen/xenfs/Makefile
@@ -0,0 +1,3 @@
+obj-$(CONFIG_XENFS) += xenfs.o
+
+xenfs-objs = super.o xenbus.o
\ No newline at end of file
diff --git a/drivers/xen/xenfs/super.c b/drivers/xen/xenfs/super.c
new file mode 100644
index 000000000000..515741a8e6b8
--- /dev/null
+++ b/drivers/xen/xenfs/super.c
@@ -0,0 +1,64 @@
+/*
+ *  xenfs.c - a filesystem for passing info between the a domain and
+ *  the hypervisor.
+ *
+ * 2008-10-07  Alex Zeffertt    Replaced /proc/xen/xenbus with xenfs filesystem
+ *                              and /proc/xen compatibility mount point.
+ *                              Turned xenfs into a loadable module.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/magic.h>
+
+#include "xenfs.h"
+
+#include <asm/xen/hypervisor.h>
+
+MODULE_DESCRIPTION("Xen filesystem");
+MODULE_LICENSE("GPL");
+
+static int xenfs_fill_super(struct super_block *sb, void *data, int silent)
+{
+        static struct tree_descr xenfs_files[] = {
+                [2] = {"xenbus", &xenbus_file_ops, S_IRUSR|S_IWUSR},
+                {""},
+        };
+
+        return simple_fill_super(sb, XENFS_SUPER_MAGIC, xenfs_files);
+}
+
+static int xenfs_get_sb(struct file_system_type *fs_type,
+                        int flags, const char *dev_name,
+                        void *data, struct vfsmount *mnt)
+{
+        return get_sb_single(fs_type, flags, data, xenfs_fill_super, mnt);
+}
+
+static struct file_system_type xenfs_type = {
+        .owner =        THIS_MODULE,
+        .name =         "xenfs",
+        .get_sb =       xenfs_get_sb,
+        .kill_sb =      kill_litter_super,
+};
+
+static int __init xenfs_init(void)
+{
+        if (xen_pv_domain())
+                return register_filesystem(&xenfs_type);
+
+        printk(KERN_INFO "XENFS: not registering filesystem on non-xen platform\n");
+        return 0;
+}
+
+static void __exit xenfs_exit(void)
+{
+        if (xen_pv_domain())
+                unregister_filesystem(&xenfs_type);
+}
+
+module_init(xenfs_init);
+module_exit(xenfs_exit);
+
diff --git a/drivers/xen/xenfs/xenbus.c b/drivers/xen/xenfs/xenbus.c
new file mode 100644
index 000000000000..875a4c59c594
--- /dev/null
+++ b/drivers/xen/xenfs/xenbus.c
@@ -0,0 +1,593 @@
+/*
+ * Driver giving user-space access to the kernel's xenbus connection
+ * to xenstore.
+ *
+ * Copyright (c) 2005, Christian Limpach
+ * Copyright (c) 2005, Rusty Russell, IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Changes:
+ * 2008-10-07  Alex Zeffertt    Replaced /proc/xen/xenbus with xenfs filesystem
+ *                              and /proc/xen compatibility mount point.
+ *                              Turned xenfs into a loadable module.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/uio.h>
+#include <linux/notifier.h>
+#include <linux/wait.h>
+#include <linux/fs.h>
+#include <linux/poll.h>
+#include <linux/mutex.h>
+#include <linux/spinlock.h>
+#include <linux/mount.h>
+#include <linux/pagemap.h>
+#include <linux/uaccess.h>
+#include <linux/init.h>
+#include <linux/namei.h>
+#include <linux/string.h>
+
+#include "xenfs.h"
+#include "../xenbus/xenbus_comms.h"
+
+#include <xen/xenbus.h>
+#include <asm/xen/hypervisor.h>
+
+/*
+ * An element of a list of outstanding transactions, for which we're
+ * still waiting a reply.
+ */
+struct xenbus_transaction_holder {
+        struct list_head list;
+        struct xenbus_transaction handle;
+};
+
+/*
+ * A buffer of data on the queue.
+ */
+struct read_buffer {
+        struct list_head list;
+        unsigned int cons;
+        unsigned int len;
+        char msg[];
+};
+
+struct xenbus_file_priv {
+        /*
+         * msgbuffer_mutex is held while partial requests are built up
+         * and complete requests are acted on.  It therefore protects
+         * the "transactions" and "watches" lists, and the partial
+         * request length and buffer.
+         *
+         * reply_mutex protects the reply being built up to return to
+         * usermode.  It nests inside msgbuffer_mutex but may be held
+         * alone during a watch callback.
+         */
+        struct mutex msgbuffer_mutex;
+
+        /* In-progress transactions */
+        struct list_head transactions;
+
+        /* Active watches. */
+        struct list_head watches;
+
+        /* Partial request. */
+        unsigned int len;
+        union {
+                struct xsd_sockmsg msg;
+                char buffer[PAGE_SIZE];
+        } u;
+
+        /* Response queue. */
+        struct mutex reply_mutex;
+        struct list_head read_buffers;
+        wait_queue_head_t read_waitq;
+
+};
+
+/* Read out any raw xenbus messages queued up. */
+static ssize_t xenbus_file_read(struct file *filp,
+                               char __user *ubuf,
+                               size_t len, loff_t *ppos)
+{
+        struct xenbus_file_priv *u = filp->private_data;
+        struct read_buffer *rb;
+        unsigned i;
+        int ret;
+
+        mutex_lock(&u->reply_mutex);
+        while (list_empty(&u->read_buffers)) {
+                mutex_unlock(&u->reply_mutex);
+                ret = wait_event_interruptible(u->read_waitq,
+                                               !list_empty(&u->read_buffers));
+                if (ret)
+                        return ret;
+                mutex_lock(&u->reply_mutex);
+        }
+
+        rb = list_entry(u->read_buffers.next, struct read_buffer, list);
+        i = 0;
+        while (i < len) {
+                unsigned sz = min((unsigned)len - i, rb->len - rb->cons);
+
+                ret = copy_to_user(ubuf + i, &rb->msg[rb->cons], sz);
+
+                i += sz - ret;
+                rb->cons += sz - ret;
+
+                if (ret != sz) {
+                        if (i == 0)
+                                i = -EFAULT;
+                        goto out;
+                }
+
+                /* Clear out buffer if it has been consumed */
+                if (rb->cons == rb->len) {
+                        list_del(&rb->list);
+                        kfree(rb);
+                        if (list_empty(&u->read_buffers))
+                                break;
+                        rb = list_entry(u->read_buffers.next,
+                                        struct read_buffer, list);
+                }
+        }
+
+out:
+        mutex_unlock(&u->reply_mutex);
+        return i;
+}
+
+/*
+ * Add a buffer to the queue.  Caller must hold the appropriate lock
+ * if the queue is not local.  (Commonly the caller will build up
+ * multiple queued buffers on a temporary local list, and then add it
+ * to the appropriate list under lock once all the buffers have een
+ * successfully allocated.)
+ */
+static int queue_reply(struct list_head *queue, const void *data, size_t len)
+{
+        struct read_buffer *rb;
+
+        if (len == 0)
+                return 0;
+
+        rb = kmalloc(sizeof(*rb) + len, GFP_KERNEL);
+        if (rb == NULL)
+                return -ENOMEM;
+
+        rb->cons = 0;
+        rb->len = len;
+
+        memcpy(rb->msg, data, len);
+
+        list_add_tail(&rb->list, queue);
+        return 0;
+}
+
+/*
+ * Free all the read_buffer s on a list.
+ * Caller must have sole reference to list.
+ */
+static void queue_cleanup(struct list_head *list)
+{
+        struct read_buffer *rb;
+
+        while (!list_empty(list)) {
+                rb = list_entry(list->next, struct read_buffer, list);
+                list_del(list->next);
+                kfree(rb);
+        }
+}
+
+struct watch_adapter {
+        struct list_head list;
+        struct xenbus_watch watch;
+        struct xenbus_file_priv *dev_data;
+        char *token;
+};
+
+static void free_watch_adapter(struct watch_adapter *watch)
+{
+        kfree(watch->watch.node);
+        kfree(watch->token);
+        kfree(watch);
+}
+
+static struct watch_adapter *alloc_watch_adapter(const char *path,
+                                                 const char *token)
+{
+        struct watch_adapter *watch;
+
+        watch = kzalloc(sizeof(*watch), GFP_KERNEL);
+        if (watch == NULL)
+                goto out_fail;
+
+        watch->watch.node = kstrdup(path, GFP_KERNEL);
+        if (watch->watch.node == NULL)
+                goto out_free;
+
+        watch->token = kstrdup(token, GFP_KERNEL);
+        if (watch->token == NULL)
+                goto out_free;
+
+        return watch;
+
+out_free:
+        free_watch_adapter(watch);
+
+out_fail:
+        return NULL;
+}
+
+static void watch_fired(struct xenbus_watch *watch,
+                        const char **vec,
+                        unsigned int len)
+{
+        struct watch_adapter *adap;
+        struct xsd_sockmsg hdr;
+        const char *path, *token;
+        int path_len, tok_len, body_len, data_len = 0;
+        int ret;
+        LIST_HEAD(staging_q);
+
+        adap = container_of(watch, struct watch_adapter, watch);
+
+        path = vec[XS_WATCH_PATH];
+        token = adap->token;
+
+        path_len = strlen(path) + 1;
+        tok_len = strlen(token) + 1;
+        if (len > 2)
+                data_len = vec[len] - vec[2] + 1;
+        body_len = path_len + tok_len + data_len;
+
+        hdr.type = XS_WATCH_EVENT;
+        hdr.len = body_len;
+
+        mutex_lock(&adap->dev_data->reply_mutex);
+
+        ret = queue_reply(&staging_q, &hdr, sizeof(hdr));
+        if (!ret)
+                ret = queue_reply(&staging_q, path, path_len);
+        if (!ret)
+                ret = queue_reply(&staging_q, token, tok_len);
+        if (!ret && len > 2)
+                ret = queue_reply(&staging_q, vec[2], data_len);
+
+        if (!ret) {
+                /* success: pass reply list onto watcher */
+                list_splice_tail(&staging_q, &adap->dev_data->read_buffers);
+                wake_up(&adap->dev_data->read_waitq);
+        } else
+                queue_cleanup(&staging_q);
+
+        mutex_unlock(&adap->dev_data->reply_mutex);
+}
+
+static int xenbus_write_transaction(unsigned msg_type,
+                                    struct xenbus_file_priv *u)
+{
+        int rc, ret;
+        void *reply;
+        struct xenbus_transaction_holder *trans = NULL;
+        LIST_HEAD(staging_q);
+
+        if (msg_type == XS_TRANSACTION_START) {
+                trans = kmalloc(sizeof(*trans), GFP_KERNEL);
+                if (!trans) {
+                        rc = -ENOMEM;
+                        goto out;
+                }
+        }
+
+        reply = xenbus_dev_request_and_reply(&u->u.msg);
+        if (IS_ERR(reply)) {
+                kfree(trans);
+                rc = PTR_ERR(reply);
+                goto out;
+        }
+
+        if (msg_type == XS_TRANSACTION_START) {
+                trans->handle.id = simple_strtoul(reply, NULL, 0);
+
+                list_add(&trans->list, &u->transactions);
+        } else if (msg_type == XS_TRANSACTION_END) {
+                list_for_each_entry(trans, &u->transactions, list)
+                        if (trans->handle.id == u->u.msg.tx_id)
+                                break;
+                BUG_ON(&trans->list == &u->transactions);
+                list_del(&trans->list);
+
+                kfree(trans);
+        }
+
+        mutex_lock(&u->reply_mutex);
+        ret = queue_reply(&staging_q, &u->u.msg, sizeof(u->u.msg));
+        if (!ret)
+                ret = queue_reply(&staging_q, reply, u->u.msg.len);
+        if (!ret) {
+                list_splice_tail(&staging_q, &u->read_buffers);
+                wake_up(&u->read_waitq);
+        } else {
+                queue_cleanup(&staging_q);
+                rc = ret;
+        }
+        mutex_unlock(&u->reply_mutex);
+
+        kfree(reply);
+
+out:
+        return rc;
+}
+
+static int xenbus_write_watch(unsigned msg_type, struct xenbus_file_priv *u)
+{
+        struct watch_adapter *watch, *tmp_watch;
+        char *path, *token;
+        int err, rc;
+        LIST_HEAD(staging_q);
+
+        path = u->u.buffer + sizeof(u->u.msg);
+        token = memchr(path, 0, u->u.msg.len);
+        if (token == NULL) {
+                rc = -EILSEQ;
+                goto out;
+        }
+        token++;
+
+        if (msg_type == XS_WATCH) {
+                watch = alloc_watch_adapter(path, token);
+                if (watch == NULL) {
+                        rc = -ENOMEM;
+                        goto out;
+                }
+
+                watch->watch.callback = watch_fired;
+                watch->dev_data = u;
+
+                err = register_xenbus_watch(&watch->watch);
+                if (err) {
+                        free_watch_adapter(watch);
+                        rc = err;
+                        goto out;
+                }
+                list_add(&watch->list, &u->watches);
+        } else {
+                list_for_each_entry_safe(watch, tmp_watch, &u->watches, list) {
+                        if (!strcmp(watch->token, token) &&
+                            !strcmp(watch->watch.node, path)) {
+                                unregister_xenbus_watch(&watch->watch);
+                                list_del(&watch->list);
+                                free_watch_adapter(watch);
+                                break;
+                        }
+                }
+        }
+
+        /* Success.  Synthesize a reply to say all is OK. */
+        {
+                struct {
+                        struct xsd_sockmsg hdr;
+                        char body[3];
+                } __packed reply = {
+                        {
+                                .type = msg_type,
+                                .len = sizeof(reply.body)
+                        },
+                        "OK"
+                };
+
+                mutex_lock(&u->reply_mutex);
+                rc = queue_reply(&u->read_buffers, &reply, sizeof(reply));
+                mutex_unlock(&u->reply_mutex);
+        }
+
+out:
+        return rc;
+}
+
+static ssize_t xenbus_file_write(struct file *filp,
+                                const char __user *ubuf,
+                                size_t len, loff_t *ppos)
+{
+        struct xenbus_file_priv *u = filp->private_data;
+        uint32_t msg_type;
+        int rc = len;
+        int ret;
+        LIST_HEAD(staging_q);
+
+        /*
+         * We're expecting usermode to be writing properly formed
+         * xenbus messages.  If they write an incomplete message we
+         * buffer it up.  Once it is complete, we act on it.
+         */
+
+        /*
+         * Make sure concurrent writers can't stomp all over each
+         * other's messages and make a mess of our partial message
+         * buffer.  We don't make any attemppt to stop multiple
+         * writers from making a mess of each other's incomplete
+         * messages; we're just trying to guarantee our own internal
+         * consistency and make sure that single writes are handled
+         * atomically.
+         */
+        mutex_lock(&u->msgbuffer_mutex);
+
+        /* Get this out of the way early to avoid confusion */
+        if (len == 0)
+                goto out;
+
+        /* Can't write a xenbus message larger we can buffer */
+        if ((len + u->len) > sizeof(u->u.buffer)) {
+                /* On error, dump existing buffer */
+                u->len = 0;
+                rc = -EINVAL;
+                goto out;
+        }
+
+        ret = copy_from_user(u->u.buffer + u->len, ubuf, len);
+
+        if (ret == len) {
+                rc = -EFAULT;
+                goto out;
+        }
+
+        /* Deal with a partial copy. */
+        len -= ret;
+        rc = len;
+
+        u->len += len;
+
+        /* Return if we haven't got a full message yet */
+        if (u->len < sizeof(u->u.msg))
+                goto out;       /* not even the header yet */
+
+        /* If we're expecting a message that's larger than we can
+           possibly send, dump what we have and return an error. */
+        if ((sizeof(u->u.msg) + u->u.msg.len) > sizeof(u->u.buffer)) {
+                rc = -E2BIG;
+                u->len = 0;
+                goto out;
+        }
+
+        if (u->len < (sizeof(u->u.msg) + u->u.msg.len))
+                goto out;       /* incomplete data portion */
+
+        /*
+         * OK, now we have a complete message.  Do something with it.
+         */
+
+        msg_type = u->u.msg.type;
+
+        switch (msg_type) {
+        case XS_TRANSACTION_START:
+        case XS_TRANSACTION_END:
+        case XS_DIRECTORY:
+        case XS_READ:
+        case XS_GET_PERMS:
+        case XS_RELEASE:
+        case XS_GET_DOMAIN_PATH:
+        case XS_WRITE:
+        case XS_MKDIR:
+        case XS_RM:
+        case XS_SET_PERMS:
+                /* Send out a transaction */
+                ret = xenbus_write_transaction(msg_type, u);
+                break;
+
+        case XS_WATCH:
+        case XS_UNWATCH:
+                /* (Un)Ask for some path to be watched for changes */
+                ret = xenbus_write_watch(msg_type, u);
+                break;
+
+        default:
+                ret = -EINVAL;
+                break;
+        }
+        if (ret != 0)
+                rc = ret;
+
+        /* Buffered message consumed */
+        u->len = 0;
+
+ out:
+        mutex_unlock(&u->msgbuffer_mutex);
+        return rc;
+}
+
+static int xenbus_file_open(struct inode *inode, struct file *filp)
+{
+        struct xenbus_file_priv *u;
+
+        if (xen_store_evtchn == 0)
+                return -ENOENT;
+
+        nonseekable_open(inode, filp);
+
+        u = kzalloc(sizeof(*u), GFP_KERNEL);
+        if (u == NULL)
+                return -ENOMEM;
+
+        INIT_LIST_HEAD(&u->transactions);
+        INIT_LIST_HEAD(&u->watches);
+        INIT_LIST_HEAD(&u->read_buffers);
+        init_waitqueue_head(&u->read_waitq);
+
+        mutex_init(&u->reply_mutex);
+        mutex_init(&u->msgbuffer_mutex);
+
+        filp->private_data = u;
+
+        return 0;
+}
+
+static int xenbus_file_release(struct inode *inode, struct file *filp)
+{
+        struct xenbus_file_priv *u = filp->private_data;
+        struct xenbus_transaction_holder *trans, *tmp;
+        struct watch_adapter *watch, *tmp_watch;
+
+        /*
+         * No need for locking here because there are no other users,
+         * by definition.
+         */
+
+        list_for_each_entry_safe(trans, tmp, &u->transactions, list) {
+                xenbus_transaction_end(trans->handle, 1);
+                list_del(&trans->list);
+                kfree(trans);
+        }
+
+        list_for_each_entry_safe(watch, tmp_watch, &u->watches, list) {
+                unregister_xenbus_watch(&watch->watch);
+                list_del(&watch->list);
+                free_watch_adapter(watch);
+        }
+
+        kfree(u);
+
+        return 0;
+}
+
+static unsigned int xenbus_file_poll(struct file *file, poll_table *wait)
+{
+        struct xenbus_file_priv *u = file->private_data;
+
+        poll_wait(file, &u->read_waitq, wait);
+        if (!list_empty(&u->read_buffers))
+                return POLLIN | POLLRDNORM;
+        return 0;
+}
+
+const struct file_operations xenbus_file_ops = {
+        .read = xenbus_file_read,
+        .write = xenbus_file_write,
+        .open = xenbus_file_open,
+        .release = xenbus_file_release,
+        .poll = xenbus_file_poll,
+};
diff --git a/drivers/xen/xenfs/xenfs.h b/drivers/xen/xenfs/xenfs.h
new file mode 100644
index 000000000000..51f08b2d0bf1
--- /dev/null
+++ b/drivers/xen/xenfs/xenfs.h
@@ -0,0 +1,6 @@
+#ifndef _XENFS_XENBUS_H
+#define _XENFS_XENBUS_H
+
+extern const struct file_operations xenbus_file_ops;
+
+#endif  /* _XENFS_XENBUS_H */
diff --git a/include/linux/magic.h b/include/linux/magic.h
index f7f3fdddbef0..439f6f3cb0c4 100644
--- a/include/linux/magic.h
+++ b/include/linux/magic.h
@@ -13,6 +13,7 @@
 #define EFS_SUPER_MAGIC         0x414A53
 #define EXT2_SUPER_MAGIC        0xEF53
 #define EXT3_SUPER_MAGIC        0xEF53
+#define XENFS_SUPER_MAGIC       0xabba1974
 #define EXT4_SUPER_MAGIC        0xEF53
 #define HPFS_SUPER_MAGIC        0xf995e849
 #define ISOFS_SUPER_MAGIC       0x9660
diff --git a/include/xen/xenbus.h b/include/xen/xenbus.h
index 6369d89c25d5..f87f9614844d 100644
--- a/include/xen/xenbus.h
+++ b/include/xen/xenbus.h
@@ -136,8 +136,6 @@ struct xenbus_transaction
 /* Nil transaction ID. */
 #define XBT_NIL ((struct xenbus_transaction) { 0 })
 
-int __init xenbus_dev_init(void);
-
 char **xenbus_directory(struct xenbus_transaction t,
                         const char *dir, const char *node, unsigned int *num);
 void *xenbus_read(struct xenbus_transaction t,