rbd: replace the rbd sysfs interface
authorYehuda Sadeh <yehuda@hq.newdream.net>
Fri, 19 Nov 2010 22:51:04 +0000 (14:51 -0800)
committerSage Weil <sage@newdream.net>
Wed, 1 Dec 2010 23:53:22 +0000 (15:53 -0800)
The new interface creates directories per mapped image
and under each it creates a subdir per available snapshot.
This allows keeping a cleaner interface within the sysfs
guidelines. The ABI documentation was updated too.

Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
Documentation/ABI/testing/sysfs-bus-rbd [new file with mode: 0644]
drivers/block/rbd.c

diff --git a/Documentation/ABI/testing/sysfs-bus-rbd b/Documentation/ABI/testing/sysfs-bus-rbd
new file mode 100644 (file)
index 0000000..90a87e2
--- /dev/null
@@ -0,0 +1,83 @@
+What:          /sys/bus/rbd/
+Date:          November 2010
+Contact:       Yehuda Sadeh <yehuda@hq.newdream.net>,
+               Sage Weil <sage@newdream.net>
+Description:
+
+Being used for adding and removing rbd block devices.
+
+Usage: <mon ip addr> <options> <pool name> <rbd image name> [snap name]
+
+ $ echo "192.168.0.1 name=admin rbd foo" > /sys/bus/rbd/add
+
+The snapshot name can be "-" or omitted to map the image read/write. A <dev-id>
+will be assigned for any registered block device. If snapshot is used, it will
+be mapped read-only.
+
+Removal of a device:
+
+  $ echo <dev-id> > /sys/bus/rbd/remove
+
+Entries under /sys/bus/rbd/devices/<dev-id>/
+--------------------------------------------
+
+client_id
+
+       The ceph unique client id that was assigned for this specific session.
+
+major
+
+       The block device major number.
+
+name
+
+       The name of the rbd image.
+
+pool
+
+       The pool where this rbd image resides. The pool-name pair is unique
+       per rados system.
+
+size
+
+       The size (in bytes) of the mapped block device.
+
+refresh
+
+       Writing to this file will reread the image header data and set
+       all relevant datastructures accordingly.
+
+current_snap
+
+       The current snapshot for which the device is mapped.
+
+create_snap
+
+       Create a snapshot:
+
+        $ echo <snap-name> > /sys/bus/rbd/devices/<dev-id>/snap_create
+
+rollback_snap
+
+       Rolls back data to the specified snapshot. This goes over the entire
+       list of rados blocks and sends a rollback command to each.
+
+        $ echo <snap-name> > /sys/bus/rbd/devices/<dev-id>/snap_rollback
+
+snap_*
+
+       A directory per each snapshot
+
+
+Entries under /sys/bus/rbd/devices/<dev-id>/snap_<snap-name>
+-------------------------------------------------------------
+
+id
+
+       The rados internal snapshot id assigned for this snapshot
+
+size
+
+       The size of the image when this snapshot was taken.
+
+
index 6ec9d53806c5e791b5d56fa0ae27dae0e903db02..008d4a00b50df57f529528b1b2eefec9b379a544 100644 (file)
 
 
 
-   Instructions for use
-   --------------------
+   For usage instructions, please refer to:
 
-   1) Map a Linux block device to an existing rbd image.
-
-      Usage: <mon ip addr> <options> <pool name> <rbd image name> [snap name]
-
-      $ echo "192.168.0.1 name=admin rbd foo" > /sys/class/rbd/add
-
-      The snapshot name can be "-" or omitted to map the image read/write.
-
-   2) List all active blkdev<->object mappings.
-
-      In this example, we have performed step #1 twice, creating two blkdevs,
-      mapped to two separate rados objects in the rados rbd pool
-
-      $ cat /sys/class/rbd/list
-      #id     major   client_name     pool    name    snap    KB
-      0       254     client4143      rbd     foo     -      1024000
-
-      The columns, in order, are:
-      - blkdev unique id
-      - blkdev assigned major
-      - rados client id
-      - rados pool name
-      - rados block device name
-      - mapped snapshot ("-" if none)
-      - device size in KB
-
-
-   3) Create a snapshot.
-
-      Usage: <blkdev id> <snapname>
-
-      $ echo "0 mysnap" > /sys/class/rbd/snap_create
-
-
-   4) Listing a snapshot.
-
-      $ cat /sys/class/rbd/snaps_list
-      #id     snap    KB
-      0       -       1024000 (*)
-      0       foo     1024000
-
-      The columns, in order, are:
-      - blkdev unique id
-      - snapshot name, '-' means none (active read/write version)
-      - size of device at time of snapshot
-      - the (*) indicates this is the active version
-
-   5) Rollback to snapshot.
-
-      Usage: <blkdev id> <snapname>
-
-      $ echo "0 mysnap" > /sys/class/rbd/snap_rollback
-
-
-   6) Mapping an image using snapshot.
-
-      A snapshot mapping is read-only. This is being done by passing
-      snap=<snapname> to the options when adding a device.
-
-      $ echo "192.168.0.1 name=admin,snap=mysnap rbd foo" > /sys/class/rbd/add
-
-
-   7) Remove an active blkdev<->rbd image mapping.
-
-      In this example, we remove the mapping with blkdev unique id 1.
-
-      $ echo 1 > /sys/class/rbd/remove
-
-
-   NOTE:  The actual creation and deletion of rados objects is outside the scope
-   of this driver.
+                 Documentation/ABI/testing/sysfs-bus-rbd
 
  */
 
@@ -163,6 +92,14 @@ struct rbd_request {
        u64                     len;
 };
 
+struct rbd_snap {
+       struct  device          dev;
+       const char              *name;
+       size_t                  size;
+       struct list_head        node;
+       u64                     id;
+};
+
 /*
  * a single device
  */
@@ -193,21 +130,60 @@ struct rbd_device {
        int read_only;
 
        struct list_head        node;
+
+       /* list of snapshots */
+       struct list_head        snaps;
+
+       /* sysfs related */
+       struct device           dev;
+};
+
+static struct bus_type rbd_bus_type = {
+       .name           = "rbd",
 };
 
 static spinlock_t node_lock;      /* protects client get/put */
 
-static struct class *class_rbd;          /* /sys/class/rbd */
 static DEFINE_MUTEX(ctl_mutex);          /* Serialize open/close/setup/teardown */
 static LIST_HEAD(rbd_dev_list);    /* devices */
 static LIST_HEAD(rbd_client_list);      /* clients */
 
+static int __rbd_init_snaps_header(struct rbd_device *rbd_dev);
+static void rbd_dev_release(struct device *dev);
+static ssize_t rbd_snap_rollback(struct device *dev,
+                                struct device_attribute *attr,
+                                const char *buf,
+                                size_t size);
+static ssize_t rbd_snap_add(struct device *dev,
+                           struct device_attribute *attr,
+                           const char *buf,
+                           size_t count);
+static void __rbd_remove_snap_dev(struct rbd_device *rbd_dev,
+                                 struct rbd_snap *snap);;
+
+
+static struct rbd_device *dev_to_rbd(struct device *dev)
+{
+       return container_of(dev, struct rbd_device, dev);
+}
+
+static struct device *rbd_get_dev(struct rbd_device *rbd_dev)
+{
+       return get_device(&rbd_dev->dev);
+}
+
+static void rbd_put_dev(struct rbd_device *rbd_dev)
+{
+       put_device(&rbd_dev->dev);
+}
 
 static int rbd_open(struct block_device *bdev, fmode_t mode)
 {
        struct gendisk *disk = bdev->bd_disk;
        struct rbd_device *rbd_dev = disk->private_data;
 
+       rbd_get_dev(rbd_dev);
+
        set_device_ro(bdev, rbd_dev->read_only);
 
        if ((mode & FMODE_WRITE) && rbd_dev->read_only)
@@ -216,9 +192,19 @@ static int rbd_open(struct block_device *bdev, fmode_t mode)
        return 0;
 }
 
+static int rbd_release(struct gendisk *disk, fmode_t mode)
+{
+       struct rbd_device *rbd_dev = disk->private_data;
+
+       rbd_put_dev(rbd_dev);
+
+       return 0;
+}
+
 static const struct block_device_operations rbd_bd_ops = {
        .owner                  = THIS_MODULE,
        .open                   = rbd_open,
+       .release                = rbd_release,
 };
 
 /*
@@ -361,7 +347,6 @@ static int rbd_header_from_disk(struct rbd_image_header *header,
        int ret = -ENOMEM;
 
        init_rwsem(&header->snap_rwsem);
-
        header->snap_names_len = le64_to_cpu(ondisk->snap_names_len);
        header->snapc = kmalloc(sizeof(struct ceph_snap_context) +
                                snap_count *
@@ -1256,10 +1241,20 @@ bad:
        return -ERANGE;
 }
 
+static void __rbd_remove_all_snaps(struct rbd_device *rbd_dev)
+{
+       struct rbd_snap *snap;
+
+       while (!list_empty(&rbd_dev->snaps)) {
+               snap = list_first_entry(&rbd_dev->snaps, struct rbd_snap, node);
+               __rbd_remove_snap_dev(rbd_dev, snap);
+       }
+}
+
 /*
  * only read the first part of the ondisk header, without the snaps info
  */
-static int rbd_update_snaps(struct rbd_device *rbd_dev)
+static int __rbd_update_snaps(struct rbd_device *rbd_dev)
 {
        int ret;
        struct rbd_image_header h;
@@ -1280,12 +1275,15 @@ static int rbd_update_snaps(struct rbd_device *rbd_dev)
        rbd_dev->header.total_snaps = h.total_snaps;
        rbd_dev->header.snapc = h.snapc;
        rbd_dev->header.snap_names = h.snap_names;
+       rbd_dev->header.snap_names_len = h.snap_names_len;
        rbd_dev->header.snap_sizes = h.snap_sizes;
        rbd_dev->header.snapc->seq = snap_seq;
 
+       ret = __rbd_init_snaps_header(rbd_dev);
+
        up_write(&rbd_dev->header.snap_rwsem);
 
-       return 0;
+       return ret;
 }
 
 static int rbd_init_disk(struct rbd_device *rbd_dev)
@@ -1300,6 +1298,11 @@ static int rbd_init_disk(struct rbd_device *rbd_dev)
        if (rc)
                return rc;
 
+       /* no need to lock here, as rbd_dev is not registered yet */
+       rc = __rbd_init_snaps_header(rbd_dev);
+       if (rc)
+               return rc;
+
        rc = rbd_header_set_snap(rbd_dev, rbd_dev->snap_name, &total_size);
        if (rc)
                return rc;
@@ -1343,54 +1346,360 @@ out:
        return rc;
 }
 
-/********************************************************************
- * /sys/class/rbd/
- *                   add       map rados objects to blkdev
- *                   remove    unmap rados objects
- *                   list      show mappings
- *******************************************************************/
+/*
+  sysfs
+*/
+
+static ssize_t rbd_size_show(struct device *dev,
+                            struct device_attribute *attr, char *buf)
+{
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+
+       return sprintf(buf, "%llu\n", (unsigned long long)rbd_dev->header.image_size);
+}
+
+static ssize_t rbd_major_show(struct device *dev,
+                             struct device_attribute *attr, char *buf)
+{
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
 
-static void class_rbd_release(struct class *cls)
+       return sprintf(buf, "%d\n", rbd_dev->major);
+}
+
+static ssize_t rbd_client_id_show(struct device *dev,
+                                 struct device_attribute *attr, char *buf)
 {
-       kfree(cls);
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+
+       return sprintf(buf, "client%lld\n", ceph_client_id(rbd_dev->client));
 }
 
-static ssize_t class_rbd_list(struct class *c,
-                             struct class_attribute *attr,
-                             char *data)
+static ssize_t rbd_pool_show(struct device *dev,
+                            struct device_attribute *attr, char *buf)
 {
-       int n = 0;
-       struct list_head *tmp;
-       int max = PAGE_SIZE;
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+
+       return sprintf(buf, "%s\n", rbd_dev->pool_name);
+}
+
+static ssize_t rbd_name_show(struct device *dev,
+                            struct device_attribute *attr, char *buf)
+{
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+
+       return sprintf(buf, "%s\n", rbd_dev->obj);
+}
+
+static ssize_t rbd_snap_show(struct device *dev,
+                            struct device_attribute *attr,
+                            char *buf)
+{
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+
+       return sprintf(buf, "%s\n", rbd_dev->snap_name);
+}
+
+static ssize_t rbd_image_refresh(struct device *dev,
+                                struct device_attribute *attr,
+                                const char *buf,
+                                size_t size)
+{
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+       int rc;
+       int ret = size;
 
        mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
 
-       n += snprintf(data, max,
-                     "#id\tmajor\tclient_name\tpool\tname\tsnap\tKB\n");
+       rc = __rbd_update_snaps(rbd_dev);
+       if (rc < 0)
+               ret = rc;
 
-       list_for_each(tmp, &rbd_dev_list) {
-               struct rbd_device *rbd_dev;
+       mutex_unlock(&ctl_mutex);
+       return ret;
+}
 
-               rbd_dev = list_entry(tmp, struct rbd_device, node);
-               n += snprintf(data+n, max-n,
-                             "%d\t%d\tclient%lld\t%s\t%s\t%s\t%lld\n",
-                             rbd_dev->id,
-                             rbd_dev->major,
-                             ceph_client_id(rbd_dev->client),
-                             rbd_dev->pool_name,
-                             rbd_dev->obj, rbd_dev->snap_name,
-                             rbd_dev->header.image_size >> 10);
-               if (n == max)
+static DEVICE_ATTR(size, S_IRUGO, rbd_size_show, NULL);
+static DEVICE_ATTR(major, S_IRUGO, rbd_major_show, NULL);
+static DEVICE_ATTR(client_id, S_IRUGO, rbd_client_id_show, NULL);
+static DEVICE_ATTR(pool, S_IRUGO, rbd_pool_show, NULL);
+static DEVICE_ATTR(name, S_IRUGO, rbd_name_show, NULL);
+static DEVICE_ATTR(refresh, S_IWUSR, NULL, rbd_image_refresh);
+static DEVICE_ATTR(current_snap, S_IRUGO, rbd_snap_show, NULL);
+static DEVICE_ATTR(create_snap, S_IWUSR, NULL, rbd_snap_add);
+static DEVICE_ATTR(rollback_snap, S_IWUSR, NULL, rbd_snap_rollback);
+
+static struct attribute *rbd_attrs[] = {
+       &dev_attr_size.attr,
+       &dev_attr_major.attr,
+       &dev_attr_client_id.attr,
+       &dev_attr_pool.attr,
+       &dev_attr_name.attr,
+       &dev_attr_current_snap.attr,
+       &dev_attr_refresh.attr,
+       &dev_attr_create_snap.attr,
+       &dev_attr_rollback_snap.attr,
+       NULL
+};
+
+static struct attribute_group rbd_attr_group = {
+       .attrs = rbd_attrs,
+};
+
+static const struct attribute_group *rbd_attr_groups[] = {
+       &rbd_attr_group,
+       NULL
+};
+
+static void rbd_sysfs_dev_release(struct device *dev)
+{
+}
+
+static struct device_type rbd_device_type = {
+       .name           = "rbd",
+       .groups         = rbd_attr_groups,
+       .release        = rbd_sysfs_dev_release,
+};
+
+
+/*
+  sysfs - snapshots
+*/
+
+static ssize_t rbd_snap_size_show(struct device *dev,
+                                 struct device_attribute *attr,
+                                 char *buf)
+{
+       struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev);
+
+       return sprintf(buf, "%lld\n", (long long)snap->size);
+}
+
+static ssize_t rbd_snap_id_show(struct device *dev,
+                               struct device_attribute *attr,
+                               char *buf)
+{
+       struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev);
+
+       return sprintf(buf, "%lld\n", (long long)snap->id);
+}
+
+static DEVICE_ATTR(snap_size, S_IRUGO, rbd_snap_size_show, NULL);
+static DEVICE_ATTR(snap_id, S_IRUGO, rbd_snap_id_show, NULL);
+
+static struct attribute *rbd_snap_attrs[] = {
+       &dev_attr_snap_size.attr,
+       &dev_attr_snap_id.attr,
+       NULL,
+};
+
+static struct attribute_group rbd_snap_attr_group = {
+       .attrs = rbd_snap_attrs,
+};
+
+static void rbd_snap_dev_release(struct device *dev)
+{
+       struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev);
+       kfree(snap->name);
+       kfree(snap);
+}
+
+static const struct attribute_group *rbd_snap_attr_groups[] = {
+       &rbd_snap_attr_group,
+       NULL
+};
+
+static struct device_type rbd_snap_device_type = {
+       .groups         = rbd_snap_attr_groups,
+       .release        = rbd_snap_dev_release,
+};
+
+static void __rbd_remove_snap_dev(struct rbd_device *rbd_dev,
+                                 struct rbd_snap *snap)
+{
+       list_del(&snap->node);
+       device_unregister(&snap->dev);
+}
+
+static int rbd_register_snap_dev(struct rbd_device *rbd_dev,
+                                 struct rbd_snap *snap,
+                                 struct device *parent)
+{
+       struct device *dev = &snap->dev;
+       int ret;
+
+       dev->type = &rbd_snap_device_type;
+       dev->parent = parent;
+       dev->release = rbd_snap_dev_release;
+       dev_set_name(dev, "snap_%s", snap->name);
+       ret = device_register(dev);
+
+       return ret;
+}
+
+static int __rbd_add_snap_dev(struct rbd_device *rbd_dev,
+                             int i, const char *name,
+                             struct rbd_snap **snapp)
+{
+       int ret;
+       struct rbd_snap *snap = kzalloc(sizeof(*snap), GFP_KERNEL);
+       if (!snap)
+               return -ENOMEM;
+       snap->name = kstrdup(name, GFP_KERNEL);
+       snap->size = rbd_dev->header.snap_sizes[i];
+       snap->id = rbd_dev->header.snapc->snaps[i];
+       if (device_is_registered(&rbd_dev->dev)) {
+               ret = rbd_register_snap_dev(rbd_dev, snap,
+                                            &rbd_dev->dev);
+               if (ret < 0)
+                       goto err;
+       }
+       *snapp = snap;
+       return 0;
+err:
+       kfree(snap->name);
+       kfree(snap);
+       return ret;
+}
+
+/*
+ * search for the previous snap in a null delimited string list
+ */
+const char *rbd_prev_snap_name(const char *name, const char *start)
+{
+       if (name < start + 2)
+               return NULL;
+
+       name -= 2;
+       while (*name) {
+               if (name == start)
+                       return start;
+               name--;
+       }
+       return name + 1;
+}
+
+/*
+ * compare the old list of snapshots that we have to what's in the header
+ * and update it accordingly. Note that the header holds the snapshots
+ * in a reverse order (from newest to oldest) and we need to go from
+ * older to new so that we don't get a duplicate snap name when
+ * doing the process (e.g., removed snapshot and recreated a new
+ * one with the same name.
+ */
+static int __rbd_init_snaps_header(struct rbd_device *rbd_dev)
+{
+       const char *name, *first_name;
+       int i = rbd_dev->header.total_snaps;
+       struct rbd_snap *snap, *old_snap = NULL;
+       int ret;
+       struct list_head *p, *n;
+
+       first_name = rbd_dev->header.snap_names;
+       name = first_name + rbd_dev->header.snap_names_len;
+
+       list_for_each_prev_safe(p, n, &rbd_dev->snaps) {
+               u64 cur_id;
+
+               old_snap = list_entry(p, struct rbd_snap, node);
+
+               if (i)
+                       cur_id = rbd_dev->header.snapc->snaps[i - 1];
+
+               if (!i || old_snap->id < cur_id) {
+                       /* old_snap->id was skipped, thus was removed */
+                       __rbd_remove_snap_dev(rbd_dev, old_snap);
+                       continue;
+               }
+               if (old_snap->id == cur_id) {
+                       /* we have this snapshot already */
+                       i--;
+                       name = rbd_prev_snap_name(name, first_name);
+                       continue;
+               }
+               for (; i > 0;
+                    i--, name = rbd_prev_snap_name(name, first_name)) {
+                       if (!name) {
+                               WARN_ON(1);
+                               return -EINVAL;
+                       }
+                       cur_id = rbd_dev->header.snapc->snaps[i];
+                       /* snapshot removal? handle it above */
+                       if (cur_id >= old_snap->id)
+                               break;
+                       /* a new snapshot */
+                       ret = __rbd_add_snap_dev(rbd_dev, i - 1, name, &snap);
+                       if (ret < 0)
+                               return ret;
+
+                       /* note that we add it backward so using n and not p */
+                       list_add(&snap->node, n);
+                       p = &snap->node;
+               }
+       }
+       /* we're done going over the old snap list, just add what's left */
+       for (; i > 0; i--) {
+               name = rbd_prev_snap_name(name, first_name);
+               if (!name) {
+                       WARN_ON(1);
+                       return -EINVAL;
+               }
+               ret = __rbd_add_snap_dev(rbd_dev, i - 1, name, &snap);
+               if (ret < 0)
+                       return ret;
+               list_add(&snap->node, &rbd_dev->snaps);
+       }
+
+       return 0;
+}
+
+
+static void rbd_root_dev_release(struct device *dev)
+{
+}
+
+static struct device rbd_root_dev = {
+       .init_name =    "rbd",
+       .release =      rbd_root_dev_release,
+};
+
+static int rbd_bus_add_dev(struct rbd_device *rbd_dev)
+{
+       int ret = -ENOMEM;
+       struct device *dev;
+       struct rbd_snap *snap;
+
+       mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
+       dev = &rbd_dev->dev;
+
+       dev->bus = &rbd_bus_type;
+       dev->type = &rbd_device_type;
+       dev->parent = &rbd_root_dev;
+       dev->release = rbd_dev_release;
+       dev_set_name(dev, "%d", rbd_dev->id);
+       ret = device_register(dev);
+       if (ret < 0)
+               goto done_free;
+
+       list_for_each_entry(snap, &rbd_dev->snaps, node) {
+               ret = rbd_register_snap_dev(rbd_dev, snap,
+                                            &rbd_dev->dev);
+               if (ret < 0)
                        break;
        }
 
        mutex_unlock(&ctl_mutex);
-       return n;
+       return 0;
+done_free:
+       mutex_unlock(&ctl_mutex);
+       return ret;
 }
 
-static ssize_t class_rbd_add(struct class *c,
-                            struct class_attribute *attr,
-                            const char *buf, size_t count)
+static void rbd_bus_del_dev(struct rbd_device *rbd_dev)
+{
+       device_unregister(&rbd_dev->dev);
+}
+
+static ssize_t rbd_add(struct bus_type *bus, const char *buf, size_t count)
 {
        struct ceph_osd_client *osdc;
        struct rbd_device *rbd_dev;
@@ -1419,6 +1728,7 @@ static ssize_t class_rbd_add(struct class *c,
        /* static rbd_device initialization */
        spin_lock_init(&rbd_dev->lock);
        INIT_LIST_HEAD(&rbd_dev->node);
+       INIT_LIST_HEAD(&rbd_dev->snaps);
 
        /* generate unique id: find highest unique id, add one */
        mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
@@ -1478,6 +1788,9 @@ static ssize_t class_rbd_add(struct class *c,
        }
        rbd_dev->major = irc;
 
+       rc = rbd_bus_add_dev(rbd_dev);
+       if (rc)
+               goto err_out_disk;
        /* set up and announce blkdev mapping */
        rc = rbd_init_disk(rbd_dev);
        if (rc)
@@ -1487,6 +1800,8 @@ static ssize_t class_rbd_add(struct class *c,
 
 err_out_blkdev:
        unregister_blkdev(rbd_dev->major, rbd_dev->name);
+err_out_disk:
+       rbd_free_disk(rbd_dev);
 err_out_client:
        rbd_put_client(rbd_dev);
        mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
@@ -1518,35 +1833,10 @@ static struct rbd_device *__rbd_get_dev(unsigned long id)
        return NULL;
 }
 
-static ssize_t class_rbd_remove(struct class *c,
-                               struct class_attribute *attr,
-                               const char *buf,
-                               size_t count)
+static void rbd_dev_release(struct device *dev)
 {
-       struct rbd_device *rbd_dev = NULL;
-       int target_id, rc;
-       unsigned long ul;
-
-       rc = strict_strtoul(buf, 10, &ul);
-       if (rc)
-               return rc;
-
-       /* convert to int; abort if we lost anything in the conversion */
-       target_id = (int) ul;
-       if (target_id != ul)
-               return -EINVAL;
-
-       /* remove object from list immediately */
-       mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
-
-       rbd_dev = __rbd_get_dev(target_id);
-       if (rbd_dev)
-               list_del_init(&rbd_dev->node);
-
-       mutex_unlock(&ctl_mutex);
-
-       if (!rbd_dev)
-               return -ENOENT;
+       struct rbd_device *rbd_dev =
+                       container_of(dev, struct rbd_device, dev);
 
        rbd_put_client(rbd_dev);
 
@@ -1557,67 +1847,11 @@ static ssize_t class_rbd_remove(struct class *c,
 
        /* release module ref */
        module_put(THIS_MODULE);
-
-       return count;
 }
 
-static ssize_t class_rbd_snaps_list(struct class *c,
-                             struct class_attribute *attr,
-                             char *data)
-{
-       struct rbd_device *rbd_dev = NULL;
-       struct list_head *tmp;
-       struct rbd_image_header *header;
-       int i, n = 0, max = PAGE_SIZE;
-       int ret;
-
-       mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
-
-       n += snprintf(data, max, "#id\tsnap\tKB\n");
-
-       list_for_each(tmp, &rbd_dev_list) {
-               char *names, *p;
-               struct ceph_snap_context *snapc;
-
-               rbd_dev = list_entry(tmp, struct rbd_device, node);
-               header = &rbd_dev->header;
-
-               down_read(&header->snap_rwsem);
-
-               names = header->snap_names;
-               snapc = header->snapc;
-
-               n += snprintf(data + n, max - n, "%d\t%s\t%lld%s\n",
-                             rbd_dev->id, RBD_SNAP_HEAD_NAME,
-                             header->image_size >> 10,
-                             (!rbd_dev->cur_snap ? " (*)" : ""));
-               if (n == max)
-                       break;
-
-               p = names;
-               for (i = 0; i < header->total_snaps; i++, p += strlen(p) + 1) {
-                       n += snprintf(data + n, max - n, "%d\t%s\t%lld%s\n",
-                             rbd_dev->id, p, header->snap_sizes[i] >> 10,
-                             (rbd_dev->cur_snap &&
-                              (snap_index(header, i) == rbd_dev->cur_snap) ?
-                              " (*)" : ""));
-                       if (n == max)
-                               break;
-               }
-
-               up_read(&header->snap_rwsem);
-       }
-
-
-       ret = n;
-       mutex_unlock(&ctl_mutex);
-       return ret;
-}
-
-static ssize_t class_rbd_snaps_refresh(struct class *c,
-                               struct class_attribute *attr,
-                               const char *buf,
-                               size_t count)
+static ssize_t rbd_remove(struct bus_type *bus,
+                         const char *buf,
+                         size_t count)
 {
        struct rbd_device *rbd_dev = NULL;
        int target_id, rc;
@@ -1641,95 +1875,70 @@ static ssize_t class_rbd_snaps_refresh(struct class *c,
                goto done;
        }
 
-       rc = rbd_update_snaps(rbd_dev);
-       if (rc < 0)
-               ret = rc;
+       list_del_init(&rbd_dev->node);
+
+       __rbd_remove_all_snaps(rbd_dev);
+       rbd_bus_del_dev(rbd_dev);
 
 done:
        mutex_unlock(&ctl_mutex);
        return ret;
 }
 
-static ssize_t class_rbd_snap_create(struct class *c,
-                               struct class_attribute *attr,
-                               const char *buf,
-                               size_t count)
+static ssize_t rbd_snap_add(struct device *dev,
+                           struct device_attribute *attr,
+                           const char *buf,
+                           size_t count)
 {
-       struct rbd_device *rbd_dev = NULL;
-       int target_id, ret;
-       char *name;
-
-       name = kmalloc(RBD_MAX_SNAP_NAME_LEN + 1, GFP_KERNEL);
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+       int ret;
+       char *name = kmalloc(count + 1, GFP_KERNEL);
        if (!name)
                return -ENOMEM;
 
-       /* parse snaps add command */
-       if (sscanf(buf, "%d "
-                  "%" __stringify(RBD_MAX_SNAP_NAME_LEN) "s",
-                  &target_id,
-                  name) != 2) {
-               ret = -EINVAL;
-               goto done;
-       }
+       snprintf(name, count, "%s", buf);
 
        mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
 
-       rbd_dev = __rbd_get_dev(target_id);
-       if (!rbd_dev) {
-               ret = -ENOENT;
-               goto done_unlock;
-       }
-
        ret = rbd_header_add_snap(rbd_dev,
                                  name, GFP_KERNEL);
        if (ret < 0)
                goto done_unlock;
 
-       ret = rbd_update_snaps(rbd_dev);
+       ret = __rbd_update_snaps(rbd_dev);
        if (ret < 0)
                goto done_unlock;
 
        ret = count;
 done_unlock:
        mutex_unlock(&ctl_mutex);
-done:
        kfree(name);
        return ret;
 }
 
-static ssize_t class_rbd_rollback(struct class *c,
-                               struct class_attribute *attr,
-                               const char *buf,
-                               size_t count)
+static ssize_t rbd_snap_rollback(struct device *dev,
+                                struct device_attribute *attr,
+                                const char *buf,
+                                size_t count)
 {
-       struct rbd_device *rbd_dev = NULL;
-       int target_id, ret;
+       struct rbd_device *rbd_dev = dev_to_rbd(dev);
+       int ret;
        u64 snapid;
-       char snap_name[RBD_MAX_SNAP_NAME_LEN];
        u64 cur_ofs;
-       char *seg_name;
+       char *seg_name = NULL;
+       char *snap_name = kmalloc(count + 1, GFP_KERNEL);
+       ret = -ENOMEM;
+       if (!snap_name)
+               return ret;
 
        /* parse snaps add command */
-       if (sscanf(buf, "%d "
-                  "%" __stringify(RBD_MAX_SNAP_NAME_LEN) "s",
-                  &target_id,
-                  snap_name) != 2) {
-               return -EINVAL;
-       }
-
-       ret = -ENOMEM;
+       snprintf(snap_name, count, "%s", buf);
        seg_name = kmalloc(RBD_MAX_SEG_NAME_LEN + 1, GFP_NOIO);
        if (!seg_name)
-               return ret;
+               goto done;
 
        mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING);
 
-       rbd_dev = __rbd_get_dev(target_id);
-       if (!rbd_dev) {
-               ret = -ENOENT;
-               goto done_unlock;
-       }
-
        ret = snap_by_name(&rbd_dev->header, snap_name, &snapid, NULL);
        if (ret < 0)
                goto done_unlock;
@@ -1750,7 +1959,7 @@ static ssize_t class_rbd_rollback(struct class *c,
                                   seg_name, ret);
        }
 
-       ret = rbd_update_snaps(rbd_dev);
+       ret = __rbd_update_snaps(rbd_dev);
        if (ret < 0)
                goto done_unlock;
 
@@ -1758,57 +1967,42 @@ static ssize_t class_rbd_rollback(struct class *c,
 
 done_unlock:
        mutex_unlock(&ctl_mutex);
+done:
        kfree(seg_name);
+       kfree(snap_name);
 
        return ret;
 }
 
-static struct class_attribute class_rbd_attrs[] = {
-       __ATTR(add,             0200, NULL, class_rbd_add),
-       __ATTR(remove,          0200, NULL, class_rbd_remove),
-       __ATTR(list,            0444, class_rbd_list, NULL),
-       __ATTR(snaps_refresh,   0200, NULL, class_rbd_snaps_refresh),
-       __ATTR(snap_create,     0200, NULL, class_rbd_snap_create),
-       __ATTR(snaps_list,      0444, class_rbd_snaps_list, NULL),
-       __ATTR(snap_rollback,   0200, NULL, class_rbd_rollback),
+static struct bus_attribute rbd_bus_attrs[] = {
+       __ATTR(add, S_IWUSR, NULL, rbd_add),
+       __ATTR(remove, S_IWUSR, NULL, rbd_remove),
        __ATTR_NULL
 };
 
 /*
  * create control files in sysfs
- * /sys/class/rbd/...
+ * /sys/bus/rbd/...
  */
 static int rbd_sysfs_init(void)
 {
-       int ret = -ENOMEM;
+       int ret;
 
-       class_rbd = kzalloc(sizeof(*class_rbd), GFP_KERNEL);
-       if (!class_rbd)
-               goto out;
+       rbd_bus_type.bus_attrs = rbd_bus_attrs;
 
-       class_rbd->name = DRV_NAME;
-       class_rbd->owner = THIS_MODULE;
-       class_rbd->class_release = class_rbd_release;
-       class_rbd->class_attrs = class_rbd_attrs;
+       ret = bus_register(&rbd_bus_type);
+        if (ret < 0)
+               return ret;
 
-       ret = class_register(class_rbd);
-       if (ret)
-               goto out_class;
-       return 0;
+       ret = device_register(&rbd_root_dev);
 
-out_class:
-       kfree(class_rbd);
-       class_rbd = NULL;
-       pr_err(DRV_NAME ": failed to create class rbd\n");
-out:
        return ret;
 }
 
 static void rbd_sysfs_cleanup(void)
 {
-       if (class_rbd)
-               class_destroy(class_rbd);
-       class_rbd = NULL;
+       device_unregister(&rbd_root_dev);
+       bus_unregister(&rbd_bus_type);
 }
 
 int __init rbd_init(void)