size_t count);
static ssize_t rbd_remove(struct bus_type *bus, const char *buf,
size_t count);
-static int rbd_dev_probe(struct rbd_device *rbd_dev);
+static int rbd_dev_image_probe(struct rbd_device *rbd_dev);
static struct bus_attribute rbd_bus_attrs[] = {
__ATTR(add, S_IWUSR, NULL, rbd_add),
# define rbd_assert(expr) ((void) 0)
#endif /* !RBD_DEBUG */
-static void rbd_img_parent_read(struct rbd_obj_request *obj_request);
static int rbd_img_obj_request_submit(struct rbd_obj_request *obj_request);
+static void rbd_img_parent_read(struct rbd_obj_request *obj_request);
+static void rbd_dev_remove_parent(struct rbd_device *rbd_dev);
static int rbd_dev_refresh(struct rbd_device *rbd_dev, u64 *hver);
static int rbd_dev_v2_refresh(struct rbd_device *rbd_dev, u64 *hver);
return NULL;
}
-static int rbd_dev_set_mapping(struct rbd_device *rbd_dev)
+static int rbd_dev_mapping_set(struct rbd_device *rbd_dev)
{
if (!memcmp(rbd_dev->spec->snap_name, RBD_SNAP_HEAD_NAME,
sizeof (RBD_SNAP_HEAD_NAME))) {
rbd_dev->mapping.features = snap->features;
rbd_dev->mapping.read_only = true;
}
- set_bit(RBD_DEV_FLAG_EXISTS, &rbd_dev->flags);
return 0;
}
+static void rbd_dev_mapping_clear(struct rbd_device *rbd_dev)
+{
+ rbd_dev->mapping.size = 0;
+ rbd_dev->mapping.features = 0;
+ rbd_dev->mapping.read_only = true;
+}
+
static void rbd_header_free(struct rbd_image_header *header)
{
kfree(header->object_prefix);
static void rbd_update_mapping_size(struct rbd_device *rbd_dev)
{
- sector_t size;
-
if (rbd_dev->spec->snap_id != CEPH_NOSNAP)
return;
- size = (sector_t) rbd_dev->header.image_size / SECTOR_SIZE;
- dout("setting size to %llu sectors", (unsigned long long) size);
- rbd_dev->mapping.size = (u64) size;
- set_capacity(rbd_dev->disk, size);
+ if (rbd_dev->mapping.size != rbd_dev->header.image_size) {
+ sector_t size;
+
+ rbd_dev->mapping.size = rbd_dev->header.image_size;
+ size = (sector_t)rbd_dev->mapping.size / SECTOR_SIZE;
+ dout("setting size to %llu sectors", (unsigned long long)size);
+ set_capacity(rbd_dev->disk, size);
+ }
}
/*
rbd_dev->disk = disk;
- set_capacity(rbd_dev->disk, rbd_dev->mapping.size / SECTOR_SIZE);
-
return 0;
out_disk:
put_disk(disk);
struct device_attribute *attr, char *buf)
{
struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
- sector_t size;
-
- down_read(&rbd_dev->header_rwsem);
- size = get_capacity(rbd_dev->disk);
- up_read(&rbd_dev->header_rwsem);
- return sprintf(buf, "%llu\n", (unsigned long long) size * SECTOR_SIZE);
+ return sprintf(buf, "%llu\n",
+ (unsigned long long)rbd_dev->mapping.size);
}
/*
struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
return sprintf(buf, "0x%016llx\n",
- (unsigned long long) rbd_dev->mapping.features);
+ (unsigned long long)rbd_dev->mapping.features);
}
static ssize_t rbd_major_show(struct device *dev,
{
struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
- return sprintf(buf, "%d\n", rbd_dev->major);
+ if (rbd_dev->major)
+ return sprintf(buf, "%d\n", rbd_dev->major);
+
+ return sprintf(buf, "(none)\n");
+
}
static ssize_t rbd_client_id_show(struct device *dev,
struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
return sprintf(buf, "%llu\n",
- (unsigned long long) rbd_dev->spec->pool_id);
+ (unsigned long long) rbd_dev->spec->pool_id);
}
static ssize_t rbd_name_show(struct device *dev,
static void rbd_dev_destroy(struct rbd_device *rbd_dev)
{
- rbd_spec_put(rbd_dev->parent_spec);
- kfree(rbd_dev->header_name);
rbd_put_client(rbd_dev->rbd_client);
rbd_spec_put(rbd_dev->spec);
kfree(rbd_dev);
}
/*
- * When a parent image gets probed, we only have the pool, image,
- * and snapshot ids but not the names of any of them. This call
- * is made later to fill in those names. It has to be done after
- * rbd_dev_snaps_update() has completed because some of the
- * information (in particular, snapshot name) is not available
- * until then.
+ * When an rbd image has a parent image, it is identified by the
+ * pool, image, and snapshot ids (not names). This function fills
+ * in the names for those ids. (It's OK if we can't figure out the
+ * name for an image id, but the pool and snapshot ids should always
+ * exist and have names.) All names in an rbd spec are dynamically
+ * allocated.
*
* When an image being mapped (not a parent) is probed, we have the
* pool name and pool id, image name and image id, and the snapshot
* name. The only thing we're missing is the snapshot id.
+ *
+ * The set of snapshots for an image is not known until they have
+ * been read by rbd_dev_snaps_update(), so we can't completely fill
+ * in this information until after that has been called.
*/
-static int rbd_dev_probe_update_spec(struct rbd_device *rbd_dev)
+static int rbd_dev_spec_update(struct rbd_device *rbd_dev)
{
- struct ceph_osd_client *osdc;
- const char *name;
- void *reply_buf = NULL;
+ struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
+ struct rbd_spec *spec = rbd_dev->spec;
+ const char *pool_name;
+ const char *image_name;
+ const char *snap_name;
int ret;
/*
* An image being mapped will have the pool name (etc.), but
* we need to look up the snapshot id.
*/
- if (rbd_dev->spec->pool_name) {
- if (strcmp(rbd_dev->spec->snap_name, RBD_SNAP_HEAD_NAME)) {
+ if (spec->pool_name) {
+ if (strcmp(spec->snap_name, RBD_SNAP_HEAD_NAME)) {
struct rbd_snap *snap;
- snap = snap_by_name(rbd_dev, rbd_dev->spec->snap_name);
+ snap = snap_by_name(rbd_dev, spec->snap_name);
if (!snap)
return -ENOENT;
- rbd_dev->spec->snap_id = snap->id;
+ spec->snap_id = snap->id;
} else {
- rbd_dev->spec->snap_id = CEPH_NOSNAP;
+ spec->snap_id = CEPH_NOSNAP;
}
return 0;
}
- /* Look up the pool name */
+ /* Get the pool name; we have to make our own copy of this */
- osdc = &rbd_dev->rbd_client->client->osdc;
- name = ceph_pg_pool_name_by_id(osdc->osdmap, rbd_dev->spec->pool_id);
- if (!name) {
- rbd_warn(rbd_dev, "there is no pool with id %llu",
- rbd_dev->spec->pool_id); /* Really a BUG() */
+ pool_name = ceph_pg_pool_name_by_id(osdc->osdmap, spec->pool_id);
+ if (!pool_name) {
+ rbd_warn(rbd_dev, "no pool with id %llu", spec->pool_id);
return -EIO;
}
-
- rbd_dev->spec->pool_name = kstrdup(name, GFP_KERNEL);
- if (!rbd_dev->spec->pool_name)
+ pool_name = kstrdup(pool_name, GFP_KERNEL);
+ if (!pool_name)
return -ENOMEM;
/* Fetch the image name; tolerate failure here */
- name = rbd_dev_image_name(rbd_dev);
- if (name)
- rbd_dev->spec->image_name = (char *)name;
- else
+ image_name = rbd_dev_image_name(rbd_dev);
+ if (!image_name)
rbd_warn(rbd_dev, "unable to get image name");
- /* Look up the snapshot name. */
+ /* Look up the snapshot name, and make a copy */
- name = rbd_snap_name(rbd_dev, rbd_dev->spec->snap_id);
- if (!name) {
- rbd_warn(rbd_dev, "no snapshot with id %llu",
- rbd_dev->spec->snap_id); /* Really a BUG() */
+ snap_name = rbd_snap_name(rbd_dev, spec->snap_id);
+ if (!snap_name) {
+ rbd_warn(rbd_dev, "no snapshot with id %llu", spec->snap_id);
ret = -EIO;
goto out_err;
}
- rbd_dev->spec->snap_name = kstrdup(name, GFP_KERNEL);
- if(!rbd_dev->spec->snap_name)
+ snap_name = kstrdup(snap_name, GFP_KERNEL);
+ if (!snap_name) {
+ ret = -ENOMEM;
goto out_err;
+ }
+
+ spec->pool_name = pool_name;
+ spec->image_name = image_name;
+ spec->snap_name = snap_name;
return 0;
out_err:
- kfree(reply_buf);
- kfree(rbd_dev->spec->pool_name);
- rbd_dev->spec->pool_name = NULL;
+ kfree(image_name);
+ kfree(pool_name);
return ret;
}
static int rbd_dev_v2_refresh(struct rbd_device *rbd_dev, u64 *hver)
{
int ret;
- __u8 obj_order;
down_write(&rbd_dev->header_rwsem);
- /* Grab old order first, to see if it changes */
-
- obj_order = rbd_dev->header.obj_order,
ret = rbd_dev_v2_image_size(rbd_dev);
if (ret)
goto out;
- if (rbd_dev->header.obj_order != obj_order) {
- ret = -EIO;
- goto out;
- }
rbd_update_mapping_size(rbd_dev);
ret = rbd_dev_v2_snap_context(rbd_dev, hver);
static int rbd_dev_v1_probe(struct rbd_device *rbd_dev)
{
int ret;
- size_t size;
-
- /* Record the header object name for this rbd image. */
-
- size = strlen(rbd_dev->spec->image_name) + sizeof (RBD_SUFFIX);
- rbd_dev->header_name = kmalloc(size, GFP_KERNEL);
- if (!rbd_dev->header_name) {
- ret = -ENOMEM;
- goto out_err;
- }
- sprintf(rbd_dev->header_name, "%s%s",
- rbd_dev->spec->image_name, RBD_SUFFIX);
/* Populate rbd image metadata */
static int rbd_dev_v2_probe(struct rbd_device *rbd_dev)
{
- size_t size;
int ret;
u64 ver = 0;
- /*
- * Image id was filled in by the caller. Record the header
- * object name for this rbd image.
- */
- size = sizeof (RBD_HEADER_PREFIX) + strlen(rbd_dev->spec->image_id);
- rbd_dev->header_name = kmalloc(size, GFP_KERNEL);
- if (!rbd_dev->header_name)
- return -ENOMEM;
- sprintf(rbd_dev->header_name, "%s%s",
- RBD_HEADER_PREFIX, rbd_dev->spec->image_id);
-
- /* Get the size and object order for the image */
ret = rbd_dev_v2_image_size(rbd_dev);
if (ret)
goto out_err;
return ret;
}
-static int rbd_dev_probe_finish(struct rbd_device *rbd_dev)
+static int rbd_dev_probe_parent(struct rbd_device *rbd_dev)
{
struct rbd_device *parent = NULL;
- struct rbd_spec *parent_spec = NULL;
- struct rbd_client *rbdc = NULL;
+ struct rbd_spec *parent_spec;
+ struct rbd_client *rbdc;
int ret;
- /* no need to lock here, as rbd_dev is not registered yet */
- ret = rbd_dev_snaps_update(rbd_dev);
- if (ret)
- return ret;
+ if (!rbd_dev->parent_spec)
+ return 0;
+ /*
+ * We need to pass a reference to the client and the parent
+ * spec when creating the parent rbd_dev. Images related by
+ * parent/child relationships always share both.
+ */
+ parent_spec = rbd_spec_get(rbd_dev->parent_spec);
+ rbdc = __rbd_get_client(rbd_dev->rbd_client);
+
+ ret = -ENOMEM;
+ parent = rbd_dev_create(rbdc, parent_spec);
+ if (!parent)
+ goto out_err;
+
+ ret = rbd_dev_image_probe(parent);
+ if (ret < 0)
+ goto out_err;
+ rbd_dev->parent = parent;
+
+ return 0;
+out_err:
+ if (parent) {
+ rbd_spec_put(rbd_dev->parent_spec);
+ kfree(rbd_dev->header_name);
+ rbd_dev_destroy(parent);
+ } else {
+ rbd_put_client(rbdc);
+ rbd_spec_put(parent_spec);
+ }
+
+ return ret;
+}
- ret = rbd_dev_probe_update_spec(rbd_dev);
+static int rbd_dev_probe_finish(struct rbd_device *rbd_dev)
+{
+ int ret;
+ int tmp;
+
+ ret = rbd_dev_header_watch_sync(rbd_dev, 1);
if (ret)
- goto err_out_snaps;
+ return ret;
- ret = rbd_dev_set_mapping(rbd_dev);
+ ret = rbd_dev_mapping_set(rbd_dev);
if (ret)
- goto err_out_snaps;
+ return ret;
/* generate unique id: find highest unique id, add one */
rbd_dev_id_get(rbd_dev);
if (ret)
goto err_out_disk;
- /*
- * At this point cleanup in the event of an error is the job
- * of the sysfs code (initiated by rbd_bus_del_dev()).
- */
- /* Probe the parent if there is one */
-
- if (rbd_dev->parent_spec) {
- /*
- * We need to pass a reference to the client and the
- * parent spec when creating the parent rbd_dev.
- * Images related by parent/child relationships
- * always share both.
- */
- parent_spec = rbd_spec_get(rbd_dev->parent_spec);
- rbdc = __rbd_get_client(rbd_dev->rbd_client);
-
- parent = rbd_dev_create(rbdc, parent_spec);
- if (!parent) {
- ret = -ENOMEM;
- goto err_out_spec;
- }
- rbdc = NULL; /* parent now owns reference */
- parent_spec = NULL; /* parent now owns reference */
- ret = rbd_dev_probe(parent);
- if (ret < 0)
- goto err_out_parent;
- rbd_dev->parent = parent;
- }
-
- ret = rbd_dev_header_watch_sync(rbd_dev, 1);
- if (ret)
- goto err_out_bus;
-
/* Everything's ready. Announce the disk to the world. */
+ set_capacity(rbd_dev->disk, rbd_dev->mapping.size / SECTOR_SIZE);
+ set_bit(RBD_DEV_FLAG_EXISTS, &rbd_dev->flags);
add_disk(rbd_dev->disk);
pr_info("%s: added with size 0x%llx\n", rbd_dev->disk->disk_name,
return ret;
-err_out_parent:
- rbd_dev_destroy(parent);
-err_out_spec:
- rbd_spec_put(parent_spec);
- rbd_put_client(rbdc);
-err_out_bus:
- /* this will also clean up rest of rbd_dev stuff */
-
- rbd_bus_del_dev(rbd_dev);
-
- return ret;
err_out_disk:
rbd_free_disk(rbd_dev);
err_out_blkdev:
unregister_blkdev(rbd_dev->major, rbd_dev->name);
err_out_id:
rbd_dev_id_put(rbd_dev);
-err_out_snaps:
- rbd_remove_all_snaps(rbd_dev);
+ tmp = rbd_dev_header_watch_sync(rbd_dev, 0);
+ if (tmp)
+ rbd_warn(rbd_dev, "failed to cancel watch event (%d)\n", ret);
+ rbd_dev_mapping_clear(rbd_dev);
return ret;
}
+static int rbd_dev_header_name(struct rbd_device *rbd_dev)
+{
+ struct rbd_spec *spec = rbd_dev->spec;
+ size_t size;
+
+ /* Record the header object name for this rbd image. */
+
+ rbd_assert(rbd_image_format_valid(rbd_dev->image_format));
+
+ if (rbd_dev->image_format == 1)
+ size = strlen(spec->image_name) + sizeof (RBD_SUFFIX);
+ else
+ size = sizeof (RBD_HEADER_PREFIX) + strlen(spec->image_id);
+
+ rbd_dev->header_name = kmalloc(size, GFP_KERNEL);
+ if (!rbd_dev->header_name)
+ return -ENOMEM;
+
+ if (rbd_dev->image_format == 1)
+ sprintf(rbd_dev->header_name, "%s%s",
+ spec->image_name, RBD_SUFFIX);
+ else
+ sprintf(rbd_dev->header_name, "%s%s",
+ RBD_HEADER_PREFIX, spec->image_id);
+ return 0;
+}
+
/*
* Probe for the existence of the header object for the given rbd
* device. For format 2 images this includes determining the image
* id.
*/
-static int rbd_dev_probe(struct rbd_device *rbd_dev)
+static int rbd_dev_image_probe(struct rbd_device *rbd_dev)
{
int ret;
rbd_assert(rbd_dev->spec->image_id);
rbd_assert(rbd_image_format_valid(rbd_dev->image_format));
+ ret = rbd_dev_header_name(rbd_dev);
+ if (ret)
+ goto err_out_format;
+
if (rbd_dev->image_format == 1)
ret = rbd_dev_v1_probe(rbd_dev);
else
ret = rbd_dev_v2_probe(rbd_dev);
if (ret)
- goto out_err;
+ goto out_header_name;
+
+ ret = rbd_dev_snaps_update(rbd_dev);
+ if (ret)
+ goto out_header_name;
+
+ ret = rbd_dev_spec_update(rbd_dev);
+ if (ret)
+ goto err_out_snaps;
+
+ ret = rbd_dev_probe_parent(rbd_dev);
+ if (ret)
+ goto err_out_snaps;
ret = rbd_dev_probe_finish(rbd_dev);
if (ret)
- rbd_header_free(&rbd_dev->header);
+ goto err_out_parent;
return ret;
-out_err:
+err_out_parent:
+ rbd_dev_remove_parent(rbd_dev);
+ rbd_header_free(&rbd_dev->header);
+err_out_snaps:
+ rbd_remove_all_snaps(rbd_dev);
+out_header_name:
+ kfree(rbd_dev->header_name);
+ rbd_dev->header_name = NULL;
+err_out_format:
+ rbd_dev->image_format = 0;
kfree(rbd_dev->spec->image_id);
rbd_dev->spec->image_id = NULL;
kfree(rbd_opts);
rbd_opts = NULL; /* done with this */
- rc = rbd_dev_probe(rbd_dev);
+ rc = rbd_dev_image_probe(rbd_dev);
if (rc < 0)
goto err_out_rbd_dev;
return count;
err_out_rbd_dev:
+ kfree(rbd_dev->header_name);
rbd_dev_destroy(rbd_dev);
err_out_client:
rbd_put_client(rbdc);
{
struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
- if (rbd_dev->watch_event)
- rbd_dev_header_watch_sync(rbd_dev, 0);
-
/* clean up and free blkdev */
rbd_free_disk(rbd_dev);
unregister_blkdev(rbd_dev->major, rbd_dev->name);
/* done with the id, and with the rbd_dev */
rbd_dev_id_put(rbd_dev);
+ rbd_dev_mapping_clear(rbd_dev);
rbd_assert(rbd_dev->rbd_client != NULL);
+ rbd_spec_put(rbd_dev->parent_spec);
+ kfree(rbd_dev->header_name);
rbd_dev_destroy(rbd_dev);
/* release module ref */
module_put(THIS_MODULE);
}
-static void __rbd_remove(struct rbd_device *rbd_dev)
+static void rbd_dev_remove_parent(struct rbd_device *rbd_dev)
{
- rbd_remove_all_snaps(rbd_dev);
- rbd_bus_del_dev(rbd_dev);
+ while (rbd_dev->parent) {
+ struct rbd_device *first = rbd_dev;
+ struct rbd_device *second = first->parent;
+ struct rbd_device *third;
+ int ret;
+
+ /*
+ * Follow to the parent with no grandparent and
+ * remove it.
+ */
+ while (second && (third = second->parent)) {
+ first = second;
+ second = third;
+ }
+ rbd_assert(second);
+ ret = rbd_dev_header_watch_sync(rbd_dev, 0);
+ if (ret)
+ rbd_warn(rbd_dev,
+ "failed to cancel watch event (%d)\n", ret);
+ rbd_remove_all_snaps(second);
+ rbd_bus_del_dev(second);
+ first->parent = NULL;
+ first->parent_overlap = 0;
+
+ rbd_assert(first->parent_spec);
+ rbd_spec_put(first->parent_spec);
+ first->parent_spec = NULL;
+ }
}
static ssize_t rbd_remove(struct bus_type *bus,
size_t count)
{
struct rbd_device *rbd_dev = NULL;
- int target_id, rc;
+ int target_id;
unsigned long ul;
- int ret = count;
+ int ret;
- rc = strict_strtoul(buf, 10, &ul);
- if (rc)
- return rc;
+ ret = strict_strtoul(buf, 10, &ul);
+ if (ret)
+ return ret;
/* convert to int; abort if we lost anything in the conversion */
target_id = (int) ul;
if (ret < 0)
goto done;
- while (rbd_dev->parent_spec) {
- struct rbd_device *first = rbd_dev;
- struct rbd_device *second = first->parent;
- struct rbd_device *third;
-
- /*
- * Follow to the parent with no grandparent and
- * remove it.
- */
- while (second && (third = second->parent)) {
- first = second;
- second = third;
- }
- __rbd_remove(second);
- rbd_spec_put(first->parent_spec);
- first->parent_spec = NULL;
- first->parent_overlap = 0;
- first->parent = NULL;
+ ret = rbd_dev_header_watch_sync(rbd_dev, 0);
+ if (ret) {
+ rbd_warn(rbd_dev, "failed to cancel watch event (%d)\n", ret);
+ clear_bit(RBD_DEV_FLAG_REMOVING, &rbd_dev->flags);
+ smp_mb();
+ return ret;
}
- __rbd_remove(rbd_dev);
+ ret = count;
+ rbd_dev_remove_parent(rbd_dev);
+
+ rbd_remove_all_snaps(rbd_dev);
+ rbd_bus_del_dev(rbd_dev);
done:
mutex_unlock(&ctl_mutex);