rbd: don't destroy ceph_opts in rbd_add()

Whether rbd_client_create() successfully creates a new client or
not, it takes responsibility for getting the ceph_opts structure
it's passed destroyed.  If successful, the structure becomes
associated with the created client; if not, rbd_client_create()
will destroy it.

Previously, rbd_get_client() would call ceph_destroy_options()
if rbd_get_client() failed, and that meant it got called twice.
That led freeing various pointers more than once, which is never a
good idea.

This resolves:
    http://tracker.ceph.com/issues/4559

Cc: stable@vger.kernel.org # 3.8+
Reported-by: Dan van der Ster <dan@vanderster.com>
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

drivers/block/rbd.c

@@ -521,8 +521,8 @@ static const struct block_device_operations rbd_bd_ops = {
 };
 
 /*
- * Initialize an rbd client instance.
- * We own *ceph_opts.
+ * Initialize an rbd client instance.  Success or not, this function
+ * consumes ceph_opts.
  */
 static struct rbd_client *rbd_client_create(struct ceph_options *ceph_opts)
 {
@@ -677,7 +677,8 @@ static int parse_rbd_opts_token(char *c, void *private)
 
 /*
  * Get a ceph client with specific addr and configuration, if one does
- * not exist create it.
+ * not exist create it.  Either way, ceph_opts is consumed by this
+ * function.
  */
 static struct rbd_client *rbd_get_client(struct ceph_options *ceph_opts)
 {
@@ -4994,7 +4995,6 @@ static ssize_t rbd_add(struct bus_type *bus,
 		rc = PTR_ERR(rbdc);
 		goto err_out_args;
 	}
-	ceph_opts = NULL;	/* rbd_dev client now owns this */
 
 	/* pick the pool */
 	osdc = &rbdc->client->osdc;
@@ -5038,8 +5038,6 @@ static ssize_t rbd_add(struct bus_type *bus,
 err_out_client:
 	rbd_put_client(rbdc);
 err_out_args:
-	if (ceph_opts)
-		ceph_destroy_options(ceph_opts);
 	kfree(rbd_opts);
 	rbd_spec_put(spec);
 err_out_module: