drbd: Fixed processing of disk-barrier, disk-flushes and disk-drain
authorPhilipp Reisner <philipp.reisner@linbit.com>
Fri, 30 Mar 2012 12:12:15 +0000 (14:12 +0200)
committerPhilipp Reisner <philipp.reisner@linbit.com>
Thu, 8 Nov 2012 15:58:25 +0000 (16:58 +0100)
Since drbd_bump_write_ordering() is called in the attaching
process while the disk state is D_ATTACHING, it was not
considering these three flags during attach.

A call to this function was missing form drbd_adm_disk_opts().

Fixed both issues.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
drivers/block/drbd/drbd_nl.c
drivers/block/drbd/drbd_receiver.c

index c5aa084..ce9f4ca 100644 (file)
@@ -1229,6 +1229,9 @@ int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info)
        }
 
        mutex_unlock(&mdev->tconn->conf_update);
+
+       drbd_bump_write_ordering(mdev->tconn, WO_bdev_flush);
+
        drbd_md_sync(mdev);
 
        if (mdev->state.conn >= C_CONNECTED)
index 786a755..e8cd4c4 100644 (file)
@@ -1229,7 +1229,7 @@ void drbd_bump_write_ordering(struct drbd_tconn *tconn, enum write_ordering_e wo
        wo = min(pwo, wo);
        rcu_read_lock();
        idr_for_each_entry(&tconn->volumes, mdev, vnr) {
-               if (!get_ldev(mdev))
+               if (!get_ldev_if_state(mdev, D_ATTACHING))
                        continue;
                dc = rcu_dereference(mdev->ldev->disk_conf);