A previous patch designed to update the sync status of a mirror was to hasty. 
The original idea was to make a machine immediately aware of a change from sync
-> out-of-sync when dealing with cluster mirrors.  However, the machines need to
discover this on their own, otherwise they cannot tell if they can switch a
primary device when it fails...   From the patch header:

We must only allow do_recovery to mark ms->in_sync as 1.  It is
the job of the fault handling code (like __bio_mark_nosync) to
mark ms->in_sync as 0, if necessary.

If do_recovery handles this, it is possible for us not to be able
to switch primary devices in the case of cluster mirroring.  The
scenario is:

0) Mirror is in-sync
1) Node1 writes to disk, but write fails to the primary device
2) Node1 increments the error count for that device
3) Node1 checks ms->in_sync to see if it is safe to switch the
   primary.  (We cannot switch the primary if other devices are
   not in-sync.  This would lead to bad data being read.)
4) Node1 switches the primary because the mirror is in-sync, then
   marks the region out-of-sync and ms->in_sync = 0.
5) Node2 writes and fails to the primary device
6) Node2 increments the error count for that device
7) Node2 checks ms->in_sync to see if it is safe to switch the
   primary.  It isn't because do_recovery has stepped in and changed
   ms->in_sync when it shouldn't have.