ceph: fix flush_dirty_caps race with caps migration

The flush_dirty_caps() used to loop over the first entry of the cap_dirty
dirty list on the assumption that after calling ceph_check_caps() it would
be removed from the list.  This isn't true for caps that are being
migrated between MDSs, where we've received the EXPORT but not the IMPORT.

Instead, do a safe list iteration, and pin the next inode on the list via
the CEPH_I_NOFLUSH flag.

Signed-off-by: Sage Weil <sage@newdream.net>
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index ff7aaa3..6a778f2 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -289,6 +289,7 @@
 #define CEPH_I_COMPLETE  1  /* we have complete directory cached */
 #define CEPH_I_NODELAY   4  /* do not delay cap release */
 #define CEPH_I_FLUSH     8  /* do not delay flush of dirty metadata */
+#define CEPH_I_NOFLUSH  16  /* do not flush dirty caps */
 
 struct ceph_inode_info {
 	struct ceph_vino i_vino;   /* ceph ino + snap */