ocfs2/dlm: Retract fix for race between purge and migrate
authorSunil Mushran <sunil.mushran@oracle.com>
Tue, 3 Feb 2009 20:37:13 +0000 (12:37 -0800)
committerMark Fasheh <mfasheh@suse.com>
Thu, 26 Feb 2009 19:51:09 +0000 (11:51 -0800)
Mainline commit d4f7e650e55af6b235871126f747da88600e8040 attempts to delay
the dlm_thread from sending the drop ref message if the lockres is being
migrated. The problem is that we make the dlm_thread wait for the migration
to complete. This causes a deadlock as dlm_thread also participates in the
lockres migration process.

A better fix for the original oss bugzilla#1012 is in testing.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Acked-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
fs/ocfs2/dlm/dlmthread.c

index d1295203029fd927131e626c44f97def0433a918..4060bb328bc8a08c22bbd77c59835d757ebdcda5 100644 (file)
@@ -181,8 +181,7 @@ static int dlm_purge_lockres(struct dlm_ctxt *dlm,
 
                spin_lock(&res->spinlock);
                /* This ensures that clear refmap is sent after the set */
-               __dlm_wait_on_lockres_flags(res, (DLM_LOCK_RES_SETREF_INPROG |
-                                                 DLM_LOCK_RES_MIGRATING));
+               __dlm_wait_on_lockres_flags(res, DLM_LOCK_RES_SETREF_INPROG);
                spin_unlock(&res->spinlock);
 
                /* clear our bit from the master's refmap, ignore errors */