Commit 756a1501 authored Apr 17, 2007 by Srinivas Eeda Committed by Mark Fasheh Apr 26, 2007

ocfs2_dlm: fix race in dlm_remaster_locks



There is a possibility that dlm_remaster_locks could overwride node->state
with DLM_RECO_NODE_DATA_REQUESTED after dlm_reco_data_done_handler sets the
node->state to DLM_RECO_NODE_DATA_DONE. This could lead to recovery getting
stuck and requires a cluster reboot. Synchronize with dlm_reco_state_lock
spinlock.

Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

parent de46c337

fs/ocfs2/dlm/dlmrecovery.c

+2 −0

Original line number	Diff line number	Diff line
		@@ -611,6 +611,7 @@ static int dlm_remaster_locks(struct dlm_ctxt *dlm, u8 dead_node)
		}
		} while (status != 0);

		spin_lock(&dlm_reco_state_lock);
		switch (ndata->state) {
		case DLM_RECO_NODE_DATA_INIT:
		case DLM_RECO_NODE_DATA_FINALIZE_SENT:
		@@ -641,6 +642,7 @@ static int dlm_remaster_locks(struct dlm_ctxt *dlm, u8 dead_node)
		ndata->node_num, dead_node);
		break;
		}
		spin_unlock(&dlm_reco_state_lock);
		}

		mlog(0, "done requesting all lock info\n");

Admin message