tipc: fix a potential deadlock
Locking dependency detected below possible unsafe locking scenario:
CPU0 CPU1
T0: tipc_named_rcv() tipc_rcv()
T1: [grab nametble write lock]* [grab node lock]*
T2: tipc_update_nametbl() tipc_node_link_up()
T3: tipc_nodesub_subscribe() tipc_nametbl_publish()
T4: [grab node lock]* [grab nametble write lock]*
The opposite order of holding nametbl write lock and node lock on
above two different paths may result in a deadlock. If we move the
the updating of the name table after link state named out of node
lock, the reverse order of holding locks will be eliminated, and
as a result, the deadlock risk.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/net/tipc/node.h b/net/tipc/node.h
index 67513c3..04e9145 100644
--- a/net/tipc/node.h
+++ b/net/tipc/node.h
@@ -53,6 +53,7 @@
* TIPC_WAIT_OWN_LINKS_DOWN: wait until peer node is declared down
* TIPC_NOTIFY_NODE_DOWN: notify node is down
* TIPC_NOTIFY_NODE_UP: notify node is up
+ * TIPC_DISTRIBUTE_NAME: publish or withdraw link state name type
*/
enum {
TIPC_WAIT_PEER_LINKS_DOWN = (1 << 1),
@@ -60,7 +61,9 @@
TIPC_NOTIFY_NODE_DOWN = (1 << 3),
TIPC_NOTIFY_NODE_UP = (1 << 4),
TIPC_WAKEUP_USERS = (1 << 5),
- TIPC_WAKEUP_BCAST_USERS = (1 << 6)
+ TIPC_WAKEUP_BCAST_USERS = (1 << 6),
+ TIPC_NOTIFY_LINK_UP = (1 << 7),
+ TIPC_NOTIFY_LINK_DOWN = (1 << 8)
};
/**
@@ -100,6 +103,7 @@
* @working_links: number of working links to node (both active and standby)
* @link_cnt: number of links to node
* @signature: node instance identifier
+ * @link_id: local and remote bearer ids of changing link, if any
* @nsub: list of "node down" subscriptions monitoring node
* @rcu: rcu struct for tipc_node
*/
@@ -116,6 +120,7 @@
int link_cnt;
int working_links;
u32 signature;
+ u32 link_id;
struct list_head nsub;
struct sk_buff_head waiting_sks;
struct list_head conn_sks;