RDMA/nes: Fix Xansation test crash on cm_node ref_count
Faisal Latif [Wed, 9 Dec 2009 23:54:18 +0000 (15:54 -0800)]
While running a Xansation test, an active side node crashed.  The
problem started on the passive side, which generated an STtag that was
0.  The passive side sent a TERMINATE instead of an MPA REJECT msg.
The active side, receives TERMINATE and sends connect_err() and set
the cm_node state to CLOSED.  The passive side sends FIN + ACK after
TERMINATE.  Active side ends up in handle_ack_pkt() and send_reset().
send_reset() consumes 1 cm_node's ref_count.  Because the cm_node is
in CLOSED state, which means that cm_node will be destroyed after
completion of the connect_err() indication, CM will crash after
send_reset().

Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

drivers/infiniband/hw/nes/nes_cm.c

index 20e21f1..a258168 100644 (file)
@@ -1610,6 +1610,7 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
                break;
        case NES_CM_STATE_CLOSED:
                cleanup_retrans_entry(cm_node);
+               add_ref_cm_node(cm_node);
                send_reset(cm_node, skb);
                break;
        case NES_CM_STATE_TSA:
@@ -1661,9 +1662,15 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
                passive_open_err(cm_node, skb, 1);
                break;
        case NES_CM_STATE_LISTENING:
+               cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq);
+               cleanup_retrans_entry(cm_node);
+               cm_node->state = NES_CM_STATE_CLOSED;
+               send_reset(cm_node, skb);
+               break;
        case NES_CM_STATE_CLOSED:
                cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq);
                cleanup_retrans_entry(cm_node);
+               add_ref_cm_node(cm_node);
                send_reset(cm_node, skb);
                break;
        case NES_CM_STATE_ESTABLISHED:
@@ -1732,8 +1739,13 @@ static int handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
                        dev_kfree_skb_any(skb);
                break;
        case NES_CM_STATE_LISTENING:
+               cleanup_retrans_entry(cm_node);
+               cm_node->state = NES_CM_STATE_CLOSED;
+               send_reset(cm_node, skb);
+               break;
        case NES_CM_STATE_CLOSED:
                cleanup_retrans_entry(cm_node);
+               add_ref_cm_node(cm_node);
                send_reset(cm_node, skb);
                break;
        case NES_CM_STATE_LAST_ACK:
@@ -2193,8 +2205,11 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod
        case NES_CM_STATE_CLOSING:
                ret = -1;
                break;
-       case NES_CM_STATE_MPAREJ_RCVD:
        case NES_CM_STATE_LISTENING:
+               cleanup_retrans_entry(cm_node);
+               send_reset(cm_node, NULL);
+               break;
+       case NES_CM_STATE_MPAREJ_RCVD:
        case NES_CM_STATE_UNKNOWN:
        case NES_CM_STATE_INITED:
        case NES_CM_STATE_CLOSED: