ibmvnic: queue reset when CRQ gets closed during reset
authorNathan Fontenot <nfont@linux.vnet.ibm.com>
Wed, 7 Feb 2018 19:00:24 +0000 (13:00 -0600)
committerDavid S. Miller <davem@davemloft.net>
Thu, 8 Feb 2018 20:11:15 +0000 (15:11 -0500)
While handling a driver reset we get a H_CLOSED return trying
to send a CRQ event. When this occurs we need to queue up another
reset attempt. Without doing this we see instances where the driver
is left in a closed state because the reset failed and there is no
further attempts to reset the driver.

Signed-off-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/ibm/ibmvnic.c

index 8dabc9d..dd4a294 100644 (file)
@@ -2914,8 +2914,12 @@ static int ibmvnic_send_crq(struct ibmvnic_adapter *adapter,
                                cpu_to_be64(u64_crq[1]));
 
        if (rc) {
-               if (rc == H_CLOSED)
+               if (rc == H_CLOSED) {
                        dev_warn(dev, "CRQ Queue closed\n");
+                       if (adapter->resetting)
+                               ibmvnic_reset(adapter, VNIC_RESET_FATAL);
+               }
+
                dev_warn(dev, "Send error (rc=%d)\n", rc);
        }