xen-netfront: fix oops when disconnected from backend

xennet_disconnect_backend() was not correctly iterating over all the
queues.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
David Vrabel 2014-06-18 10:47:27 +01:00 committed by David S. Miller
parent a705a906c6
commit 765418694b
1 changed files with 3 additions and 2 deletions

View File

@ -1287,7 +1287,7 @@ static irqreturn_t xennet_rx_interrupt(int irq, void *dev_id)
if (likely(netif_carrier_ok(dev) &&
RING_HAS_UNCONSUMED_RESPONSES(&queue->rx)))
napi_schedule(&queue->napi);
napi_schedule(&queue->napi);
return IRQ_HANDLED;
}
@ -1437,10 +1437,11 @@ static void xennet_end_access(int ref, void *page)
static void xennet_disconnect_backend(struct netfront_info *info)
{
unsigned int i = 0;
struct netfront_queue *queue = NULL;
unsigned int num_queues = info->netdev->real_num_tx_queues;
for (i = 0; i < num_queues; ++i) {
struct netfront_queue *queue = &info->queues[i];
/* Stop old i/f to prevent errors whilst we rebuild the state. */
spin_lock_bh(&queue->rx_lock);
spin_lock_irq(&queue->tx_lock);