xen-netback: remove loop waiting function
authorWei Liu <wei.liu2@citrix.com>
Tue, 12 Aug 2014 10:48:08 +0000 (11:48 +0100)
committerDavid S. Miller <davem@davemloft.net>
Thu, 14 Aug 2014 03:07:44 +0000 (20:07 -0700)
The original implementation relies on a loop to check if all inflight
packets are freed. Now we have proper reference counting, there's no
need to use loop anymore.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Zoltan Kiss <zoltan.kiss@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/xen-netback/interface.c

index 0aaca90..e29e15d 100644 (file)
@@ -660,25 +660,6 @@ void xenvif_carrier_off(struct xenvif *vif)
        rtnl_unlock();
 }
 
-static void xenvif_wait_unmap_timeout(struct xenvif_queue *queue,
-                                     unsigned int worst_case_skb_lifetime)
-{
-       int i, unmap_timeout = 0;
-
-       for (i = 0; i < MAX_PENDING_REQS; ++i) {
-               if (queue->grant_tx_handle[i] != NETBACK_INVALID_HANDLE) {
-                       unmap_timeout++;
-                       schedule_timeout(msecs_to_jiffies(1000));
-                       if (unmap_timeout > worst_case_skb_lifetime &&
-                           net_ratelimit())
-                               netdev_err(queue->vif->dev,
-                                          "Page still granted! Index: %x\n",
-                                          i);
-                       i = -1;
-               }
-       }
-}
-
 void xenvif_disconnect(struct xenvif *vif)
 {
        struct xenvif_queue *queue = NULL;
@@ -731,21 +712,11 @@ void xenvif_free(struct xenvif *vif)
        struct xenvif_queue *queue = NULL;
        unsigned int num_queues = vif->num_queues;
        unsigned int queue_index;
-       /* Here we want to avoid timeout messages if an skb can be legitimately
-        * stuck somewhere else. Realistically this could be an another vif's
-        * internal or QDisc queue. That another vif also has this
-        * rx_drain_timeout_msecs timeout, so give it time to drain out.
-        * Although if that other guest wakes up just before its timeout happens
-        * and takes only one skb from QDisc, it can hold onto other skbs for a
-        * longer period.
-        */
-       unsigned int worst_case_skb_lifetime = (rx_drain_timeout_msecs/1000);
 
        unregister_netdev(vif->dev);
 
        for (queue_index = 0; queue_index < num_queues; ++queue_index) {
                queue = &vif->queues[queue_index];
-               xenvif_wait_unmap_timeout(queue, worst_case_skb_lifetime);
                xenvif_deinit_queue(queue);
        }