aoe: Fix OOPS after SKB queue changes.

Reported by Thomas Graf.

If we don't unlink the SKB from the queue when we send it
out in aoenet_xmit(), dev_hard_start_xmit() will see skb->next
as non-NULL and interpret this to mean the SKB is part of a
GSO segment list.

Add __skb_unlink() call to fix that.

Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/block/aoe/aoenet.c b/drivers/block/aoe/aoenet.c
index 8fb26030eba8540b3ad7c3c9fd7a995e96b4da1e..9157d64270cb041ba8b92bc462913439b12018e0 100644 (file)
--- a/drivers/block/aoe/aoenet.c
+++ b/drivers/block/aoe/aoenet.c
@@ -99,8 +99,10 @@ aoenet_xmit(struct sk_buff_head *queue)
 {
        struct sk_buff *skb, *tmp;
 
-       skb_queue_walk_safe(queue, skb, tmp)
+       skb_queue_walk_safe(queue, skb, tmp) {
+               __skb_unlink(skb, queue);
                dev_queue_xmit(skb);
+       }
 }
 
 /*