Subject: net: fix race in process_backlog The recent NAPI rework (4fa57c9ea9f36f9ca852f3a88ca5d2f1aebbc960) ... processing to drop dead. netif_rx() process_backlog() irq_disable(); skb = __skb_dequeue(); irq_enable(); irq_disable(); ...

linux-kernel - Peter Zijlstra - Oct 3 2007 - 11:44

... zijlstra@chello.nl> wrote: > Subject: net: fix race in process_backlog > > The recent NAPI ... processing to drop dead. > > netif_rx() process_backlog() > > irq_disable(); > skb = __skb_dequeue(); > irq_enable(); > > irq_disable(); ...

linux-kernel - Stephen Hemminger - Oct 3 2007 - 12:15

... 2.6/net/core/dev.c > > @@ -2095,11 +2095,11 @@ static int process_backlog(struct napi_s > > > > local_irq_disable(); > > skb = __skb_dequeue(&queue->input_pkt_queue ... looks like a better fix. the irq_enable is needed in both cases. --- ...

linux-kernel - Stephen Hemminger - Oct 3 2007 - 18:05

... -2.6/net/core/dev.c > > > @@ -2095,11 +2095,11 @@ static int process_backlog(struct napi_s > > > > > > local_irq_disable(); > > > skb = __skb_dequeue(&queue-> ... This looks like a better fix. the irq_enable is needed in both cases. Yep, ...

linux-kernel - David Miller - Oct 3 2007 - 19:39

... -2.6/net/core/dev.c > @@ -2095,11 +2095,11 @@ static int process_backlog(struct napi_s ... > + local_irq_enable(); What re-enables interrupts in the !skb path? - To unsubscribe from ... : send the line "unsubscribe linux-kernel" in the body of a message to ...

linux-kernel - David Miller - Oct 3 2007 - 17:58