On Mon, Nov 13, 2017 at 3:08 PM, John Fastabend <john.fastab...@gmail.com> wrote: > sch_direct_xmit() uses qdisc_qlen as a return value but all call sites > of the routine only check if it is zero or not. Simplify the logic so > that we don't need to return an actual queue length value. > > This introduces a case now where sch_direct_xmit would have returned > a qlen of zero but now it returns true.
You mean the first case, when the likely(skb) branch failed? Can that return false, then? > However in this case all > call sites of sch_direct_xmit will implement a dequeue() and get > a null skb and abort. This trades tracking qlen in the hotpath for > an extra dequeue operation. Overall this seems to be good for > performance. > > Signed-off-by: John Fastabend <john.fastab...@gmail.com> > @@ -198,18 +198,19 @@ int sch_direct_xmit(struct sk_buff *skb, struct Qdisc > *q, > > if (dev_xmit_complete(ret)) { > /* Driver sent out skb successfully or skb was consumed */ > - ret = qdisc_qlen(q); > + ret = true; > } else { > /* Driver returned NETDEV_TX_BUSY - requeue skb */ > if (unlikely(ret != NETDEV_TX_BUSY)) > net_warn_ratelimited("BUG %s code %d qlen %d\n", > dev->name, ret, q->q.qlen); > > - ret = dev_requeue_skb(skb, q); > + dev_requeue_skb(skb, q); > + ret = false; > } > > if (ret && netif_xmit_frozen_or_stopped(txq)) > - ret = 0; > + ret = false; > > return ret; > } This can now become if (!dev_queue_complete(ret)) { if (unlikely(ret != NETDEV_TX_BUSY)) ... dev_requeue_skb(skb, q); return false; } if (netif_xmit_frozen_or_stopped(txq)) return false; return true;