On Thu, Sep 11, 2008 at 09:00:35PM +1000, Herbert Xu wrote:

On Thu, Sep 11, 2008 at 03:49:55AM -0700, David Miller wrote:

quoted

Well some kind of check has to be there.

I _did_ remove it during my initial implementation, and that
turned into a reported performance regression.

I see.  How about looking at the queue that the head-of-qdisc
packet maps to? That should be fairly cheap to compute.

Very good point: "That can't be right"! But I'm not sure I got the
above idea: e.g. this new sch_multiq can have different packets
depending just on some unstopped tx_queue.

IMHO, the most reasonable test here is for all tx_queues of a qdisc
beeing stopped, but since this is quite heavy, probably we need an
additional qdisc flag for such an occasion.

Jarek P.