Jarek Poplawski wrote:

On Tue, Apr 29, 2008 at 10:53:41PM +0200, Jarek Poplawski wrote:
...

quoted

So it's really by design! (I don't know why I missed this, and why

...Hmm..., actually this is the place:

sfq_enqueue(struct sk_buff *skb, struct Qdisc *sch)
{
...
	x = q->ht[hash];
	if (x == SFQ_DEPTH) {
		q->ht[hash] = x = q->dep[SFQ_DEPTH].next;
...

sfq counts on free slot always available here.

IIRC thats why the limit is capped at SFQ_DEPTH - 1.