[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: dsmark

On Tuesday 21 November 2000 19:32, Rui Prior wrote:
> I've found something in the design of dsmark qdisc which I find suspicious.
> When tcindex is initialized via default_index, it is initialized with a
> value between 0 and indices-1, supposedly one possible value for each of
> dsmark's classes. If tcindex is set after successful classification though,
> tcindex is initialized with the minor number of the class: also one
> possible value for each of dsmark's classes, but increased by 1.
> Probably there's a reason for this, but if the two cases were to be always
> distiguishable they should have disjoint ranges of possible values, right?

Two other possibilities would be:
- allowing the default_index to have any value
- allowing the default_index to be the minor of any class or zero

Rui Prior