Re: [PATCH] netem: fix logic bug in reorder conditional

[Julio Kriger <juliokriger@gmail.com>]

On 5/24/05, Stephen Hemminger <shemminger@osdl.org> wrote:
> On Tue, 24 May 2005 12:41:11 -0300
> Julio Kriger <juliokriger@gmail.com> wrote:
> 
> > > > 2) If I set latency = 50ms and a jitter = 300ms, tabledist can give me
> > > > a negative number. This value is addes to cb->time_to_send, so it
> > > > could change it to a negative value. Should we only accept positives
> > > > number before add it to cb->time_to_send? or will
> > > > q->qdisc->enqueue(skb, q->qdisc) put the package on the queue in a
> > > > special way so it will be handled "before" other packages alrealy on
> > > > the queue but with gretaer time_to_send?
> > >
> > > probably should bound the value to 0 before the addition, to avoid large
> > > wraparound problems, but since enqueue checks for for time it will work
> > > as long as delta less than 2^32/2.
> > >
> >
> > I think the value should be restricted to be positive and greater than
> > zero. Becuase if a negative number is allowed we will be "losing"
> > packages to be reordered, hence we will not be reordering, say 25%, of
> > packages instead we will be reordering about 15%.
> > In other words, packages that should be reordered will not be
> > reordered because its new time to send will be the same as the old
> > time to send.
> > Regards,
> > Julio
> 
> The problem is that the user specification (latency 50ms +/- 300ms with reordering)
> is problematic. Just like specifying reordering without delay (and a fast connection).

I agree with you. Maybe changing the actual parameters to a "range"
could be a better solution. Say "I want delay a package between 50 and
300 ms with correlation of 50% and normal distribution". The code
should not change too much, I think.

Regards,
Julio


-- 
----------------------------
Julio Kriger
mailto:juliokriger@gmail.com