Re: [Madwifi-devel] ibss beacon backoff

[bruno r. <br...@th...>]

On Tuesday 22 January 2008 12:00:48 Benoit PAPILLAULT wrote:
> bruno randolf a =E9crit :
> > hello!
> >
> > i found some dubious code with regards to beacon backoff and i would
> > appreciate your opinion on it.
> >
> > in "11.1.2.2 Beacon generation in an IBSS" the IEEE802.11 standard says,
> > each STA should...
> >
> > "b) Calculate a random delay uniformly distributed in the range between
> > zero and twice aCWmin =D7 aSlotTime,"
> >
> > in ath_beaconq_config() where we set the cwmin and cwmax values of the
> > beacon queue i think this is applied wrongly to set cwmin =3D 2 * cwmin=
=2E i
> > rather think the following patch would be a more correct implementation
> > of the standard. using this change beacon backoff still works for me, b=
ut
> > have i missed something else?
> >
> > greetings,
> > bruno
>
> Hi Bruno,
>
> I was also wondering this part but this is a bit beyond my skills :-).

mine too - but that has never stopped us, right ;)

> Reading the standard and your patch, looks good to me. Is there a way to
> measure the results? Would it be possible to see appropriate change in a
> monitor trace for instance?

i have checked with wireshark and i see correct beacon backoff in both case=
s,=20
so it doesn't seem to make a big difference, but that was on an otherwise=20
silent channel...

in theory (as far as i understand it) if we use 2*cw_min for beacons and=20
1*cw_min for data packets that would cause beacons give precedence to data=
=20
packets, which is not how it is supposed to be. also if we have cw_max of=20
1023 (as now) that might delay the beacons too much (1023 * slotTime could =
be=20
larger than the beacon interval) - not sure what's going to happen in that=
=20
case.

i will try to trace the register writes for both settings, i need to do tha=
t=20
anyways since i suspect a bug in ath5k setting of these values.

bruno