Re: [Madwifi-devel] CWmin / CWmax variables in the code?

[Amit K. <aku...@me...>]

On Tue, 2004-03-02 at 20:11, Sam Leffler wrote:

> >
> > I dont want to alter the calculation of CW. I only want to read it so
> > that i can generate a random delay between zero and CW before the ad=20
> > hoc
> > node tries to send a beacon.
>=20
> There are changes coming to support WME that'll (presumably) let you=20
> read these values.  Until then use 15 for CWmin (31 when operating in=20
> 11b) and 1023 for CWmax where these numbers are in units of slots.
>=20
> Regardless, I'm not sure what you can do as it appears the current=20
> scheme for handling beacons in adhoc mode is botched.  That is, when I=20

I thought that the newer HAL that Greg sent me partially fixed the
problem, so that the ad hoc nodes are now generating the beacons albeit
only after a 2 minute delay. I have finished some code that I will be
testing with the new HAL for distributed beacon generation.

> wrote the code I assumed the SWBA would be suppressed when operating in=20
> adhoc mode and the node was not to transmit a beacon (because it heard=20
> a beacon from another node).  It appears this assumption was wrong and=20
> the only way to do things right is to use the h/w xmit queue directly=20
> (as the Atheros driver does).  I've got no time to chase this but=20
> beware that there's probably a better (and simpler) solution that does=20
> away with host-generated beacons entirely.

Not generating beacons on the host? I am not quite sure I get that.

> It would be nice if some of the other folks that have hal source code=20
> and an interest in adhoc networks would contribute their results (hint=20
> hint).
>=20
> 	Sam

Yeah I was hoping for some help from the others too, since a lot of
answers seem to lie in the HAL.

Regards,
Amit

--=20
     .| Amit Kucheria                       |.
   ...| Wireless Systems Engineer           |...
 .....| Metric Systems Corp., San Diego, CA |.....
......| www.metricsystems.com               |......