On Fri, 2004-02-27 at 12:37, Francis Bogsanyi wrote:
> On 26-Feb-04, at 1:55 PM, Sam Leffler wrote:
> >> 2. In the ad hoc mode, currently, both my test nodes are transmitting
> >> beacons. AFAIK, only one node in the IBSS transmits beacons (like an=20
> >> AP)
> >> and the others 'tune' into it (like stations). I traced this problem
> >> down to if_ath.c/ath_beacon_config() where only STAs are taken care of
> >> while everything else is treated as a AP.
> > Sigh, you're correct beacon generation should only happen for the IBSS=20
> > master.
> I don't have the standard in front of me, but the following quote from=20
> Atsushi Onoe on the NetBSD tech-net list seems to suggest otherwise:
> > Subject: wi IBSS mode
> > To: None <tech-net@...>
> > From: Atsushi Onoe <onoe@...>
> > List: tech-net
> > Date: 08/18/2002 12:37:33
> > I'm in vacation and I've not yet tried new wi driver. But the=20
> > description
> > in wi(4) looks a little bit strage to me.
> > | Recent versions of Lucent and PRISM-II firmware support IBSS=20
> > creation.
> > | IBSS is the standard IEEE 802.11 ad-hoc mode. In this mode,=20
> > one station
> > | should be configured as the IBSS ``master''. The nwid is not=20
> > ignored in
> > | IBSS mode.
> > The IEEE 802.11 standard doesn't define 'master' for IBSS mode. All=20
> > the
> > node in the IBSS should speak beacon regardless the existence of other
> > nodes. If two or more BSS in the same IBSS, i.e. different bssid with
> > the same SSID, are seen, they should be merged according to the=20
> > timestamp
> > in the beacon. So according to the standard, ALL the station in the=20
> > IBSS
> > should be configured as the IBSS master.
Ok. I read that standard and Atsushi is right. Pg 123-125 of the
802.11-1999 standard deal with synchronization issues. IBSS requires
distributed beacon generation. Meaning, each node in the IBSS maintains
its own TSF timer. If no beacon arrives during the beacon period of the
IBSS + random backoff, then the node sends a beacon. If a beacon does
arrive, it resets its "beacon-arrival" timer.
So the question to Sam is: How can we tell the chip to *NOT* generate a
beacon frame when the "beacon-arrival" timer has not expired?
In other words: Only generate the beacon when the timer does expire.
.| Amit Kucheria |.
...| Wireless Systems Engineer |...
.....| Metric Systems Corp., San Diego, CA |.....
......| http://www.metricsystems.com |......