Re: [Madwifi-users] Re: Beacon Miss Interrupts with 20030802

[Greg C. <ma...@at...>]

Jean Lorchat wrote:
> I thought the beacons were prioritized traffic in comparison to simple 
this is correct.
> data frames. Isn't it the case ? or do the traffic flow too fast for 
> them to preempt data frames ? 
it is possible for normal traffic to severely delay the sending of the 
beacon
because there is nothing in the 802.11 standard that requires stations 
to stop
transmitting at the beacon time.

Because missed beacons as well as delayed
> beacons can have very bad consequences on power saving stations since 
> the TIM is announced in beacons.
this is also correct.
but we have a noisy and imprecise channel
and must deal with the consequences.


> 
> Jean
> 
> Greg Chesson wrote:
> 
>> In our stress testing with the drivers we have suspected that it is 
>> possible
>> for the channel to become sufficiently busy that the beacons aren't 
>> transmitted
>> or are hugely delayed.    This effect is compounded by rate adaptation 
>> responding
>> by selecting lower phy rates after collision which just makes things 
>> worse.
>>
>> I don't think the bmisshack should be reinstated as it was a 
>> workaround for a bug.
>>
>> The technique of doing a probe request after the bmiss seems like a 
>> good compromise.
>>
>> g
>>
>> William S. Kish wrote:
>>
>>> I'm doing some fairly aggressive performance/stress testing with 5211
>>> cards and seeing some sub-nominal behavior with 20030802 that I didn't
>>> see with the previous release.   One thing I notice is that my station
>>> is now regularly getting "missed beacon" interrupts (HAL_INT_BMISS)
>>> during the testing.   I see that if_ath.c previously had some
>>> "bmisshack" stuff that essentially filtered out these interrupts while
>>> 20030802 release has HAL_INT_BMISS fully enabled. These BMISS 
>>> interrupts really bring things down hard as they force a
>>> change to SCAN state. If I go back to masking these interrupts in
>>> ath_intr(), my tests continue on blissfully. I'm getting most of my
>>> packets through from the AP to the station so its kinda hard to believe
>>> that I'm dropping so many beacons. (Currently I see bmissthreshold gets
>>> set to 7).
>>>
>>> So do we really believe HAL_INT_BMISS now?   Is there maybe something
>>> wrong in the BMISS reporting?  Or is my test really stressing something
>>> else on the AP (also a madwifi 5211) or the station to the point of
>>> missing beacons?
>>>
>>> Thanks,
>>> Bill
>>>
>>>
>>>
>>>
>>
>>
>>
>> -------------------------------------------------------
>> This SF.Net email sponsored by: Free pre-built ASP.NET sites including
>> Data Reports, E-commerce, Portals, and Forums are available now.
>> Download today and enter to win an XBOX or Visual Studio .NET.
>> http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 
>>
>> _______________________________________________
>> Madwifi-users mailing list
>> Mad...@li...
>> https://lists.sourceforge.net/lists/listinfo/madwifi-users
>>
> 
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email sponsored by: Free pre-built ASP.NET sites including
> Data Reports, E-commerce, Portals, and Forums are available now.
> Download today and enter to win an XBOX or Visual Studio .NET.
> http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
> _______________________________________________
> Madwifi-users mailing list
> Mad...@li...
> https://lists.sourceforge.net/lists/listinfo/madwifi-users